Linear.Projection:inverseInfinitePerspective from linear-1.19.1.3

Percentage Accurate: 89.7% → 100.0%

Time: 7.6s

Alternatives: 7

Speedup: 1.4×

• 75.3% of points produce a very large (infinite) output. You may want to add a precondition.

Specification

Math

\[\left(x \cdot y - z \cdot y\right) \cdot t \]

FPCore

(FPCore (x y z t)
  :precision binary64
  (* (- (* x y) (* z y)) t))

C

double code(double x, double y, double z, double t) {
	return ((x * y) - (z * y)) * t;
}

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(x, y, z, t)
use fmin_fmax_functions
    real(8), intent (in) :: x
    real(8), intent (in) :: y
    real(8), intent (in) :: z
    real(8), intent (in) :: t
    code = ((x * y) - (z * y)) * t
end function

Java

public static double code(double x, double y, double z, double t) {
	return ((x * y) - (z * y)) * t;
}

Python

def code(x, y, z, t):
	return ((x * y) - (z * y)) * t

Julia

function code(x, y, z, t)
	return Float64(Float64(Float64(x * y) - Float64(z * y)) * t)
end

MATLAB

function tmp = code(x, y, z, t)
	tmp = ((x * y) - (z * y)) * t;
end

Wolfram

code[x_, y_, z_, t_] := N[(N[(N[(x * y), $MachinePrecision] - N[(z * y), $MachinePrecision]), $MachinePrecision] * t), $MachinePrecision]

Initial Program: 89.7% accurate, 1.0× speedup

Math

\[\left(x \cdot y - z \cdot y\right) \cdot t \]

FPCore

(FPCore (x y z t)
  :precision binary64
  (* (- (* x y) (* z y)) t))

C

double code(double x, double y, double z, double t) {
	return ((x * y) - (z * y)) * t;
}

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(x, y, z, t)
use fmin_fmax_functions
    real(8), intent (in) :: x
    real(8), intent (in) :: y
    real(8), intent (in) :: z
    real(8), intent (in) :: t
    code = ((x * y) - (z * y)) * t
end function

Java

public static double code(double x, double y, double z, double t) {
	return ((x * y) - (z * y)) * t;
}

Python

def code(x, y, z, t):
	return ((x * y) - (z * y)) * t

Julia

function code(x, y, z, t)
	return Float64(Float64(Float64(x * y) - Float64(z * y)) * t)
end

MATLAB

function tmp = code(x, y, z, t)
	tmp = ((x * y) - (z * y)) * t;
end

Wolfram

code[x_, y_, z_, t_] := N[(N[(N[(x * y), $MachinePrecision] - N[(z * y), $MachinePrecision]), $MachinePrecision] * t), $MachinePrecision]

Alternative 1: 100.0% accurate, 0.9× speedup

Math

\[\mathsf{134\_z0z1z2z3z4}\left(y, x, t, z, t\right) \]

FPCore

(FPCore (x y z t)
  :precision binary64
  (134-z0z1z2z3z4 y x t z t))

Derivation

1. Initial program 89.7%

   \[\left(x \cdot y - z \cdot y\right) \cdot t \]
2. Step-by-step derivation

   1. lift-*.f64 N/A

      \[\leadsto \color{blue}{\left(x \cdot y - z \cdot y\right) \cdot t} \]

   2. *-commutative N/A

      \[\leadsto \color{blue}{t \cdot \left(x \cdot y - z \cdot y\right)} \]

   3. lift--.f64 N/A

      \[\leadsto t \cdot \color{blue}{\left(x \cdot y - z \cdot y\right)} \]

   4. distribute-lft-out-- N/A

      \[\leadsto \color{blue}{t \cdot \left(x \cdot y\right) - t \cdot \left(z \cdot y\right)} \]

   5. lift-*.f64 N/A

      \[\leadsto t \cdot \color{blue}{\left(x \cdot y\right)} - t \cdot \left(z \cdot y\right) \]

   6. associate-*r* N/A

      \[\leadsto \color{blue}{\left(t \cdot x\right) \cdot y} - t \cdot \left(z \cdot y\right) \]

   7. lift-*.f64 N/A

      \[\leadsto \left(t \cdot x\right) \cdot y - t \cdot \color{blue}{\left(z \cdot y\right)} \]

   8. associate-*r* N/A

      \[\leadsto \left(t \cdot x\right) \cdot y - \color{blue}{\left(t \cdot z\right) \cdot y} \]

   9. distribute-rgt-out-- N/A

      \[\leadsto \color{blue}{y \cdot \left(t \cdot x - t \cdot z\right)} \]

   10. sub-flip-reverse N/A

       \[\leadsto y \cdot \color{blue}{\left(t \cdot x + \left(\mathsf{neg}\left(t \cdot z\right)\right)\right)} \]

   11. distribute-rgt-neg-out N/A

       \[\leadsto y \cdot \left(t \cdot x + \color{blue}{t \cdot \left(\mathsf{neg}\left(z\right)\right)}\right) \]

   12. distribute-lft-in N/A

       \[\leadsto y \cdot \color{blue}{\left(t \cdot \left(x + \left(\mathsf{neg}\left(z\right)\right)\right)\right)} \]

   13. sub-flip N/A

       \[\leadsto y \cdot \left(t \cdot \color{blue}{\left(x - z\right)}\right) \]

   14. distribute-rgt-out-- N/A

       \[\leadsto y \cdot \color{blue}{\left(x \cdot t - z \cdot t\right)} \]

   15. lower-134-z0z1z2z3z4 100.0%

       \[\leadsto \color{blue}{\mathsf{134\_z0z1z2z3z4}\left(y, x, t, z, t\right)} \]

3. Applied rewrites 100.0%

   \[\leadsto \color{blue}{\mathsf{134\_z0z1z2z3z4}\left(y, x, t, z, t\right)} \]
4. Add Preprocessing

Alternative 2: 99.6% accurate, 0.0× speedup

Math

\[\begin{array}{l} t_1 := \mathsf{min}\left(\left|y\right|, \left|t\right|\right)\\ t_2 := \mathsf{max}\left(\left|y\right|, \left|t\right|\right)\\ t_3 := \left(\left(x - z\right) \cdot t\_1\right) \cdot t\_2\\ t_4 := \left(x \cdot t\_1 - z \cdot t\_1\right) \cdot t\_2\\ \mathsf{copysign}\left(1, y\right) \cdot \left(\mathsf{copysign}\left(1, t\right) \cdot \begin{array}{l} \mathbf{if}\;t\_4 \leq \frac{-6090821257124999}{609082125712499942522086399242199269429764178599687970429244153575809293172901631404100941617625641201581557264463041761466198116575193377911124206019540838720704856247279564366924353468128353022049974592451148679605349870337179684109147725966810350801733675194017346692614286874494631936}:\\ \;\;\;\;t\_3\\ \mathbf{elif}\;t\_4 \leq \frac{554533938824163}{2772669694120814859578414184143083703436437075375816575170479580614621307805625623039974406104139578097391210961403571828974157824}:\\ \;\;\;\;\left(\left(x - z\right) \cdot t\_2\right) \cdot t\_1\\ \mathbf{else}:\\ \;\;\;\;t\_3\\ \end{array}\right) \end{array} \]

FPCore

(FPCore (x y z t)
  :precision binary64
  (let* ((t_1 (fmin (fabs y) (fabs t)))
       (t_2 (fmax (fabs y) (fabs t)))
       (t_3 (* (* (- x z) t_1) t_2))
       (t_4 (* (- (* x t_1) (* z t_1)) t_2)))
  (*
   (copysign 1 y)
   (*
    (copysign 1 t)
    (if (<=
         t_4
         -6090821257124999/609082125712499942522086399242199269429764178599687970429244153575809293172901631404100941617625641201581557264463041761466198116575193377911124206019540838720704856247279564366924353468128353022049974592451148679605349870337179684109147725966810350801733675194017346692614286874494631936)
      t_3
      (if (<=
           t_4
           554533938824163/2772669694120814859578414184143083703436437075375816575170479580614621307805625623039974406104139578097391210961403571828974157824)
        (* (* (- x z) t_2) t_1)
        t_3))))))

C

double code(double x, double y, double z, double t) {
	double t_1 = fmin(fabs(y), fabs(t));
	double t_2 = fmax(fabs(y), fabs(t));
	double t_3 = ((x - z) * t_1) * t_2;
	double t_4 = ((x * t_1) - (z * t_1)) * t_2;
	double tmp;
	if (t_4 <= -1e-272) {
		tmp = t_3;
	} else if (t_4 <= 2e-115) {
		tmp = ((x - z) * t_2) * t_1;
	} else {
		tmp = t_3;
	}
	return copysign(1.0, y) * (copysign(1.0, t) * tmp);
}

Java

public static double code(double x, double y, double z, double t) {
	double t_1 = fmin(Math.abs(y), Math.abs(t));
	double t_2 = fmax(Math.abs(y), Math.abs(t));
	double t_3 = ((x - z) * t_1) * t_2;
	double t_4 = ((x * t_1) - (z * t_1)) * t_2;
	double tmp;
	if (t_4 <= -1e-272) {
		tmp = t_3;
	} else if (t_4 <= 2e-115) {
		tmp = ((x - z) * t_2) * t_1;
	} else {
		tmp = t_3;
	}
	return Math.copySign(1.0, y) * (Math.copySign(1.0, t) * tmp);
}

Python

def code(x, y, z, t):
	t_1 = fmin(math.fabs(y), math.fabs(t))
	t_2 = fmax(math.fabs(y), math.fabs(t))
	t_3 = ((x - z) * t_1) * t_2
	t_4 = ((x * t_1) - (z * t_1)) * t_2
	tmp = 0
	if t_4 <= -1e-272:
		tmp = t_3
	elif t_4 <= 2e-115:
		tmp = ((x - z) * t_2) * t_1
	else:
		tmp = t_3
	return math.copysign(1.0, y) * (math.copysign(1.0, t) * tmp)

Julia

function code(x, y, z, t)
	t_1 = fmin(abs(y), abs(t))
	t_2 = fmax(abs(y), abs(t))
	t_3 = Float64(Float64(Float64(x - z) * t_1) * t_2)
	t_4 = Float64(Float64(Float64(x * t_1) - Float64(z * t_1)) * t_2)
	tmp = 0.0
	if (t_4 <= -1e-272)
		tmp = t_3;
	elseif (t_4 <= 2e-115)
		tmp = Float64(Float64(Float64(x - z) * t_2) * t_1);
	else
		tmp = t_3;
	end
	return Float64(copysign(1.0, y) * Float64(copysign(1.0, t) * tmp))
end

MATLAB

function tmp_2 = code(x, y, z, t)
	t_1 = min(abs(y), abs(t));
	t_2 = max(abs(y), abs(t));
	t_3 = ((x - z) * t_1) * t_2;
	t_4 = ((x * t_1) - (z * t_1)) * t_2;
	tmp = 0.0;
	if (t_4 <= -1e-272)
		tmp = t_3;
	elseif (t_4 <= 2e-115)
		tmp = ((x - z) * t_2) * t_1;
	else
		tmp = t_3;
	end
	tmp_2 = (sign(y) * abs(1.0)) * ((sign(t) * abs(1.0)) * tmp);
end

Wolfram

code[x_, y_, z_, t_] := Block[{t$95$1 = N[Min[N[Abs[y], $MachinePrecision], N[Abs[t], $MachinePrecision]], $MachinePrecision]}, Block[{t$95$2 = N[Max[N[Abs[y], $MachinePrecision], N[Abs[t], $MachinePrecision]], $MachinePrecision]}, Block[{t$95$3 = N[(N[(N[(x - z), $MachinePrecision] * t$95$1), $MachinePrecision] * t$95$2), $MachinePrecision]}, Block[{t$95$4 = N[(N[(N[(x * t$95$1), $MachinePrecision] - N[(z * t$95$1), $MachinePrecision]), $MachinePrecision] * t$95$2), $MachinePrecision]}, N[(N[With[{TMP1 = Abs[1], TMP2 = Sign[y]}, TMP1 * If[TMP2 == 0, 1, TMP2]], $MachinePrecision] * N[(N[With[{TMP1 = Abs[1], TMP2 = Sign[t]}, TMP1 * If[TMP2 == 0, 1, TMP2]], $MachinePrecision] * If[LessEqual[t$95$4, -6090821257124999/609082125712499942522086399242199269429764178599687970429244153575809293172901631404100941617625641201581557264463041761466198116575193377911124206019540838720704856247279564366924353468128353022049974592451148679605349870337179684109147725966810350801733675194017346692614286874494631936], t$95$3, If[LessEqual[t$95$4, 554533938824163/2772669694120814859578414184143083703436437075375816575170479580614621307805625623039974406104139578097391210961403571828974157824], N[(N[(N[(x - z), $MachinePrecision] * t$95$2), $MachinePrecision] * t$95$1), $MachinePrecision], t$95$3]]), $MachinePrecision]), $MachinePrecision]]]]]

Derivation

1. Split input into 2 regimes

2. if (*.f64 (-.f64 (*.f64 x y) (*.f64 z y)) t) < -9.9999999999999993e-273 or 2.0000000000000001e-115 < (*.f64 (-.f64 (*.f64 x y) (*.f64 z y)) t)

   1. Initial program 89.7%

      \[\left(x \cdot y - z \cdot y\right) \cdot t \]
   2. Step-by-step derivation

      1. lift--.f64 N/A

         \[\leadsto \color{blue}{\left(x \cdot y - z \cdot y\right)} \cdot t \]

      2. lift-*.f64 N/A

         \[\leadsto \left(\color{blue}{x \cdot y} - z \cdot y\right) \cdot t \]

      3. lift-*.f64 N/A

         \[\leadsto \left(x \cdot y - \color{blue}{z \cdot y}\right) \cdot t \]

      4. distribute-rgt-out-- N/A

         \[\leadsto \color{blue}{\left(y \cdot \left(x - z\right)\right)} \cdot t \]

      5. *-commutative N/A

         \[\leadsto \color{blue}{\left(\left(x - z\right) \cdot y\right)} \cdot t \]

      6. lower-*.f64 N/A

         \[\leadsto \color{blue}{\left(\left(x - z\right) \cdot y\right)} \cdot t \]

      7. lower--.f64 91.7%

         \[\leadsto \left(\color{blue}{\left(x - z\right)} \cdot y\right) \cdot t \]

   3. Applied rewrites 91.7%

      \[\leadsto \color{blue}{\left(\left(x - z\right) \cdot y\right)} \cdot t \]

   if -9.9999999999999993e-273 < (*.f64 (-.f64 (*.f64 x y) (*.f64 z y)) t) < 2.0000000000000001e-115

   1. Initial program 89.7%

      \[\left(x \cdot y - z \cdot y\right) \cdot t \]
   2. Step-by-step derivation

      1. lift-*.f64 N/A

         \[\leadsto \color{blue}{\left(x \cdot y - z \cdot y\right) \cdot t} \]

      2. lift--.f64 N/A

         \[\leadsto \color{blue}{\left(x \cdot y - z \cdot y\right)} \cdot t \]

      3. lift-*.f64 N/A

         \[\leadsto \left(\color{blue}{x \cdot y} - z \cdot y\right) \cdot t \]

      4. lift-*.f64 N/A

         \[\leadsto \left(x \cdot y - \color{blue}{z \cdot y}\right) \cdot t \]

      5. distribute-rgt-out-- N/A

         \[\leadsto \color{blue}{\left(y \cdot \left(x - z\right)\right)} \cdot t \]

      6. associate-*l* N/A

         \[\leadsto \color{blue}{y \cdot \left(\left(x - z\right) \cdot t\right)} \]

      7. *-commutative N/A

         \[\leadsto \color{blue}{\left(\left(x - z\right) \cdot t\right) \cdot y} \]

      8. lower-*.f64 N/A

         \[\leadsto \color{blue}{\left(\left(x - z\right) \cdot t\right) \cdot y} \]

      9. lower-*.f64 N/A

         \[\leadsto \color{blue}{\left(\left(x - z\right) \cdot t\right)} \cdot y \]

      10. lower--.f64 91.7%

          \[\leadsto \left(\color{blue}{\left(x - z\right)} \cdot t\right) \cdot y \]

   3. Applied rewrites 91.7%

      \[\leadsto \color{blue}{\left(\left(x - z\right) \cdot t\right) \cdot y} \]
3. Recombined 2 regimes into one program.
4. Add Preprocessing

Alternative 3: 98.1% accurate, 0.0× speedup

Math

\[\begin{array}{l} t_1 := \mathsf{min}\left(\left|y\right|, \left|t\right|\right)\\ t_2 := \mathsf{max}\left(\left|y\right|, \left|t\right|\right)\\ \mathsf{copysign}\left(1, y\right) \cdot \left(\mathsf{copysign}\left(1, t\right) \cdot \begin{array}{l} \mathbf{if}\;t\_2 \leq \frac{7423025276069149}{11972621413014756705924586149611790497021399392059392}:\\ \;\;\;\;\left(x \cdot t\_1 - z \cdot t\_1\right) \cdot t\_2\\ \mathbf{else}:\\ \;\;\;\;\left(x - z\right) \cdot \left(t\_2 \cdot t\_1\right)\\ \end{array}\right) \end{array} \]

FPCore

(FPCore (x y z t)
  :precision binary64
  (let* ((t_1 (fmin (fabs y) (fabs t))) (t_2 (fmax (fabs y) (fabs t))))
  (*
   (copysign 1 y)
   (*
    (copysign 1 t)
    (if (<=
         t_2
         7423025276069149/11972621413014756705924586149611790497021399392059392)
      (* (- (* x t_1) (* z t_1)) t_2)
      (* (- x z) (* t_2 t_1)))))))

C

double code(double x, double y, double z, double t) {
	double t_1 = fmin(fabs(y), fabs(t));
	double t_2 = fmax(fabs(y), fabs(t));
	double tmp;
	if (t_2 <= 6.2e-37) {
		tmp = ((x * t_1) - (z * t_1)) * t_2;
	} else {
		tmp = (x - z) * (t_2 * t_1);
	}
	return copysign(1.0, y) * (copysign(1.0, t) * tmp);
}

Java

public static double code(double x, double y, double z, double t) {
	double t_1 = fmin(Math.abs(y), Math.abs(t));
	double t_2 = fmax(Math.abs(y), Math.abs(t));
	double tmp;
	if (t_2 <= 6.2e-37) {
		tmp = ((x * t_1) - (z * t_1)) * t_2;
	} else {
		tmp = (x - z) * (t_2 * t_1);
	}
	return Math.copySign(1.0, y) * (Math.copySign(1.0, t) * tmp);
}

Python

def code(x, y, z, t):
	t_1 = fmin(math.fabs(y), math.fabs(t))
	t_2 = fmax(math.fabs(y), math.fabs(t))
	tmp = 0
	if t_2 <= 6.2e-37:
		tmp = ((x * t_1) - (z * t_1)) * t_2
	else:
		tmp = (x - z) * (t_2 * t_1)
	return math.copysign(1.0, y) * (math.copysign(1.0, t) * tmp)

Julia

function code(x, y, z, t)
	t_1 = fmin(abs(y), abs(t))
	t_2 = fmax(abs(y), abs(t))
	tmp = 0.0
	if (t_2 <= 6.2e-37)
		tmp = Float64(Float64(Float64(x * t_1) - Float64(z * t_1)) * t_2);
	else
		tmp = Float64(Float64(x - z) * Float64(t_2 * t_1));
	end
	return Float64(copysign(1.0, y) * Float64(copysign(1.0, t) * tmp))
end

MATLAB

function tmp_2 = code(x, y, z, t)
	t_1 = min(abs(y), abs(t));
	t_2 = max(abs(y), abs(t));
	tmp = 0.0;
	if (t_2 <= 6.2e-37)
		tmp = ((x * t_1) - (z * t_1)) * t_2;
	else
		tmp = (x - z) * (t_2 * t_1);
	end
	tmp_2 = (sign(y) * abs(1.0)) * ((sign(t) * abs(1.0)) * tmp);
end

Wolfram

code[x_, y_, z_, t_] := Block[{t$95$1 = N[Min[N[Abs[y], $MachinePrecision], N[Abs[t], $MachinePrecision]], $MachinePrecision]}, Block[{t$95$2 = N[Max[N[Abs[y], $MachinePrecision], N[Abs[t], $MachinePrecision]], $MachinePrecision]}, N[(N[With[{TMP1 = Abs[1], TMP2 = Sign[y]}, TMP1 * If[TMP2 == 0, 1, TMP2]], $MachinePrecision] * N[(N[With[{TMP1 = Abs[1], TMP2 = Sign[t]}, TMP1 * If[TMP2 == 0, 1, TMP2]], $MachinePrecision] * If[LessEqual[t$95$2, 7423025276069149/11972621413014756705924586149611790497021399392059392], N[(N[(N[(x * t$95$1), $MachinePrecision] - N[(z * t$95$1), $MachinePrecision]), $MachinePrecision] * t$95$2), $MachinePrecision], N[(N[(x - z), $MachinePrecision] * N[(t$95$2 * t$95$1), $MachinePrecision]), $MachinePrecision]]), $MachinePrecision]), $MachinePrecision]]]

Derivation

1. Split input into 2 regimes

2. if t < 6.1999999999999999e-37

   1. Initial program 89.7%

      \[\left(x \cdot y - z \cdot y\right) \cdot t \]

   if 6.1999999999999999e-37 < t

   1. Initial program 89.7%

      \[\left(x \cdot y - z \cdot y\right) \cdot t \]
   2. Step-by-step derivation

      1. lift-*.f64 N/A

         \[\leadsto \color{blue}{\left(x \cdot y - z \cdot y\right) \cdot t} \]

      2. *-commutative N/A

         \[\leadsto \color{blue}{t \cdot \left(x \cdot y - z \cdot y\right)} \]

      3. lift--.f64 N/A

         \[\leadsto t \cdot \color{blue}{\left(x \cdot y - z \cdot y\right)} \]

      4. lift-*.f64 N/A

         \[\leadsto t \cdot \left(\color{blue}{x \cdot y} - z \cdot y\right) \]

      5. lift-*.f64 N/A

         \[\leadsto t \cdot \left(x \cdot y - \color{blue}{z \cdot y}\right) \]

      6. distribute-rgt-out-- N/A

         \[\leadsto t \cdot \color{blue}{\left(y \cdot \left(x - z\right)\right)} \]

      7. associate-*r* N/A

         \[\leadsto \color{blue}{\left(t \cdot y\right) \cdot \left(x - z\right)} \]

      8. *-commutative N/A

         \[\leadsto \color{blue}{\left(x - z\right) \cdot \left(t \cdot y\right)} \]

      9. lower-*.f64 N/A

         \[\leadsto \color{blue}{\left(x - z\right) \cdot \left(t \cdot y\right)} \]

      10. lower--.f64 N/A

          \[\leadsto \color{blue}{\left(x - z\right)} \cdot \left(t \cdot y\right) \]

      11. lower-*.f64 92.1%

          \[\leadsto \left(x - z\right) \cdot \color{blue}{\left(t \cdot y\right)} \]

   3. Applied rewrites 92.1%

      \[\leadsto \color{blue}{\left(x - z\right) \cdot \left(t \cdot y\right)} \]
3. Recombined 2 regimes into one program.
4. Add Preprocessing

Alternative 4: 98.1% accurate, 0.0× speedup

Math

\[\begin{array}{l} t_1 := \mathsf{min}\left(\left|y\right|, \left|t\right|\right)\\ t_2 := \mathsf{max}\left(\left|y\right|, \left|t\right|\right)\\ \mathsf{copysign}\left(1, y\right) \cdot \left(\mathsf{copysign}\left(1, t\right) \cdot \begin{array}{l} \mathbf{if}\;t\_2 \leq \frac{2993155353253689}{5986310706507378352962293074805895248510699696029696}:\\ \;\;\;\;\left(\left(x - z\right) \cdot t\_2\right) \cdot t\_1\\ \mathbf{else}:\\ \;\;\;\;\left(x - z\right) \cdot \left(t\_2 \cdot t\_1\right)\\ \end{array}\right) \end{array} \]

FPCore

(FPCore (x y z t)
  :precision binary64
  (let* ((t_1 (fmin (fabs y) (fabs t))) (t_2 (fmax (fabs y) (fabs t))))
  (*
   (copysign 1 y)
   (*
    (copysign 1 t)
    (if (<=
         t_2
         2993155353253689/5986310706507378352962293074805895248510699696029696)
      (* (* (- x z) t_2) t_1)
      (* (- x z) (* t_2 t_1)))))))

C

double code(double x, double y, double z, double t) {
	double t_1 = fmin(fabs(y), fabs(t));
	double t_2 = fmax(fabs(y), fabs(t));
	double tmp;
	if (t_2 <= 5e-37) {
		tmp = ((x - z) * t_2) * t_1;
	} else {
		tmp = (x - z) * (t_2 * t_1);
	}
	return copysign(1.0, y) * (copysign(1.0, t) * tmp);
}

Java

public static double code(double x, double y, double z, double t) {
	double t_1 = fmin(Math.abs(y), Math.abs(t));
	double t_2 = fmax(Math.abs(y), Math.abs(t));
	double tmp;
	if (t_2 <= 5e-37) {
		tmp = ((x - z) * t_2) * t_1;
	} else {
		tmp = (x - z) * (t_2 * t_1);
	}
	return Math.copySign(1.0, y) * (Math.copySign(1.0, t) * tmp);
}

Python

def code(x, y, z, t):
	t_1 = fmin(math.fabs(y), math.fabs(t))
	t_2 = fmax(math.fabs(y), math.fabs(t))
	tmp = 0
	if t_2 <= 5e-37:
		tmp = ((x - z) * t_2) * t_1
	else:
		tmp = (x - z) * (t_2 * t_1)
	return math.copysign(1.0, y) * (math.copysign(1.0, t) * tmp)

Julia

function code(x, y, z, t)
	t_1 = fmin(abs(y), abs(t))
	t_2 = fmax(abs(y), abs(t))
	tmp = 0.0
	if (t_2 <= 5e-37)
		tmp = Float64(Float64(Float64(x - z) * t_2) * t_1);
	else
		tmp = Float64(Float64(x - z) * Float64(t_2 * t_1));
	end
	return Float64(copysign(1.0, y) * Float64(copysign(1.0, t) * tmp))
end

MATLAB

function tmp_2 = code(x, y, z, t)
	t_1 = min(abs(y), abs(t));
	t_2 = max(abs(y), abs(t));
	tmp = 0.0;
	if (t_2 <= 5e-37)
		tmp = ((x - z) * t_2) * t_1;
	else
		tmp = (x - z) * (t_2 * t_1);
	end
	tmp_2 = (sign(y) * abs(1.0)) * ((sign(t) * abs(1.0)) * tmp);
end

Wolfram

code[x_, y_, z_, t_] := Block[{t$95$1 = N[Min[N[Abs[y], $MachinePrecision], N[Abs[t], $MachinePrecision]], $MachinePrecision]}, Block[{t$95$2 = N[Max[N[Abs[y], $MachinePrecision], N[Abs[t], $MachinePrecision]], $MachinePrecision]}, N[(N[With[{TMP1 = Abs[1], TMP2 = Sign[y]}, TMP1 * If[TMP2 == 0, 1, TMP2]], $MachinePrecision] * N[(N[With[{TMP1 = Abs[1], TMP2 = Sign[t]}, TMP1 * If[TMP2 == 0, 1, TMP2]], $MachinePrecision] * If[LessEqual[t$95$2, 2993155353253689/5986310706507378352962293074805895248510699696029696], N[(N[(N[(x - z), $MachinePrecision] * t$95$2), $MachinePrecision] * t$95$1), $MachinePrecision], N[(N[(x - z), $MachinePrecision] * N[(t$95$2 * t$95$1), $MachinePrecision]), $MachinePrecision]]), $MachinePrecision]), $MachinePrecision]]]

Derivation

1. Split input into 2 regimes

2. if t < 4.9999999999999997e-37

   1. Initial program 89.7%

      \[\left(x \cdot y - z \cdot y\right) \cdot t \]
   2. Step-by-step derivation

      1. lift-*.f64 N/A

         \[\leadsto \color{blue}{\left(x \cdot y - z \cdot y\right) \cdot t} \]

      2. lift--.f64 N/A

         \[\leadsto \color{blue}{\left(x \cdot y - z \cdot y\right)} \cdot t \]

      3. lift-*.f64 N/A

         \[\leadsto \left(\color{blue}{x \cdot y} - z \cdot y\right) \cdot t \]

      4. lift-*.f64 N/A

         \[\leadsto \left(x \cdot y - \color{blue}{z \cdot y}\right) \cdot t \]

      5. distribute-rgt-out-- N/A

         \[\leadsto \color{blue}{\left(y \cdot \left(x - z\right)\right)} \cdot t \]

      6. associate-*l* N/A

         \[\leadsto \color{blue}{y \cdot \left(\left(x - z\right) \cdot t\right)} \]

      7. *-commutative N/A

         \[\leadsto \color{blue}{\left(\left(x - z\right) \cdot t\right) \cdot y} \]

      8. lower-*.f64 N/A

         \[\leadsto \color{blue}{\left(\left(x - z\right) \cdot t\right) \cdot y} \]

      9. lower-*.f64 N/A

         \[\leadsto \color{blue}{\left(\left(x - z\right) \cdot t\right)} \cdot y \]

      10. lower--.f64 91.7%

          \[\leadsto \left(\color{blue}{\left(x - z\right)} \cdot t\right) \cdot y \]

   3. Applied rewrites 91.7%

      \[\leadsto \color{blue}{\left(\left(x - z\right) \cdot t\right) \cdot y} \]

   if 4.9999999999999997e-37 < t

   1. Initial program 89.7%

      \[\left(x \cdot y - z \cdot y\right) \cdot t \]
   2. Step-by-step derivation

      1. lift-*.f64 N/A

         \[\leadsto \color{blue}{\left(x \cdot y - z \cdot y\right) \cdot t} \]

      2. *-commutative N/A

         \[\leadsto \color{blue}{t \cdot \left(x \cdot y - z \cdot y\right)} \]

      3. lift--.f64 N/A

         \[\leadsto t \cdot \color{blue}{\left(x \cdot y - z \cdot y\right)} \]

      4. lift-*.f64 N/A

         \[\leadsto t \cdot \left(\color{blue}{x \cdot y} - z \cdot y\right) \]

      5. lift-*.f64 N/A

         \[\leadsto t \cdot \left(x \cdot y - \color{blue}{z \cdot y}\right) \]

      6. distribute-rgt-out-- N/A

         \[\leadsto t \cdot \color{blue}{\left(y \cdot \left(x - z\right)\right)} \]

      7. associate-*r* N/A

         \[\leadsto \color{blue}{\left(t \cdot y\right) \cdot \left(x - z\right)} \]

      8. *-commutative N/A

         \[\leadsto \color{blue}{\left(x - z\right) \cdot \left(t \cdot y\right)} \]

      9. lower-*.f64 N/A

         \[\leadsto \color{blue}{\left(x - z\right) \cdot \left(t \cdot y\right)} \]

      10. lower--.f64 N/A

          \[\leadsto \color{blue}{\left(x - z\right)} \cdot \left(t \cdot y\right) \]

      11. lower-*.f64 92.1%

          \[\leadsto \left(x - z\right) \cdot \color{blue}{\left(t \cdot y\right)} \]

   3. Applied rewrites 92.1%

      \[\leadsto \color{blue}{\left(x - z\right) \cdot \left(t \cdot y\right)} \]
3. Recombined 2 regimes into one program.
4. Add Preprocessing

Alternative 5: 91.7% accurate, 1.4× speedup

Math

\[\left(\left(x - z\right) \cdot t\right) \cdot y \]

FPCore

(FPCore (x y z t)
  :precision binary64
  (* (* (- x z) t) y))

C

double code(double x, double y, double z, double t) {
	return ((x - z) * t) * y;
}

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(x, y, z, t)
use fmin_fmax_functions
    real(8), intent (in) :: x
    real(8), intent (in) :: y
    real(8), intent (in) :: z
    real(8), intent (in) :: t
    code = ((x - z) * t) * y
end function

Java

public static double code(double x, double y, double z, double t) {
	return ((x - z) * t) * y;
}

Python

def code(x, y, z, t):
	return ((x - z) * t) * y

Julia

function code(x, y, z, t)
	return Float64(Float64(Float64(x - z) * t) * y)
end

MATLAB

function tmp = code(x, y, z, t)
	tmp = ((x - z) * t) * y;
end

Wolfram

code[x_, y_, z_, t_] := N[(N[(N[(x - z), $MachinePrecision] * t), $MachinePrecision] * y), $MachinePrecision]

Derivation

1. Initial program 89.7%

   \[\left(x \cdot y - z \cdot y\right) \cdot t \]
2. Step-by-step derivation

   1. lift-*.f64 N/A

      \[\leadsto \color{blue}{\left(x \cdot y - z \cdot y\right) \cdot t} \]

   2. lift--.f64 N/A

      \[\leadsto \color{blue}{\left(x \cdot y - z \cdot y\right)} \cdot t \]

   3. lift-*.f64 N/A

      \[\leadsto \left(\color{blue}{x \cdot y} - z \cdot y\right) \cdot t \]

   4. lift-*.f64 N/A

      \[\leadsto \left(x \cdot y - \color{blue}{z \cdot y}\right) \cdot t \]

   5. distribute-rgt-out-- N/A

      \[\leadsto \color{blue}{\left(y \cdot \left(x - z\right)\right)} \cdot t \]

   6. associate-*l* N/A

      \[\leadsto \color{blue}{y \cdot \left(\left(x - z\right) \cdot t\right)} \]

   7. *-commutative N/A

      \[\leadsto \color{blue}{\left(\left(x - z\right) \cdot t\right) \cdot y} \]

   8. lower-*.f64 N/A

      \[\leadsto \color{blue}{\left(\left(x - z\right) \cdot t\right) \cdot y} \]

   9. lower-*.f64 N/A

      \[\leadsto \color{blue}{\left(\left(x - z\right) \cdot t\right)} \cdot y \]

   10. lower--.f64 91.7%

       \[\leadsto \left(\color{blue}{\left(x - z\right)} \cdot t\right) \cdot y \]

3. Applied rewrites 91.7%

   \[\leadsto \color{blue}{\left(\left(x - z\right) \cdot t\right) \cdot y} \]
4. Add Preprocessing

Alternative 6: 58.0% accurate, 0.0× speedup

Math

\[\begin{array}{l} t_1 := \mathsf{min}\left(\left|y\right|, \left|t\right|\right)\\ t_2 := \mathsf{max}\left(\left|y\right|, \left|t\right|\right)\\ \mathsf{copysign}\left(1, y\right) \cdot \left(\mathsf{copysign}\left(1, t\right) \cdot \begin{array}{l} \mathbf{if}\;t\_2 \leq \frac{1915619426082361}{95780971304118053647396689196894323976171195136475136}:\\ \;\;\;\;t\_2 \cdot \left(x \cdot t\_1\right)\\ \mathbf{else}:\\ \;\;\;\;\left(t\_2 \cdot t\_1\right) \cdot x\\ \end{array}\right) \end{array} \]

FPCore

(FPCore (x y z t)
  :precision binary64
  (let* ((t_1 (fmin (fabs y) (fabs t))) (t_2 (fmax (fabs y) (fabs t))))
  (*
   (copysign 1 y)
   (*
    (copysign 1 t)
    (if (<=
         t_2
         1915619426082361/95780971304118053647396689196894323976171195136475136)
      (* t_2 (* x t_1))
      (* (* t_2 t_1) x))))))

C

double code(double x, double y, double z, double t) {
	double t_1 = fmin(fabs(y), fabs(t));
	double t_2 = fmax(fabs(y), fabs(t));
	double tmp;
	if (t_2 <= 2e-38) {
		tmp = t_2 * (x * t_1);
	} else {
		tmp = (t_2 * t_1) * x;
	}
	return copysign(1.0, y) * (copysign(1.0, t) * tmp);
}

Java

public static double code(double x, double y, double z, double t) {
	double t_1 = fmin(Math.abs(y), Math.abs(t));
	double t_2 = fmax(Math.abs(y), Math.abs(t));
	double tmp;
	if (t_2 <= 2e-38) {
		tmp = t_2 * (x * t_1);
	} else {
		tmp = (t_2 * t_1) * x;
	}
	return Math.copySign(1.0, y) * (Math.copySign(1.0, t) * tmp);
}

Python

def code(x, y, z, t):
	t_1 = fmin(math.fabs(y), math.fabs(t))
	t_2 = fmax(math.fabs(y), math.fabs(t))
	tmp = 0
	if t_2 <= 2e-38:
		tmp = t_2 * (x * t_1)
	else:
		tmp = (t_2 * t_1) * x
	return math.copysign(1.0, y) * (math.copysign(1.0, t) * tmp)

Julia

function code(x, y, z, t)
	t_1 = fmin(abs(y), abs(t))
	t_2 = fmax(abs(y), abs(t))
	tmp = 0.0
	if (t_2 <= 2e-38)
		tmp = Float64(t_2 * Float64(x * t_1));
	else
		tmp = Float64(Float64(t_2 * t_1) * x);
	end
	return Float64(copysign(1.0, y) * Float64(copysign(1.0, t) * tmp))
end

MATLAB

function tmp_2 = code(x, y, z, t)
	t_1 = min(abs(y), abs(t));
	t_2 = max(abs(y), abs(t));
	tmp = 0.0;
	if (t_2 <= 2e-38)
		tmp = t_2 * (x * t_1);
	else
		tmp = (t_2 * t_1) * x;
	end
	tmp_2 = (sign(y) * abs(1.0)) * ((sign(t) * abs(1.0)) * tmp);
end

Wolfram

code[x_, y_, z_, t_] := Block[{t$95$1 = N[Min[N[Abs[y], $MachinePrecision], N[Abs[t], $MachinePrecision]], $MachinePrecision]}, Block[{t$95$2 = N[Max[N[Abs[y], $MachinePrecision], N[Abs[t], $MachinePrecision]], $MachinePrecision]}, N[(N[With[{TMP1 = Abs[1], TMP2 = Sign[y]}, TMP1 * If[TMP2 == 0, 1, TMP2]], $MachinePrecision] * N[(N[With[{TMP1 = Abs[1], TMP2 = Sign[t]}, TMP1 * If[TMP2 == 0, 1, TMP2]], $MachinePrecision] * If[LessEqual[t$95$2, 1915619426082361/95780971304118053647396689196894323976171195136475136], N[(t$95$2 * N[(x * t$95$1), $MachinePrecision]), $MachinePrecision], N[(N[(t$95$2 * t$95$1), $MachinePrecision] * x), $MachinePrecision]]), $MachinePrecision]), $MachinePrecision]]]

Derivation

1. Split input into 2 regimes

2. if t < 1.9999999999999999e-38

   1. Initial program 89.7%

      \[\left(x \cdot y - z \cdot y\right) \cdot t \]

   2. Taylor expanded in x around inf

      \[\leadsto \color{blue}{t \cdot \left(x \cdot y\right)} \]
   3. Step-by-step derivation

      1. lower-*.f64 N/A

         \[\leadsto t \cdot \color{blue}{\left(x \cdot y\right)} \]

      2. lower-*.f64 53.4%

         \[\leadsto t \cdot \left(x \cdot \color{blue}{y}\right) \]

   4. Applied rewrites 53.4%

      \[\leadsto \color{blue}{t \cdot \left(x \cdot y\right)} \]

   if 1.9999999999999999e-38 < t

   1. Initial program 89.7%

      \[\left(x \cdot y - z \cdot y\right) \cdot t \]

   2. Taylor expanded in x around inf

      \[\leadsto \color{blue}{t \cdot \left(x \cdot y\right)} \]
   3. Step-by-step derivation

      1. lower-*.f64 N/A

         \[\leadsto t \cdot \color{blue}{\left(x \cdot y\right)} \]

      2. lower-*.f64 53.4%

         \[\leadsto t \cdot \left(x \cdot \color{blue}{y}\right) \]

   4. Applied rewrites 53.4%

      \[\leadsto \color{blue}{t \cdot \left(x \cdot y\right)} \]
   5. Step-by-step derivation

      1. lift-*.f64 N/A

         \[\leadsto t \cdot \color{blue}{\left(x \cdot y\right)} \]

      2. remove-double-neg N/A

         \[\leadsto t \cdot \left(\mathsf{neg}\left(\left(\mathsf{neg}\left(x \cdot y\right)\right)\right)\right) \]

      3. distribute-rgt-neg-out N/A

         \[\leadsto \mathsf{neg}\left(t \cdot \left(\mathsf{neg}\left(x \cdot y\right)\right)\right) \]

      4. distribute-lft-neg-out N/A

         \[\leadsto \left(\mathsf{neg}\left(t\right)\right) \cdot \color{blue}{\left(\mathsf{neg}\left(x \cdot y\right)\right)} \]

      5. lift-*.f64 N/A

         \[\leadsto \left(\mathsf{neg}\left(t\right)\right) \cdot \left(\mathsf{neg}\left(x \cdot y\right)\right) \]

      6. *-commutative N/A

         \[\leadsto \left(\mathsf{neg}\left(t\right)\right) \cdot \left(\mathsf{neg}\left(y \cdot x\right)\right) \]

      7. distribute-lft-neg-out N/A

         \[\leadsto \left(\mathsf{neg}\left(t\right)\right) \cdot \left(\left(\mathsf{neg}\left(y\right)\right) \cdot \color{blue}{x}\right) \]

      8. associate-*r* N/A

         \[\leadsto \left(\left(\mathsf{neg}\left(t\right)\right) \cdot \left(\mathsf{neg}\left(y\right)\right)\right) \cdot \color{blue}{x} \]

      9. distribute-rgt-neg-out N/A

         \[\leadsto \left(\mathsf{neg}\left(\left(\mathsf{neg}\left(t\right)\right) \cdot y\right)\right) \cdot x \]

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

          \[\leadsto \left(\mathsf{neg}\left(\left(\mathsf{neg}\left(t \cdot y\right)\right)\right)\right) \cdot x \]

      11. lift-*.f64 N/A

          \[\leadsto \left(\mathsf{neg}\left(\left(\mathsf{neg}\left(t \cdot y\right)\right)\right)\right) \cdot x \]

      12. lower-*.f64 N/A

          \[\leadsto \left(\mathsf{neg}\left(\left(\mathsf{neg}\left(t \cdot y\right)\right)\right)\right) \cdot \color{blue}{x} \]

      13. lift-*.f64 N/A

          \[\leadsto \left(\mathsf{neg}\left(\left(\mathsf{neg}\left(t \cdot y\right)\right)\right)\right) \cdot x \]

      14. distribute-lft-neg-out N/A

          \[\leadsto \left(\mathsf{neg}\left(\left(\mathsf{neg}\left(t\right)\right) \cdot y\right)\right) \cdot x \]

      15. distribute-lft-neg-out N/A

          \[\leadsto \left(\left(\mathsf{neg}\left(\left(\mathsf{neg}\left(t\right)\right)\right)\right) \cdot y\right) \cdot x \]

      16. remove-double-neg N/A

          \[\leadsto \left(t \cdot y\right) \cdot x \]

      17. lift-*.f64 55.0%

          \[\leadsto \left(t \cdot y\right) \cdot x \]

   6. Applied rewrites 55.0%

      \[\leadsto \left(t \cdot y\right) \cdot \color{blue}{x} \]
3. Recombined 2 regimes into one program.
4. Add Preprocessing

Alternative 7: 54.4% accurate, 0.1× speedup

Math

\[\mathsf{copysign}\left(1, t\right) \cdot \left(\mathsf{max}\left(y, \left|t\right|\right) \cdot \left(x \cdot \mathsf{min}\left(y, \left|t\right|\right)\right)\right) \]

FPCore

(FPCore (x y z t)
  :precision binary64
  (* (copysign 1 t) (* (fmax y (fabs t)) (* x (fmin y (fabs t))))))

C

double code(double x, double y, double z, double t) {
	return copysign(1.0, t) * (fmax(y, fabs(t)) * (x * fmin(y, fabs(t))));
}

Java

public static double code(double x, double y, double z, double t) {
	return Math.copySign(1.0, t) * (fmax(y, Math.abs(t)) * (x * fmin(y, Math.abs(t))));
}

Python

def code(x, y, z, t):
	return math.copysign(1.0, t) * (fmax(y, math.fabs(t)) * (x * fmin(y, math.fabs(t))))

Julia

function code(x, y, z, t)
	return Float64(copysign(1.0, t) * Float64(fmax(y, abs(t)) * Float64(x * fmin(y, abs(t)))))
end

MATLAB

function tmp = code(x, y, z, t)
	tmp = (sign(t) * abs(1.0)) * (max(y, abs(t)) * (x * min(y, abs(t))));
end

Wolfram

code[x_, y_, z_, t_] := N[(N[With[{TMP1 = Abs[1], TMP2 = Sign[t]}, TMP1 * If[TMP2 == 0, 1, TMP2]], $MachinePrecision] * N[(N[Max[y, N[Abs[t], $MachinePrecision]], $MachinePrecision] * N[(x * N[Min[y, N[Abs[t], $MachinePrecision]], $MachinePrecision]), $MachinePrecision]), $MachinePrecision]), $MachinePrecision]

Derivation

1. Initial program 89.7%

   \[\left(x \cdot y - z \cdot y\right) \cdot t \]

2. Taylor expanded in x around inf

   \[\leadsto \color{blue}{t \cdot \left(x \cdot y\right)} \]
3. Step-by-step derivation

   1. lower-*.f64 N/A

      \[\leadsto t \cdot \color{blue}{\left(x \cdot y\right)} \]

   2. lower-*.f64 53.4%

      \[\leadsto t \cdot \left(x \cdot \color{blue}{y}\right) \]

4. Applied rewrites 53.4%

   \[\leadsto \color{blue}{t \cdot \left(x \cdot y\right)} \]
5. Add Preprocessing

Reproduce

herbie shell --seed 2025271 -o generate:evaluate
(FPCore (x y z t)
  :name "Linear.Projection:inverseInfinitePerspective from linear-1.19.1.3"
  :precision binary64
  (* (- (* x y) (* z y)) t))