math.cos on complex, imaginary part

Percentage Accurate: 64.7% → 99.8%

Time: 6.8s

Alternatives: 18

Speedup: 0.9×

• 48.3% of points produce a very large (infinite) output. You may want to add a precondition.

Specification

Math

\[\begin{array}{l} \\ \left(0.5 \cdot \sin re\right) \cdot \left(e^{-im} - e^{im}\right) \end{array} \]

FPCore

(FPCore (re im)
 :precision binary64
 (* (* 0.5 (sin re)) (- (exp (- im)) (exp im))))

C

double code(double re, double im) {
	return (0.5 * sin(re)) * (exp(-im) - exp(im));
}

Fortran

module fmin_fmax_functions
    implicit none
    private
    public fmax
    public fmin

    interface fmax
        module procedure fmax88
        module procedure fmax44
        module procedure fmax84
        module procedure fmax48
    end interface
    interface fmin
        module procedure fmin88
        module procedure fmin44
        module procedure fmin84
        module procedure fmin48
    end interface
contains
    real(8) function fmax88(x, y) result (res)
        real(8), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(x, max(x, y), y /= y), x /= x)
    end function
    real(4) function fmax44(x, y) result (res)
        real(4), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(y, merge(x, max(x, y), y /= y), x /= x)
    end function
    real(8) function fmax84(x, y) result(res)
        real(8), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(dble(y), merge(x, max(x, dble(y)), y /= y), x /= x)
    end function
    real(8) function fmax48(x, y) result(res)
        real(4), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(dble(x), max(dble(x), y), y /= y), x /= x)
    end function
    real(8) function fmin88(x, y) result (res)
        real(8), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(x, min(x, y), y /= y), x /= x)
    end function
    real(4) function fmin44(x, y) result (res)
        real(4), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(y, merge(x, min(x, y), y /= y), x /= x)
    end function
    real(8) function fmin84(x, y) result(res)
        real(8), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(dble(y), merge(x, min(x, dble(y)), y /= y), x /= x)
    end function
    real(8) function fmin48(x, y) result(res)
        real(4), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(dble(x), min(dble(x), y), y /= y), x /= x)
    end function
end module

real(8) function code(re, im)
use fmin_fmax_functions
    real(8), intent (in) :: re
    real(8), intent (in) :: im
    code = (0.5d0 * sin(re)) * (exp(-im) - exp(im))
end function

Java

public static double code(double re, double im) {
	return (0.5 * Math.sin(re)) * (Math.exp(-im) - Math.exp(im));
}

Python

def code(re, im):
	return (0.5 * math.sin(re)) * (math.exp(-im) - math.exp(im))

Julia

function code(re, im)
	return Float64(Float64(0.5 * sin(re)) * Float64(exp(Float64(-im)) - exp(im)))
end

MATLAB

function tmp = code(re, im)
	tmp = (0.5 * sin(re)) * (exp(-im) - exp(im));
end

Wolfram

code[re_, im_] := N[(N[(0.5 * N[Sin[re], $MachinePrecision]), $MachinePrecision] * N[(N[Exp[(-im)], $MachinePrecision] - N[Exp[im], $MachinePrecision]), $MachinePrecision]), $MachinePrecision]

Initial Program: 64.7% accurate, 1.0× speedup

Math

\[\begin{array}{l} \\ \left(0.5 \cdot \sin re\right) \cdot \left(e^{-im} - e^{im}\right) \end{array} \]

FPCore

(FPCore (re im)
 :precision binary64
 (* (* 0.5 (sin re)) (- (exp (- im)) (exp im))))

C

double code(double re, double im) {
	return (0.5 * sin(re)) * (exp(-im) - exp(im));
}

Fortran

module fmin_fmax_functions
    implicit none
    private
    public fmax
    public fmin

    interface fmax
        module procedure fmax88
        module procedure fmax44
        module procedure fmax84
        module procedure fmax48
    end interface
    interface fmin
        module procedure fmin88
        module procedure fmin44
        module procedure fmin84
        module procedure fmin48
    end interface
contains
    real(8) function fmax88(x, y) result (res)
        real(8), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(x, max(x, y), y /= y), x /= x)
    end function
    real(4) function fmax44(x, y) result (res)
        real(4), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(y, merge(x, max(x, y), y /= y), x /= x)
    end function
    real(8) function fmax84(x, y) result(res)
        real(8), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(dble(y), merge(x, max(x, dble(y)), y /= y), x /= x)
    end function
    real(8) function fmax48(x, y) result(res)
        real(4), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(dble(x), max(dble(x), y), y /= y), x /= x)
    end function
    real(8) function fmin88(x, y) result (res)
        real(8), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(x, min(x, y), y /= y), x /= x)
    end function
    real(4) function fmin44(x, y) result (res)
        real(4), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(y, merge(x, min(x, y), y /= y), x /= x)
    end function
    real(8) function fmin84(x, y) result(res)
        real(8), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(dble(y), merge(x, min(x, dble(y)), y /= y), x /= x)
    end function
    real(8) function fmin48(x, y) result(res)
        real(4), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(dble(x), min(dble(x), y), y /= y), x /= x)
    end function
end module

real(8) function code(re, im)
use fmin_fmax_functions
    real(8), intent (in) :: re
    real(8), intent (in) :: im
    code = (0.5d0 * sin(re)) * (exp(-im) - exp(im))
end function

Java

public static double code(double re, double im) {
	return (0.5 * Math.sin(re)) * (Math.exp(-im) - Math.exp(im));
}

Python

def code(re, im):
	return (0.5 * math.sin(re)) * (math.exp(-im) - math.exp(im))

Julia

function code(re, im)
	return Float64(Float64(0.5 * sin(re)) * Float64(exp(Float64(-im)) - exp(im)))
end

MATLAB

function tmp = code(re, im)
	tmp = (0.5 * sin(re)) * (exp(-im) - exp(im));
end

Wolfram

code[re_, im_] := N[(N[(0.5 * N[Sin[re], $MachinePrecision]), $MachinePrecision] * N[(N[Exp[(-im)], $MachinePrecision] - N[Exp[im], $MachinePrecision]), $MachinePrecision]), $MachinePrecision]

Alternative 1: 99.8% accurate, 0.9× speedup

Math

\[\begin{array}{l} im\_m = \left|im\right| \\ im\_s = \mathsf{copysign}\left(1, im\right) \\ im\_s \cdot \begin{array}{l} \mathbf{if}\;im\_m \leq 6 \cdot 10^{-6}:\\ \;\;\;\;\left(-im\_m\right) \cdot \sin re\\ \mathbf{else}:\\ \;\;\;\;\left(0.5 \cdot \sin re\right) \cdot \left(e^{-im\_m} - e^{im\_m}\right)\\ \end{array} \end{array} \]

FPCore

im\_m = (fabs.f64 im)
im\_s = (copysign.f64 #s(literal 1 binary64) im)
(FPCore (im_s re im_m)
 :precision binary64
 (*
  im_s
  (if (<= im_m 6e-6)
    (* (- im_m) (sin re))
    (* (* 0.5 (sin re)) (- (exp (- im_m)) (exp im_m))))))

C

im\_m = fabs(im);
im\_s = copysign(1.0, im);
double code(double im_s, double re, double im_m) {
	double tmp;
	if (im_m <= 6e-6) {
		tmp = -im_m * sin(re);
	} else {
		tmp = (0.5 * sin(re)) * (exp(-im_m) - exp(im_m));
	}
	return im_s * tmp;
}

Fortran

im\_m =     private
im\_s =     private
module fmin_fmax_functions
    implicit none
    private
    public fmax
    public fmin

    interface fmax
        module procedure fmax88
        module procedure fmax44
        module procedure fmax84
        module procedure fmax48
    end interface
    interface fmin
        module procedure fmin88
        module procedure fmin44
        module procedure fmin84
        module procedure fmin48
    end interface
contains
    real(8) function fmax88(x, y) result (res)
        real(8), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(x, max(x, y), y /= y), x /= x)
    end function
    real(4) function fmax44(x, y) result (res)
        real(4), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(y, merge(x, max(x, y), y /= y), x /= x)
    end function
    real(8) function fmax84(x, y) result(res)
        real(8), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(dble(y), merge(x, max(x, dble(y)), y /= y), x /= x)
    end function
    real(8) function fmax48(x, y) result(res)
        real(4), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(dble(x), max(dble(x), y), y /= y), x /= x)
    end function
    real(8) function fmin88(x, y) result (res)
        real(8), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(x, min(x, y), y /= y), x /= x)
    end function
    real(4) function fmin44(x, y) result (res)
        real(4), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(y, merge(x, min(x, y), y /= y), x /= x)
    end function
    real(8) function fmin84(x, y) result(res)
        real(8), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(dble(y), merge(x, min(x, dble(y)), y /= y), x /= x)
    end function
    real(8) function fmin48(x, y) result(res)
        real(4), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(dble(x), min(dble(x), y), y /= y), x /= x)
    end function
end module

real(8) function code(im_s, re, im_m)
use fmin_fmax_functions
    real(8), intent (in) :: im_s
    real(8), intent (in) :: re
    real(8), intent (in) :: im_m
    real(8) :: tmp
    if (im_m <= 6d-6) then
        tmp = -im_m * sin(re)
    else
        tmp = (0.5d0 * sin(re)) * (exp(-im_m) - exp(im_m))
    end if
    code = im_s * tmp
end function

Java

im\_m = Math.abs(im);
im\_s = Math.copySign(1.0, im);
public static double code(double im_s, double re, double im_m) {
	double tmp;
	if (im_m <= 6e-6) {
		tmp = -im_m * Math.sin(re);
	} else {
		tmp = (0.5 * Math.sin(re)) * (Math.exp(-im_m) - Math.exp(im_m));
	}
	return im_s * tmp;
}

Python

im\_m = math.fabs(im)
im\_s = math.copysign(1.0, im)
def code(im_s, re, im_m):
	tmp = 0
	if im_m <= 6e-6:
		tmp = -im_m * math.sin(re)
	else:
		tmp = (0.5 * math.sin(re)) * (math.exp(-im_m) - math.exp(im_m))
	return im_s * tmp

Julia

im\_m = abs(im)
im\_s = copysign(1.0, im)
function code(im_s, re, im_m)
	tmp = 0.0
	if (im_m <= 6e-6)
		tmp = Float64(Float64(-im_m) * sin(re));
	else
		tmp = Float64(Float64(0.5 * sin(re)) * Float64(exp(Float64(-im_m)) - exp(im_m)));
	end
	return Float64(im_s * tmp)
end

MATLAB

im\_m = abs(im);
im\_s = sign(im) * abs(1.0);
function tmp_2 = code(im_s, re, im_m)
	tmp = 0.0;
	if (im_m <= 6e-6)
		tmp = -im_m * sin(re);
	else
		tmp = (0.5 * sin(re)) * (exp(-im_m) - exp(im_m));
	end
	tmp_2 = im_s * tmp;
end

Wolfram

im\_m = N[Abs[im], $MachinePrecision]
im\_s = N[With[{TMP1 = Abs[1.0], TMP2 = Sign[im]}, TMP1 * If[TMP2 == 0, 1, TMP2]], $MachinePrecision]
code[im$95$s_, re_, im$95$m_] := N[(im$95$s * If[LessEqual[im$95$m, 6e-6], N[((-im$95$m) * N[Sin[re], $MachinePrecision]), $MachinePrecision], N[(N[(0.5 * N[Sin[re], $MachinePrecision]), $MachinePrecision] * N[(N[Exp[(-im$95$m)], $MachinePrecision] - N[Exp[im$95$m], $MachinePrecision]), $MachinePrecision]), $MachinePrecision]]), $MachinePrecision]

Derivation

1. Split input into 2 regimes

2. if im < 6.0000000000000002e-6

   1. Initial program 30.4%

      \[\left(0.5 \cdot \sin re\right) \cdot \left(e^{-im} - e^{im}\right) \]

   2. Taylor expanded in im around 0

      \[\leadsto \color{blue}{-1 \cdot \left(im \cdot \sin re\right)} \]
   3. Step-by-step derivation

      1. associate-*r* N/A

         \[\leadsto \left(-1 \cdot im\right) \cdot \color{blue}{\sin re} \]

      2. lower-*.f64 N/A

         \[\leadsto \left(-1 \cdot im\right) \cdot \color{blue}{\sin re} \]

      3. mul-1-neg N/A

         \[\leadsto \left(\mathsf{neg}\left(im\right)\right) \cdot \sin \color{blue}{re} \]

      4. lift-neg.f64 N/A

         \[\leadsto \left(-im\right) \cdot \sin \color{blue}{re} \]

      5. lift-sin.f64 99.7

         \[\leadsto \left(-im\right) \cdot \sin re \]

   4. Applied rewrites 99.7%

      \[\leadsto \color{blue}{\left(-im\right) \cdot \sin re} \]

   if 6.0000000000000002e-6 < im

   1. Initial program 99.9%

      \[\left(0.5 \cdot \sin re\right) \cdot \left(e^{-im} - e^{im}\right) \]
3. Recombined 2 regimes into one program.
4. Add Preprocessing

Alternative 2: 85.5% accurate, 0.4× speedup

Math

\[\begin{array}{l} im\_m = \left|im\right| \\ im\_s = \mathsf{copysign}\left(1, im\right) \\ \begin{array}{l} t_0 := \left(0.5 \cdot \sin re\right) \cdot \left(e^{-im\_m} - e^{im\_m}\right)\\ im\_s \cdot \begin{array}{l} \mathbf{if}\;t\_0 \leq -1 \cdot 10^{-49}:\\ \;\;\;\;\left(\left(-2 \cdot \sinh im\_m\right) \cdot re\right) \cdot 0.5\\ \mathbf{elif}\;t\_0 \leq 0:\\ \;\;\;\;\left(-im\_m\right) \cdot \sin re\\ \mathbf{else}:\\ \;\;\;\;\left(\mathsf{fma}\left(re \cdot re, -0.08333333333333333, 0.5\right) \cdot re\right) \cdot \left(1 - e^{im\_m}\right)\\ \end{array} \end{array} \end{array} \]

FPCore

im\_m = (fabs.f64 im)
im\_s = (copysign.f64 #s(literal 1 binary64) im)
(FPCore (im_s re im_m)
 :precision binary64
 (let* ((t_0 (* (* 0.5 (sin re)) (- (exp (- im_m)) (exp im_m)))))
   (*
    im_s
    (if (<= t_0 -1e-49)
      (* (* (* -2.0 (sinh im_m)) re) 0.5)
      (if (<= t_0 0.0)
        (* (- im_m) (sin re))
        (*
         (* (fma (* re re) -0.08333333333333333 0.5) re)
         (- 1.0 (exp im_m))))))))

C

im\_m = fabs(im);
im\_s = copysign(1.0, im);
double code(double im_s, double re, double im_m) {
	double t_0 = (0.5 * sin(re)) * (exp(-im_m) - exp(im_m));
	double tmp;
	if (t_0 <= -1e-49) {
		tmp = ((-2.0 * sinh(im_m)) * re) * 0.5;
	} else if (t_0 <= 0.0) {
		tmp = -im_m * sin(re);
	} else {
		tmp = (fma((re * re), -0.08333333333333333, 0.5) * re) * (1.0 - exp(im_m));
	}
	return im_s * tmp;
}

Julia

im\_m = abs(im)
im\_s = copysign(1.0, im)
function code(im_s, re, im_m)
	t_0 = Float64(Float64(0.5 * sin(re)) * Float64(exp(Float64(-im_m)) - exp(im_m)))
	tmp = 0.0
	if (t_0 <= -1e-49)
		tmp = Float64(Float64(Float64(-2.0 * sinh(im_m)) * re) * 0.5);
	elseif (t_0 <= 0.0)
		tmp = Float64(Float64(-im_m) * sin(re));
	else
		tmp = Float64(Float64(fma(Float64(re * re), -0.08333333333333333, 0.5) * re) * Float64(1.0 - exp(im_m)));
	end
	return Float64(im_s * tmp)
end

Wolfram

im\_m = N[Abs[im], $MachinePrecision]
im\_s = N[With[{TMP1 = Abs[1.0], TMP2 = Sign[im]}, TMP1 * If[TMP2 == 0, 1, TMP2]], $MachinePrecision]
code[im$95$s_, re_, im$95$m_] := Block[{t$95$0 = N[(N[(0.5 * N[Sin[re], $MachinePrecision]), $MachinePrecision] * N[(N[Exp[(-im$95$m)], $MachinePrecision] - N[Exp[im$95$m], $MachinePrecision]), $MachinePrecision]), $MachinePrecision]}, N[(im$95$s * If[LessEqual[t$95$0, -1e-49], N[(N[(N[(-2.0 * N[Sinh[im$95$m], $MachinePrecision]), $MachinePrecision] * re), $MachinePrecision] * 0.5), $MachinePrecision], If[LessEqual[t$95$0, 0.0], N[((-im$95$m) * N[Sin[re], $MachinePrecision]), $MachinePrecision], N[(N[(N[(N[(re * re), $MachinePrecision] * -0.08333333333333333 + 0.5), $MachinePrecision] * re), $MachinePrecision] * N[(1.0 - N[Exp[im$95$m], $MachinePrecision]), $MachinePrecision]), $MachinePrecision]]]), $MachinePrecision]]

Derivation

1. Split input into 3 regimes

2. if (*.f64 (*.f64 #s(literal 1/2 binary64) (sin.f64 re)) (-.f64 (exp.f64 (neg.f64 im)) (exp.f64 im))) < -9.99999999999999936e-50

   1. Initial program 98.8%

      \[\left(0.5 \cdot \sin re\right) \cdot \left(e^{-im} - e^{im}\right) \]

   2. Taylor expanded in re around 0

      \[\leadsto \color{blue}{\frac{1}{2} \cdot \left(re \cdot \left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right)\right)} \]
   3. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(re \cdot \left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right)\right) \cdot \color{blue}{\frac{1}{2}} \]

      2. lower-*.f64 N/A

         \[\leadsto \left(re \cdot \left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right)\right) \cdot \color{blue}{\frac{1}{2}} \]

      3. *-commutative N/A

         \[\leadsto \left(\left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right) \cdot re\right) \cdot \frac{1}{2} \]

      4. lower-*.f64 N/A

         \[\leadsto \left(\left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right) \cdot re\right) \cdot \frac{1}{2} \]

      5. sub-negate N/A

         \[\leadsto \left(\left(\mathsf{neg}\left(\left(e^{im} - e^{\mathsf{neg}\left(im\right)}\right)\right)\right) \cdot re\right) \cdot \frac{1}{2} \]

      6. lower-neg.f64 N/A

         \[\leadsto \left(\left(-\left(e^{im} - e^{\mathsf{neg}\left(im\right)}\right)\right) \cdot re\right) \cdot \frac{1}{2} \]

      7. sinh-undef N/A

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot \frac{1}{2} \]

      8. lower-*.f64 N/A

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot \frac{1}{2} \]

      9. lower-sinh.f64 72.5

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot 0.5 \]

   4. Applied rewrites 72.5%

      \[\leadsto \color{blue}{\left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot 0.5} \]
   5. Step-by-step derivation

      1. flip-- 72.5

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot \color{blue}{re}\right) \cdot 0.5 \]

      2. pow2 72.5

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot 0.5 \]

      3. lift-neg.f64 N/A

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot \frac{1}{2} \]

      4. exp-prod N/A

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot \frac{1}{2} \]

      5. lift-neg.f64 72.5

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot 0.5 \]

      6. exp-sum 72.5

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot 0.5 \]

   6. Applied rewrites 72.5%

      \[\leadsto \color{blue}{\left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot 0.5} \]

   if -9.99999999999999936e-50 < (*.f64 (*.f64 #s(literal 1/2 binary64) (sin.f64 re)) (-.f64 (exp.f64 (neg.f64 im)) (exp.f64 im))) < 0.0

   1. Initial program 30.2%

      \[\left(0.5 \cdot \sin re\right) \cdot \left(e^{-im} - e^{im}\right) \]

   2. Taylor expanded in im around 0

      \[\leadsto \color{blue}{-1 \cdot \left(im \cdot \sin re\right)} \]
   3. Step-by-step derivation

      1. associate-*r* N/A

         \[\leadsto \left(-1 \cdot im\right) \cdot \color{blue}{\sin re} \]

      2. lower-*.f64 N/A

         \[\leadsto \left(-1 \cdot im\right) \cdot \color{blue}{\sin re} \]

      3. mul-1-neg N/A

         \[\leadsto \left(\mathsf{neg}\left(im\right)\right) \cdot \sin \color{blue}{re} \]

      4. lift-neg.f64 N/A

         \[\leadsto \left(-im\right) \cdot \sin \color{blue}{re} \]

      5. lift-sin.f64 99.4

         \[\leadsto \left(-im\right) \cdot \sin re \]

   4. Applied rewrites 99.4%

      \[\leadsto \color{blue}{\left(-im\right) \cdot \sin re} \]

   if 0.0 < (*.f64 (*.f64 #s(literal 1/2 binary64) (sin.f64 re)) (-.f64 (exp.f64 (neg.f64 im)) (exp.f64 im)))

   1. Initial program 98.0%

      \[\left(0.5 \cdot \sin re\right) \cdot \left(e^{-im} - e^{im}\right) \]

   2. Taylor expanded in re around 0

      \[\leadsto \color{blue}{\left(re \cdot \left(\frac{1}{2} + \frac{-1}{12} \cdot {re}^{2}\right)\right)} \cdot \left(e^{-im} - e^{im}\right) \]
   3. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(\left(\frac{1}{2} + \frac{-1}{12} \cdot {re}^{2}\right) \cdot \color{blue}{re}\right) \cdot \left(e^{-im} - e^{im}\right) \]

      2. lower-*.f64 N/A

         \[\leadsto \left(\left(\frac{1}{2} + \frac{-1}{12} \cdot {re}^{2}\right) \cdot \color{blue}{re}\right) \cdot \left(e^{-im} - e^{im}\right) \]

      3. +-commutative N/A

         \[\leadsto \left(\left(\frac{-1}{12} \cdot {re}^{2} + \frac{1}{2}\right) \cdot re\right) \cdot \left(e^{-im} - e^{im}\right) \]

      4. *-commutative N/A

         \[\leadsto \left(\left({re}^{2} \cdot \frac{-1}{12} + \frac{1}{2}\right) \cdot re\right) \cdot \left(e^{-im} - e^{im}\right) \]

      5. lower-fma.f64 N/A

         \[\leadsto \left(\mathsf{fma}\left({re}^{2}, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \cdot \left(e^{-im} - e^{im}\right) \]

      6. unpow2 N/A

         \[\leadsto \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \cdot \left(e^{-im} - e^{im}\right) \]

      7. lower-*.f64 72.6

         \[\leadsto \left(\mathsf{fma}\left(re \cdot re, -0.08333333333333333, 0.5\right) \cdot re\right) \cdot \left(e^{-im} - e^{im}\right) \]

   4. Applied rewrites 72.6%

      \[\leadsto \color{blue}{\left(\mathsf{fma}\left(re \cdot re, -0.08333333333333333, 0.5\right) \cdot re\right)} \cdot \left(e^{-im} - e^{im}\right) \]

   5. Taylor expanded in im around 0

      \[\leadsto \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \cdot \left(\color{blue}{1} - e^{im}\right) \]
   6. Step-by-step derivation

   7. Applied rewrites 71.5%

      \[\leadsto \left(\mathsf{fma}\left(re \cdot re, -0.08333333333333333, 0.5\right) \cdot re\right) \cdot \left(\color{blue}{1} - e^{im}\right) \]
3. Recombined 3 regimes into one program.
4. Add Preprocessing

Alternative 3: 62.7% accurate, 0.9× speedup

Math

\[\begin{array}{l} im\_m = \left|im\right| \\ im\_s = \mathsf{copysign}\left(1, im\right) \\ \begin{array}{l} t_0 := -2 \cdot \sinh im\_m\\ im\_s \cdot \begin{array}{l} \mathbf{if}\;0.5 \cdot \sin re \leq 5 \cdot 10^{-186}:\\ \;\;\;\;t\_0 \cdot \left(\mathsf{fma}\left(re \cdot re, -0.08333333333333333, 0.5\right) \cdot re\right)\\ \mathbf{else}:\\ \;\;\;\;\left(t\_0 \cdot re\right) \cdot 0.5\\ \end{array} \end{array} \end{array} \]

FPCore

im\_m = (fabs.f64 im)
im\_s = (copysign.f64 #s(literal 1 binary64) im)
(FPCore (im_s re im_m)
 :precision binary64
 (let* ((t_0 (* -2.0 (sinh im_m))))
   (*
    im_s
    (if (<= (* 0.5 (sin re)) 5e-186)
      (* t_0 (* (fma (* re re) -0.08333333333333333 0.5) re))
      (* (* t_0 re) 0.5)))))

C

im\_m = fabs(im);
im\_s = copysign(1.0, im);
double code(double im_s, double re, double im_m) {
	double t_0 = -2.0 * sinh(im_m);
	double tmp;
	if ((0.5 * sin(re)) <= 5e-186) {
		tmp = t_0 * (fma((re * re), -0.08333333333333333, 0.5) * re);
	} else {
		tmp = (t_0 * re) * 0.5;
	}
	return im_s * tmp;
}

Julia

im\_m = abs(im)
im\_s = copysign(1.0, im)
function code(im_s, re, im_m)
	t_0 = Float64(-2.0 * sinh(im_m))
	tmp = 0.0
	if (Float64(0.5 * sin(re)) <= 5e-186)
		tmp = Float64(t_0 * Float64(fma(Float64(re * re), -0.08333333333333333, 0.5) * re));
	else
		tmp = Float64(Float64(t_0 * re) * 0.5);
	end
	return Float64(im_s * tmp)
end

Wolfram

im\_m = N[Abs[im], $MachinePrecision]
im\_s = N[With[{TMP1 = Abs[1.0], TMP2 = Sign[im]}, TMP1 * If[TMP2 == 0, 1, TMP2]], $MachinePrecision]
code[im$95$s_, re_, im$95$m_] := Block[{t$95$0 = N[(-2.0 * N[Sinh[im$95$m], $MachinePrecision]), $MachinePrecision]}, N[(im$95$s * If[LessEqual[N[(0.5 * N[Sin[re], $MachinePrecision]), $MachinePrecision], 5e-186], N[(t$95$0 * N[(N[(N[(re * re), $MachinePrecision] * -0.08333333333333333 + 0.5), $MachinePrecision] * re), $MachinePrecision]), $MachinePrecision], N[(N[(t$95$0 * re), $MachinePrecision] * 0.5), $MachinePrecision]]), $MachinePrecision]]

Derivation

1. Split input into 2 regimes

2. if (*.f64 #s(literal 1/2 binary64) (sin.f64 re)) < 5e-186

   1. Initial program 69.3%

      \[\left(0.5 \cdot \sin re\right) \cdot \left(e^{-im} - e^{im}\right) \]

   2. Taylor expanded in re around 0

      \[\leadsto \color{blue}{\left(re \cdot \left(\frac{1}{2} + \frac{-1}{12} \cdot {re}^{2}\right)\right)} \cdot \left(e^{-im} - e^{im}\right) \]
   3. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(\left(\frac{1}{2} + \frac{-1}{12} \cdot {re}^{2}\right) \cdot \color{blue}{re}\right) \cdot \left(e^{-im} - e^{im}\right) \]

      2. lower-*.f64 N/A

         \[\leadsto \left(\left(\frac{1}{2} + \frac{-1}{12} \cdot {re}^{2}\right) \cdot \color{blue}{re}\right) \cdot \left(e^{-im} - e^{im}\right) \]

      3. +-commutative N/A

         \[\leadsto \left(\left(\frac{-1}{12} \cdot {re}^{2} + \frac{1}{2}\right) \cdot re\right) \cdot \left(e^{-im} - e^{im}\right) \]

      4. *-commutative N/A

         \[\leadsto \left(\left({re}^{2} \cdot \frac{-1}{12} + \frac{1}{2}\right) \cdot re\right) \cdot \left(e^{-im} - e^{im}\right) \]

      5. lower-fma.f64 N/A

         \[\leadsto \left(\mathsf{fma}\left({re}^{2}, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \cdot \left(e^{-im} - e^{im}\right) \]

      6. unpow2 N/A

         \[\leadsto \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \cdot \left(e^{-im} - e^{im}\right) \]

      7. lower-*.f64 57.6

         \[\leadsto \left(\mathsf{fma}\left(re \cdot re, -0.08333333333333333, 0.5\right) \cdot re\right) \cdot \left(e^{-im} - e^{im}\right) \]

   4. Applied rewrites 57.6%

      \[\leadsto \color{blue}{\left(\mathsf{fma}\left(re \cdot re, -0.08333333333333333, 0.5\right) \cdot re\right)} \cdot \left(e^{-im} - e^{im}\right) \]
   5. Step-by-step derivation

      1. lift-*.f64 N/A

         \[\leadsto \color{blue}{\left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \cdot \left(e^{-im} - e^{im}\right)} \]

      2. lift--.f64 N/A

         \[\leadsto \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \cdot \color{blue}{\left(e^{-im} - e^{im}\right)} \]

      3. lift-exp.f64 N/A

         \[\leadsto \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \cdot \left(\color{blue}{e^{-im}} - e^{im}\right) \]

      4. lift-neg.f64 N/A

         \[\leadsto \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \cdot \left(e^{\color{blue}{\mathsf{neg}\left(im\right)}} - e^{im}\right) \]

      5. lift-exp.f64 N/A

         \[\leadsto \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \cdot \left(e^{\mathsf{neg}\left(im\right)} - \color{blue}{e^{im}}\right) \]

      6. *-commutative N/A

         \[\leadsto \color{blue}{\left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right) \cdot \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right)} \]

      7. lower-*.f64 N/A

         \[\leadsto \color{blue}{\left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right) \cdot \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right)} \]

      8. sub-negate N/A

         \[\leadsto \color{blue}{\left(\mathsf{neg}\left(\left(e^{im} - e^{\mathsf{neg}\left(im\right)}\right)\right)\right)} \cdot \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \]

      9. sinh-undef-rev N/A

         \[\leadsto \left(\mathsf{neg}\left(\color{blue}{2 \cdot \sinh im}\right)\right) \cdot \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \]

      10. distribute-lft-neg-in N/A

          \[\leadsto \color{blue}{\left(\left(\mathsf{neg}\left(2\right)\right) \cdot \sinh im\right)} \cdot \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \]

      11. metadata-eval N/A

          \[\leadsto \left(\color{blue}{-2} \cdot \sinh im\right) \cdot \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \]

      12. lower-*.f64 N/A

          \[\leadsto \color{blue}{\left(-2 \cdot \sinh im\right)} \cdot \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \]

      13. lift-sinh.f64 69.4

          \[\leadsto \left(-2 \cdot \color{blue}{\sinh im}\right) \cdot \left(\mathsf{fma}\left(re \cdot re, -0.08333333333333333, 0.5\right) \cdot re\right) \]

   6. Applied rewrites 69.4%

      \[\leadsto \color{blue}{\left(-2 \cdot \sinh im\right) \cdot \left(\mathsf{fma}\left(re \cdot re, -0.08333333333333333, 0.5\right) \cdot re\right)} \]

   if 5e-186 < (*.f64 #s(literal 1/2 binary64) (sin.f64 re))

   1. Initial program 58.0%

      \[\left(0.5 \cdot \sin re\right) \cdot \left(e^{-im} - e^{im}\right) \]

   2. Taylor expanded in re around 0

      \[\leadsto \color{blue}{\frac{1}{2} \cdot \left(re \cdot \left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right)\right)} \]
   3. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(re \cdot \left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right)\right) \cdot \color{blue}{\frac{1}{2}} \]

      2. lower-*.f64 N/A

         \[\leadsto \left(re \cdot \left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right)\right) \cdot \color{blue}{\frac{1}{2}} \]

      3. *-commutative N/A

         \[\leadsto \left(\left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right) \cdot re\right) \cdot \frac{1}{2} \]

      4. lower-*.f64 N/A

         \[\leadsto \left(\left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right) \cdot re\right) \cdot \frac{1}{2} \]

      5. sub-negate N/A

         \[\leadsto \left(\left(\mathsf{neg}\left(\left(e^{im} - e^{\mathsf{neg}\left(im\right)}\right)\right)\right) \cdot re\right) \cdot \frac{1}{2} \]

      6. lower-neg.f64 N/A

         \[\leadsto \left(\left(-\left(e^{im} - e^{\mathsf{neg}\left(im\right)}\right)\right) \cdot re\right) \cdot \frac{1}{2} \]

      7. sinh-undef N/A

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot \frac{1}{2} \]

      8. lower-*.f64 N/A

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot \frac{1}{2} \]

      9. lower-sinh.f64 53.0

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot 0.5 \]

   4. Applied rewrites 53.0%

      \[\leadsto \color{blue}{\left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot 0.5} \]
   5. Step-by-step derivation

      1. flip-- 53.0

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot \color{blue}{re}\right) \cdot 0.5 \]

      2. pow2 53.0

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot 0.5 \]

      3. lift-neg.f64 N/A

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot \frac{1}{2} \]

      4. exp-prod N/A

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot \frac{1}{2} \]

      5. lift-neg.f64 53.0

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot 0.5 \]

      6. exp-sum 53.0

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot 0.5 \]

   6. Applied rewrites 53.0%

      \[\leadsto \color{blue}{\left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot 0.5} \]
3. Recombined 2 regimes into one program.
4. Add Preprocessing

Alternative 4: 62.7% accurate, 0.9× speedup

Math

\[\begin{array}{l} im\_m = \left|im\right| \\ im\_s = \mathsf{copysign}\left(1, im\right) \\ \begin{array}{l} t_0 := -2 \cdot \sinh im\_m\\ im\_s \cdot \begin{array}{l} \mathbf{if}\;0.5 \cdot \sin re \leq -0.02:\\ \;\;\;\;t\_0 \cdot \left(\left(\left(re \cdot re\right) \cdot re\right) \cdot -0.08333333333333333\right)\\ \mathbf{else}:\\ \;\;\;\;\left(t\_0 \cdot re\right) \cdot 0.5\\ \end{array} \end{array} \end{array} \]

FPCore

im\_m = (fabs.f64 im)
im\_s = (copysign.f64 #s(literal 1 binary64) im)
(FPCore (im_s re im_m)
 :precision binary64
 (let* ((t_0 (* -2.0 (sinh im_m))))
   (*
    im_s
    (if (<= (* 0.5 (sin re)) -0.02)
      (* t_0 (* (* (* re re) re) -0.08333333333333333))
      (* (* t_0 re) 0.5)))))

C

im\_m = fabs(im);
im\_s = copysign(1.0, im);
double code(double im_s, double re, double im_m) {
	double t_0 = -2.0 * sinh(im_m);
	double tmp;
	if ((0.5 * sin(re)) <= -0.02) {
		tmp = t_0 * (((re * re) * re) * -0.08333333333333333);
	} else {
		tmp = (t_0 * re) * 0.5;
	}
	return im_s * tmp;
}

Fortran

im\_m =     private
im\_s =     private
module fmin_fmax_functions
    implicit none
    private
    public fmax
    public fmin

    interface fmax
        module procedure fmax88
        module procedure fmax44
        module procedure fmax84
        module procedure fmax48
    end interface
    interface fmin
        module procedure fmin88
        module procedure fmin44
        module procedure fmin84
        module procedure fmin48
    end interface
contains
    real(8) function fmax88(x, y) result (res)
        real(8), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(x, max(x, y), y /= y), x /= x)
    end function
    real(4) function fmax44(x, y) result (res)
        real(4), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(y, merge(x, max(x, y), y /= y), x /= x)
    end function
    real(8) function fmax84(x, y) result(res)
        real(8), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(dble(y), merge(x, max(x, dble(y)), y /= y), x /= x)
    end function
    real(8) function fmax48(x, y) result(res)
        real(4), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(dble(x), max(dble(x), y), y /= y), x /= x)
    end function
    real(8) function fmin88(x, y) result (res)
        real(8), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(x, min(x, y), y /= y), x /= x)
    end function
    real(4) function fmin44(x, y) result (res)
        real(4), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(y, merge(x, min(x, y), y /= y), x /= x)
    end function
    real(8) function fmin84(x, y) result(res)
        real(8), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(dble(y), merge(x, min(x, dble(y)), y /= y), x /= x)
    end function
    real(8) function fmin48(x, y) result(res)
        real(4), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(dble(x), min(dble(x), y), y /= y), x /= x)
    end function
end module

real(8) function code(im_s, re, im_m)
use fmin_fmax_functions
    real(8), intent (in) :: im_s
    real(8), intent (in) :: re
    real(8), intent (in) :: im_m
    real(8) :: t_0
    real(8) :: tmp
    t_0 = (-2.0d0) * sinh(im_m)
    if ((0.5d0 * sin(re)) <= (-0.02d0)) then
        tmp = t_0 * (((re * re) * re) * (-0.08333333333333333d0))
    else
        tmp = (t_0 * re) * 0.5d0
    end if
    code = im_s * tmp
end function

Java

im\_m = Math.abs(im);
im\_s = Math.copySign(1.0, im);
public static double code(double im_s, double re, double im_m) {
	double t_0 = -2.0 * Math.sinh(im_m);
	double tmp;
	if ((0.5 * Math.sin(re)) <= -0.02) {
		tmp = t_0 * (((re * re) * re) * -0.08333333333333333);
	} else {
		tmp = (t_0 * re) * 0.5;
	}
	return im_s * tmp;
}

Python

im\_m = math.fabs(im)
im\_s = math.copysign(1.0, im)
def code(im_s, re, im_m):
	t_0 = -2.0 * math.sinh(im_m)
	tmp = 0
	if (0.5 * math.sin(re)) <= -0.02:
		tmp = t_0 * (((re * re) * re) * -0.08333333333333333)
	else:
		tmp = (t_0 * re) * 0.5
	return im_s * tmp

Julia

im\_m = abs(im)
im\_s = copysign(1.0, im)
function code(im_s, re, im_m)
	t_0 = Float64(-2.0 * sinh(im_m))
	tmp = 0.0
	if (Float64(0.5 * sin(re)) <= -0.02)
		tmp = Float64(t_0 * Float64(Float64(Float64(re * re) * re) * -0.08333333333333333));
	else
		tmp = Float64(Float64(t_0 * re) * 0.5);
	end
	return Float64(im_s * tmp)
end

MATLAB

im\_m = abs(im);
im\_s = sign(im) * abs(1.0);
function tmp_2 = code(im_s, re, im_m)
	t_0 = -2.0 * sinh(im_m);
	tmp = 0.0;
	if ((0.5 * sin(re)) <= -0.02)
		tmp = t_0 * (((re * re) * re) * -0.08333333333333333);
	else
		tmp = (t_0 * re) * 0.5;
	end
	tmp_2 = im_s * tmp;
end

Wolfram

im\_m = N[Abs[im], $MachinePrecision]
im\_s = N[With[{TMP1 = Abs[1.0], TMP2 = Sign[im]}, TMP1 * If[TMP2 == 0, 1, TMP2]], $MachinePrecision]
code[im$95$s_, re_, im$95$m_] := Block[{t$95$0 = N[(-2.0 * N[Sinh[im$95$m], $MachinePrecision]), $MachinePrecision]}, N[(im$95$s * If[LessEqual[N[(0.5 * N[Sin[re], $MachinePrecision]), $MachinePrecision], -0.02], N[(t$95$0 * N[(N[(N[(re * re), $MachinePrecision] * re), $MachinePrecision] * -0.08333333333333333), $MachinePrecision]), $MachinePrecision], N[(N[(t$95$0 * re), $MachinePrecision] * 0.5), $MachinePrecision]]), $MachinePrecision]]

Derivation

1. Split input into 2 regimes

2. if (*.f64 #s(literal 1/2 binary64) (sin.f64 re)) < -0.0200000000000000004

   1. Initial program 53.7%

      \[\left(0.5 \cdot \sin re\right) \cdot \left(e^{-im} - e^{im}\right) \]

   2. Taylor expanded in re around 0

      \[\leadsto \color{blue}{\left(re \cdot \left(\frac{1}{2} + \frac{-1}{12} \cdot {re}^{2}\right)\right)} \cdot \left(e^{-im} - e^{im}\right) \]
   3. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(\left(\frac{1}{2} + \frac{-1}{12} \cdot {re}^{2}\right) \cdot \color{blue}{re}\right) \cdot \left(e^{-im} - e^{im}\right) \]

      2. lower-*.f64 N/A

         \[\leadsto \left(\left(\frac{1}{2} + \frac{-1}{12} \cdot {re}^{2}\right) \cdot \color{blue}{re}\right) \cdot \left(e^{-im} - e^{im}\right) \]

      3. +-commutative N/A

         \[\leadsto \left(\left(\frac{-1}{12} \cdot {re}^{2} + \frac{1}{2}\right) \cdot re\right) \cdot \left(e^{-im} - e^{im}\right) \]

      4. *-commutative N/A

         \[\leadsto \left(\left({re}^{2} \cdot \frac{-1}{12} + \frac{1}{2}\right) \cdot re\right) \cdot \left(e^{-im} - e^{im}\right) \]

      5. lower-fma.f64 N/A

         \[\leadsto \left(\mathsf{fma}\left({re}^{2}, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \cdot \left(e^{-im} - e^{im}\right) \]

      6. unpow2 N/A

         \[\leadsto \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \cdot \left(e^{-im} - e^{im}\right) \]

      7. lower-*.f64 25.5

         \[\leadsto \left(\mathsf{fma}\left(re \cdot re, -0.08333333333333333, 0.5\right) \cdot re\right) \cdot \left(e^{-im} - e^{im}\right) \]

   4. Applied rewrites 25.5%

      \[\leadsto \color{blue}{\left(\mathsf{fma}\left(re \cdot re, -0.08333333333333333, 0.5\right) \cdot re\right)} \cdot \left(e^{-im} - e^{im}\right) \]
   5. Step-by-step derivation

      1. lift-*.f64 N/A

         \[\leadsto \color{blue}{\left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \cdot \left(e^{-im} - e^{im}\right)} \]

      2. lift--.f64 N/A

         \[\leadsto \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \cdot \color{blue}{\left(e^{-im} - e^{im}\right)} \]

      3. lift-exp.f64 N/A

         \[\leadsto \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \cdot \left(\color{blue}{e^{-im}} - e^{im}\right) \]

      4. lift-neg.f64 N/A

         \[\leadsto \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \cdot \left(e^{\color{blue}{\mathsf{neg}\left(im\right)}} - e^{im}\right) \]

      5. lift-exp.f64 N/A

         \[\leadsto \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \cdot \left(e^{\mathsf{neg}\left(im\right)} - \color{blue}{e^{im}}\right) \]

      6. *-commutative N/A

         \[\leadsto \color{blue}{\left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right) \cdot \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right)} \]

      7. lower-*.f64 N/A

         \[\leadsto \color{blue}{\left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right) \cdot \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right)} \]

      8. sub-negate N/A

         \[\leadsto \color{blue}{\left(\mathsf{neg}\left(\left(e^{im} - e^{\mathsf{neg}\left(im\right)}\right)\right)\right)} \cdot \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \]

      9. sinh-undef-rev N/A

         \[\leadsto \left(\mathsf{neg}\left(\color{blue}{2 \cdot \sinh im}\right)\right) \cdot \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \]

      10. distribute-lft-neg-in N/A

          \[\leadsto \color{blue}{\left(\left(\mathsf{neg}\left(2\right)\right) \cdot \sinh im\right)} \cdot \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \]

      11. metadata-eval N/A

          \[\leadsto \left(\color{blue}{-2} \cdot \sinh im\right) \cdot \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \]

      12. lower-*.f64 N/A

          \[\leadsto \color{blue}{\left(-2 \cdot \sinh im\right)} \cdot \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \]

      13. lift-sinh.f64 25.9

          \[\leadsto \left(-2 \cdot \color{blue}{\sinh im}\right) \cdot \left(\mathsf{fma}\left(re \cdot re, -0.08333333333333333, 0.5\right) \cdot re\right) \]

   6. Applied rewrites 25.9%

      \[\leadsto \color{blue}{\left(-2 \cdot \sinh im\right) \cdot \left(\mathsf{fma}\left(re \cdot re, -0.08333333333333333, 0.5\right) \cdot re\right)} \]

   7. Taylor expanded in re around inf

      \[\leadsto \left(-2 \cdot \sinh im\right) \cdot \left(\frac{-1}{12} \cdot \color{blue}{{re}^{3}}\right) \]
   8. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(-2 \cdot \sinh im\right) \cdot \left({re}^{3} \cdot \frac{-1}{12}\right) \]

      2. lower-*.f64 N/A

         \[\leadsto \left(-2 \cdot \sinh im\right) \cdot \left({re}^{3} \cdot \frac{-1}{12}\right) \]

      3. unpow3 N/A

         \[\leadsto \left(-2 \cdot \sinh im\right) \cdot \left(\left(\left(re \cdot re\right) \cdot re\right) \cdot \frac{-1}{12}\right) \]

      4. pow2 N/A

         \[\leadsto \left(-2 \cdot \sinh im\right) \cdot \left(\left({re}^{2} \cdot re\right) \cdot \frac{-1}{12}\right) \]

      5. lower-*.f64 N/A

         \[\leadsto \left(-2 \cdot \sinh im\right) \cdot \left(\left({re}^{2} \cdot re\right) \cdot \frac{-1}{12}\right) \]

      6. pow2 N/A

         \[\leadsto \left(-2 \cdot \sinh im\right) \cdot \left(\left(\left(re \cdot re\right) \cdot re\right) \cdot \frac{-1}{12}\right) \]

      7. lift-*.f64 25.6

         \[\leadsto \left(-2 \cdot \sinh im\right) \cdot \left(\left(\left(re \cdot re\right) \cdot re\right) \cdot -0.08333333333333333\right) \]

   9. Applied rewrites 25.6%

      \[\leadsto \left(-2 \cdot \sinh im\right) \cdot \left(\left(\left(re \cdot re\right) \cdot re\right) \cdot \color{blue}{-0.08333333333333333}\right) \]

   if -0.0200000000000000004 < (*.f64 #s(literal 1/2 binary64) (sin.f64 re))

   1. Initial program 68.2%

      \[\left(0.5 \cdot \sin re\right) \cdot \left(e^{-im} - e^{im}\right) \]

   2. Taylor expanded in re around 0

      \[\leadsto \color{blue}{\frac{1}{2} \cdot \left(re \cdot \left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right)\right)} \]
   3. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(re \cdot \left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right)\right) \cdot \color{blue}{\frac{1}{2}} \]

      2. lower-*.f64 N/A

         \[\leadsto \left(re \cdot \left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right)\right) \cdot \color{blue}{\frac{1}{2}} \]

      3. *-commutative N/A

         \[\leadsto \left(\left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right) \cdot re\right) \cdot \frac{1}{2} \]

      4. lower-*.f64 N/A

         \[\leadsto \left(\left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right) \cdot re\right) \cdot \frac{1}{2} \]

      5. sub-negate N/A

         \[\leadsto \left(\left(\mathsf{neg}\left(\left(e^{im} - e^{\mathsf{neg}\left(im\right)}\right)\right)\right) \cdot re\right) \cdot \frac{1}{2} \]

      6. lower-neg.f64 N/A

         \[\leadsto \left(\left(-\left(e^{im} - e^{\mathsf{neg}\left(im\right)}\right)\right) \cdot re\right) \cdot \frac{1}{2} \]

      7. sinh-undef N/A

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot \frac{1}{2} \]

      8. lower-*.f64 N/A

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot \frac{1}{2} \]

      9. lower-sinh.f64 74.2

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot 0.5 \]

   4. Applied rewrites 74.2%

      \[\leadsto \color{blue}{\left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot 0.5} \]
   5. Step-by-step derivation

      1. flip-- 74.2

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot \color{blue}{re}\right) \cdot 0.5 \]

      2. pow2 74.2

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot 0.5 \]

      3. lift-neg.f64 N/A

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot \frac{1}{2} \]

      4. exp-prod N/A

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot \frac{1}{2} \]

      5. lift-neg.f64 74.2

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot 0.5 \]

      6. exp-sum 74.2

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot 0.5 \]

   6. Applied rewrites 74.2%

      \[\leadsto \color{blue}{\left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot 0.5} \]
3. Recombined 2 regimes into one program.
4. Add Preprocessing

Alternative 5: 62.3% accurate, 0.7× speedup

Math

\[\begin{array}{l} im\_m = \left|im\right| \\ im\_s = \mathsf{copysign}\left(1, im\right) \\ im\_s \cdot \begin{array}{l} \mathbf{if}\;\left(0.5 \cdot \sin re\right) \cdot \left(e^{-im\_m} - e^{im\_m}\right) \leq 0:\\ \;\;\;\;\left(\left(-2 \cdot \sinh im\_m\right) \cdot re\right) \cdot 0.5\\ \mathbf{else}:\\ \;\;\;\;\left(\mathsf{fma}\left(re \cdot re, -0.08333333333333333, 0.5\right) \cdot re\right) \cdot \left(1 - e^{im\_m}\right)\\ \end{array} \end{array} \]

FPCore

im\_m = (fabs.f64 im)
im\_s = (copysign.f64 #s(literal 1 binary64) im)
(FPCore (im_s re im_m)
 :precision binary64
 (*
  im_s
  (if (<= (* (* 0.5 (sin re)) (- (exp (- im_m)) (exp im_m))) 0.0)
    (* (* (* -2.0 (sinh im_m)) re) 0.5)
    (* (* (fma (* re re) -0.08333333333333333 0.5) re) (- 1.0 (exp im_m))))))

C

im\_m = fabs(im);
im\_s = copysign(1.0, im);
double code(double im_s, double re, double im_m) {
	double tmp;
	if (((0.5 * sin(re)) * (exp(-im_m) - exp(im_m))) <= 0.0) {
		tmp = ((-2.0 * sinh(im_m)) * re) * 0.5;
	} else {
		tmp = (fma((re * re), -0.08333333333333333, 0.5) * re) * (1.0 - exp(im_m));
	}
	return im_s * tmp;
}

Julia

im\_m = abs(im)
im\_s = copysign(1.0, im)
function code(im_s, re, im_m)
	tmp = 0.0
	if (Float64(Float64(0.5 * sin(re)) * Float64(exp(Float64(-im_m)) - exp(im_m))) <= 0.0)
		tmp = Float64(Float64(Float64(-2.0 * sinh(im_m)) * re) * 0.5);
	else
		tmp = Float64(Float64(fma(Float64(re * re), -0.08333333333333333, 0.5) * re) * Float64(1.0 - exp(im_m)));
	end
	return Float64(im_s * tmp)
end

Wolfram

im\_m = N[Abs[im], $MachinePrecision]
im\_s = N[With[{TMP1 = Abs[1.0], TMP2 = Sign[im]}, TMP1 * If[TMP2 == 0, 1, TMP2]], $MachinePrecision]
code[im$95$s_, re_, im$95$m_] := N[(im$95$s * If[LessEqual[N[(N[(0.5 * N[Sin[re], $MachinePrecision]), $MachinePrecision] * N[(N[Exp[(-im$95$m)], $MachinePrecision] - N[Exp[im$95$m], $MachinePrecision]), $MachinePrecision]), $MachinePrecision], 0.0], N[(N[(N[(-2.0 * N[Sinh[im$95$m], $MachinePrecision]), $MachinePrecision] * re), $MachinePrecision] * 0.5), $MachinePrecision], N[(N[(N[(N[(re * re), $MachinePrecision] * -0.08333333333333333 + 0.5), $MachinePrecision] * re), $MachinePrecision] * N[(1.0 - N[Exp[im$95$m], $MachinePrecision]), $MachinePrecision]), $MachinePrecision]]), $MachinePrecision]

Derivation

1. Split input into 2 regimes

2. if (*.f64 (*.f64 #s(literal 1/2 binary64) (sin.f64 re)) (-.f64 (exp.f64 (neg.f64 im)) (exp.f64 im))) < 0.0

   1. Initial program 53.3%

      \[\left(0.5 \cdot \sin re\right) \cdot \left(e^{-im} - e^{im}\right) \]

   2. Taylor expanded in re around 0

      \[\leadsto \color{blue}{\frac{1}{2} \cdot \left(re \cdot \left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right)\right)} \]
   3. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(re \cdot \left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right)\right) \cdot \color{blue}{\frac{1}{2}} \]

      2. lower-*.f64 N/A

         \[\leadsto \left(re \cdot \left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right)\right) \cdot \color{blue}{\frac{1}{2}} \]

      3. *-commutative N/A

         \[\leadsto \left(\left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right) \cdot re\right) \cdot \frac{1}{2} \]

      4. lower-*.f64 N/A

         \[\leadsto \left(\left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right) \cdot re\right) \cdot \frac{1}{2} \]

      5. sub-negate N/A

         \[\leadsto \left(\left(\mathsf{neg}\left(\left(e^{im} - e^{\mathsf{neg}\left(im\right)}\right)\right)\right) \cdot re\right) \cdot \frac{1}{2} \]

      6. lower-neg.f64 N/A

         \[\leadsto \left(\left(-\left(e^{im} - e^{\mathsf{neg}\left(im\right)}\right)\right) \cdot re\right) \cdot \frac{1}{2} \]

      7. sinh-undef N/A

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot \frac{1}{2} \]

      8. lower-*.f64 N/A

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot \frac{1}{2} \]

      9. lower-sinh.f64 59.1

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot 0.5 \]

   4. Applied rewrites 59.1%

      \[\leadsto \color{blue}{\left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot 0.5} \]
   5. Step-by-step derivation

      1. flip-- 59.1

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot \color{blue}{re}\right) \cdot 0.5 \]

      2. pow2 59.1

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot 0.5 \]

      3. lift-neg.f64 N/A

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot \frac{1}{2} \]

      4. exp-prod N/A

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot \frac{1}{2} \]

      5. lift-neg.f64 59.1

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot 0.5 \]

      6. exp-sum 59.1

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot 0.5 \]

   6. Applied rewrites 59.1%

      \[\leadsto \color{blue}{\left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot 0.5} \]

   if 0.0 < (*.f64 (*.f64 #s(literal 1/2 binary64) (sin.f64 re)) (-.f64 (exp.f64 (neg.f64 im)) (exp.f64 im)))

   1. Initial program 98.0%

      \[\left(0.5 \cdot \sin re\right) \cdot \left(e^{-im} - e^{im}\right) \]

   2. Taylor expanded in re around 0

      \[\leadsto \color{blue}{\left(re \cdot \left(\frac{1}{2} + \frac{-1}{12} \cdot {re}^{2}\right)\right)} \cdot \left(e^{-im} - e^{im}\right) \]
   3. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(\left(\frac{1}{2} + \frac{-1}{12} \cdot {re}^{2}\right) \cdot \color{blue}{re}\right) \cdot \left(e^{-im} - e^{im}\right) \]

      2. lower-*.f64 N/A

         \[\leadsto \left(\left(\frac{1}{2} + \frac{-1}{12} \cdot {re}^{2}\right) \cdot \color{blue}{re}\right) \cdot \left(e^{-im} - e^{im}\right) \]

      3. +-commutative N/A

         \[\leadsto \left(\left(\frac{-1}{12} \cdot {re}^{2} + \frac{1}{2}\right) \cdot re\right) \cdot \left(e^{-im} - e^{im}\right) \]

      4. *-commutative N/A

         \[\leadsto \left(\left({re}^{2} \cdot \frac{-1}{12} + \frac{1}{2}\right) \cdot re\right) \cdot \left(e^{-im} - e^{im}\right) \]

      5. lower-fma.f64 N/A

         \[\leadsto \left(\mathsf{fma}\left({re}^{2}, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \cdot \left(e^{-im} - e^{im}\right) \]

      6. unpow2 N/A

         \[\leadsto \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \cdot \left(e^{-im} - e^{im}\right) \]

      7. lower-*.f64 72.6

         \[\leadsto \left(\mathsf{fma}\left(re \cdot re, -0.08333333333333333, 0.5\right) \cdot re\right) \cdot \left(e^{-im} - e^{im}\right) \]

   4. Applied rewrites 72.6%

      \[\leadsto \color{blue}{\left(\mathsf{fma}\left(re \cdot re, -0.08333333333333333, 0.5\right) \cdot re\right)} \cdot \left(e^{-im} - e^{im}\right) \]

   5. Taylor expanded in im around 0

      \[\leadsto \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \cdot \left(\color{blue}{1} - e^{im}\right) \]
   6. Step-by-step derivation

   7. Applied rewrites 71.5%

      \[\leadsto \left(\mathsf{fma}\left(re \cdot re, -0.08333333333333333, 0.5\right) \cdot re\right) \cdot \left(\color{blue}{1} - e^{im}\right) \]
3. Recombined 2 regimes into one program.
4. Add Preprocessing

Alternative 6: 62.3% accurate, 1.0× speedup

Math

\[\begin{array}{l} im\_m = \left|im\right| \\ im\_s = \mathsf{copysign}\left(1, im\right) \\ im\_s \cdot \begin{array}{l} \mathbf{if}\;0.5 \cdot \sin re \leq -0.02:\\ \;\;\;\;\left(\mathsf{fma}\left(-0.3333333333333333, im\_m \cdot im\_m, -2\right) \cdot im\_m\right) \cdot \left(\mathsf{fma}\left(re \cdot re, -0.08333333333333333, 0.5\right) \cdot re\right)\\ \mathbf{else}:\\ \;\;\;\;\left(\left(-2 \cdot \sinh im\_m\right) \cdot re\right) \cdot 0.5\\ \end{array} \end{array} \]

FPCore

im\_m = (fabs.f64 im)
im\_s = (copysign.f64 #s(literal 1 binary64) im)
(FPCore (im_s re im_m)
 :precision binary64
 (*
  im_s
  (if (<= (* 0.5 (sin re)) -0.02)
    (*
     (* (fma -0.3333333333333333 (* im_m im_m) -2.0) im_m)
     (* (fma (* re re) -0.08333333333333333 0.5) re))
    (* (* (* -2.0 (sinh im_m)) re) 0.5))))

C

im\_m = fabs(im);
im\_s = copysign(1.0, im);
double code(double im_s, double re, double im_m) {
	double tmp;
	if ((0.5 * sin(re)) <= -0.02) {
		tmp = (fma(-0.3333333333333333, (im_m * im_m), -2.0) * im_m) * (fma((re * re), -0.08333333333333333, 0.5) * re);
	} else {
		tmp = ((-2.0 * sinh(im_m)) * re) * 0.5;
	}
	return im_s * tmp;
}

Julia

im\_m = abs(im)
im\_s = copysign(1.0, im)
function code(im_s, re, im_m)
	tmp = 0.0
	if (Float64(0.5 * sin(re)) <= -0.02)
		tmp = Float64(Float64(fma(-0.3333333333333333, Float64(im_m * im_m), -2.0) * im_m) * Float64(fma(Float64(re * re), -0.08333333333333333, 0.5) * re));
	else
		tmp = Float64(Float64(Float64(-2.0 * sinh(im_m)) * re) * 0.5);
	end
	return Float64(im_s * tmp)
end

Wolfram

im\_m = N[Abs[im], $MachinePrecision]
im\_s = N[With[{TMP1 = Abs[1.0], TMP2 = Sign[im]}, TMP1 * If[TMP2 == 0, 1, TMP2]], $MachinePrecision]
code[im$95$s_, re_, im$95$m_] := N[(im$95$s * If[LessEqual[N[(0.5 * N[Sin[re], $MachinePrecision]), $MachinePrecision], -0.02], N[(N[(N[(-0.3333333333333333 * N[(im$95$m * im$95$m), $MachinePrecision] + -2.0), $MachinePrecision] * im$95$m), $MachinePrecision] * N[(N[(N[(re * re), $MachinePrecision] * -0.08333333333333333 + 0.5), $MachinePrecision] * re), $MachinePrecision]), $MachinePrecision], N[(N[(N[(-2.0 * N[Sinh[im$95$m], $MachinePrecision]), $MachinePrecision] * re), $MachinePrecision] * 0.5), $MachinePrecision]]), $MachinePrecision]

Derivation

1. Split input into 2 regimes

2. if (*.f64 #s(literal 1/2 binary64) (sin.f64 re)) < -0.0200000000000000004

   1. Initial program 53.7%

      \[\left(0.5 \cdot \sin re\right) \cdot \left(e^{-im} - e^{im}\right) \]

   2. Taylor expanded in re around 0

      \[\leadsto \color{blue}{\left(re \cdot \left(\frac{1}{2} + \frac{-1}{12} \cdot {re}^{2}\right)\right)} \cdot \left(e^{-im} - e^{im}\right) \]
   3. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(\left(\frac{1}{2} + \frac{-1}{12} \cdot {re}^{2}\right) \cdot \color{blue}{re}\right) \cdot \left(e^{-im} - e^{im}\right) \]

      2. lower-*.f64 N/A

         \[\leadsto \left(\left(\frac{1}{2} + \frac{-1}{12} \cdot {re}^{2}\right) \cdot \color{blue}{re}\right) \cdot \left(e^{-im} - e^{im}\right) \]

      3. +-commutative N/A

         \[\leadsto \left(\left(\frac{-1}{12} \cdot {re}^{2} + \frac{1}{2}\right) \cdot re\right) \cdot \left(e^{-im} - e^{im}\right) \]

      4. *-commutative N/A

         \[\leadsto \left(\left({re}^{2} \cdot \frac{-1}{12} + \frac{1}{2}\right) \cdot re\right) \cdot \left(e^{-im} - e^{im}\right) \]

      5. lower-fma.f64 N/A

         \[\leadsto \left(\mathsf{fma}\left({re}^{2}, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \cdot \left(e^{-im} - e^{im}\right) \]

      6. unpow2 N/A

         \[\leadsto \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \cdot \left(e^{-im} - e^{im}\right) \]

      7. lower-*.f64 25.5

         \[\leadsto \left(\mathsf{fma}\left(re \cdot re, -0.08333333333333333, 0.5\right) \cdot re\right) \cdot \left(e^{-im} - e^{im}\right) \]

   4. Applied rewrites 25.5%

      \[\leadsto \color{blue}{\left(\mathsf{fma}\left(re \cdot re, -0.08333333333333333, 0.5\right) \cdot re\right)} \cdot \left(e^{-im} - e^{im}\right) \]
   5. Step-by-step derivation

      1. lift-*.f64 N/A

         \[\leadsto \color{blue}{\left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \cdot \left(e^{-im} - e^{im}\right)} \]

      2. lift--.f64 N/A

         \[\leadsto \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \cdot \color{blue}{\left(e^{-im} - e^{im}\right)} \]

      3. lift-exp.f64 N/A

         \[\leadsto \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \cdot \left(\color{blue}{e^{-im}} - e^{im}\right) \]

      4. lift-neg.f64 N/A

         \[\leadsto \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \cdot \left(e^{\color{blue}{\mathsf{neg}\left(im\right)}} - e^{im}\right) \]

      5. lift-exp.f64 N/A

         \[\leadsto \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \cdot \left(e^{\mathsf{neg}\left(im\right)} - \color{blue}{e^{im}}\right) \]

      6. *-commutative N/A

         \[\leadsto \color{blue}{\left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right) \cdot \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right)} \]

      7. lower-*.f64 N/A

         \[\leadsto \color{blue}{\left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right) \cdot \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right)} \]

      8. sub-negate N/A

         \[\leadsto \color{blue}{\left(\mathsf{neg}\left(\left(e^{im} - e^{\mathsf{neg}\left(im\right)}\right)\right)\right)} \cdot \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \]

      9. sinh-undef-rev N/A

         \[\leadsto \left(\mathsf{neg}\left(\color{blue}{2 \cdot \sinh im}\right)\right) \cdot \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \]

      10. distribute-lft-neg-in N/A

          \[\leadsto \color{blue}{\left(\left(\mathsf{neg}\left(2\right)\right) \cdot \sinh im\right)} \cdot \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \]

      11. metadata-eval N/A

          \[\leadsto \left(\color{blue}{-2} \cdot \sinh im\right) \cdot \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \]

      12. lower-*.f64 N/A

          \[\leadsto \color{blue}{\left(-2 \cdot \sinh im\right)} \cdot \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \]

      13. lift-sinh.f64 25.9

          \[\leadsto \left(-2 \cdot \color{blue}{\sinh im}\right) \cdot \left(\mathsf{fma}\left(re \cdot re, -0.08333333333333333, 0.5\right) \cdot re\right) \]

   6. Applied rewrites 25.9%

      \[\leadsto \color{blue}{\left(-2 \cdot \sinh im\right) \cdot \left(\mathsf{fma}\left(re \cdot re, -0.08333333333333333, 0.5\right) \cdot re\right)} \]

   7. Taylor expanded in im around 0

      \[\leadsto \color{blue}{\left(im \cdot \left(\frac{-1}{3} \cdot {im}^{2} - 2\right)\right)} \cdot \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \]
   8. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(\left(\frac{-1}{3} \cdot {im}^{2} - 2\right) \cdot \color{blue}{im}\right) \cdot \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \]

      2. lower-*.f64 N/A

         \[\leadsto \left(\left(\frac{-1}{3} \cdot {im}^{2} - 2\right) \cdot \color{blue}{im}\right) \cdot \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \]

      3. sub-flip N/A

         \[\leadsto \left(\left(\frac{-1}{3} \cdot {im}^{2} + \left(\mathsf{neg}\left(2\right)\right)\right) \cdot im\right) \cdot \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \]

      4. metadata-eval N/A

         \[\leadsto \left(\left(\frac{-1}{3} \cdot {im}^{2} + -2\right) \cdot im\right) \cdot \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \]

      5. lower-fma.f64 N/A

         \[\leadsto \left(\mathsf{fma}\left(\frac{-1}{3}, {im}^{2}, -2\right) \cdot im\right) \cdot \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \]

      6. pow2 N/A

         \[\leadsto \left(\mathsf{fma}\left(\frac{-1}{3}, im \cdot im, -2\right) \cdot im\right) \cdot \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \]

      7. lift-*.f64 24.3

         \[\leadsto \left(\mathsf{fma}\left(-0.3333333333333333, im \cdot im, -2\right) \cdot im\right) \cdot \left(\mathsf{fma}\left(re \cdot re, -0.08333333333333333, 0.5\right) \cdot re\right) \]

   9. Applied rewrites 24.3%

      \[\leadsto \color{blue}{\left(\mathsf{fma}\left(-0.3333333333333333, im \cdot im, -2\right) \cdot im\right)} \cdot \left(\mathsf{fma}\left(re \cdot re, -0.08333333333333333, 0.5\right) \cdot re\right) \]

   if -0.0200000000000000004 < (*.f64 #s(literal 1/2 binary64) (sin.f64 re))

   1. Initial program 68.2%

      \[\left(0.5 \cdot \sin re\right) \cdot \left(e^{-im} - e^{im}\right) \]

   2. Taylor expanded in re around 0

      \[\leadsto \color{blue}{\frac{1}{2} \cdot \left(re \cdot \left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right)\right)} \]
   3. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(re \cdot \left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right)\right) \cdot \color{blue}{\frac{1}{2}} \]

      2. lower-*.f64 N/A

         \[\leadsto \left(re \cdot \left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right)\right) \cdot \color{blue}{\frac{1}{2}} \]

      3. *-commutative N/A

         \[\leadsto \left(\left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right) \cdot re\right) \cdot \frac{1}{2} \]

      4. lower-*.f64 N/A

         \[\leadsto \left(\left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right) \cdot re\right) \cdot \frac{1}{2} \]

      5. sub-negate N/A

         \[\leadsto \left(\left(\mathsf{neg}\left(\left(e^{im} - e^{\mathsf{neg}\left(im\right)}\right)\right)\right) \cdot re\right) \cdot \frac{1}{2} \]

      6. lower-neg.f64 N/A

         \[\leadsto \left(\left(-\left(e^{im} - e^{\mathsf{neg}\left(im\right)}\right)\right) \cdot re\right) \cdot \frac{1}{2} \]

      7. sinh-undef N/A

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot \frac{1}{2} \]

      8. lower-*.f64 N/A

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot \frac{1}{2} \]

      9. lower-sinh.f64 74.2

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot 0.5 \]

   4. Applied rewrites 74.2%

      \[\leadsto \color{blue}{\left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot 0.5} \]
   5. Step-by-step derivation

      1. flip-- 74.2

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot \color{blue}{re}\right) \cdot 0.5 \]

      2. pow2 74.2

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot 0.5 \]

      3. lift-neg.f64 N/A

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot \frac{1}{2} \]

      4. exp-prod N/A

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot \frac{1}{2} \]

      5. lift-neg.f64 74.2

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot 0.5 \]

      6. exp-sum 74.2

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot 0.5 \]

   6. Applied rewrites 74.2%

      \[\leadsto \color{blue}{\left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot 0.5} \]
3. Recombined 2 regimes into one program.
4. Add Preprocessing

Alternative 7: 62.3% accurate, 1.0× speedup

Math

\[\begin{array}{l} im\_m = \left|im\right| \\ im\_s = \mathsf{copysign}\left(1, im\right) \\ im\_s \cdot \begin{array}{l} \mathbf{if}\;0.5 \cdot \sin re \leq -0.02:\\ \;\;\;\;\left(\left(\left(im\_m \cdot im\_m\right) \cdot im\_m\right) \cdot -0.3333333333333333\right) \cdot \left(\mathsf{fma}\left(re \cdot re, -0.08333333333333333, 0.5\right) \cdot re\right)\\ \mathbf{else}:\\ \;\;\;\;\left(\left(-2 \cdot \sinh im\_m\right) \cdot re\right) \cdot 0.5\\ \end{array} \end{array} \]

FPCore

im\_m = (fabs.f64 im)
im\_s = (copysign.f64 #s(literal 1 binary64) im)
(FPCore (im_s re im_m)
 :precision binary64
 (*
  im_s
  (if (<= (* 0.5 (sin re)) -0.02)
    (*
     (* (* (* im_m im_m) im_m) -0.3333333333333333)
     (* (fma (* re re) -0.08333333333333333 0.5) re))
    (* (* (* -2.0 (sinh im_m)) re) 0.5))))

C

im\_m = fabs(im);
im\_s = copysign(1.0, im);
double code(double im_s, double re, double im_m) {
	double tmp;
	if ((0.5 * sin(re)) <= -0.02) {
		tmp = (((im_m * im_m) * im_m) * -0.3333333333333333) * (fma((re * re), -0.08333333333333333, 0.5) * re);
	} else {
		tmp = ((-2.0 * sinh(im_m)) * re) * 0.5;
	}
	return im_s * tmp;
}

Julia

im\_m = abs(im)
im\_s = copysign(1.0, im)
function code(im_s, re, im_m)
	tmp = 0.0
	if (Float64(0.5 * sin(re)) <= -0.02)
		tmp = Float64(Float64(Float64(Float64(im_m * im_m) * im_m) * -0.3333333333333333) * Float64(fma(Float64(re * re), -0.08333333333333333, 0.5) * re));
	else
		tmp = Float64(Float64(Float64(-2.0 * sinh(im_m)) * re) * 0.5);
	end
	return Float64(im_s * tmp)
end

Wolfram

im\_m = N[Abs[im], $MachinePrecision]
im\_s = N[With[{TMP1 = Abs[1.0], TMP2 = Sign[im]}, TMP1 * If[TMP2 == 0, 1, TMP2]], $MachinePrecision]
code[im$95$s_, re_, im$95$m_] := N[(im$95$s * If[LessEqual[N[(0.5 * N[Sin[re], $MachinePrecision]), $MachinePrecision], -0.02], N[(N[(N[(N[(im$95$m * im$95$m), $MachinePrecision] * im$95$m), $MachinePrecision] * -0.3333333333333333), $MachinePrecision] * N[(N[(N[(re * re), $MachinePrecision] * -0.08333333333333333 + 0.5), $MachinePrecision] * re), $MachinePrecision]), $MachinePrecision], N[(N[(N[(-2.0 * N[Sinh[im$95$m], $MachinePrecision]), $MachinePrecision] * re), $MachinePrecision] * 0.5), $MachinePrecision]]), $MachinePrecision]

Derivation

1. Split input into 2 regimes

2. if (*.f64 #s(literal 1/2 binary64) (sin.f64 re)) < -0.0200000000000000004

   1. Initial program 53.7%

      \[\left(0.5 \cdot \sin re\right) \cdot \left(e^{-im} - e^{im}\right) \]

   2. Taylor expanded in re around 0

      \[\leadsto \color{blue}{\left(re \cdot \left(\frac{1}{2} + \frac{-1}{12} \cdot {re}^{2}\right)\right)} \cdot \left(e^{-im} - e^{im}\right) \]
   3. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(\left(\frac{1}{2} + \frac{-1}{12} \cdot {re}^{2}\right) \cdot \color{blue}{re}\right) \cdot \left(e^{-im} - e^{im}\right) \]

      2. lower-*.f64 N/A

         \[\leadsto \left(\left(\frac{1}{2} + \frac{-1}{12} \cdot {re}^{2}\right) \cdot \color{blue}{re}\right) \cdot \left(e^{-im} - e^{im}\right) \]

      3. +-commutative N/A

         \[\leadsto \left(\left(\frac{-1}{12} \cdot {re}^{2} + \frac{1}{2}\right) \cdot re\right) \cdot \left(e^{-im} - e^{im}\right) \]

      4. *-commutative N/A

         \[\leadsto \left(\left({re}^{2} \cdot \frac{-1}{12} + \frac{1}{2}\right) \cdot re\right) \cdot \left(e^{-im} - e^{im}\right) \]

      5. lower-fma.f64 N/A

         \[\leadsto \left(\mathsf{fma}\left({re}^{2}, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \cdot \left(e^{-im} - e^{im}\right) \]

      6. unpow2 N/A

         \[\leadsto \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \cdot \left(e^{-im} - e^{im}\right) \]

      7. lower-*.f64 25.5

         \[\leadsto \left(\mathsf{fma}\left(re \cdot re, -0.08333333333333333, 0.5\right) \cdot re\right) \cdot \left(e^{-im} - e^{im}\right) \]

   4. Applied rewrites 25.5%

      \[\leadsto \color{blue}{\left(\mathsf{fma}\left(re \cdot re, -0.08333333333333333, 0.5\right) \cdot re\right)} \cdot \left(e^{-im} - e^{im}\right) \]
   5. Step-by-step derivation

      1. lift-*.f64 N/A

         \[\leadsto \color{blue}{\left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \cdot \left(e^{-im} - e^{im}\right)} \]

      2. lift--.f64 N/A

         \[\leadsto \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \cdot \color{blue}{\left(e^{-im} - e^{im}\right)} \]

      3. lift-exp.f64 N/A

         \[\leadsto \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \cdot \left(\color{blue}{e^{-im}} - e^{im}\right) \]

      4. lift-neg.f64 N/A

         \[\leadsto \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \cdot \left(e^{\color{blue}{\mathsf{neg}\left(im\right)}} - e^{im}\right) \]

      5. lift-exp.f64 N/A

         \[\leadsto \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \cdot \left(e^{\mathsf{neg}\left(im\right)} - \color{blue}{e^{im}}\right) \]

      6. *-commutative N/A

         \[\leadsto \color{blue}{\left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right) \cdot \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right)} \]

      7. lower-*.f64 N/A

         \[\leadsto \color{blue}{\left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right) \cdot \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right)} \]

      8. sub-negate N/A

         \[\leadsto \color{blue}{\left(\mathsf{neg}\left(\left(e^{im} - e^{\mathsf{neg}\left(im\right)}\right)\right)\right)} \cdot \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \]

      9. sinh-undef-rev N/A

         \[\leadsto \left(\mathsf{neg}\left(\color{blue}{2 \cdot \sinh im}\right)\right) \cdot \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \]

      10. distribute-lft-neg-in N/A

          \[\leadsto \color{blue}{\left(\left(\mathsf{neg}\left(2\right)\right) \cdot \sinh im\right)} \cdot \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \]

      11. metadata-eval N/A

          \[\leadsto \left(\color{blue}{-2} \cdot \sinh im\right) \cdot \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \]

      12. lower-*.f64 N/A

          \[\leadsto \color{blue}{\left(-2 \cdot \sinh im\right)} \cdot \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \]

      13. lift-sinh.f64 25.9

          \[\leadsto \left(-2 \cdot \color{blue}{\sinh im}\right) \cdot \left(\mathsf{fma}\left(re \cdot re, -0.08333333333333333, 0.5\right) \cdot re\right) \]

   6. Applied rewrites 25.9%

      \[\leadsto \color{blue}{\left(-2 \cdot \sinh im\right) \cdot \left(\mathsf{fma}\left(re \cdot re, -0.08333333333333333, 0.5\right) \cdot re\right)} \]

   7. Taylor expanded in im around 0

      \[\leadsto \color{blue}{\left(im \cdot \left(\frac{-1}{3} \cdot {im}^{2} - 2\right)\right)} \cdot \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \]
   8. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(\left(\frac{-1}{3} \cdot {im}^{2} - 2\right) \cdot \color{blue}{im}\right) \cdot \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \]

      2. lower-*.f64 N/A

         \[\leadsto \left(\left(\frac{-1}{3} \cdot {im}^{2} - 2\right) \cdot \color{blue}{im}\right) \cdot \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \]

      3. sub-flip N/A

         \[\leadsto \left(\left(\frac{-1}{3} \cdot {im}^{2} + \left(\mathsf{neg}\left(2\right)\right)\right) \cdot im\right) \cdot \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \]

      4. metadata-eval N/A

         \[\leadsto \left(\left(\frac{-1}{3} \cdot {im}^{2} + -2\right) \cdot im\right) \cdot \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \]

      5. lower-fma.f64 N/A

         \[\leadsto \left(\mathsf{fma}\left(\frac{-1}{3}, {im}^{2}, -2\right) \cdot im\right) \cdot \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \]

      6. pow2 N/A

         \[\leadsto \left(\mathsf{fma}\left(\frac{-1}{3}, im \cdot im, -2\right) \cdot im\right) \cdot \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \]

      7. lift-*.f64 24.3

         \[\leadsto \left(\mathsf{fma}\left(-0.3333333333333333, im \cdot im, -2\right) \cdot im\right) \cdot \left(\mathsf{fma}\left(re \cdot re, -0.08333333333333333, 0.5\right) \cdot re\right) \]

   9. Applied rewrites 24.3%

      \[\leadsto \color{blue}{\left(\mathsf{fma}\left(-0.3333333333333333, im \cdot im, -2\right) \cdot im\right)} \cdot \left(\mathsf{fma}\left(re \cdot re, -0.08333333333333333, 0.5\right) \cdot re\right) \]

   10. Taylor expanded in im around inf

       \[\leadsto \left(\frac{-1}{3} \cdot \color{blue}{{im}^{3}}\right) \cdot \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \]
   11. Step-by-step derivation

       1. *-commutative N/A

          \[\leadsto \left({im}^{3} \cdot \frac{-1}{3}\right) \cdot \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \]

       2. lower-*.f64 N/A

          \[\leadsto \left({im}^{3} \cdot \frac{-1}{3}\right) \cdot \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \]

       3. unpow3 N/A

          \[\leadsto \left(\left(\left(im \cdot im\right) \cdot im\right) \cdot \frac{-1}{3}\right) \cdot \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \]

       4. pow2 N/A

          \[\leadsto \left(\left({im}^{2} \cdot im\right) \cdot \frac{-1}{3}\right) \cdot \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \]

       5. lower-*.f64 N/A

          \[\leadsto \left(\left({im}^{2} \cdot im\right) \cdot \frac{-1}{3}\right) \cdot \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \]

       6. pow2 N/A

          \[\leadsto \left(\left(\left(im \cdot im\right) \cdot im\right) \cdot \frac{-1}{3}\right) \cdot \left(\mathsf{fma}\left(re \cdot re, \frac{-1}{12}, \frac{1}{2}\right) \cdot re\right) \]

       7. lift-*.f64 24.2

          \[\leadsto \left(\left(\left(im \cdot im\right) \cdot im\right) \cdot -0.3333333333333333\right) \cdot \left(\mathsf{fma}\left(re \cdot re, -0.08333333333333333, 0.5\right) \cdot re\right) \]

   12. Applied rewrites 24.2%

       \[\leadsto \left(\left(\left(im \cdot im\right) \cdot im\right) \cdot \color{blue}{-0.3333333333333333}\right) \cdot \left(\mathsf{fma}\left(re \cdot re, -0.08333333333333333, 0.5\right) \cdot re\right) \]

   if -0.0200000000000000004 < (*.f64 #s(literal 1/2 binary64) (sin.f64 re))

   1. Initial program 68.2%

      \[\left(0.5 \cdot \sin re\right) \cdot \left(e^{-im} - e^{im}\right) \]

   2. Taylor expanded in re around 0

      \[\leadsto \color{blue}{\frac{1}{2} \cdot \left(re \cdot \left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right)\right)} \]
   3. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(re \cdot \left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right)\right) \cdot \color{blue}{\frac{1}{2}} \]

      2. lower-*.f64 N/A

         \[\leadsto \left(re \cdot \left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right)\right) \cdot \color{blue}{\frac{1}{2}} \]

      3. *-commutative N/A

         \[\leadsto \left(\left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right) \cdot re\right) \cdot \frac{1}{2} \]

      4. lower-*.f64 N/A

         \[\leadsto \left(\left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right) \cdot re\right) \cdot \frac{1}{2} \]

      5. sub-negate N/A

         \[\leadsto \left(\left(\mathsf{neg}\left(\left(e^{im} - e^{\mathsf{neg}\left(im\right)}\right)\right)\right) \cdot re\right) \cdot \frac{1}{2} \]

      6. lower-neg.f64 N/A

         \[\leadsto \left(\left(-\left(e^{im} - e^{\mathsf{neg}\left(im\right)}\right)\right) \cdot re\right) \cdot \frac{1}{2} \]

      7. sinh-undef N/A

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot \frac{1}{2} \]

      8. lower-*.f64 N/A

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot \frac{1}{2} \]

      9. lower-sinh.f64 74.2

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot 0.5 \]

   4. Applied rewrites 74.2%

      \[\leadsto \color{blue}{\left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot 0.5} \]
   5. Step-by-step derivation

      1. flip-- 74.2

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot \color{blue}{re}\right) \cdot 0.5 \]

      2. pow2 74.2

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot 0.5 \]

      3. lift-neg.f64 N/A

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot \frac{1}{2} \]

      4. exp-prod N/A

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot \frac{1}{2} \]

      5. lift-neg.f64 74.2

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot 0.5 \]

      6. exp-sum 74.2

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot 0.5 \]

   6. Applied rewrites 74.2%

      \[\leadsto \color{blue}{\left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot 0.5} \]
3. Recombined 2 regimes into one program.
4. Add Preprocessing

Alternative 8: 61.5% accurate, 1.0× speedup

Math

\[\begin{array}{l} im\_m = \left|im\right| \\ im\_s = \mathsf{copysign}\left(1, im\right) \\ im\_s \cdot \begin{array}{l} \mathbf{if}\;0.5 \cdot \sin re \leq -0.004:\\ \;\;\;\;\left(0.5 \cdot re\right) \cdot \left(\mathsf{fma}\left(\mathsf{fma}\left(0.5, im\_m, -1\right), im\_m, 1\right) - 1\right)\\ \mathbf{else}:\\ \;\;\;\;\left(\left(-2 \cdot \sinh im\_m\right) \cdot re\right) \cdot 0.5\\ \end{array} \end{array} \]

FPCore

im\_m = (fabs.f64 im)
im\_s = (copysign.f64 #s(literal 1 binary64) im)
(FPCore (im_s re im_m)
 :precision binary64
 (*
  im_s
  (if (<= (* 0.5 (sin re)) -0.004)
    (* (* 0.5 re) (- (fma (fma 0.5 im_m -1.0) im_m 1.0) 1.0))
    (* (* (* -2.0 (sinh im_m)) re) 0.5))))

C

im\_m = fabs(im);
im\_s = copysign(1.0, im);
double code(double im_s, double re, double im_m) {
	double tmp;
	if ((0.5 * sin(re)) <= -0.004) {
		tmp = (0.5 * re) * (fma(fma(0.5, im_m, -1.0), im_m, 1.0) - 1.0);
	} else {
		tmp = ((-2.0 * sinh(im_m)) * re) * 0.5;
	}
	return im_s * tmp;
}

Julia

im\_m = abs(im)
im\_s = copysign(1.0, im)
function code(im_s, re, im_m)
	tmp = 0.0
	if (Float64(0.5 * sin(re)) <= -0.004)
		tmp = Float64(Float64(0.5 * re) * Float64(fma(fma(0.5, im_m, -1.0), im_m, 1.0) - 1.0));
	else
		tmp = Float64(Float64(Float64(-2.0 * sinh(im_m)) * re) * 0.5);
	end
	return Float64(im_s * tmp)
end

Wolfram

im\_m = N[Abs[im], $MachinePrecision]
im\_s = N[With[{TMP1 = Abs[1.0], TMP2 = Sign[im]}, TMP1 * If[TMP2 == 0, 1, TMP2]], $MachinePrecision]
code[im$95$s_, re_, im$95$m_] := N[(im$95$s * If[LessEqual[N[(0.5 * N[Sin[re], $MachinePrecision]), $MachinePrecision], -0.004], N[(N[(0.5 * re), $MachinePrecision] * N[(N[(N[(0.5 * im$95$m + -1.0), $MachinePrecision] * im$95$m + 1.0), $MachinePrecision] - 1.0), $MachinePrecision]), $MachinePrecision], N[(N[(N[(-2.0 * N[Sinh[im$95$m], $MachinePrecision]), $MachinePrecision] * re), $MachinePrecision] * 0.5), $MachinePrecision]]), $MachinePrecision]

Derivation

1. Split input into 2 regimes

2. if (*.f64 #s(literal 1/2 binary64) (sin.f64 re)) < -0.0040000000000000001

   1. Initial program 53.7%

      \[\left(0.5 \cdot \sin re\right) \cdot \left(e^{-im} - e^{im}\right) \]

   2. Taylor expanded in re around 0

      \[\leadsto \left(\frac{1}{2} \cdot \color{blue}{re}\right) \cdot \left(e^{-im} - e^{im}\right) \]
   3. Step-by-step derivation

   4. Applied rewrites 27.3%

      \[\leadsto \left(0.5 \cdot \color{blue}{re}\right) \cdot \left(e^{-im} - e^{im}\right) \]

   5. Taylor expanded in im around 0

      \[\leadsto \left(\frac{1}{2} \cdot re\right) \cdot \left(\color{blue}{1} - e^{im}\right) \]
   6. Step-by-step derivation

   7. Applied rewrites 27.3%

      \[\leadsto \left(0.5 \cdot re\right) \cdot \left(\color{blue}{1} - e^{im}\right) \]

   8. Taylor expanded in im around 0

      \[\leadsto \left(\frac{1}{2} \cdot re\right) \cdot \left(1 - \color{blue}{1}\right) \]
   9. Step-by-step derivation

   10. Applied rewrites 3.5%

       \[\leadsto \left(0.5 \cdot re\right) \cdot \left(1 - \color{blue}{1}\right) \]

   11. Taylor expanded in im around 0

       \[\leadsto \left(\frac{1}{2} \cdot re\right) \cdot \left(\color{blue}{\left(1 + im \cdot \left(\frac{1}{2} \cdot im - 1\right)\right)} - 1\right) \]
   12. Step-by-step derivation

       1. +-commutative N/A

          \[\leadsto \left(\frac{1}{2} \cdot re\right) \cdot \left(\left(im \cdot \left(\frac{1}{2} \cdot im - 1\right) + \color{blue}{1}\right) - 1\right) \]

       2. *-commutative N/A

          \[\leadsto \left(\frac{1}{2} \cdot re\right) \cdot \left(\left(\left(\frac{1}{2} \cdot im - 1\right) \cdot im + 1\right) - 1\right) \]

       3. lower-fma.f64 N/A

          \[\leadsto \left(\frac{1}{2} \cdot re\right) \cdot \left(\mathsf{fma}\left(\frac{1}{2} \cdot im - 1, \color{blue}{im}, 1\right) - 1\right) \]

       4. sub-flip N/A

          \[\leadsto \left(\frac{1}{2} \cdot re\right) \cdot \left(\mathsf{fma}\left(\frac{1}{2} \cdot im + \left(\mathsf{neg}\left(1\right)\right), im, 1\right) - 1\right) \]

       5. metadata-eval N/A

          \[\leadsto \left(\frac{1}{2} \cdot re\right) \cdot \left(\mathsf{fma}\left(\frac{1}{2} \cdot im + -1, im, 1\right) - 1\right) \]

       6. lower-fma.f64 21.1

          \[\leadsto \left(0.5 \cdot re\right) \cdot \left(\mathsf{fma}\left(\mathsf{fma}\left(0.5, im, -1\right), im, 1\right) - 1\right) \]

   13. Applied rewrites 21.1%

       \[\leadsto \left(0.5 \cdot re\right) \cdot \left(\color{blue}{\mathsf{fma}\left(\mathsf{fma}\left(0.5, im, -1\right), im, 1\right)} - 1\right) \]

   if -0.0040000000000000001 < (*.f64 #s(literal 1/2 binary64) (sin.f64 re))

   1. Initial program 68.3%

      \[\left(0.5 \cdot \sin re\right) \cdot \left(e^{-im} - e^{im}\right) \]

   2. Taylor expanded in re around 0

      \[\leadsto \color{blue}{\frac{1}{2} \cdot \left(re \cdot \left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right)\right)} \]
   3. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(re \cdot \left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right)\right) \cdot \color{blue}{\frac{1}{2}} \]

      2. lower-*.f64 N/A

         \[\leadsto \left(re \cdot \left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right)\right) \cdot \color{blue}{\frac{1}{2}} \]

      3. *-commutative N/A

         \[\leadsto \left(\left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right) \cdot re\right) \cdot \frac{1}{2} \]

      4. lower-*.f64 N/A

         \[\leadsto \left(\left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right) \cdot re\right) \cdot \frac{1}{2} \]

      5. sub-negate N/A

         \[\leadsto \left(\left(\mathsf{neg}\left(\left(e^{im} - e^{\mathsf{neg}\left(im\right)}\right)\right)\right) \cdot re\right) \cdot \frac{1}{2} \]

      6. lower-neg.f64 N/A

         \[\leadsto \left(\left(-\left(e^{im} - e^{\mathsf{neg}\left(im\right)}\right)\right) \cdot re\right) \cdot \frac{1}{2} \]

      7. sinh-undef N/A

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot \frac{1}{2} \]

      8. lower-*.f64 N/A

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot \frac{1}{2} \]

      9. lower-sinh.f64 74.5

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot 0.5 \]

   4. Applied rewrites 74.5%

      \[\leadsto \color{blue}{\left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot 0.5} \]
   5. Step-by-step derivation

      1. flip-- 74.5

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot \color{blue}{re}\right) \cdot 0.5 \]

      2. pow2 74.5

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot 0.5 \]

      3. lift-neg.f64 N/A

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot \frac{1}{2} \]

      4. exp-prod N/A

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot \frac{1}{2} \]

      5. lift-neg.f64 74.5

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot 0.5 \]

      6. exp-sum 74.5

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot 0.5 \]

   6. Applied rewrites 74.5%

      \[\leadsto \color{blue}{\left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot 0.5} \]
3. Recombined 2 regimes into one program.
4. Add Preprocessing

Alternative 9: 53.3% accurate, 0.7× speedup

Math

\[\begin{array}{l} im\_m = \left|im\right| \\ im\_s = \mathsf{copysign}\left(1, im\right) \\ im\_s \cdot \begin{array}{l} \mathbf{if}\;\left(0.5 \cdot \sin re\right) \cdot \left(e^{-im\_m} - e^{im\_m}\right) \leq -1 \cdot 10^{-49}:\\ \;\;\;\;0.5 \cdot \left(re \cdot \left(1 - e^{im\_m}\right)\right)\\ \mathbf{else}:\\ \;\;\;\;\left(\mathsf{fma}\left(re \cdot re, 0.16666666666666666, -1\right) \cdot im\_m\right) \cdot re\\ \end{array} \end{array} \]

FPCore

im\_m = (fabs.f64 im)
im\_s = (copysign.f64 #s(literal 1 binary64) im)
(FPCore (im_s re im_m)
 :precision binary64
 (*
  im_s
  (if (<= (* (* 0.5 (sin re)) (- (exp (- im_m)) (exp im_m))) -1e-49)
    (* 0.5 (* re (- 1.0 (exp im_m))))
    (* (* (fma (* re re) 0.16666666666666666 -1.0) im_m) re))))

C

im\_m = fabs(im);
im\_s = copysign(1.0, im);
double code(double im_s, double re, double im_m) {
	double tmp;
	if (((0.5 * sin(re)) * (exp(-im_m) - exp(im_m))) <= -1e-49) {
		tmp = 0.5 * (re * (1.0 - exp(im_m)));
	} else {
		tmp = (fma((re * re), 0.16666666666666666, -1.0) * im_m) * re;
	}
	return im_s * tmp;
}

Julia

im\_m = abs(im)
im\_s = copysign(1.0, im)
function code(im_s, re, im_m)
	tmp = 0.0
	if (Float64(Float64(0.5 * sin(re)) * Float64(exp(Float64(-im_m)) - exp(im_m))) <= -1e-49)
		tmp = Float64(0.5 * Float64(re * Float64(1.0 - exp(im_m))));
	else
		tmp = Float64(Float64(fma(Float64(re * re), 0.16666666666666666, -1.0) * im_m) * re);
	end
	return Float64(im_s * tmp)
end

Wolfram

im\_m = N[Abs[im], $MachinePrecision]
im\_s = N[With[{TMP1 = Abs[1.0], TMP2 = Sign[im]}, TMP1 * If[TMP2 == 0, 1, TMP2]], $MachinePrecision]
code[im$95$s_, re_, im$95$m_] := N[(im$95$s * If[LessEqual[N[(N[(0.5 * N[Sin[re], $MachinePrecision]), $MachinePrecision] * N[(N[Exp[(-im$95$m)], $MachinePrecision] - N[Exp[im$95$m], $MachinePrecision]), $MachinePrecision]), $MachinePrecision], -1e-49], N[(0.5 * N[(re * N[(1.0 - N[Exp[im$95$m], $MachinePrecision]), $MachinePrecision]), $MachinePrecision]), $MachinePrecision], N[(N[(N[(N[(re * re), $MachinePrecision] * 0.16666666666666666 + -1.0), $MachinePrecision] * im$95$m), $MachinePrecision] * re), $MachinePrecision]]), $MachinePrecision]

Derivation

1. Split input into 2 regimes

2. if (*.f64 (*.f64 #s(literal 1/2 binary64) (sin.f64 re)) (-.f64 (exp.f64 (neg.f64 im)) (exp.f64 im))) < -9.99999999999999936e-50

   1. Initial program 98.8%

      \[\left(0.5 \cdot \sin re\right) \cdot \left(e^{-im} - e^{im}\right) \]

   2. Taylor expanded in re around 0

      \[\leadsto \left(\frac{1}{2} \cdot \color{blue}{re}\right) \cdot \left(e^{-im} - e^{im}\right) \]
   3. Step-by-step derivation

   4. Applied rewrites 72.4%

      \[\leadsto \left(0.5 \cdot \color{blue}{re}\right) \cdot \left(e^{-im} - e^{im}\right) \]

   5. Taylor expanded in im around 0

      \[\leadsto \left(\frac{1}{2} \cdot re\right) \cdot \left(\color{blue}{1} - e^{im}\right) \]
   6. Step-by-step derivation

   7. Applied rewrites 72.2%

      \[\leadsto \left(0.5 \cdot re\right) \cdot \left(\color{blue}{1} - e^{im}\right) \]
   8. Step-by-step derivation

      1. lift-*.f64 N/A

         \[\leadsto \color{blue}{\left(\frac{1}{2} \cdot re\right) \cdot \left(1 - e^{im}\right)} \]

      2. lift-*.f64 N/A

         \[\leadsto \color{blue}{\left(\frac{1}{2} \cdot re\right)} \cdot \left(1 - e^{im}\right) \]

      3. associate-*l* N/A

         \[\leadsto \color{blue}{\frac{1}{2} \cdot \left(re \cdot \left(1 - e^{im}\right)\right)} \]

      4. lower-*.f64 N/A

         \[\leadsto \color{blue}{\frac{1}{2} \cdot \left(re \cdot \left(1 - e^{im}\right)\right)} \]

      5. lower-*.f64 72.2

         \[\leadsto 0.5 \cdot \color{blue}{\left(re \cdot \left(1 - e^{im}\right)\right)} \]

   9. Applied rewrites 72.2%

      \[\leadsto \color{blue}{0.5 \cdot \left(re \cdot \left(1 - e^{im}\right)\right)} \]

   if -9.99999999999999936e-50 < (*.f64 (*.f64 #s(literal 1/2 binary64) (sin.f64 re)) (-.f64 (exp.f64 (neg.f64 im)) (exp.f64 im)))

   1. Initial program 53.4%

      \[\left(0.5 \cdot \sin re\right) \cdot \left(e^{-im} - e^{im}\right) \]

   2. Taylor expanded in im around 0

      \[\leadsto \color{blue}{-1 \cdot \left(im \cdot \sin re\right)} \]
   3. Step-by-step derivation

      1. associate-*r* N/A

         \[\leadsto \left(-1 \cdot im\right) \cdot \color{blue}{\sin re} \]

      2. lower-*.f64 N/A

         \[\leadsto \left(-1 \cdot im\right) \cdot \color{blue}{\sin re} \]

      3. mul-1-neg N/A

         \[\leadsto \left(\mathsf{neg}\left(im\right)\right) \cdot \sin \color{blue}{re} \]

      4. lift-neg.f64 N/A

         \[\leadsto \left(-im\right) \cdot \sin \color{blue}{re} \]

      5. lift-sin.f64 68.4

         \[\leadsto \left(-im\right) \cdot \sin re \]

   4. Applied rewrites 68.4%

      \[\leadsto \color{blue}{\left(-im\right) \cdot \sin re} \]

   5. Taylor expanded in re around 0

      \[\leadsto re \cdot \color{blue}{\left(-1 \cdot im + \frac{1}{6} \cdot \left(im \cdot {re}^{2}\right)\right)} \]
   6. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(-1 \cdot im + \frac{1}{6} \cdot \left(im \cdot {re}^{2}\right)\right) \cdot re \]

      2. lower-*.f64 N/A

         \[\leadsto \left(-1 \cdot im + \frac{1}{6} \cdot \left(im \cdot {re}^{2}\right)\right) \cdot re \]

      3. +-commutative N/A

         \[\leadsto \left(\frac{1}{6} \cdot \left(im \cdot {re}^{2}\right) + -1 \cdot im\right) \cdot re \]

      4. *-commutative N/A

         \[\leadsto \left(\left(im \cdot {re}^{2}\right) \cdot \frac{1}{6} + -1 \cdot im\right) \cdot re \]

      5. lower-fma.f64 N/A

         \[\leadsto \mathsf{fma}\left(im \cdot {re}^{2}, \frac{1}{6}, -1 \cdot im\right) \cdot re \]

      6. *-commutative N/A

         \[\leadsto \mathsf{fma}\left({re}^{2} \cdot im, \frac{1}{6}, -1 \cdot im\right) \cdot re \]

      7. lower-*.f64 N/A

         \[\leadsto \mathsf{fma}\left({re}^{2} \cdot im, \frac{1}{6}, -1 \cdot im\right) \cdot re \]

      8. pow2 N/A

         \[\leadsto \mathsf{fma}\left(\left(re \cdot re\right) \cdot im, \frac{1}{6}, -1 \cdot im\right) \cdot re \]

      9. lift-*.f64 N/A

         \[\leadsto \mathsf{fma}\left(\left(re \cdot re\right) \cdot im, \frac{1}{6}, -1 \cdot im\right) \cdot re \]

      10. mul-1-neg N/A

          \[\leadsto \mathsf{fma}\left(\left(re \cdot re\right) \cdot im, \frac{1}{6}, \mathsf{neg}\left(im\right)\right) \cdot re \]

      11. lift-neg.f64 42.0

          \[\leadsto \mathsf{fma}\left(\left(re \cdot re\right) \cdot im, 0.16666666666666666, -im\right) \cdot re \]

   7. Applied rewrites 42.0%

      \[\leadsto \mathsf{fma}\left(\left(re \cdot re\right) \cdot im, 0.16666666666666666, -im\right) \cdot \color{blue}{re} \]

   8. Taylor expanded in im around 0

      \[\leadsto \left(im \cdot \left(\frac{1}{6} \cdot {re}^{2} - 1\right)\right) \cdot re \]
   9. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(\left(\frac{1}{6} \cdot {re}^{2} - 1\right) \cdot im\right) \cdot re \]

      2. lower-*.f64 N/A

         \[\leadsto \left(\left(\frac{1}{6} \cdot {re}^{2} - 1\right) \cdot im\right) \cdot re \]

      3. sub-flip N/A

         \[\leadsto \left(\left(\frac{1}{6} \cdot {re}^{2} + \left(\mathsf{neg}\left(1\right)\right)\right) \cdot im\right) \cdot re \]

      4. *-commutative N/A

         \[\leadsto \left(\left({re}^{2} \cdot \frac{1}{6} + \left(\mathsf{neg}\left(1\right)\right)\right) \cdot im\right) \cdot re \]

      5. metadata-eval N/A

         \[\leadsto \left(\left({re}^{2} \cdot \frac{1}{6} + -1\right) \cdot im\right) \cdot re \]

      6. lower-fma.f64 N/A

         \[\leadsto \left(\mathsf{fma}\left({re}^{2}, \frac{1}{6}, -1\right) \cdot im\right) \cdot re \]

      7. pow2 N/A

         \[\leadsto \left(\mathsf{fma}\left(re \cdot re, \frac{1}{6}, -1\right) \cdot im\right) \cdot re \]

      8. lift-*.f64 42.0

         \[\leadsto \left(\mathsf{fma}\left(re \cdot re, 0.16666666666666666, -1\right) \cdot im\right) \cdot re \]

   10. Applied rewrites 42.0%

       \[\leadsto \left(\mathsf{fma}\left(re \cdot re, 0.16666666666666666, -1\right) \cdot im\right) \cdot re \]
3. Recombined 2 regimes into one program.
4. Add Preprocessing

Alternative 10: 53.3% accurate, 1.1× speedup

Math

\[\begin{array}{l} im\_m = \left|im\right| \\ im\_s = \mathsf{copysign}\left(1, im\right) \\ im\_s \cdot \begin{array}{l} \mathbf{if}\;0.5 \cdot \sin re \leq -0.004:\\ \;\;\;\;\left(0.5 \cdot re\right) \cdot \left(\mathsf{fma}\left(\mathsf{fma}\left(0.5, im\_m, -1\right), im\_m, 1\right) - 1\right)\\ \mathbf{else}:\\ \;\;\;\;\left(0.5 \cdot re\right) \cdot \left(\mathsf{fma}\left(im\_m \cdot im\_m, -0.3333333333333333, -2\right) \cdot im\_m\right)\\ \end{array} \end{array} \]

FPCore

im\_m = (fabs.f64 im)
im\_s = (copysign.f64 #s(literal 1 binary64) im)
(FPCore (im_s re im_m)
 :precision binary64
 (*
  im_s
  (if (<= (* 0.5 (sin re)) -0.004)
    (* (* 0.5 re) (- (fma (fma 0.5 im_m -1.0) im_m 1.0) 1.0))
    (* (* 0.5 re) (* (fma (* im_m im_m) -0.3333333333333333 -2.0) im_m)))))

C

im\_m = fabs(im);
im\_s = copysign(1.0, im);
double code(double im_s, double re, double im_m) {
	double tmp;
	if ((0.5 * sin(re)) <= -0.004) {
		tmp = (0.5 * re) * (fma(fma(0.5, im_m, -1.0), im_m, 1.0) - 1.0);
	} else {
		tmp = (0.5 * re) * (fma((im_m * im_m), -0.3333333333333333, -2.0) * im_m);
	}
	return im_s * tmp;
}

Julia

im\_m = abs(im)
im\_s = copysign(1.0, im)
function code(im_s, re, im_m)
	tmp = 0.0
	if (Float64(0.5 * sin(re)) <= -0.004)
		tmp = Float64(Float64(0.5 * re) * Float64(fma(fma(0.5, im_m, -1.0), im_m, 1.0) - 1.0));
	else
		tmp = Float64(Float64(0.5 * re) * Float64(fma(Float64(im_m * im_m), -0.3333333333333333, -2.0) * im_m));
	end
	return Float64(im_s * tmp)
end

Wolfram

im\_m = N[Abs[im], $MachinePrecision]
im\_s = N[With[{TMP1 = Abs[1.0], TMP2 = Sign[im]}, TMP1 * If[TMP2 == 0, 1, TMP2]], $MachinePrecision]
code[im$95$s_, re_, im$95$m_] := N[(im$95$s * If[LessEqual[N[(0.5 * N[Sin[re], $MachinePrecision]), $MachinePrecision], -0.004], N[(N[(0.5 * re), $MachinePrecision] * N[(N[(N[(0.5 * im$95$m + -1.0), $MachinePrecision] * im$95$m + 1.0), $MachinePrecision] - 1.0), $MachinePrecision]), $MachinePrecision], N[(N[(0.5 * re), $MachinePrecision] * N[(N[(N[(im$95$m * im$95$m), $MachinePrecision] * -0.3333333333333333 + -2.0), $MachinePrecision] * im$95$m), $MachinePrecision]), $MachinePrecision]]), $MachinePrecision]

Derivation

1. Split input into 2 regimes

2. if (*.f64 #s(literal 1/2 binary64) (sin.f64 re)) < -0.0040000000000000001

   1. Initial program 53.7%

      \[\left(0.5 \cdot \sin re\right) \cdot \left(e^{-im} - e^{im}\right) \]

   2. Taylor expanded in re around 0

      \[\leadsto \left(\frac{1}{2} \cdot \color{blue}{re}\right) \cdot \left(e^{-im} - e^{im}\right) \]
   3. Step-by-step derivation

   4. Applied rewrites 27.3%

      \[\leadsto \left(0.5 \cdot \color{blue}{re}\right) \cdot \left(e^{-im} - e^{im}\right) \]

   5. Taylor expanded in im around 0

      \[\leadsto \left(\frac{1}{2} \cdot re\right) \cdot \left(\color{blue}{1} - e^{im}\right) \]
   6. Step-by-step derivation

   7. Applied rewrites 27.3%

      \[\leadsto \left(0.5 \cdot re\right) \cdot \left(\color{blue}{1} - e^{im}\right) \]

   8. Taylor expanded in im around 0

      \[\leadsto \left(\frac{1}{2} \cdot re\right) \cdot \left(1 - \color{blue}{1}\right) \]
   9. Step-by-step derivation

   10. Applied rewrites 3.5%

       \[\leadsto \left(0.5 \cdot re\right) \cdot \left(1 - \color{blue}{1}\right) \]

   11. Taylor expanded in im around 0

       \[\leadsto \left(\frac{1}{2} \cdot re\right) \cdot \left(\color{blue}{\left(1 + im \cdot \left(\frac{1}{2} \cdot im - 1\right)\right)} - 1\right) \]
   12. Step-by-step derivation

       1. +-commutative N/A

          \[\leadsto \left(\frac{1}{2} \cdot re\right) \cdot \left(\left(im \cdot \left(\frac{1}{2} \cdot im - 1\right) + \color{blue}{1}\right) - 1\right) \]

       2. *-commutative N/A

          \[\leadsto \left(\frac{1}{2} \cdot re\right) \cdot \left(\left(\left(\frac{1}{2} \cdot im - 1\right) \cdot im + 1\right) - 1\right) \]

       3. lower-fma.f64 N/A

          \[\leadsto \left(\frac{1}{2} \cdot re\right) \cdot \left(\mathsf{fma}\left(\frac{1}{2} \cdot im - 1, \color{blue}{im}, 1\right) - 1\right) \]

       4. sub-flip N/A

          \[\leadsto \left(\frac{1}{2} \cdot re\right) \cdot \left(\mathsf{fma}\left(\frac{1}{2} \cdot im + \left(\mathsf{neg}\left(1\right)\right), im, 1\right) - 1\right) \]

       5. metadata-eval N/A

          \[\leadsto \left(\frac{1}{2} \cdot re\right) \cdot \left(\mathsf{fma}\left(\frac{1}{2} \cdot im + -1, im, 1\right) - 1\right) \]

       6. lower-fma.f64 21.1

          \[\leadsto \left(0.5 \cdot re\right) \cdot \left(\mathsf{fma}\left(\mathsf{fma}\left(0.5, im, -1\right), im, 1\right) - 1\right) \]

   13. Applied rewrites 21.1%

       \[\leadsto \left(0.5 \cdot re\right) \cdot \left(\color{blue}{\mathsf{fma}\left(\mathsf{fma}\left(0.5, im, -1\right), im, 1\right)} - 1\right) \]

   if -0.0040000000000000001 < (*.f64 #s(literal 1/2 binary64) (sin.f64 re))

   1. Initial program 68.3%

      \[\left(0.5 \cdot \sin re\right) \cdot \left(e^{-im} - e^{im}\right) \]

   2. Taylor expanded in re around 0

      \[\leadsto \left(\frac{1}{2} \cdot \color{blue}{re}\right) \cdot \left(e^{-im} - e^{im}\right) \]
   3. Step-by-step derivation

   4. Applied rewrites 59.5%

      \[\leadsto \left(0.5 \cdot \color{blue}{re}\right) \cdot \left(e^{-im} - e^{im}\right) \]

   5. Taylor expanded in im around 0

      \[\leadsto \left(\frac{1}{2} \cdot re\right) \cdot \left(\color{blue}{1} - e^{im}\right) \]
   6. Step-by-step derivation

   7. Applied rewrites 58.6%

      \[\leadsto \left(0.5 \cdot re\right) \cdot \left(\color{blue}{1} - e^{im}\right) \]

   8. Taylor expanded in im around 0

      \[\leadsto \left(\frac{1}{2} \cdot re\right) \cdot \left(1 - \color{blue}{1}\right) \]
   9. Step-by-step derivation

   10. Applied rewrites 19.1%

       \[\leadsto \left(0.5 \cdot re\right) \cdot \left(1 - \color{blue}{1}\right) \]

   11. Taylor expanded in im around 0

       \[\leadsto \left(\frac{1}{2} \cdot re\right) \cdot \color{blue}{\left(im \cdot \left(\frac{-1}{3} \cdot {im}^{2} - 2\right)\right)} \]
   12. Step-by-step derivation

       1. *-commutative N/A

          \[\leadsto \left(\frac{1}{2} \cdot re\right) \cdot \left(\left(\frac{-1}{3} \cdot {im}^{2} - 2\right) \cdot \color{blue}{im}\right) \]

       2. sub-flip N/A

          \[\leadsto \left(\frac{1}{2} \cdot re\right) \cdot \left(\left(\frac{-1}{3} \cdot {im}^{2} + \left(\mathsf{neg}\left(2\right)\right)\right) \cdot im\right) \]

       3. pow2 N/A

          \[\leadsto \left(\frac{1}{2} \cdot re\right) \cdot \left(\left(\frac{-1}{3} \cdot \left(im \cdot im\right) + \left(\mathsf{neg}\left(2\right)\right)\right) \cdot im\right) \]

       4. metadata-eval N/A

          \[\leadsto \left(\frac{1}{2} \cdot re\right) \cdot \left(\left(\frac{-1}{3} \cdot \left(im \cdot im\right) + -2\right) \cdot im\right) \]

       5. lower-*.f64 N/A

          \[\leadsto \left(\frac{1}{2} \cdot re\right) \cdot \left(\left(\frac{-1}{3} \cdot \left(im \cdot im\right) + -2\right) \cdot \color{blue}{im}\right) \]

       6. pow2 N/A

          \[\leadsto \left(\frac{1}{2} \cdot re\right) \cdot \left(\left(\frac{-1}{3} \cdot {im}^{2} + -2\right) \cdot im\right) \]

       7. *-commutative N/A

          \[\leadsto \left(\frac{1}{2} \cdot re\right) \cdot \left(\left({im}^{2} \cdot \frac{-1}{3} + -2\right) \cdot im\right) \]

       8. lower-fma.f64 N/A

          \[\leadsto \left(\frac{1}{2} \cdot re\right) \cdot \left(\mathsf{fma}\left({im}^{2}, \frac{-1}{3}, -2\right) \cdot im\right) \]

       9. pow2 N/A

          \[\leadsto \left(\frac{1}{2} \cdot re\right) \cdot \left(\mathsf{fma}\left(im \cdot im, \frac{-1}{3}, -2\right) \cdot im\right) \]

       10. lift-*.f64 63.3

           \[\leadsto \left(0.5 \cdot re\right) \cdot \left(\mathsf{fma}\left(im \cdot im, -0.3333333333333333, -2\right) \cdot im\right) \]

   13. Applied rewrites 63.3%

       \[\leadsto \left(0.5 \cdot re\right) \cdot \color{blue}{\left(\mathsf{fma}\left(im \cdot im, -0.3333333333333333, -2\right) \cdot im\right)} \]
3. Recombined 2 regimes into one program.
4. Add Preprocessing

Alternative 11: 53.0% accurate, 1.1× speedup

Math

\[\begin{array}{l} im\_m = \left|im\right| \\ im\_s = \mathsf{copysign}\left(1, im\right) \\ im\_s \cdot \begin{array}{l} \mathbf{if}\;0.5 \cdot \sin re \leq -0.02:\\ \;\;\;\;\left(\left(\left(re \cdot re\right) \cdot im\_m\right) \cdot 0.16666666666666666\right) \cdot re\\ \mathbf{else}:\\ \;\;\;\;\left(0.5 \cdot re\right) \cdot \left(\mathsf{fma}\left(im\_m \cdot im\_m, -0.3333333333333333, -2\right) \cdot im\_m\right)\\ \end{array} \end{array} \]

FPCore

im\_m = (fabs.f64 im)
im\_s = (copysign.f64 #s(literal 1 binary64) im)
(FPCore (im_s re im_m)
 :precision binary64
 (*
  im_s
  (if (<= (* 0.5 (sin re)) -0.02)
    (* (* (* (* re re) im_m) 0.16666666666666666) re)
    (* (* 0.5 re) (* (fma (* im_m im_m) -0.3333333333333333 -2.0) im_m)))))

C

im\_m = fabs(im);
im\_s = copysign(1.0, im);
double code(double im_s, double re, double im_m) {
	double tmp;
	if ((0.5 * sin(re)) <= -0.02) {
		tmp = (((re * re) * im_m) * 0.16666666666666666) * re;
	} else {
		tmp = (0.5 * re) * (fma((im_m * im_m), -0.3333333333333333, -2.0) * im_m);
	}
	return im_s * tmp;
}

Julia

im\_m = abs(im)
im\_s = copysign(1.0, im)
function code(im_s, re, im_m)
	tmp = 0.0
	if (Float64(0.5 * sin(re)) <= -0.02)
		tmp = Float64(Float64(Float64(Float64(re * re) * im_m) * 0.16666666666666666) * re);
	else
		tmp = Float64(Float64(0.5 * re) * Float64(fma(Float64(im_m * im_m), -0.3333333333333333, -2.0) * im_m));
	end
	return Float64(im_s * tmp)
end

Wolfram

im\_m = N[Abs[im], $MachinePrecision]
im\_s = N[With[{TMP1 = Abs[1.0], TMP2 = Sign[im]}, TMP1 * If[TMP2 == 0, 1, TMP2]], $MachinePrecision]
code[im$95$s_, re_, im$95$m_] := N[(im$95$s * If[LessEqual[N[(0.5 * N[Sin[re], $MachinePrecision]), $MachinePrecision], -0.02], N[(N[(N[(N[(re * re), $MachinePrecision] * im$95$m), $MachinePrecision] * 0.16666666666666666), $MachinePrecision] * re), $MachinePrecision], N[(N[(0.5 * re), $MachinePrecision] * N[(N[(N[(im$95$m * im$95$m), $MachinePrecision] * -0.3333333333333333 + -2.0), $MachinePrecision] * im$95$m), $MachinePrecision]), $MachinePrecision]]), $MachinePrecision]

Derivation

1. Split input into 2 regimes

2. if (*.f64 #s(literal 1/2 binary64) (sin.f64 re)) < -0.0200000000000000004

   1. Initial program 53.7%

      \[\left(0.5 \cdot \sin re\right) \cdot \left(e^{-im} - e^{im}\right) \]

   2. Taylor expanded in im around 0

      \[\leadsto \color{blue}{-1 \cdot \left(im \cdot \sin re\right)} \]
   3. Step-by-step derivation

      1. associate-*r* N/A

         \[\leadsto \left(-1 \cdot im\right) \cdot \color{blue}{\sin re} \]

      2. lower-*.f64 N/A

         \[\leadsto \left(-1 \cdot im\right) \cdot \color{blue}{\sin re} \]

      3. mul-1-neg N/A

         \[\leadsto \left(\mathsf{neg}\left(im\right)\right) \cdot \sin \color{blue}{re} \]

      4. lift-neg.f64 N/A

         \[\leadsto \left(-im\right) \cdot \sin \color{blue}{re} \]

      5. lift-sin.f64 52.8

         \[\leadsto \left(-im\right) \cdot \sin re \]

   4. Applied rewrites 52.8%

      \[\leadsto \color{blue}{\left(-im\right) \cdot \sin re} \]

   5. Taylor expanded in re around 0

      \[\leadsto re \cdot \color{blue}{\left(-1 \cdot im + \frac{1}{6} \cdot \left(im \cdot {re}^{2}\right)\right)} \]
   6. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(-1 \cdot im + \frac{1}{6} \cdot \left(im \cdot {re}^{2}\right)\right) \cdot re \]

      2. lower-*.f64 N/A

         \[\leadsto \left(-1 \cdot im + \frac{1}{6} \cdot \left(im \cdot {re}^{2}\right)\right) \cdot re \]

      3. +-commutative N/A

         \[\leadsto \left(\frac{1}{6} \cdot \left(im \cdot {re}^{2}\right) + -1 \cdot im\right) \cdot re \]

      4. *-commutative N/A

         \[\leadsto \left(\left(im \cdot {re}^{2}\right) \cdot \frac{1}{6} + -1 \cdot im\right) \cdot re \]

      5. lower-fma.f64 N/A

         \[\leadsto \mathsf{fma}\left(im \cdot {re}^{2}, \frac{1}{6}, -1 \cdot im\right) \cdot re \]

      6. *-commutative N/A

         \[\leadsto \mathsf{fma}\left({re}^{2} \cdot im, \frac{1}{6}, -1 \cdot im\right) \cdot re \]

      7. lower-*.f64 N/A

         \[\leadsto \mathsf{fma}\left({re}^{2} \cdot im, \frac{1}{6}, -1 \cdot im\right) \cdot re \]

      8. pow2 N/A

         \[\leadsto \mathsf{fma}\left(\left(re \cdot re\right) \cdot im, \frac{1}{6}, -1 \cdot im\right) \cdot re \]

      9. lift-*.f64 N/A

         \[\leadsto \mathsf{fma}\left(\left(re \cdot re\right) \cdot im, \frac{1}{6}, -1 \cdot im\right) \cdot re \]

      10. mul-1-neg N/A

          \[\leadsto \mathsf{fma}\left(\left(re \cdot re\right) \cdot im, \frac{1}{6}, \mathsf{neg}\left(im\right)\right) \cdot re \]

      11. lift-neg.f64 22.2

          \[\leadsto \mathsf{fma}\left(\left(re \cdot re\right) \cdot im, 0.16666666666666666, -im\right) \cdot re \]

   7. Applied rewrites 22.2%

      \[\leadsto \mathsf{fma}\left(\left(re \cdot re\right) \cdot im, 0.16666666666666666, -im\right) \cdot \color{blue}{re} \]

   8. Taylor expanded in re around inf

      \[\leadsto \left(\frac{1}{6} \cdot \left(im \cdot {re}^{2}\right)\right) \cdot re \]
   9. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(\left(im \cdot {re}^{2}\right) \cdot \frac{1}{6}\right) \cdot re \]

      2. lower-*.f64 N/A

         \[\leadsto \left(\left(im \cdot {re}^{2}\right) \cdot \frac{1}{6}\right) \cdot re \]

      3. *-commutative N/A

         \[\leadsto \left(\left({re}^{2} \cdot im\right) \cdot \frac{1}{6}\right) \cdot re \]

      4. pow2 N/A

         \[\leadsto \left(\left(\left(re \cdot re\right) \cdot im\right) \cdot \frac{1}{6}\right) \cdot re \]

      5. lift-*.f64 N/A

         \[\leadsto \left(\left(\left(re \cdot re\right) \cdot im\right) \cdot \frac{1}{6}\right) \cdot re \]

      6. lift-*.f64 22.1

         \[\leadsto \left(\left(\left(re \cdot re\right) \cdot im\right) \cdot 0.16666666666666666\right) \cdot re \]

   10. Applied rewrites 22.1%

       \[\leadsto \left(\left(\left(re \cdot re\right) \cdot im\right) \cdot 0.16666666666666666\right) \cdot re \]

   if -0.0200000000000000004 < (*.f64 #s(literal 1/2 binary64) (sin.f64 re))

   1. Initial program 68.2%

      \[\left(0.5 \cdot \sin re\right) \cdot \left(e^{-im} - e^{im}\right) \]

   2. Taylor expanded in re around 0

      \[\leadsto \left(\frac{1}{2} \cdot \color{blue}{re}\right) \cdot \left(e^{-im} - e^{im}\right) \]
   3. Step-by-step derivation

   4. Applied rewrites 59.3%

      \[\leadsto \left(0.5 \cdot \color{blue}{re}\right) \cdot \left(e^{-im} - e^{im}\right) \]

   5. Taylor expanded in im around 0

      \[\leadsto \left(\frac{1}{2} \cdot re\right) \cdot \left(\color{blue}{1} - e^{im}\right) \]
   6. Step-by-step derivation

   7. Applied rewrites 58.4%

      \[\leadsto \left(0.5 \cdot re\right) \cdot \left(\color{blue}{1} - e^{im}\right) \]

   8. Taylor expanded in im around 0

      \[\leadsto \left(\frac{1}{2} \cdot re\right) \cdot \left(1 - \color{blue}{1}\right) \]
   9. Step-by-step derivation

   10. Applied rewrites 19.0%

       \[\leadsto \left(0.5 \cdot re\right) \cdot \left(1 - \color{blue}{1}\right) \]

   11. Taylor expanded in im around 0

       \[\leadsto \left(\frac{1}{2} \cdot re\right) \cdot \color{blue}{\left(im \cdot \left(\frac{-1}{3} \cdot {im}^{2} - 2\right)\right)} \]
   12. Step-by-step derivation

       1. *-commutative N/A

          \[\leadsto \left(\frac{1}{2} \cdot re\right) \cdot \left(\left(\frac{-1}{3} \cdot {im}^{2} - 2\right) \cdot \color{blue}{im}\right) \]

       2. sub-flip N/A

          \[\leadsto \left(\frac{1}{2} \cdot re\right) \cdot \left(\left(\frac{-1}{3} \cdot {im}^{2} + \left(\mathsf{neg}\left(2\right)\right)\right) \cdot im\right) \]

       3. pow2 N/A

          \[\leadsto \left(\frac{1}{2} \cdot re\right) \cdot \left(\left(\frac{-1}{3} \cdot \left(im \cdot im\right) + \left(\mathsf{neg}\left(2\right)\right)\right) \cdot im\right) \]

       4. metadata-eval N/A

          \[\leadsto \left(\frac{1}{2} \cdot re\right) \cdot \left(\left(\frac{-1}{3} \cdot \left(im \cdot im\right) + -2\right) \cdot im\right) \]

       5. lower-*.f64 N/A

          \[\leadsto \left(\frac{1}{2} \cdot re\right) \cdot \left(\left(\frac{-1}{3} \cdot \left(im \cdot im\right) + -2\right) \cdot \color{blue}{im}\right) \]

       6. pow2 N/A

          \[\leadsto \left(\frac{1}{2} \cdot re\right) \cdot \left(\left(\frac{-1}{3} \cdot {im}^{2} + -2\right) \cdot im\right) \]

       7. *-commutative N/A

          \[\leadsto \left(\frac{1}{2} \cdot re\right) \cdot \left(\left({im}^{2} \cdot \frac{-1}{3} + -2\right) \cdot im\right) \]

       8. lower-fma.f64 N/A

          \[\leadsto \left(\frac{1}{2} \cdot re\right) \cdot \left(\mathsf{fma}\left({im}^{2}, \frac{-1}{3}, -2\right) \cdot im\right) \]

       9. pow2 N/A

          \[\leadsto \left(\frac{1}{2} \cdot re\right) \cdot \left(\mathsf{fma}\left(im \cdot im, \frac{-1}{3}, -2\right) \cdot im\right) \]

       10. lift-*.f64 63.1

           \[\leadsto \left(0.5 \cdot re\right) \cdot \left(\mathsf{fma}\left(im \cdot im, -0.3333333333333333, -2\right) \cdot im\right) \]

   13. Applied rewrites 63.1%

       \[\leadsto \left(0.5 \cdot re\right) \cdot \color{blue}{\left(\mathsf{fma}\left(im \cdot im, -0.3333333333333333, -2\right) \cdot im\right)} \]
3. Recombined 2 regimes into one program.
4. Add Preprocessing

Alternative 12: 49.5% accurate, 1.1× speedup

Math

\[\begin{array}{l} im\_m = \left|im\right| \\ im\_s = \mathsf{copysign}\left(1, im\right) \\ im\_s \cdot \begin{array}{l} \mathbf{if}\;0.5 \cdot \sin re \leq -0.02:\\ \;\;\;\;\left(\left(\left(re \cdot re\right) \cdot im\_m\right) \cdot 0.16666666666666666\right) \cdot re\\ \mathbf{else}:\\ \;\;\;\;\left(\left(\mathsf{fma}\left(-0.3333333333333333, im\_m \cdot im\_m, -2\right) \cdot im\_m\right) \cdot re\right) \cdot 0.5\\ \end{array} \end{array} \]

FPCore

im\_m = (fabs.f64 im)
im\_s = (copysign.f64 #s(literal 1 binary64) im)
(FPCore (im_s re im_m)
 :precision binary64
 (*
  im_s
  (if (<= (* 0.5 (sin re)) -0.02)
    (* (* (* (* re re) im_m) 0.16666666666666666) re)
    (* (* (* (fma -0.3333333333333333 (* im_m im_m) -2.0) im_m) re) 0.5))))

C

im\_m = fabs(im);
im\_s = copysign(1.0, im);
double code(double im_s, double re, double im_m) {
	double tmp;
	if ((0.5 * sin(re)) <= -0.02) {
		tmp = (((re * re) * im_m) * 0.16666666666666666) * re;
	} else {
		tmp = ((fma(-0.3333333333333333, (im_m * im_m), -2.0) * im_m) * re) * 0.5;
	}
	return im_s * tmp;
}

Julia

im\_m = abs(im)
im\_s = copysign(1.0, im)
function code(im_s, re, im_m)
	tmp = 0.0
	if (Float64(0.5 * sin(re)) <= -0.02)
		tmp = Float64(Float64(Float64(Float64(re * re) * im_m) * 0.16666666666666666) * re);
	else
		tmp = Float64(Float64(Float64(fma(-0.3333333333333333, Float64(im_m * im_m), -2.0) * im_m) * re) * 0.5);
	end
	return Float64(im_s * tmp)
end

Wolfram

im\_m = N[Abs[im], $MachinePrecision]
im\_s = N[With[{TMP1 = Abs[1.0], TMP2 = Sign[im]}, TMP1 * If[TMP2 == 0, 1, TMP2]], $MachinePrecision]
code[im$95$s_, re_, im$95$m_] := N[(im$95$s * If[LessEqual[N[(0.5 * N[Sin[re], $MachinePrecision]), $MachinePrecision], -0.02], N[(N[(N[(N[(re * re), $MachinePrecision] * im$95$m), $MachinePrecision] * 0.16666666666666666), $MachinePrecision] * re), $MachinePrecision], N[(N[(N[(N[(-0.3333333333333333 * N[(im$95$m * im$95$m), $MachinePrecision] + -2.0), $MachinePrecision] * im$95$m), $MachinePrecision] * re), $MachinePrecision] * 0.5), $MachinePrecision]]), $MachinePrecision]

Derivation

1. Split input into 2 regimes

2. if (*.f64 #s(literal 1/2 binary64) (sin.f64 re)) < -0.0200000000000000004

   1. Initial program 53.7%

      \[\left(0.5 \cdot \sin re\right) \cdot \left(e^{-im} - e^{im}\right) \]

   2. Taylor expanded in im around 0

      \[\leadsto \color{blue}{-1 \cdot \left(im \cdot \sin re\right)} \]
   3. Step-by-step derivation

      1. associate-*r* N/A

         \[\leadsto \left(-1 \cdot im\right) \cdot \color{blue}{\sin re} \]

      2. lower-*.f64 N/A

         \[\leadsto \left(-1 \cdot im\right) \cdot \color{blue}{\sin re} \]

      3. mul-1-neg N/A

         \[\leadsto \left(\mathsf{neg}\left(im\right)\right) \cdot \sin \color{blue}{re} \]

      4. lift-neg.f64 N/A

         \[\leadsto \left(-im\right) \cdot \sin \color{blue}{re} \]

      5. lift-sin.f64 52.8

         \[\leadsto \left(-im\right) \cdot \sin re \]

   4. Applied rewrites 52.8%

      \[\leadsto \color{blue}{\left(-im\right) \cdot \sin re} \]

   5. Taylor expanded in re around 0

      \[\leadsto re \cdot \color{blue}{\left(-1 \cdot im + \frac{1}{6} \cdot \left(im \cdot {re}^{2}\right)\right)} \]
   6. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(-1 \cdot im + \frac{1}{6} \cdot \left(im \cdot {re}^{2}\right)\right) \cdot re \]

      2. lower-*.f64 N/A

         \[\leadsto \left(-1 \cdot im + \frac{1}{6} \cdot \left(im \cdot {re}^{2}\right)\right) \cdot re \]

      3. +-commutative N/A

         \[\leadsto \left(\frac{1}{6} \cdot \left(im \cdot {re}^{2}\right) + -1 \cdot im\right) \cdot re \]

      4. *-commutative N/A

         \[\leadsto \left(\left(im \cdot {re}^{2}\right) \cdot \frac{1}{6} + -1 \cdot im\right) \cdot re \]

      5. lower-fma.f64 N/A

         \[\leadsto \mathsf{fma}\left(im \cdot {re}^{2}, \frac{1}{6}, -1 \cdot im\right) \cdot re \]

      6. *-commutative N/A

         \[\leadsto \mathsf{fma}\left({re}^{2} \cdot im, \frac{1}{6}, -1 \cdot im\right) \cdot re \]

      7. lower-*.f64 N/A

         \[\leadsto \mathsf{fma}\left({re}^{2} \cdot im, \frac{1}{6}, -1 \cdot im\right) \cdot re \]

      8. pow2 N/A

         \[\leadsto \mathsf{fma}\left(\left(re \cdot re\right) \cdot im, \frac{1}{6}, -1 \cdot im\right) \cdot re \]

      9. lift-*.f64 N/A

         \[\leadsto \mathsf{fma}\left(\left(re \cdot re\right) \cdot im, \frac{1}{6}, -1 \cdot im\right) \cdot re \]

      10. mul-1-neg N/A

          \[\leadsto \mathsf{fma}\left(\left(re \cdot re\right) \cdot im, \frac{1}{6}, \mathsf{neg}\left(im\right)\right) \cdot re \]

      11. lift-neg.f64 22.2

          \[\leadsto \mathsf{fma}\left(\left(re \cdot re\right) \cdot im, 0.16666666666666666, -im\right) \cdot re \]

   7. Applied rewrites 22.2%

      \[\leadsto \mathsf{fma}\left(\left(re \cdot re\right) \cdot im, 0.16666666666666666, -im\right) \cdot \color{blue}{re} \]

   8. Taylor expanded in re around inf

      \[\leadsto \left(\frac{1}{6} \cdot \left(im \cdot {re}^{2}\right)\right) \cdot re \]
   9. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(\left(im \cdot {re}^{2}\right) \cdot \frac{1}{6}\right) \cdot re \]

      2. lower-*.f64 N/A

         \[\leadsto \left(\left(im \cdot {re}^{2}\right) \cdot \frac{1}{6}\right) \cdot re \]

      3. *-commutative N/A

         \[\leadsto \left(\left({re}^{2} \cdot im\right) \cdot \frac{1}{6}\right) \cdot re \]

      4. pow2 N/A

         \[\leadsto \left(\left(\left(re \cdot re\right) \cdot im\right) \cdot \frac{1}{6}\right) \cdot re \]

      5. lift-*.f64 N/A

         \[\leadsto \left(\left(\left(re \cdot re\right) \cdot im\right) \cdot \frac{1}{6}\right) \cdot re \]

      6. lift-*.f64 22.1

         \[\leadsto \left(\left(\left(re \cdot re\right) \cdot im\right) \cdot 0.16666666666666666\right) \cdot re \]

   10. Applied rewrites 22.1%

       \[\leadsto \left(\left(\left(re \cdot re\right) \cdot im\right) \cdot 0.16666666666666666\right) \cdot re \]

   if -0.0200000000000000004 < (*.f64 #s(literal 1/2 binary64) (sin.f64 re))

   1. Initial program 68.2%

      \[\left(0.5 \cdot \sin re\right) \cdot \left(e^{-im} - e^{im}\right) \]

   2. Taylor expanded in re around 0

      \[\leadsto \color{blue}{\frac{1}{2} \cdot \left(re \cdot \left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right)\right)} \]
   3. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(re \cdot \left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right)\right) \cdot \color{blue}{\frac{1}{2}} \]

      2. lower-*.f64 N/A

         \[\leadsto \left(re \cdot \left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right)\right) \cdot \color{blue}{\frac{1}{2}} \]

      3. *-commutative N/A

         \[\leadsto \left(\left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right) \cdot re\right) \cdot \frac{1}{2} \]

      4. lower-*.f64 N/A

         \[\leadsto \left(\left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right) \cdot re\right) \cdot \frac{1}{2} \]

      5. sub-negate N/A

         \[\leadsto \left(\left(\mathsf{neg}\left(\left(e^{im} - e^{\mathsf{neg}\left(im\right)}\right)\right)\right) \cdot re\right) \cdot \frac{1}{2} \]

      6. lower-neg.f64 N/A

         \[\leadsto \left(\left(-\left(e^{im} - e^{\mathsf{neg}\left(im\right)}\right)\right) \cdot re\right) \cdot \frac{1}{2} \]

      7. sinh-undef N/A

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot \frac{1}{2} \]

      8. lower-*.f64 N/A

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot \frac{1}{2} \]

      9. lower-sinh.f64 74.2

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot 0.5 \]

   4. Applied rewrites 74.2%

      \[\leadsto \color{blue}{\left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot 0.5} \]

   5. Taylor expanded in im around 0

      \[\leadsto \left(\left(im \cdot \left(\frac{-1}{3} \cdot {im}^{2} - 2\right)\right) \cdot re\right) \cdot \frac{1}{2} \]
   6. Step-by-step derivation

      1. sinh-undef-rev N/A

         \[\leadsto \left(\left(im \cdot \left(\frac{-1}{3} \cdot {im}^{2} - 2\right)\right) \cdot re\right) \cdot \frac{1}{2} \]

      2. sub-negate N/A

         \[\leadsto \left(\left(im \cdot \left(\frac{-1}{3} \cdot {im}^{2} - 2\right)\right) \cdot re\right) \cdot \frac{1}{2} \]

      3. flip-- N/A

         \[\leadsto \left(\left(im \cdot \left(\frac{-1}{3} \cdot {im}^{2} - 2\right)\right) \cdot re\right) \cdot \frac{1}{2} \]

      4. pow2 N/A

         \[\leadsto \left(\left(im \cdot \left(\frac{-1}{3} \cdot {im}^{2} - 2\right)\right) \cdot re\right) \cdot \frac{1}{2} \]

      5. lift-neg.f64 N/A

         \[\leadsto \left(\left(im \cdot \left(\frac{-1}{3} \cdot {im}^{2} - 2\right)\right) \cdot re\right) \cdot \frac{1}{2} \]

      6. exp-prod N/A

         \[\leadsto \left(\left(im \cdot \left(\frac{-1}{3} \cdot {im}^{2} - 2\right)\right) \cdot re\right) \cdot \frac{1}{2} \]

      7. lift-neg.f64 N/A

         \[\leadsto \left(\left(im \cdot \left(\frac{-1}{3} \cdot {im}^{2} - 2\right)\right) \cdot re\right) \cdot \frac{1}{2} \]

      8. exp-sum N/A

         \[\leadsto \left(\left(im \cdot \left(\frac{-1}{3} \cdot {im}^{2} - 2\right)\right) \cdot re\right) \cdot \frac{1}{2} \]

      9. *-commutative N/A

         \[\leadsto \left(\left(\left(\frac{-1}{3} \cdot {im}^{2} - 2\right) \cdot im\right) \cdot re\right) \cdot \frac{1}{2} \]

      10. lower-*.f64 N/A

          \[\leadsto \left(\left(\left(\frac{-1}{3} \cdot {im}^{2} - 2\right) \cdot im\right) \cdot re\right) \cdot \frac{1}{2} \]

      11. sub-flip N/A

          \[\leadsto \left(\left(\left(\frac{-1}{3} \cdot {im}^{2} + \left(\mathsf{neg}\left(2\right)\right)\right) \cdot im\right) \cdot re\right) \cdot \frac{1}{2} \]

      12. metadata-eval N/A

          \[\leadsto \left(\left(\left(\frac{-1}{3} \cdot {im}^{2} + -2\right) \cdot im\right) \cdot re\right) \cdot \frac{1}{2} \]

      13. lower-fma.f64 N/A

          \[\leadsto \left(\left(\mathsf{fma}\left(\frac{-1}{3}, {im}^{2}, -2\right) \cdot im\right) \cdot re\right) \cdot \frac{1}{2} \]

      14. unpow2 N/A

          \[\leadsto \left(\left(\mathsf{fma}\left(\frac{-1}{3}, im \cdot im, -2\right) \cdot im\right) \cdot re\right) \cdot \frac{1}{2} \]

      15. lower-*.f64 63.1

          \[\leadsto \left(\left(\mathsf{fma}\left(-0.3333333333333333, im \cdot im, -2\right) \cdot im\right) \cdot re\right) \cdot 0.5 \]

   7. Applied rewrites 63.1%

      \[\leadsto \left(\left(\mathsf{fma}\left(-0.3333333333333333, im \cdot im, -2\right) \cdot im\right) \cdot re\right) \cdot 0.5 \]
3. Recombined 2 regimes into one program.
4. Add Preprocessing

Alternative 13: 45.1% accurate, 0.4× speedup

Math

\[\begin{array}{l} im\_m = \left|im\right| \\ im\_s = \mathsf{copysign}\left(1, im\right) \\ \begin{array}{l} t_0 := \left(0.5 \cdot \sin re\right) \cdot \left(e^{-im\_m} - e^{im\_m}\right)\\ im\_s \cdot \begin{array}{l} \mathbf{if}\;t\_0 \leq -1 \cdot 10^{-49}:\\ \;\;\;\;\left(\left(\left(im\_m \cdot im\_m\right) \cdot im\_m\right) \cdot re\right) \cdot -0.16666666666666666\\ \mathbf{elif}\;t\_0 \leq 0:\\ \;\;\;\;\left(-im\_m\right) \cdot re\\ \mathbf{else}:\\ \;\;\;\;\left(\left(\left(re \cdot re\right) \cdot re\right) \cdot im\_m\right) \cdot 0.16666666666666666\\ \end{array} \end{array} \end{array} \]

FPCore

im\_m = (fabs.f64 im)
im\_s = (copysign.f64 #s(literal 1 binary64) im)
(FPCore (im_s re im_m)
 :precision binary64
 (let* ((t_0 (* (* 0.5 (sin re)) (- (exp (- im_m)) (exp im_m)))))
   (*
    im_s
    (if (<= t_0 -1e-49)
      (* (* (* (* im_m im_m) im_m) re) -0.16666666666666666)
      (if (<= t_0 0.0)
        (* (- im_m) re)
        (* (* (* (* re re) re) im_m) 0.16666666666666666))))))

C

im\_m = fabs(im);
im\_s = copysign(1.0, im);
double code(double im_s, double re, double im_m) {
	double t_0 = (0.5 * sin(re)) * (exp(-im_m) - exp(im_m));
	double tmp;
	if (t_0 <= -1e-49) {
		tmp = (((im_m * im_m) * im_m) * re) * -0.16666666666666666;
	} else if (t_0 <= 0.0) {
		tmp = -im_m * re;
	} else {
		tmp = (((re * re) * re) * im_m) * 0.16666666666666666;
	}
	return im_s * tmp;
}

Fortran

im\_m =     private
im\_s =     private
module fmin_fmax_functions
    implicit none
    private
    public fmax
    public fmin

    interface fmax
        module procedure fmax88
        module procedure fmax44
        module procedure fmax84
        module procedure fmax48
    end interface
    interface fmin
        module procedure fmin88
        module procedure fmin44
        module procedure fmin84
        module procedure fmin48
    end interface
contains
    real(8) function fmax88(x, y) result (res)
        real(8), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(x, max(x, y), y /= y), x /= x)
    end function
    real(4) function fmax44(x, y) result (res)
        real(4), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(y, merge(x, max(x, y), y /= y), x /= x)
    end function
    real(8) function fmax84(x, y) result(res)
        real(8), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(dble(y), merge(x, max(x, dble(y)), y /= y), x /= x)
    end function
    real(8) function fmax48(x, y) result(res)
        real(4), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(dble(x), max(dble(x), y), y /= y), x /= x)
    end function
    real(8) function fmin88(x, y) result (res)
        real(8), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(x, min(x, y), y /= y), x /= x)
    end function
    real(4) function fmin44(x, y) result (res)
        real(4), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(y, merge(x, min(x, y), y /= y), x /= x)
    end function
    real(8) function fmin84(x, y) result(res)
        real(8), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(dble(y), merge(x, min(x, dble(y)), y /= y), x /= x)
    end function
    real(8) function fmin48(x, y) result(res)
        real(4), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(dble(x), min(dble(x), y), y /= y), x /= x)
    end function
end module

real(8) function code(im_s, re, im_m)
use fmin_fmax_functions
    real(8), intent (in) :: im_s
    real(8), intent (in) :: re
    real(8), intent (in) :: im_m
    real(8) :: t_0
    real(8) :: tmp
    t_0 = (0.5d0 * sin(re)) * (exp(-im_m) - exp(im_m))
    if (t_0 <= (-1d-49)) then
        tmp = (((im_m * im_m) * im_m) * re) * (-0.16666666666666666d0)
    else if (t_0 <= 0.0d0) then
        tmp = -im_m * re
    else
        tmp = (((re * re) * re) * im_m) * 0.16666666666666666d0
    end if
    code = im_s * tmp
end function

Java

im\_m = Math.abs(im);
im\_s = Math.copySign(1.0, im);
public static double code(double im_s, double re, double im_m) {
	double t_0 = (0.5 * Math.sin(re)) * (Math.exp(-im_m) - Math.exp(im_m));
	double tmp;
	if (t_0 <= -1e-49) {
		tmp = (((im_m * im_m) * im_m) * re) * -0.16666666666666666;
	} else if (t_0 <= 0.0) {
		tmp = -im_m * re;
	} else {
		tmp = (((re * re) * re) * im_m) * 0.16666666666666666;
	}
	return im_s * tmp;
}

Python

im\_m = math.fabs(im)
im\_s = math.copysign(1.0, im)
def code(im_s, re, im_m):
	t_0 = (0.5 * math.sin(re)) * (math.exp(-im_m) - math.exp(im_m))
	tmp = 0
	if t_0 <= -1e-49:
		tmp = (((im_m * im_m) * im_m) * re) * -0.16666666666666666
	elif t_0 <= 0.0:
		tmp = -im_m * re
	else:
		tmp = (((re * re) * re) * im_m) * 0.16666666666666666
	return im_s * tmp

Julia

im\_m = abs(im)
im\_s = copysign(1.0, im)
function code(im_s, re, im_m)
	t_0 = Float64(Float64(0.5 * sin(re)) * Float64(exp(Float64(-im_m)) - exp(im_m)))
	tmp = 0.0
	if (t_0 <= -1e-49)
		tmp = Float64(Float64(Float64(Float64(im_m * im_m) * im_m) * re) * -0.16666666666666666);
	elseif (t_0 <= 0.0)
		tmp = Float64(Float64(-im_m) * re);
	else
		tmp = Float64(Float64(Float64(Float64(re * re) * re) * im_m) * 0.16666666666666666);
	end
	return Float64(im_s * tmp)
end

MATLAB

im\_m = abs(im);
im\_s = sign(im) * abs(1.0);
function tmp_2 = code(im_s, re, im_m)
	t_0 = (0.5 * sin(re)) * (exp(-im_m) - exp(im_m));
	tmp = 0.0;
	if (t_0 <= -1e-49)
		tmp = (((im_m * im_m) * im_m) * re) * -0.16666666666666666;
	elseif (t_0 <= 0.0)
		tmp = -im_m * re;
	else
		tmp = (((re * re) * re) * im_m) * 0.16666666666666666;
	end
	tmp_2 = im_s * tmp;
end

Wolfram

im\_m = N[Abs[im], $MachinePrecision]
im\_s = N[With[{TMP1 = Abs[1.0], TMP2 = Sign[im]}, TMP1 * If[TMP2 == 0, 1, TMP2]], $MachinePrecision]
code[im$95$s_, re_, im$95$m_] := Block[{t$95$0 = N[(N[(0.5 * N[Sin[re], $MachinePrecision]), $MachinePrecision] * N[(N[Exp[(-im$95$m)], $MachinePrecision] - N[Exp[im$95$m], $MachinePrecision]), $MachinePrecision]), $MachinePrecision]}, N[(im$95$s * If[LessEqual[t$95$0, -1e-49], N[(N[(N[(N[(im$95$m * im$95$m), $MachinePrecision] * im$95$m), $MachinePrecision] * re), $MachinePrecision] * -0.16666666666666666), $MachinePrecision], If[LessEqual[t$95$0, 0.0], N[((-im$95$m) * re), $MachinePrecision], N[(N[(N[(N[(re * re), $MachinePrecision] * re), $MachinePrecision] * im$95$m), $MachinePrecision] * 0.16666666666666666), $MachinePrecision]]]), $MachinePrecision]]

Derivation

1. Split input into 3 regimes

2. if (*.f64 (*.f64 #s(literal 1/2 binary64) (sin.f64 re)) (-.f64 (exp.f64 (neg.f64 im)) (exp.f64 im))) < -9.99999999999999936e-50

   1. Initial program 98.8%

      \[\left(0.5 \cdot \sin re\right) \cdot \left(e^{-im} - e^{im}\right) \]

   2. Taylor expanded in re around 0

      \[\leadsto \color{blue}{\frac{1}{2} \cdot \left(re \cdot \left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right)\right)} \]
   3. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(re \cdot \left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right)\right) \cdot \color{blue}{\frac{1}{2}} \]

      2. lower-*.f64 N/A

         \[\leadsto \left(re \cdot \left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right)\right) \cdot \color{blue}{\frac{1}{2}} \]

      3. *-commutative N/A

         \[\leadsto \left(\left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right) \cdot re\right) \cdot \frac{1}{2} \]

      4. lower-*.f64 N/A

         \[\leadsto \left(\left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right) \cdot re\right) \cdot \frac{1}{2} \]

      5. sub-negate N/A

         \[\leadsto \left(\left(\mathsf{neg}\left(\left(e^{im} - e^{\mathsf{neg}\left(im\right)}\right)\right)\right) \cdot re\right) \cdot \frac{1}{2} \]

      6. lower-neg.f64 N/A

         \[\leadsto \left(\left(-\left(e^{im} - e^{\mathsf{neg}\left(im\right)}\right)\right) \cdot re\right) \cdot \frac{1}{2} \]

      7. sinh-undef N/A

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot \frac{1}{2} \]

      8. lower-*.f64 N/A

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot \frac{1}{2} \]

      9. lower-sinh.f64 72.5

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot 0.5 \]

   4. Applied rewrites 72.5%

      \[\leadsto \color{blue}{\left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot 0.5} \]

   5. Taylor expanded in im around 0

      \[\leadsto im \cdot \color{blue}{\left(-1 \cdot re + \frac{-1}{6} \cdot \left({im}^{2} \cdot re\right)\right)} \]
   6. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(-1 \cdot re + \frac{-1}{6} \cdot \left({im}^{2} \cdot re\right)\right) \cdot im \]

      2. lower-*.f64 N/A

         \[\leadsto \left(-1 \cdot re + \frac{-1}{6} \cdot \left({im}^{2} \cdot re\right)\right) \cdot im \]

      3. +-commutative N/A

         \[\leadsto \left(\frac{-1}{6} \cdot \left({im}^{2} \cdot re\right) + -1 \cdot re\right) \cdot im \]

      4. *-commutative N/A

         \[\leadsto \left(\left({im}^{2} \cdot re\right) \cdot \frac{-1}{6} + -1 \cdot re\right) \cdot im \]

      5. lower-fma.f64 N/A

         \[\leadsto \mathsf{fma}\left({im}^{2} \cdot re, \frac{-1}{6}, -1 \cdot re\right) \cdot im \]

      6. lower-*.f64 N/A

         \[\leadsto \mathsf{fma}\left({im}^{2} \cdot re, \frac{-1}{6}, -1 \cdot re\right) \cdot im \]

      7. unpow2 N/A

         \[\leadsto \mathsf{fma}\left(\left(im \cdot im\right) \cdot re, \frac{-1}{6}, -1 \cdot re\right) \cdot im \]

      8. lower-*.f64 N/A

         \[\leadsto \mathsf{fma}\left(\left(im \cdot im\right) \cdot re, \frac{-1}{6}, -1 \cdot re\right) \cdot im \]

      9. mul-1-neg N/A

         \[\leadsto \mathsf{fma}\left(\left(im \cdot im\right) \cdot re, \frac{-1}{6}, \mathsf{neg}\left(re\right)\right) \cdot im \]

      10. lower-neg.f64 49.1

          \[\leadsto \mathsf{fma}\left(\left(im \cdot im\right) \cdot re, -0.16666666666666666, -re\right) \cdot im \]

   7. Applied rewrites 49.1%

      \[\leadsto \mathsf{fma}\left(\left(im \cdot im\right) \cdot re, -0.16666666666666666, -re\right) \cdot \color{blue}{im} \]

   8. Taylor expanded in im around inf

      \[\leadsto \frac{-1}{6} \cdot \left({im}^{3} \cdot \color{blue}{re}\right) \]
   9. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left({im}^{3} \cdot re\right) \cdot \frac{-1}{6} \]

      2. lower-*.f64 N/A

         \[\leadsto \left({im}^{3} \cdot re\right) \cdot \frac{-1}{6} \]

      3. lower-*.f64 N/A

         \[\leadsto \left({im}^{3} \cdot re\right) \cdot \frac{-1}{6} \]

      4. unpow3 N/A

         \[\leadsto \left(\left(\left(im \cdot im\right) \cdot im\right) \cdot re\right) \cdot \frac{-1}{6} \]

      5. pow2 N/A

         \[\leadsto \left(\left({im}^{2} \cdot im\right) \cdot re\right) \cdot \frac{-1}{6} \]

      6. lower-*.f64 N/A

         \[\leadsto \left(\left({im}^{2} \cdot im\right) \cdot re\right) \cdot \frac{-1}{6} \]

      7. pow2 N/A

         \[\leadsto \left(\left(\left(im \cdot im\right) \cdot im\right) \cdot re\right) \cdot \frac{-1}{6} \]

      8. lift-*.f64 54.5

         \[\leadsto \left(\left(\left(im \cdot im\right) \cdot im\right) \cdot re\right) \cdot -0.16666666666666666 \]

   10. Applied rewrites 54.5%

       \[\leadsto \left(\left(\left(im \cdot im\right) \cdot im\right) \cdot re\right) \cdot -0.16666666666666666 \]

   if -9.99999999999999936e-50 < (*.f64 (*.f64 #s(literal 1/2 binary64) (sin.f64 re)) (-.f64 (exp.f64 (neg.f64 im)) (exp.f64 im))) < 0.0

   1. Initial program 30.2%

      \[\left(0.5 \cdot \sin re\right) \cdot \left(e^{-im} - e^{im}\right) \]

   2. Taylor expanded in re around 0

      \[\leadsto \color{blue}{\frac{1}{2} \cdot \left(re \cdot \left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right)\right)} \]
   3. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(re \cdot \left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right)\right) \cdot \color{blue}{\frac{1}{2}} \]

      2. lower-*.f64 N/A

         \[\leadsto \left(re \cdot \left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right)\right) \cdot \color{blue}{\frac{1}{2}} \]

      3. *-commutative N/A

         \[\leadsto \left(\left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right) \cdot re\right) \cdot \frac{1}{2} \]

      4. lower-*.f64 N/A

         \[\leadsto \left(\left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right) \cdot re\right) \cdot \frac{1}{2} \]

      5. sub-negate N/A

         \[\leadsto \left(\left(\mathsf{neg}\left(\left(e^{im} - e^{\mathsf{neg}\left(im\right)}\right)\right)\right) \cdot re\right) \cdot \frac{1}{2} \]

      6. lower-neg.f64 N/A

         \[\leadsto \left(\left(-\left(e^{im} - e^{\mathsf{neg}\left(im\right)}\right)\right) \cdot re\right) \cdot \frac{1}{2} \]

      7. sinh-undef N/A

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot \frac{1}{2} \]

      8. lower-*.f64 N/A

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot \frac{1}{2} \]

      9. lower-sinh.f64 52.3

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot 0.5 \]

   4. Applied rewrites 52.3%

      \[\leadsto \color{blue}{\left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot 0.5} \]

   5. Taylor expanded in im around 0

      \[\leadsto -1 \cdot \color{blue}{\left(im \cdot re\right)} \]
   6. Step-by-step derivation

      1. associate-*r* N/A

         \[\leadsto \left(-1 \cdot im\right) \cdot re \]

      2. lower-*.f64 N/A

         \[\leadsto \left(-1 \cdot im\right) \cdot re \]

      3. mul-1-neg N/A

         \[\leadsto \left(\mathsf{neg}\left(im\right)\right) \cdot re \]

      4. lift-neg.f64 51.9

         \[\leadsto \left(-im\right) \cdot re \]

   7. Applied rewrites 51.9%

      \[\leadsto \left(-im\right) \cdot \color{blue}{re} \]

   if 0.0 < (*.f64 (*.f64 #s(literal 1/2 binary64) (sin.f64 re)) (-.f64 (exp.f64 (neg.f64 im)) (exp.f64 im)))

   1. Initial program 98.0%

      \[\left(0.5 \cdot \sin re\right) \cdot \left(e^{-im} - e^{im}\right) \]

   2. Taylor expanded in im around 0

      \[\leadsto \color{blue}{-1 \cdot \left(im \cdot \sin re\right)} \]
   3. Step-by-step derivation

      1. associate-*r* N/A

         \[\leadsto \left(-1 \cdot im\right) \cdot \color{blue}{\sin re} \]

      2. lower-*.f64 N/A

         \[\leadsto \left(-1 \cdot im\right) \cdot \color{blue}{\sin re} \]

      3. mul-1-neg N/A

         \[\leadsto \left(\mathsf{neg}\left(im\right)\right) \cdot \sin \color{blue}{re} \]

      4. lift-neg.f64 N/A

         \[\leadsto \left(-im\right) \cdot \sin \color{blue}{re} \]

      5. lift-sin.f64 8.6

         \[\leadsto \left(-im\right) \cdot \sin re \]

   4. Applied rewrites 8.6%

      \[\leadsto \color{blue}{\left(-im\right) \cdot \sin re} \]

   5. Taylor expanded in re around 0

      \[\leadsto re \cdot \color{blue}{\left(-1 \cdot im + \frac{1}{6} \cdot \left(im \cdot {re}^{2}\right)\right)} \]
   6. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(-1 \cdot im + \frac{1}{6} \cdot \left(im \cdot {re}^{2}\right)\right) \cdot re \]

      2. lower-*.f64 N/A

         \[\leadsto \left(-1 \cdot im + \frac{1}{6} \cdot \left(im \cdot {re}^{2}\right)\right) \cdot re \]

      3. +-commutative N/A

         \[\leadsto \left(\frac{1}{6} \cdot \left(im \cdot {re}^{2}\right) + -1 \cdot im\right) \cdot re \]

      4. *-commutative N/A

         \[\leadsto \left(\left(im \cdot {re}^{2}\right) \cdot \frac{1}{6} + -1 \cdot im\right) \cdot re \]

      5. lower-fma.f64 N/A

         \[\leadsto \mathsf{fma}\left(im \cdot {re}^{2}, \frac{1}{6}, -1 \cdot im\right) \cdot re \]

      6. *-commutative N/A

         \[\leadsto \mathsf{fma}\left({re}^{2} \cdot im, \frac{1}{6}, -1 \cdot im\right) \cdot re \]

      7. lower-*.f64 N/A

         \[\leadsto \mathsf{fma}\left({re}^{2} \cdot im, \frac{1}{6}, -1 \cdot im\right) \cdot re \]

      8. pow2 N/A

         \[\leadsto \mathsf{fma}\left(\left(re \cdot re\right) \cdot im, \frac{1}{6}, -1 \cdot im\right) \cdot re \]

      9. lift-*.f64 N/A

         \[\leadsto \mathsf{fma}\left(\left(re \cdot re\right) \cdot im, \frac{1}{6}, -1 \cdot im\right) \cdot re \]

      10. mul-1-neg N/A

          \[\leadsto \mathsf{fma}\left(\left(re \cdot re\right) \cdot im, \frac{1}{6}, \mathsf{neg}\left(im\right)\right) \cdot re \]

      11. lift-neg.f64 23.6

          \[\leadsto \mathsf{fma}\left(\left(re \cdot re\right) \cdot im, 0.16666666666666666, -im\right) \cdot re \]

   7. Applied rewrites 23.6%

      \[\leadsto \mathsf{fma}\left(\left(re \cdot re\right) \cdot im, 0.16666666666666666, -im\right) \cdot \color{blue}{re} \]

   8. Taylor expanded in re around inf

      \[\leadsto \frac{1}{6} \cdot \left(im \cdot \color{blue}{{re}^{3}}\right) \]
   9. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(im \cdot {re}^{3}\right) \cdot \frac{1}{6} \]

      2. lower-*.f64 N/A

         \[\leadsto \left(im \cdot {re}^{3}\right) \cdot \frac{1}{6} \]

      3. *-commutative N/A

         \[\leadsto \left({re}^{3} \cdot im\right) \cdot \frac{1}{6} \]

      4. lower-*.f64 N/A

         \[\leadsto \left({re}^{3} \cdot im\right) \cdot \frac{1}{6} \]

      5. unpow3 N/A

         \[\leadsto \left(\left(\left(re \cdot re\right) \cdot re\right) \cdot im\right) \cdot \frac{1}{6} \]

      6. pow2 N/A

         \[\leadsto \left(\left({re}^{2} \cdot re\right) \cdot im\right) \cdot \frac{1}{6} \]

      7. lower-*.f64 N/A

         \[\leadsto \left(\left({re}^{2} \cdot re\right) \cdot im\right) \cdot \frac{1}{6} \]

      8. pow2 N/A

         \[\leadsto \left(\left(\left(re \cdot re\right) \cdot re\right) \cdot im\right) \cdot \frac{1}{6} \]

      9. lift-*.f64 20.6

         \[\leadsto \left(\left(\left(re \cdot re\right) \cdot re\right) \cdot im\right) \cdot 0.16666666666666666 \]

   10. Applied rewrites 20.6%

       \[\leadsto \left(\left(\left(re \cdot re\right) \cdot re\right) \cdot im\right) \cdot 0.16666666666666666 \]
3. Recombined 3 regimes into one program.
4. Add Preprocessing

Alternative 14: 45.1% accurate, 0.8× speedup

Math

\[\begin{array}{l} im\_m = \left|im\right| \\ im\_s = \mathsf{copysign}\left(1, im\right) \\ im\_s \cdot \begin{array}{l} \mathbf{if}\;\left(0.5 \cdot \sin re\right) \cdot \left(e^{-im\_m} - e^{im\_m}\right) \leq -1 \cdot 10^{-49}:\\ \;\;\;\;\left(\left(\left(im\_m \cdot im\_m\right) \cdot im\_m\right) \cdot re\right) \cdot -0.16666666666666666\\ \mathbf{else}:\\ \;\;\;\;\left(\mathsf{fma}\left(re \cdot re, 0.16666666666666666, -1\right) \cdot im\_m\right) \cdot re\\ \end{array} \end{array} \]

FPCore

im\_m = (fabs.f64 im)
im\_s = (copysign.f64 #s(literal 1 binary64) im)
(FPCore (im_s re im_m)
 :precision binary64
 (*
  im_s
  (if (<= (* (* 0.5 (sin re)) (- (exp (- im_m)) (exp im_m))) -1e-49)
    (* (* (* (* im_m im_m) im_m) re) -0.16666666666666666)
    (* (* (fma (* re re) 0.16666666666666666 -1.0) im_m) re))))

C

im\_m = fabs(im);
im\_s = copysign(1.0, im);
double code(double im_s, double re, double im_m) {
	double tmp;
	if (((0.5 * sin(re)) * (exp(-im_m) - exp(im_m))) <= -1e-49) {
		tmp = (((im_m * im_m) * im_m) * re) * -0.16666666666666666;
	} else {
		tmp = (fma((re * re), 0.16666666666666666, -1.0) * im_m) * re;
	}
	return im_s * tmp;
}

Julia

im\_m = abs(im)
im\_s = copysign(1.0, im)
function code(im_s, re, im_m)
	tmp = 0.0
	if (Float64(Float64(0.5 * sin(re)) * Float64(exp(Float64(-im_m)) - exp(im_m))) <= -1e-49)
		tmp = Float64(Float64(Float64(Float64(im_m * im_m) * im_m) * re) * -0.16666666666666666);
	else
		tmp = Float64(Float64(fma(Float64(re * re), 0.16666666666666666, -1.0) * im_m) * re);
	end
	return Float64(im_s * tmp)
end

Wolfram

im\_m = N[Abs[im], $MachinePrecision]
im\_s = N[With[{TMP1 = Abs[1.0], TMP2 = Sign[im]}, TMP1 * If[TMP2 == 0, 1, TMP2]], $MachinePrecision]
code[im$95$s_, re_, im$95$m_] := N[(im$95$s * If[LessEqual[N[(N[(0.5 * N[Sin[re], $MachinePrecision]), $MachinePrecision] * N[(N[Exp[(-im$95$m)], $MachinePrecision] - N[Exp[im$95$m], $MachinePrecision]), $MachinePrecision]), $MachinePrecision], -1e-49], N[(N[(N[(N[(im$95$m * im$95$m), $MachinePrecision] * im$95$m), $MachinePrecision] * re), $MachinePrecision] * -0.16666666666666666), $MachinePrecision], N[(N[(N[(N[(re * re), $MachinePrecision] * 0.16666666666666666 + -1.0), $MachinePrecision] * im$95$m), $MachinePrecision] * re), $MachinePrecision]]), $MachinePrecision]

Derivation

1. Split input into 2 regimes

2. if (*.f64 (*.f64 #s(literal 1/2 binary64) (sin.f64 re)) (-.f64 (exp.f64 (neg.f64 im)) (exp.f64 im))) < -9.99999999999999936e-50

   1. Initial program 98.8%

      \[\left(0.5 \cdot \sin re\right) \cdot \left(e^{-im} - e^{im}\right) \]

   2. Taylor expanded in re around 0

      \[\leadsto \color{blue}{\frac{1}{2} \cdot \left(re \cdot \left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right)\right)} \]
   3. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(re \cdot \left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right)\right) \cdot \color{blue}{\frac{1}{2}} \]

      2. lower-*.f64 N/A

         \[\leadsto \left(re \cdot \left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right)\right) \cdot \color{blue}{\frac{1}{2}} \]

      3. *-commutative N/A

         \[\leadsto \left(\left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right) \cdot re\right) \cdot \frac{1}{2} \]

      4. lower-*.f64 N/A

         \[\leadsto \left(\left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right) \cdot re\right) \cdot \frac{1}{2} \]

      5. sub-negate N/A

         \[\leadsto \left(\left(\mathsf{neg}\left(\left(e^{im} - e^{\mathsf{neg}\left(im\right)}\right)\right)\right) \cdot re\right) \cdot \frac{1}{2} \]

      6. lower-neg.f64 N/A

         \[\leadsto \left(\left(-\left(e^{im} - e^{\mathsf{neg}\left(im\right)}\right)\right) \cdot re\right) \cdot \frac{1}{2} \]

      7. sinh-undef N/A

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot \frac{1}{2} \]

      8. lower-*.f64 N/A

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot \frac{1}{2} \]

      9. lower-sinh.f64 72.5

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot 0.5 \]

   4. Applied rewrites 72.5%

      \[\leadsto \color{blue}{\left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot 0.5} \]

   5. Taylor expanded in im around 0

      \[\leadsto im \cdot \color{blue}{\left(-1 \cdot re + \frac{-1}{6} \cdot \left({im}^{2} \cdot re\right)\right)} \]
   6. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(-1 \cdot re + \frac{-1}{6} \cdot \left({im}^{2} \cdot re\right)\right) \cdot im \]

      2. lower-*.f64 N/A

         \[\leadsto \left(-1 \cdot re + \frac{-1}{6} \cdot \left({im}^{2} \cdot re\right)\right) \cdot im \]

      3. +-commutative N/A

         \[\leadsto \left(\frac{-1}{6} \cdot \left({im}^{2} \cdot re\right) + -1 \cdot re\right) \cdot im \]

      4. *-commutative N/A

         \[\leadsto \left(\left({im}^{2} \cdot re\right) \cdot \frac{-1}{6} + -1 \cdot re\right) \cdot im \]

      5. lower-fma.f64 N/A

         \[\leadsto \mathsf{fma}\left({im}^{2} \cdot re, \frac{-1}{6}, -1 \cdot re\right) \cdot im \]

      6. lower-*.f64 N/A

         \[\leadsto \mathsf{fma}\left({im}^{2} \cdot re, \frac{-1}{6}, -1 \cdot re\right) \cdot im \]

      7. unpow2 N/A

         \[\leadsto \mathsf{fma}\left(\left(im \cdot im\right) \cdot re, \frac{-1}{6}, -1 \cdot re\right) \cdot im \]

      8. lower-*.f64 N/A

         \[\leadsto \mathsf{fma}\left(\left(im \cdot im\right) \cdot re, \frac{-1}{6}, -1 \cdot re\right) \cdot im \]

      9. mul-1-neg N/A

         \[\leadsto \mathsf{fma}\left(\left(im \cdot im\right) \cdot re, \frac{-1}{6}, \mathsf{neg}\left(re\right)\right) \cdot im \]

      10. lower-neg.f64 49.1

          \[\leadsto \mathsf{fma}\left(\left(im \cdot im\right) \cdot re, -0.16666666666666666, -re\right) \cdot im \]

   7. Applied rewrites 49.1%

      \[\leadsto \mathsf{fma}\left(\left(im \cdot im\right) \cdot re, -0.16666666666666666, -re\right) \cdot \color{blue}{im} \]

   8. Taylor expanded in im around inf

      \[\leadsto \frac{-1}{6} \cdot \left({im}^{3} \cdot \color{blue}{re}\right) \]
   9. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left({im}^{3} \cdot re\right) \cdot \frac{-1}{6} \]

      2. lower-*.f64 N/A

         \[\leadsto \left({im}^{3} \cdot re\right) \cdot \frac{-1}{6} \]

      3. lower-*.f64 N/A

         \[\leadsto \left({im}^{3} \cdot re\right) \cdot \frac{-1}{6} \]

      4. unpow3 N/A

         \[\leadsto \left(\left(\left(im \cdot im\right) \cdot im\right) \cdot re\right) \cdot \frac{-1}{6} \]

      5. pow2 N/A

         \[\leadsto \left(\left({im}^{2} \cdot im\right) \cdot re\right) \cdot \frac{-1}{6} \]

      6. lower-*.f64 N/A

         \[\leadsto \left(\left({im}^{2} \cdot im\right) \cdot re\right) \cdot \frac{-1}{6} \]

      7. pow2 N/A

         \[\leadsto \left(\left(\left(im \cdot im\right) \cdot im\right) \cdot re\right) \cdot \frac{-1}{6} \]

      8. lift-*.f64 54.5

         \[\leadsto \left(\left(\left(im \cdot im\right) \cdot im\right) \cdot re\right) \cdot -0.16666666666666666 \]

   10. Applied rewrites 54.5%

       \[\leadsto \left(\left(\left(im \cdot im\right) \cdot im\right) \cdot re\right) \cdot -0.16666666666666666 \]

   if -9.99999999999999936e-50 < (*.f64 (*.f64 #s(literal 1/2 binary64) (sin.f64 re)) (-.f64 (exp.f64 (neg.f64 im)) (exp.f64 im)))

   1. Initial program 53.4%

      \[\left(0.5 \cdot \sin re\right) \cdot \left(e^{-im} - e^{im}\right) \]

   2. Taylor expanded in im around 0

      \[\leadsto \color{blue}{-1 \cdot \left(im \cdot \sin re\right)} \]
   3. Step-by-step derivation

      1. associate-*r* N/A

         \[\leadsto \left(-1 \cdot im\right) \cdot \color{blue}{\sin re} \]

      2. lower-*.f64 N/A

         \[\leadsto \left(-1 \cdot im\right) \cdot \color{blue}{\sin re} \]

      3. mul-1-neg N/A

         \[\leadsto \left(\mathsf{neg}\left(im\right)\right) \cdot \sin \color{blue}{re} \]

      4. lift-neg.f64 N/A

         \[\leadsto \left(-im\right) \cdot \sin \color{blue}{re} \]

      5. lift-sin.f64 68.4

         \[\leadsto \left(-im\right) \cdot \sin re \]

   4. Applied rewrites 68.4%

      \[\leadsto \color{blue}{\left(-im\right) \cdot \sin re} \]

   5. Taylor expanded in re around 0

      \[\leadsto re \cdot \color{blue}{\left(-1 \cdot im + \frac{1}{6} \cdot \left(im \cdot {re}^{2}\right)\right)} \]
   6. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(-1 \cdot im + \frac{1}{6} \cdot \left(im \cdot {re}^{2}\right)\right) \cdot re \]

      2. lower-*.f64 N/A

         \[\leadsto \left(-1 \cdot im + \frac{1}{6} \cdot \left(im \cdot {re}^{2}\right)\right) \cdot re \]

      3. +-commutative N/A

         \[\leadsto \left(\frac{1}{6} \cdot \left(im \cdot {re}^{2}\right) + -1 \cdot im\right) \cdot re \]

      4. *-commutative N/A

         \[\leadsto \left(\left(im \cdot {re}^{2}\right) \cdot \frac{1}{6} + -1 \cdot im\right) \cdot re \]

      5. lower-fma.f64 N/A

         \[\leadsto \mathsf{fma}\left(im \cdot {re}^{2}, \frac{1}{6}, -1 \cdot im\right) \cdot re \]

      6. *-commutative N/A

         \[\leadsto \mathsf{fma}\left({re}^{2} \cdot im, \frac{1}{6}, -1 \cdot im\right) \cdot re \]

      7. lower-*.f64 N/A

         \[\leadsto \mathsf{fma}\left({re}^{2} \cdot im, \frac{1}{6}, -1 \cdot im\right) \cdot re \]

      8. pow2 N/A

         \[\leadsto \mathsf{fma}\left(\left(re \cdot re\right) \cdot im, \frac{1}{6}, -1 \cdot im\right) \cdot re \]

      9. lift-*.f64 N/A

         \[\leadsto \mathsf{fma}\left(\left(re \cdot re\right) \cdot im, \frac{1}{6}, -1 \cdot im\right) \cdot re \]

      10. mul-1-neg N/A

          \[\leadsto \mathsf{fma}\left(\left(re \cdot re\right) \cdot im, \frac{1}{6}, \mathsf{neg}\left(im\right)\right) \cdot re \]

      11. lift-neg.f64 42.0

          \[\leadsto \mathsf{fma}\left(\left(re \cdot re\right) \cdot im, 0.16666666666666666, -im\right) \cdot re \]

   7. Applied rewrites 42.0%

      \[\leadsto \mathsf{fma}\left(\left(re \cdot re\right) \cdot im, 0.16666666666666666, -im\right) \cdot \color{blue}{re} \]

   8. Taylor expanded in im around 0

      \[\leadsto \left(im \cdot \left(\frac{1}{6} \cdot {re}^{2} - 1\right)\right) \cdot re \]
   9. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(\left(\frac{1}{6} \cdot {re}^{2} - 1\right) \cdot im\right) \cdot re \]

      2. lower-*.f64 N/A

         \[\leadsto \left(\left(\frac{1}{6} \cdot {re}^{2} - 1\right) \cdot im\right) \cdot re \]

      3. sub-flip N/A

         \[\leadsto \left(\left(\frac{1}{6} \cdot {re}^{2} + \left(\mathsf{neg}\left(1\right)\right)\right) \cdot im\right) \cdot re \]

      4. *-commutative N/A

         \[\leadsto \left(\left({re}^{2} \cdot \frac{1}{6} + \left(\mathsf{neg}\left(1\right)\right)\right) \cdot im\right) \cdot re \]

      5. metadata-eval N/A

         \[\leadsto \left(\left({re}^{2} \cdot \frac{1}{6} + -1\right) \cdot im\right) \cdot re \]

      6. lower-fma.f64 N/A

         \[\leadsto \left(\mathsf{fma}\left({re}^{2}, \frac{1}{6}, -1\right) \cdot im\right) \cdot re \]

      7. pow2 N/A

         \[\leadsto \left(\mathsf{fma}\left(re \cdot re, \frac{1}{6}, -1\right) \cdot im\right) \cdot re \]

      8. lift-*.f64 42.0

         \[\leadsto \left(\mathsf{fma}\left(re \cdot re, 0.16666666666666666, -1\right) \cdot im\right) \cdot re \]

   10. Applied rewrites 42.0%

       \[\leadsto \left(\mathsf{fma}\left(re \cdot re, 0.16666666666666666, -1\right) \cdot im\right) \cdot re \]
3. Recombined 2 regimes into one program.
4. Add Preprocessing

Alternative 15: 44.6% accurate, 0.8× speedup

Math

\[\begin{array}{l} im\_m = \left|im\right| \\ im\_s = \mathsf{copysign}\left(1, im\right) \\ im\_s \cdot \begin{array}{l} \mathbf{if}\;\left(0.5 \cdot \sin re\right) \cdot \left(e^{-im\_m} - e^{im\_m}\right) \leq -1 \cdot 10^{-49}:\\ \;\;\;\;\left(\left(\left(im\_m \cdot im\_m\right) \cdot im\_m\right) \cdot re\right) \cdot -0.16666666666666666\\ \mathbf{else}:\\ \;\;\;\;\left(\mathsf{fma}\left(re \cdot re, 0.16666666666666666, -1\right) \cdot re\right) \cdot im\_m\\ \end{array} \end{array} \]

FPCore

im\_m = (fabs.f64 im)
im\_s = (copysign.f64 #s(literal 1 binary64) im)
(FPCore (im_s re im_m)
 :precision binary64
 (*
  im_s
  (if (<= (* (* 0.5 (sin re)) (- (exp (- im_m)) (exp im_m))) -1e-49)
    (* (* (* (* im_m im_m) im_m) re) -0.16666666666666666)
    (* (* (fma (* re re) 0.16666666666666666 -1.0) re) im_m))))

C

im\_m = fabs(im);
im\_s = copysign(1.0, im);
double code(double im_s, double re, double im_m) {
	double tmp;
	if (((0.5 * sin(re)) * (exp(-im_m) - exp(im_m))) <= -1e-49) {
		tmp = (((im_m * im_m) * im_m) * re) * -0.16666666666666666;
	} else {
		tmp = (fma((re * re), 0.16666666666666666, -1.0) * re) * im_m;
	}
	return im_s * tmp;
}

Julia

im\_m = abs(im)
im\_s = copysign(1.0, im)
function code(im_s, re, im_m)
	tmp = 0.0
	if (Float64(Float64(0.5 * sin(re)) * Float64(exp(Float64(-im_m)) - exp(im_m))) <= -1e-49)
		tmp = Float64(Float64(Float64(Float64(im_m * im_m) * im_m) * re) * -0.16666666666666666);
	else
		tmp = Float64(Float64(fma(Float64(re * re), 0.16666666666666666, -1.0) * re) * im_m);
	end
	return Float64(im_s * tmp)
end

Wolfram

im\_m = N[Abs[im], $MachinePrecision]
im\_s = N[With[{TMP1 = Abs[1.0], TMP2 = Sign[im]}, TMP1 * If[TMP2 == 0, 1, TMP2]], $MachinePrecision]
code[im$95$s_, re_, im$95$m_] := N[(im$95$s * If[LessEqual[N[(N[(0.5 * N[Sin[re], $MachinePrecision]), $MachinePrecision] * N[(N[Exp[(-im$95$m)], $MachinePrecision] - N[Exp[im$95$m], $MachinePrecision]), $MachinePrecision]), $MachinePrecision], -1e-49], N[(N[(N[(N[(im$95$m * im$95$m), $MachinePrecision] * im$95$m), $MachinePrecision] * re), $MachinePrecision] * -0.16666666666666666), $MachinePrecision], N[(N[(N[(N[(re * re), $MachinePrecision] * 0.16666666666666666 + -1.0), $MachinePrecision] * re), $MachinePrecision] * im$95$m), $MachinePrecision]]), $MachinePrecision]

Derivation

1. Split input into 2 regimes

2. if (*.f64 (*.f64 #s(literal 1/2 binary64) (sin.f64 re)) (-.f64 (exp.f64 (neg.f64 im)) (exp.f64 im))) < -9.99999999999999936e-50

   1. Initial program 98.8%

      \[\left(0.5 \cdot \sin re\right) \cdot \left(e^{-im} - e^{im}\right) \]

   2. Taylor expanded in re around 0

      \[\leadsto \color{blue}{\frac{1}{2} \cdot \left(re \cdot \left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right)\right)} \]
   3. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(re \cdot \left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right)\right) \cdot \color{blue}{\frac{1}{2}} \]

      2. lower-*.f64 N/A

         \[\leadsto \left(re \cdot \left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right)\right) \cdot \color{blue}{\frac{1}{2}} \]

      3. *-commutative N/A

         \[\leadsto \left(\left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right) \cdot re\right) \cdot \frac{1}{2} \]

      4. lower-*.f64 N/A

         \[\leadsto \left(\left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right) \cdot re\right) \cdot \frac{1}{2} \]

      5. sub-negate N/A

         \[\leadsto \left(\left(\mathsf{neg}\left(\left(e^{im} - e^{\mathsf{neg}\left(im\right)}\right)\right)\right) \cdot re\right) \cdot \frac{1}{2} \]

      6. lower-neg.f64 N/A

         \[\leadsto \left(\left(-\left(e^{im} - e^{\mathsf{neg}\left(im\right)}\right)\right) \cdot re\right) \cdot \frac{1}{2} \]

      7. sinh-undef N/A

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot \frac{1}{2} \]

      8. lower-*.f64 N/A

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot \frac{1}{2} \]

      9. lower-sinh.f64 72.5

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot 0.5 \]

   4. Applied rewrites 72.5%

      \[\leadsto \color{blue}{\left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot 0.5} \]

   5. Taylor expanded in im around 0

      \[\leadsto im \cdot \color{blue}{\left(-1 \cdot re + \frac{-1}{6} \cdot \left({im}^{2} \cdot re\right)\right)} \]
   6. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(-1 \cdot re + \frac{-1}{6} \cdot \left({im}^{2} \cdot re\right)\right) \cdot im \]

      2. lower-*.f64 N/A

         \[\leadsto \left(-1 \cdot re + \frac{-1}{6} \cdot \left({im}^{2} \cdot re\right)\right) \cdot im \]

      3. +-commutative N/A

         \[\leadsto \left(\frac{-1}{6} \cdot \left({im}^{2} \cdot re\right) + -1 \cdot re\right) \cdot im \]

      4. *-commutative N/A

         \[\leadsto \left(\left({im}^{2} \cdot re\right) \cdot \frac{-1}{6} + -1 \cdot re\right) \cdot im \]

      5. lower-fma.f64 N/A

         \[\leadsto \mathsf{fma}\left({im}^{2} \cdot re, \frac{-1}{6}, -1 \cdot re\right) \cdot im \]

      6. lower-*.f64 N/A

         \[\leadsto \mathsf{fma}\left({im}^{2} \cdot re, \frac{-1}{6}, -1 \cdot re\right) \cdot im \]

      7. unpow2 N/A

         \[\leadsto \mathsf{fma}\left(\left(im \cdot im\right) \cdot re, \frac{-1}{6}, -1 \cdot re\right) \cdot im \]

      8. lower-*.f64 N/A

         \[\leadsto \mathsf{fma}\left(\left(im \cdot im\right) \cdot re, \frac{-1}{6}, -1 \cdot re\right) \cdot im \]

      9. mul-1-neg N/A

         \[\leadsto \mathsf{fma}\left(\left(im \cdot im\right) \cdot re, \frac{-1}{6}, \mathsf{neg}\left(re\right)\right) \cdot im \]

      10. lower-neg.f64 49.1

          \[\leadsto \mathsf{fma}\left(\left(im \cdot im\right) \cdot re, -0.16666666666666666, -re\right) \cdot im \]

   7. Applied rewrites 49.1%

      \[\leadsto \mathsf{fma}\left(\left(im \cdot im\right) \cdot re, -0.16666666666666666, -re\right) \cdot \color{blue}{im} \]

   8. Taylor expanded in im around inf

      \[\leadsto \frac{-1}{6} \cdot \left({im}^{3} \cdot \color{blue}{re}\right) \]
   9. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left({im}^{3} \cdot re\right) \cdot \frac{-1}{6} \]

      2. lower-*.f64 N/A

         \[\leadsto \left({im}^{3} \cdot re\right) \cdot \frac{-1}{6} \]

      3. lower-*.f64 N/A

         \[\leadsto \left({im}^{3} \cdot re\right) \cdot \frac{-1}{6} \]

      4. unpow3 N/A

         \[\leadsto \left(\left(\left(im \cdot im\right) \cdot im\right) \cdot re\right) \cdot \frac{-1}{6} \]

      5. pow2 N/A

         \[\leadsto \left(\left({im}^{2} \cdot im\right) \cdot re\right) \cdot \frac{-1}{6} \]

      6. lower-*.f64 N/A

         \[\leadsto \left(\left({im}^{2} \cdot im\right) \cdot re\right) \cdot \frac{-1}{6} \]

      7. pow2 N/A

         \[\leadsto \left(\left(\left(im \cdot im\right) \cdot im\right) \cdot re\right) \cdot \frac{-1}{6} \]

      8. lift-*.f64 54.5

         \[\leadsto \left(\left(\left(im \cdot im\right) \cdot im\right) \cdot re\right) \cdot -0.16666666666666666 \]

   10. Applied rewrites 54.5%

       \[\leadsto \left(\left(\left(im \cdot im\right) \cdot im\right) \cdot re\right) \cdot -0.16666666666666666 \]

   if -9.99999999999999936e-50 < (*.f64 (*.f64 #s(literal 1/2 binary64) (sin.f64 re)) (-.f64 (exp.f64 (neg.f64 im)) (exp.f64 im)))

   1. Initial program 53.4%

      \[\left(0.5 \cdot \sin re\right) \cdot \left(e^{-im} - e^{im}\right) \]

   2. Taylor expanded in im around 0

      \[\leadsto \color{blue}{-1 \cdot \left(im \cdot \sin re\right)} \]
   3. Step-by-step derivation

      1. associate-*r* N/A

         \[\leadsto \left(-1 \cdot im\right) \cdot \color{blue}{\sin re} \]

      2. lower-*.f64 N/A

         \[\leadsto \left(-1 \cdot im\right) \cdot \color{blue}{\sin re} \]

      3. mul-1-neg N/A

         \[\leadsto \left(\mathsf{neg}\left(im\right)\right) \cdot \sin \color{blue}{re} \]

      4. lift-neg.f64 N/A

         \[\leadsto \left(-im\right) \cdot \sin \color{blue}{re} \]

      5. lift-sin.f64 68.4

         \[\leadsto \left(-im\right) \cdot \sin re \]

   4. Applied rewrites 68.4%

      \[\leadsto \color{blue}{\left(-im\right) \cdot \sin re} \]

   5. Taylor expanded in re around 0

      \[\leadsto re \cdot \color{blue}{\left(-1 \cdot im + \frac{1}{6} \cdot \left(im \cdot {re}^{2}\right)\right)} \]
   6. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(-1 \cdot im + \frac{1}{6} \cdot \left(im \cdot {re}^{2}\right)\right) \cdot re \]

      2. lower-*.f64 N/A

         \[\leadsto \left(-1 \cdot im + \frac{1}{6} \cdot \left(im \cdot {re}^{2}\right)\right) \cdot re \]

      3. +-commutative N/A

         \[\leadsto \left(\frac{1}{6} \cdot \left(im \cdot {re}^{2}\right) + -1 \cdot im\right) \cdot re \]

      4. *-commutative N/A

         \[\leadsto \left(\left(im \cdot {re}^{2}\right) \cdot \frac{1}{6} + -1 \cdot im\right) \cdot re \]

      5. lower-fma.f64 N/A

         \[\leadsto \mathsf{fma}\left(im \cdot {re}^{2}, \frac{1}{6}, -1 \cdot im\right) \cdot re \]

      6. *-commutative N/A

         \[\leadsto \mathsf{fma}\left({re}^{2} \cdot im, \frac{1}{6}, -1 \cdot im\right) \cdot re \]

      7. lower-*.f64 N/A

         \[\leadsto \mathsf{fma}\left({re}^{2} \cdot im, \frac{1}{6}, -1 \cdot im\right) \cdot re \]

      8. pow2 N/A

         \[\leadsto \mathsf{fma}\left(\left(re \cdot re\right) \cdot im, \frac{1}{6}, -1 \cdot im\right) \cdot re \]

      9. lift-*.f64 N/A

         \[\leadsto \mathsf{fma}\left(\left(re \cdot re\right) \cdot im, \frac{1}{6}, -1 \cdot im\right) \cdot re \]

      10. mul-1-neg N/A

          \[\leadsto \mathsf{fma}\left(\left(re \cdot re\right) \cdot im, \frac{1}{6}, \mathsf{neg}\left(im\right)\right) \cdot re \]

      11. lift-neg.f64 42.0

          \[\leadsto \mathsf{fma}\left(\left(re \cdot re\right) \cdot im, 0.16666666666666666, -im\right) \cdot re \]

   7. Applied rewrites 42.0%

      \[\leadsto \mathsf{fma}\left(\left(re \cdot re\right) \cdot im, 0.16666666666666666, -im\right) \cdot \color{blue}{re} \]

   8. Taylor expanded in im around 0

      \[\leadsto im \cdot \left(re \cdot \color{blue}{\left(\frac{1}{6} \cdot {re}^{2} - 1\right)}\right) \]
   9. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(re \cdot \left(\frac{1}{6} \cdot {re}^{2} - 1\right)\right) \cdot im \]

      2. lower-*.f64 N/A

         \[\leadsto \left(re \cdot \left(\frac{1}{6} \cdot {re}^{2} - 1\right)\right) \cdot im \]

      3. *-commutative N/A

         \[\leadsto \left(\left(\frac{1}{6} \cdot {re}^{2} - 1\right) \cdot re\right) \cdot im \]

      4. lower-*.f64 N/A

         \[\leadsto \left(\left(\frac{1}{6} \cdot {re}^{2} - 1\right) \cdot re\right) \cdot im \]

      5. sub-flip N/A

         \[\leadsto \left(\left(\frac{1}{6} \cdot {re}^{2} + \left(\mathsf{neg}\left(1\right)\right)\right) \cdot re\right) \cdot im \]

      6. *-commutative N/A

         \[\leadsto \left(\left({re}^{2} \cdot \frac{1}{6} + \left(\mathsf{neg}\left(1\right)\right)\right) \cdot re\right) \cdot im \]

      7. metadata-eval N/A

         \[\leadsto \left(\left({re}^{2} \cdot \frac{1}{6} + -1\right) \cdot re\right) \cdot im \]

      8. lower-fma.f64 N/A

         \[\leadsto \left(\mathsf{fma}\left({re}^{2}, \frac{1}{6}, -1\right) \cdot re\right) \cdot im \]

      9. pow2 N/A

         \[\leadsto \left(\mathsf{fma}\left(re \cdot re, \frac{1}{6}, -1\right) \cdot re\right) \cdot im \]

      10. lift-*.f64 42.0

          \[\leadsto \left(\mathsf{fma}\left(re \cdot re, 0.16666666666666666, -1\right) \cdot re\right) \cdot im \]

   10. Applied rewrites 42.0%

       \[\leadsto \left(\mathsf{fma}\left(re \cdot re, 0.16666666666666666, -1\right) \cdot re\right) \cdot im \]
3. Recombined 2 regimes into one program.
4. Add Preprocessing

Alternative 16: 35.1% accurate, 1.2× speedup

Math

\[\begin{array}{l} im\_m = \left|im\right| \\ im\_s = \mathsf{copysign}\left(1, im\right) \\ im\_s \cdot \begin{array}{l} \mathbf{if}\;0.5 \cdot \sin re \leq -0.02:\\ \;\;\;\;\left(\left(\left(re \cdot re\right) \cdot im\_m\right) \cdot 0.16666666666666666\right) \cdot re\\ \mathbf{else}:\\ \;\;\;\;\left(-im\_m\right) \cdot re\\ \end{array} \end{array} \]

FPCore

im\_m = (fabs.f64 im)
im\_s = (copysign.f64 #s(literal 1 binary64) im)
(FPCore (im_s re im_m)
 :precision binary64
 (*
  im_s
  (if (<= (* 0.5 (sin re)) -0.02)
    (* (* (* (* re re) im_m) 0.16666666666666666) re)
    (* (- im_m) re))))

C

im\_m = fabs(im);
im\_s = copysign(1.0, im);
double code(double im_s, double re, double im_m) {
	double tmp;
	if ((0.5 * sin(re)) <= -0.02) {
		tmp = (((re * re) * im_m) * 0.16666666666666666) * re;
	} else {
		tmp = -im_m * re;
	}
	return im_s * tmp;
}

Fortran

im\_m =     private
im\_s =     private
module fmin_fmax_functions
    implicit none
    private
    public fmax
    public fmin

    interface fmax
        module procedure fmax88
        module procedure fmax44
        module procedure fmax84
        module procedure fmax48
    end interface
    interface fmin
        module procedure fmin88
        module procedure fmin44
        module procedure fmin84
        module procedure fmin48
    end interface
contains
    real(8) function fmax88(x, y) result (res)
        real(8), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(x, max(x, y), y /= y), x /= x)
    end function
    real(4) function fmax44(x, y) result (res)
        real(4), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(y, merge(x, max(x, y), y /= y), x /= x)
    end function
    real(8) function fmax84(x, y) result(res)
        real(8), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(dble(y), merge(x, max(x, dble(y)), y /= y), x /= x)
    end function
    real(8) function fmax48(x, y) result(res)
        real(4), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(dble(x), max(dble(x), y), y /= y), x /= x)
    end function
    real(8) function fmin88(x, y) result (res)
        real(8), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(x, min(x, y), y /= y), x /= x)
    end function
    real(4) function fmin44(x, y) result (res)
        real(4), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(y, merge(x, min(x, y), y /= y), x /= x)
    end function
    real(8) function fmin84(x, y) result(res)
        real(8), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(dble(y), merge(x, min(x, dble(y)), y /= y), x /= x)
    end function
    real(8) function fmin48(x, y) result(res)
        real(4), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(dble(x), min(dble(x), y), y /= y), x /= x)
    end function
end module

real(8) function code(im_s, re, im_m)
use fmin_fmax_functions
    real(8), intent (in) :: im_s
    real(8), intent (in) :: re
    real(8), intent (in) :: im_m
    real(8) :: tmp
    if ((0.5d0 * sin(re)) <= (-0.02d0)) then
        tmp = (((re * re) * im_m) * 0.16666666666666666d0) * re
    else
        tmp = -im_m * re
    end if
    code = im_s * tmp
end function

Java

im\_m = Math.abs(im);
im\_s = Math.copySign(1.0, im);
public static double code(double im_s, double re, double im_m) {
	double tmp;
	if ((0.5 * Math.sin(re)) <= -0.02) {
		tmp = (((re * re) * im_m) * 0.16666666666666666) * re;
	} else {
		tmp = -im_m * re;
	}
	return im_s * tmp;
}

Python

im\_m = math.fabs(im)
im\_s = math.copysign(1.0, im)
def code(im_s, re, im_m):
	tmp = 0
	if (0.5 * math.sin(re)) <= -0.02:
		tmp = (((re * re) * im_m) * 0.16666666666666666) * re
	else:
		tmp = -im_m * re
	return im_s * tmp

Julia

im\_m = abs(im)
im\_s = copysign(1.0, im)
function code(im_s, re, im_m)
	tmp = 0.0
	if (Float64(0.5 * sin(re)) <= -0.02)
		tmp = Float64(Float64(Float64(Float64(re * re) * im_m) * 0.16666666666666666) * re);
	else
		tmp = Float64(Float64(-im_m) * re);
	end
	return Float64(im_s * tmp)
end

MATLAB

im\_m = abs(im);
im\_s = sign(im) * abs(1.0);
function tmp_2 = code(im_s, re, im_m)
	tmp = 0.0;
	if ((0.5 * sin(re)) <= -0.02)
		tmp = (((re * re) * im_m) * 0.16666666666666666) * re;
	else
		tmp = -im_m * re;
	end
	tmp_2 = im_s * tmp;
end

Wolfram

im\_m = N[Abs[im], $MachinePrecision]
im\_s = N[With[{TMP1 = Abs[1.0], TMP2 = Sign[im]}, TMP1 * If[TMP2 == 0, 1, TMP2]], $MachinePrecision]
code[im$95$s_, re_, im$95$m_] := N[(im$95$s * If[LessEqual[N[(0.5 * N[Sin[re], $MachinePrecision]), $MachinePrecision], -0.02], N[(N[(N[(N[(re * re), $MachinePrecision] * im$95$m), $MachinePrecision] * 0.16666666666666666), $MachinePrecision] * re), $MachinePrecision], N[((-im$95$m) * re), $MachinePrecision]]), $MachinePrecision]

Derivation

1. Split input into 2 regimes

2. if (*.f64 #s(literal 1/2 binary64) (sin.f64 re)) < -0.0200000000000000004

   1. Initial program 53.7%

      \[\left(0.5 \cdot \sin re\right) \cdot \left(e^{-im} - e^{im}\right) \]

   2. Taylor expanded in im around 0

      \[\leadsto \color{blue}{-1 \cdot \left(im \cdot \sin re\right)} \]
   3. Step-by-step derivation

      1. associate-*r* N/A

         \[\leadsto \left(-1 \cdot im\right) \cdot \color{blue}{\sin re} \]

      2. lower-*.f64 N/A

         \[\leadsto \left(-1 \cdot im\right) \cdot \color{blue}{\sin re} \]

      3. mul-1-neg N/A

         \[\leadsto \left(\mathsf{neg}\left(im\right)\right) \cdot \sin \color{blue}{re} \]

      4. lift-neg.f64 N/A

         \[\leadsto \left(-im\right) \cdot \sin \color{blue}{re} \]

      5. lift-sin.f64 52.8

         \[\leadsto \left(-im\right) \cdot \sin re \]

   4. Applied rewrites 52.8%

      \[\leadsto \color{blue}{\left(-im\right) \cdot \sin re} \]

   5. Taylor expanded in re around 0

      \[\leadsto re \cdot \color{blue}{\left(-1 \cdot im + \frac{1}{6} \cdot \left(im \cdot {re}^{2}\right)\right)} \]
   6. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(-1 \cdot im + \frac{1}{6} \cdot \left(im \cdot {re}^{2}\right)\right) \cdot re \]

      2. lower-*.f64 N/A

         \[\leadsto \left(-1 \cdot im + \frac{1}{6} \cdot \left(im \cdot {re}^{2}\right)\right) \cdot re \]

      3. +-commutative N/A

         \[\leadsto \left(\frac{1}{6} \cdot \left(im \cdot {re}^{2}\right) + -1 \cdot im\right) \cdot re \]

      4. *-commutative N/A

         \[\leadsto \left(\left(im \cdot {re}^{2}\right) \cdot \frac{1}{6} + -1 \cdot im\right) \cdot re \]

      5. lower-fma.f64 N/A

         \[\leadsto \mathsf{fma}\left(im \cdot {re}^{2}, \frac{1}{6}, -1 \cdot im\right) \cdot re \]

      6. *-commutative N/A

         \[\leadsto \mathsf{fma}\left({re}^{2} \cdot im, \frac{1}{6}, -1 \cdot im\right) \cdot re \]

      7. lower-*.f64 N/A

         \[\leadsto \mathsf{fma}\left({re}^{2} \cdot im, \frac{1}{6}, -1 \cdot im\right) \cdot re \]

      8. pow2 N/A

         \[\leadsto \mathsf{fma}\left(\left(re \cdot re\right) \cdot im, \frac{1}{6}, -1 \cdot im\right) \cdot re \]

      9. lift-*.f64 N/A

         \[\leadsto \mathsf{fma}\left(\left(re \cdot re\right) \cdot im, \frac{1}{6}, -1 \cdot im\right) \cdot re \]

      10. mul-1-neg N/A

          \[\leadsto \mathsf{fma}\left(\left(re \cdot re\right) \cdot im, \frac{1}{6}, \mathsf{neg}\left(im\right)\right) \cdot re \]

      11. lift-neg.f64 22.2

          \[\leadsto \mathsf{fma}\left(\left(re \cdot re\right) \cdot im, 0.16666666666666666, -im\right) \cdot re \]

   7. Applied rewrites 22.2%

      \[\leadsto \mathsf{fma}\left(\left(re \cdot re\right) \cdot im, 0.16666666666666666, -im\right) \cdot \color{blue}{re} \]

   8. Taylor expanded in re around inf

      \[\leadsto \left(\frac{1}{6} \cdot \left(im \cdot {re}^{2}\right)\right) \cdot re \]
   9. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(\left(im \cdot {re}^{2}\right) \cdot \frac{1}{6}\right) \cdot re \]

      2. lower-*.f64 N/A

         \[\leadsto \left(\left(im \cdot {re}^{2}\right) \cdot \frac{1}{6}\right) \cdot re \]

      3. *-commutative N/A

         \[\leadsto \left(\left({re}^{2} \cdot im\right) \cdot \frac{1}{6}\right) \cdot re \]

      4. pow2 N/A

         \[\leadsto \left(\left(\left(re \cdot re\right) \cdot im\right) \cdot \frac{1}{6}\right) \cdot re \]

      5. lift-*.f64 N/A

         \[\leadsto \left(\left(\left(re \cdot re\right) \cdot im\right) \cdot \frac{1}{6}\right) \cdot re \]

      6. lift-*.f64 22.1

         \[\leadsto \left(\left(\left(re \cdot re\right) \cdot im\right) \cdot 0.16666666666666666\right) \cdot re \]

   10. Applied rewrites 22.1%

       \[\leadsto \left(\left(\left(re \cdot re\right) \cdot im\right) \cdot 0.16666666666666666\right) \cdot re \]

   if -0.0200000000000000004 < (*.f64 #s(literal 1/2 binary64) (sin.f64 re))

   1. Initial program 68.2%

      \[\left(0.5 \cdot \sin re\right) \cdot \left(e^{-im} - e^{im}\right) \]

   2. Taylor expanded in re around 0

      \[\leadsto \color{blue}{\frac{1}{2} \cdot \left(re \cdot \left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right)\right)} \]
   3. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(re \cdot \left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right)\right) \cdot \color{blue}{\frac{1}{2}} \]

      2. lower-*.f64 N/A

         \[\leadsto \left(re \cdot \left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right)\right) \cdot \color{blue}{\frac{1}{2}} \]

      3. *-commutative N/A

         \[\leadsto \left(\left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right) \cdot re\right) \cdot \frac{1}{2} \]

      4. lower-*.f64 N/A

         \[\leadsto \left(\left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right) \cdot re\right) \cdot \frac{1}{2} \]

      5. sub-negate N/A

         \[\leadsto \left(\left(\mathsf{neg}\left(\left(e^{im} - e^{\mathsf{neg}\left(im\right)}\right)\right)\right) \cdot re\right) \cdot \frac{1}{2} \]

      6. lower-neg.f64 N/A

         \[\leadsto \left(\left(-\left(e^{im} - e^{\mathsf{neg}\left(im\right)}\right)\right) \cdot re\right) \cdot \frac{1}{2} \]

      7. sinh-undef N/A

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot \frac{1}{2} \]

      8. lower-*.f64 N/A

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot \frac{1}{2} \]

      9. lower-sinh.f64 74.2

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot 0.5 \]

   4. Applied rewrites 74.2%

      \[\leadsto \color{blue}{\left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot 0.5} \]

   5. Taylor expanded in im around 0

      \[\leadsto -1 \cdot \color{blue}{\left(im \cdot re\right)} \]
   6. Step-by-step derivation

      1. associate-*r* N/A

         \[\leadsto \left(-1 \cdot im\right) \cdot re \]

      2. lower-*.f64 N/A

         \[\leadsto \left(-1 \cdot im\right) \cdot re \]

      3. mul-1-neg N/A

         \[\leadsto \left(\mathsf{neg}\left(im\right)\right) \cdot re \]

      4. lift-neg.f64 39.1

         \[\leadsto \left(-im\right) \cdot re \]

   7. Applied rewrites 39.1%

      \[\leadsto \left(-im\right) \cdot \color{blue}{re} \]
3. Recombined 2 regimes into one program.
4. Add Preprocessing

Alternative 17: 35.1% accurate, 1.2× speedup

Math

\[\begin{array}{l} im\_m = \left|im\right| \\ im\_s = \mathsf{copysign}\left(1, im\right) \\ im\_s \cdot \begin{array}{l} \mathbf{if}\;0.5 \cdot \sin re \leq -0.02:\\ \;\;\;\;\left(\left(\left(re \cdot re\right) \cdot re\right) \cdot im\_m\right) \cdot 0.16666666666666666\\ \mathbf{else}:\\ \;\;\;\;\left(-im\_m\right) \cdot re\\ \end{array} \end{array} \]

FPCore

im\_m = (fabs.f64 im)
im\_s = (copysign.f64 #s(literal 1 binary64) im)
(FPCore (im_s re im_m)
 :precision binary64
 (*
  im_s
  (if (<= (* 0.5 (sin re)) -0.02)
    (* (* (* (* re re) re) im_m) 0.16666666666666666)
    (* (- im_m) re))))

C

im\_m = fabs(im);
im\_s = copysign(1.0, im);
double code(double im_s, double re, double im_m) {
	double tmp;
	if ((0.5 * sin(re)) <= -0.02) {
		tmp = (((re * re) * re) * im_m) * 0.16666666666666666;
	} else {
		tmp = -im_m * re;
	}
	return im_s * tmp;
}

Fortran

im\_m =     private
im\_s =     private
module fmin_fmax_functions
    implicit none
    private
    public fmax
    public fmin

    interface fmax
        module procedure fmax88
        module procedure fmax44
        module procedure fmax84
        module procedure fmax48
    end interface
    interface fmin
        module procedure fmin88
        module procedure fmin44
        module procedure fmin84
        module procedure fmin48
    end interface
contains
    real(8) function fmax88(x, y) result (res)
        real(8), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(x, max(x, y), y /= y), x /= x)
    end function
    real(4) function fmax44(x, y) result (res)
        real(4), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(y, merge(x, max(x, y), y /= y), x /= x)
    end function
    real(8) function fmax84(x, y) result(res)
        real(8), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(dble(y), merge(x, max(x, dble(y)), y /= y), x /= x)
    end function
    real(8) function fmax48(x, y) result(res)
        real(4), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(dble(x), max(dble(x), y), y /= y), x /= x)
    end function
    real(8) function fmin88(x, y) result (res)
        real(8), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(x, min(x, y), y /= y), x /= x)
    end function
    real(4) function fmin44(x, y) result (res)
        real(4), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(y, merge(x, min(x, y), y /= y), x /= x)
    end function
    real(8) function fmin84(x, y) result(res)
        real(8), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(dble(y), merge(x, min(x, dble(y)), y /= y), x /= x)
    end function
    real(8) function fmin48(x, y) result(res)
        real(4), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(dble(x), min(dble(x), y), y /= y), x /= x)
    end function
end module

real(8) function code(im_s, re, im_m)
use fmin_fmax_functions
    real(8), intent (in) :: im_s
    real(8), intent (in) :: re
    real(8), intent (in) :: im_m
    real(8) :: tmp
    if ((0.5d0 * sin(re)) <= (-0.02d0)) then
        tmp = (((re * re) * re) * im_m) * 0.16666666666666666d0
    else
        tmp = -im_m * re
    end if
    code = im_s * tmp
end function

Java

im\_m = Math.abs(im);
im\_s = Math.copySign(1.0, im);
public static double code(double im_s, double re, double im_m) {
	double tmp;
	if ((0.5 * Math.sin(re)) <= -0.02) {
		tmp = (((re * re) * re) * im_m) * 0.16666666666666666;
	} else {
		tmp = -im_m * re;
	}
	return im_s * tmp;
}

Python

im\_m = math.fabs(im)
im\_s = math.copysign(1.0, im)
def code(im_s, re, im_m):
	tmp = 0
	if (0.5 * math.sin(re)) <= -0.02:
		tmp = (((re * re) * re) * im_m) * 0.16666666666666666
	else:
		tmp = -im_m * re
	return im_s * tmp

Julia

im\_m = abs(im)
im\_s = copysign(1.0, im)
function code(im_s, re, im_m)
	tmp = 0.0
	if (Float64(0.5 * sin(re)) <= -0.02)
		tmp = Float64(Float64(Float64(Float64(re * re) * re) * im_m) * 0.16666666666666666);
	else
		tmp = Float64(Float64(-im_m) * re);
	end
	return Float64(im_s * tmp)
end

MATLAB

im\_m = abs(im);
im\_s = sign(im) * abs(1.0);
function tmp_2 = code(im_s, re, im_m)
	tmp = 0.0;
	if ((0.5 * sin(re)) <= -0.02)
		tmp = (((re * re) * re) * im_m) * 0.16666666666666666;
	else
		tmp = -im_m * re;
	end
	tmp_2 = im_s * tmp;
end

Wolfram

im\_m = N[Abs[im], $MachinePrecision]
im\_s = N[With[{TMP1 = Abs[1.0], TMP2 = Sign[im]}, TMP1 * If[TMP2 == 0, 1, TMP2]], $MachinePrecision]
code[im$95$s_, re_, im$95$m_] := N[(im$95$s * If[LessEqual[N[(0.5 * N[Sin[re], $MachinePrecision]), $MachinePrecision], -0.02], N[(N[(N[(N[(re * re), $MachinePrecision] * re), $MachinePrecision] * im$95$m), $MachinePrecision] * 0.16666666666666666), $MachinePrecision], N[((-im$95$m) * re), $MachinePrecision]]), $MachinePrecision]

Derivation

1. Split input into 2 regimes

2. if (*.f64 #s(literal 1/2 binary64) (sin.f64 re)) < -0.0200000000000000004

   1. Initial program 53.7%

      \[\left(0.5 \cdot \sin re\right) \cdot \left(e^{-im} - e^{im}\right) \]

   2. Taylor expanded in im around 0

      \[\leadsto \color{blue}{-1 \cdot \left(im \cdot \sin re\right)} \]
   3. Step-by-step derivation

      1. associate-*r* N/A

         \[\leadsto \left(-1 \cdot im\right) \cdot \color{blue}{\sin re} \]

      2. lower-*.f64 N/A

         \[\leadsto \left(-1 \cdot im\right) \cdot \color{blue}{\sin re} \]

      3. mul-1-neg N/A

         \[\leadsto \left(\mathsf{neg}\left(im\right)\right) \cdot \sin \color{blue}{re} \]

      4. lift-neg.f64 N/A

         \[\leadsto \left(-im\right) \cdot \sin \color{blue}{re} \]

      5. lift-sin.f64 52.8

         \[\leadsto \left(-im\right) \cdot \sin re \]

   4. Applied rewrites 52.8%

      \[\leadsto \color{blue}{\left(-im\right) \cdot \sin re} \]

   5. Taylor expanded in re around 0

      \[\leadsto re \cdot \color{blue}{\left(-1 \cdot im + \frac{1}{6} \cdot \left(im \cdot {re}^{2}\right)\right)} \]
   6. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(-1 \cdot im + \frac{1}{6} \cdot \left(im \cdot {re}^{2}\right)\right) \cdot re \]

      2. lower-*.f64 N/A

         \[\leadsto \left(-1 \cdot im + \frac{1}{6} \cdot \left(im \cdot {re}^{2}\right)\right) \cdot re \]

      3. +-commutative N/A

         \[\leadsto \left(\frac{1}{6} \cdot \left(im \cdot {re}^{2}\right) + -1 \cdot im\right) \cdot re \]

      4. *-commutative N/A

         \[\leadsto \left(\left(im \cdot {re}^{2}\right) \cdot \frac{1}{6} + -1 \cdot im\right) \cdot re \]

      5. lower-fma.f64 N/A

         \[\leadsto \mathsf{fma}\left(im \cdot {re}^{2}, \frac{1}{6}, -1 \cdot im\right) \cdot re \]

      6. *-commutative N/A

         \[\leadsto \mathsf{fma}\left({re}^{2} \cdot im, \frac{1}{6}, -1 \cdot im\right) \cdot re \]

      7. lower-*.f64 N/A

         \[\leadsto \mathsf{fma}\left({re}^{2} \cdot im, \frac{1}{6}, -1 \cdot im\right) \cdot re \]

      8. pow2 N/A

         \[\leadsto \mathsf{fma}\left(\left(re \cdot re\right) \cdot im, \frac{1}{6}, -1 \cdot im\right) \cdot re \]

      9. lift-*.f64 N/A

         \[\leadsto \mathsf{fma}\left(\left(re \cdot re\right) \cdot im, \frac{1}{6}, -1 \cdot im\right) \cdot re \]

      10. mul-1-neg N/A

          \[\leadsto \mathsf{fma}\left(\left(re \cdot re\right) \cdot im, \frac{1}{6}, \mathsf{neg}\left(im\right)\right) \cdot re \]

      11. lift-neg.f64 22.2

          \[\leadsto \mathsf{fma}\left(\left(re \cdot re\right) \cdot im, 0.16666666666666666, -im\right) \cdot re \]

   7. Applied rewrites 22.2%

      \[\leadsto \mathsf{fma}\left(\left(re \cdot re\right) \cdot im, 0.16666666666666666, -im\right) \cdot \color{blue}{re} \]

   8. Taylor expanded in re around inf

      \[\leadsto \frac{1}{6} \cdot \left(im \cdot \color{blue}{{re}^{3}}\right) \]
   9. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(im \cdot {re}^{3}\right) \cdot \frac{1}{6} \]

      2. lower-*.f64 N/A

         \[\leadsto \left(im \cdot {re}^{3}\right) \cdot \frac{1}{6} \]

      3. *-commutative N/A

         \[\leadsto \left({re}^{3} \cdot im\right) \cdot \frac{1}{6} \]

      4. lower-*.f64 N/A

         \[\leadsto \left({re}^{3} \cdot im\right) \cdot \frac{1}{6} \]

      5. unpow3 N/A

         \[\leadsto \left(\left(\left(re \cdot re\right) \cdot re\right) \cdot im\right) \cdot \frac{1}{6} \]

      6. pow2 N/A

         \[\leadsto \left(\left({re}^{2} \cdot re\right) \cdot im\right) \cdot \frac{1}{6} \]

      7. lower-*.f64 N/A

         \[\leadsto \left(\left({re}^{2} \cdot re\right) \cdot im\right) \cdot \frac{1}{6} \]

      8. pow2 N/A

         \[\leadsto \left(\left(\left(re \cdot re\right) \cdot re\right) \cdot im\right) \cdot \frac{1}{6} \]

      9. lift-*.f64 22.0

         \[\leadsto \left(\left(\left(re \cdot re\right) \cdot re\right) \cdot im\right) \cdot 0.16666666666666666 \]

   10. Applied rewrites 22.0%

       \[\leadsto \left(\left(\left(re \cdot re\right) \cdot re\right) \cdot im\right) \cdot 0.16666666666666666 \]

   if -0.0200000000000000004 < (*.f64 #s(literal 1/2 binary64) (sin.f64 re))

   1. Initial program 68.2%

      \[\left(0.5 \cdot \sin re\right) \cdot \left(e^{-im} - e^{im}\right) \]

   2. Taylor expanded in re around 0

      \[\leadsto \color{blue}{\frac{1}{2} \cdot \left(re \cdot \left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right)\right)} \]
   3. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(re \cdot \left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right)\right) \cdot \color{blue}{\frac{1}{2}} \]

      2. lower-*.f64 N/A

         \[\leadsto \left(re \cdot \left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right)\right) \cdot \color{blue}{\frac{1}{2}} \]

      3. *-commutative N/A

         \[\leadsto \left(\left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right) \cdot re\right) \cdot \frac{1}{2} \]

      4. lower-*.f64 N/A

         \[\leadsto \left(\left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right) \cdot re\right) \cdot \frac{1}{2} \]

      5. sub-negate N/A

         \[\leadsto \left(\left(\mathsf{neg}\left(\left(e^{im} - e^{\mathsf{neg}\left(im\right)}\right)\right)\right) \cdot re\right) \cdot \frac{1}{2} \]

      6. lower-neg.f64 N/A

         \[\leadsto \left(\left(-\left(e^{im} - e^{\mathsf{neg}\left(im\right)}\right)\right) \cdot re\right) \cdot \frac{1}{2} \]

      7. sinh-undef N/A

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot \frac{1}{2} \]

      8. lower-*.f64 N/A

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot \frac{1}{2} \]

      9. lower-sinh.f64 74.2

         \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot 0.5 \]

   4. Applied rewrites 74.2%

      \[\leadsto \color{blue}{\left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot 0.5} \]

   5. Taylor expanded in im around 0

      \[\leadsto -1 \cdot \color{blue}{\left(im \cdot re\right)} \]
   6. Step-by-step derivation

      1. associate-*r* N/A

         \[\leadsto \left(-1 \cdot im\right) \cdot re \]

      2. lower-*.f64 N/A

         \[\leadsto \left(-1 \cdot im\right) \cdot re \]

      3. mul-1-neg N/A

         \[\leadsto \left(\mathsf{neg}\left(im\right)\right) \cdot re \]

      4. lift-neg.f64 39.1

         \[\leadsto \left(-im\right) \cdot re \]

   7. Applied rewrites 39.1%

      \[\leadsto \left(-im\right) \cdot \color{blue}{re} \]
3. Recombined 2 regimes into one program.
4. Add Preprocessing

Alternative 18: 33.4% accurate, 12.7× speedup

Math

\[\begin{array}{l} im\_m = \left|im\right| \\ im\_s = \mathsf{copysign}\left(1, im\right) \\ im\_s \cdot \left(\left(-im\_m\right) \cdot re\right) \end{array} \]

FPCore

im\_m = (fabs.f64 im)
im\_s = (copysign.f64 #s(literal 1 binary64) im)
(FPCore (im_s re im_m) :precision binary64 (* im_s (* (- im_m) re)))

C

im\_m = fabs(im);
im\_s = copysign(1.0, im);
double code(double im_s, double re, double im_m) {
	return im_s * (-im_m * re);
}

Fortran

im\_m =     private
im\_s =     private
module fmin_fmax_functions
    implicit none
    private
    public fmax
    public fmin

    interface fmax
        module procedure fmax88
        module procedure fmax44
        module procedure fmax84
        module procedure fmax48
    end interface
    interface fmin
        module procedure fmin88
        module procedure fmin44
        module procedure fmin84
        module procedure fmin48
    end interface
contains
    real(8) function fmax88(x, y) result (res)
        real(8), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(x, max(x, y), y /= y), x /= x)
    end function
    real(4) function fmax44(x, y) result (res)
        real(4), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(y, merge(x, max(x, y), y /= y), x /= x)
    end function
    real(8) function fmax84(x, y) result(res)
        real(8), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(dble(y), merge(x, max(x, dble(y)), y /= y), x /= x)
    end function
    real(8) function fmax48(x, y) result(res)
        real(4), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(dble(x), max(dble(x), y), y /= y), x /= x)
    end function
    real(8) function fmin88(x, y) result (res)
        real(8), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(x, min(x, y), y /= y), x /= x)
    end function
    real(4) function fmin44(x, y) result (res)
        real(4), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(y, merge(x, min(x, y), y /= y), x /= x)
    end function
    real(8) function fmin84(x, y) result(res)
        real(8), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(dble(y), merge(x, min(x, dble(y)), y /= y), x /= x)
    end function
    real(8) function fmin48(x, y) result(res)
        real(4), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(dble(x), min(dble(x), y), y /= y), x /= x)
    end function
end module

real(8) function code(im_s, re, im_m)
use fmin_fmax_functions
    real(8), intent (in) :: im_s
    real(8), intent (in) :: re
    real(8), intent (in) :: im_m
    code = im_s * (-im_m * re)
end function

Java

im\_m = Math.abs(im);
im\_s = Math.copySign(1.0, im);
public static double code(double im_s, double re, double im_m) {
	return im_s * (-im_m * re);
}

Python

im\_m = math.fabs(im)
im\_s = math.copysign(1.0, im)
def code(im_s, re, im_m):
	return im_s * (-im_m * re)

Julia

im\_m = abs(im)
im\_s = copysign(1.0, im)
function code(im_s, re, im_m)
	return Float64(im_s * Float64(Float64(-im_m) * re))
end

MATLAB

im\_m = abs(im);
im\_s = sign(im) * abs(1.0);
function tmp = code(im_s, re, im_m)
	tmp = im_s * (-im_m * re);
end

Wolfram

im\_m = N[Abs[im], $MachinePrecision]
im\_s = N[With[{TMP1 = Abs[1.0], TMP2 = Sign[im]}, TMP1 * If[TMP2 == 0, 1, TMP2]], $MachinePrecision]
code[im$95$s_, re_, im$95$m_] := N[(im$95$s * N[((-im$95$m) * re), $MachinePrecision]), $MachinePrecision]

Derivation

1. Initial program 64.7%

   \[\left(0.5 \cdot \sin re\right) \cdot \left(e^{-im} - e^{im}\right) \]

2. Taylor expanded in re around 0

   \[\leadsto \color{blue}{\frac{1}{2} \cdot \left(re \cdot \left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right)\right)} \]
3. Step-by-step derivation

   1. *-commutative N/A

      \[\leadsto \left(re \cdot \left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right)\right) \cdot \color{blue}{\frac{1}{2}} \]

   2. lower-*.f64 N/A

      \[\leadsto \left(re \cdot \left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right)\right) \cdot \color{blue}{\frac{1}{2}} \]

   3. *-commutative N/A

      \[\leadsto \left(\left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right) \cdot re\right) \cdot \frac{1}{2} \]

   4. lower-*.f64 N/A

      \[\leadsto \left(\left(e^{\mathsf{neg}\left(im\right)} - e^{im}\right) \cdot re\right) \cdot \frac{1}{2} \]

   5. sub-negate N/A

      \[\leadsto \left(\left(\mathsf{neg}\left(\left(e^{im} - e^{\mathsf{neg}\left(im\right)}\right)\right)\right) \cdot re\right) \cdot \frac{1}{2} \]

   6. lower-neg.f64 N/A

      \[\leadsto \left(\left(-\left(e^{im} - e^{\mathsf{neg}\left(im\right)}\right)\right) \cdot re\right) \cdot \frac{1}{2} \]

   7. sinh-undef N/A

      \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot \frac{1}{2} \]

   8. lower-*.f64 N/A

      \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot \frac{1}{2} \]

   9. lower-sinh.f64 62.8

      \[\leadsto \left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot 0.5 \]

4. Applied rewrites 62.8%

   \[\leadsto \color{blue}{\left(\left(-2 \cdot \sinh im\right) \cdot re\right) \cdot 0.5} \]

5. Taylor expanded in im around 0

   \[\leadsto -1 \cdot \color{blue}{\left(im \cdot re\right)} \]
6. Step-by-step derivation

   1. associate-*r* N/A

      \[\leadsto \left(-1 \cdot im\right) \cdot re \]

   2. lower-*.f64 N/A

      \[\leadsto \left(-1 \cdot im\right) \cdot re \]

   3. mul-1-neg N/A

      \[\leadsto \left(\mathsf{neg}\left(im\right)\right) \cdot re \]

   4. lift-neg.f64 33.4

      \[\leadsto \left(-im\right) \cdot re \]

7. Applied rewrites 33.4%

   \[\leadsto \left(-im\right) \cdot \color{blue}{re} \]
8. Add Preprocessing

Reproduce

herbie shell --seed 2025122 
(FPCore (re im)
  :name "math.cos on complex, imaginary part"
  :precision binary64
  (* (* 0.5 (sin re)) (- (exp (- im)) (exp im))))