Toniolo and Linder, Equation (10-)

Percentage Accurate: 35.4% → 94.2%

Time: 10.6s

Alternatives: 16

Speedup: 14.4×

• 39.3% of points produce a very large (infinite) output. You may want to add a precondition.

Specification

Math

\[\begin{array}{l} \\ \frac{2}{\left(\left(\frac{{t}^{3}}{\ell \cdot \ell} \cdot \sin k\right) \cdot \tan k\right) \cdot \left(\left(1 + {\left(\frac{k}{t}\right)}^{2}\right) - 1\right)} \end{array} \]

FPCore

(FPCore (t l k)
 :precision binary64
 (/
  2.0
  (*
   (* (* (/ (pow t 3.0) (* l l)) (sin k)) (tan k))
   (- (+ 1.0 (pow (/ k t) 2.0)) 1.0))))

C

double code(double t, double l, double k) {
	return 2.0 / ((((pow(t, 3.0) / (l * l)) * sin(k)) * tan(k)) * ((1.0 + pow((k / t), 2.0)) - 1.0));
}

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(t, l, k)
use fmin_fmax_functions
    real(8), intent (in) :: t
    real(8), intent (in) :: l
    real(8), intent (in) :: k
    code = 2.0d0 / (((((t ** 3.0d0) / (l * l)) * sin(k)) * tan(k)) * ((1.0d0 + ((k / t) ** 2.0d0)) - 1.0d0))
end function

Java

public static double code(double t, double l, double k) {
	return 2.0 / ((((Math.pow(t, 3.0) / (l * l)) * Math.sin(k)) * Math.tan(k)) * ((1.0 + Math.pow((k / t), 2.0)) - 1.0));
}

Python

def code(t, l, k):
	return 2.0 / ((((math.pow(t, 3.0) / (l * l)) * math.sin(k)) * math.tan(k)) * ((1.0 + math.pow((k / t), 2.0)) - 1.0))

Julia

function code(t, l, k)
	return Float64(2.0 / Float64(Float64(Float64(Float64((t ^ 3.0) / Float64(l * l)) * sin(k)) * tan(k)) * Float64(Float64(1.0 + (Float64(k / t) ^ 2.0)) - 1.0)))
end

MATLAB

function tmp = code(t, l, k)
	tmp = 2.0 / (((((t ^ 3.0) / (l * l)) * sin(k)) * tan(k)) * ((1.0 + ((k / t) ^ 2.0)) - 1.0));
end

Wolfram

code[t_, l_, k_] := N[(2.0 / N[(N[(N[(N[(N[Power[t, 3.0], $MachinePrecision] / N[(l * l), $MachinePrecision]), $MachinePrecision] * N[Sin[k], $MachinePrecision]), $MachinePrecision] * N[Tan[k], $MachinePrecision]), $MachinePrecision] * N[(N[(1.0 + N[Power[N[(k / t), $MachinePrecision], 2.0], $MachinePrecision]), $MachinePrecision] - 1.0), $MachinePrecision]), $MachinePrecision]), $MachinePrecision]

Initial Program: 35.4% accurate, 1.0× speedup

Math

\[\begin{array}{l} \\ \frac{2}{\left(\left(\frac{{t}^{3}}{\ell \cdot \ell} \cdot \sin k\right) \cdot \tan k\right) \cdot \left(\left(1 + {\left(\frac{k}{t}\right)}^{2}\right) - 1\right)} \end{array} \]

FPCore

(FPCore (t l k)
 :precision binary64
 (/
  2.0
  (*
   (* (* (/ (pow t 3.0) (* l l)) (sin k)) (tan k))
   (- (+ 1.0 (pow (/ k t) 2.0)) 1.0))))

C

double code(double t, double l, double k) {
	return 2.0 / ((((pow(t, 3.0) / (l * l)) * sin(k)) * tan(k)) * ((1.0 + pow((k / t), 2.0)) - 1.0));
}

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(t, l, k)
use fmin_fmax_functions
    real(8), intent (in) :: t
    real(8), intent (in) :: l
    real(8), intent (in) :: k
    code = 2.0d0 / (((((t ** 3.0d0) / (l * l)) * sin(k)) * tan(k)) * ((1.0d0 + ((k / t) ** 2.0d0)) - 1.0d0))
end function

Java

public static double code(double t, double l, double k) {
	return 2.0 / ((((Math.pow(t, 3.0) / (l * l)) * Math.sin(k)) * Math.tan(k)) * ((1.0 + Math.pow((k / t), 2.0)) - 1.0));
}

Python

def code(t, l, k):
	return 2.0 / ((((math.pow(t, 3.0) / (l * l)) * math.sin(k)) * math.tan(k)) * ((1.0 + math.pow((k / t), 2.0)) - 1.0))

Julia

function code(t, l, k)
	return Float64(2.0 / Float64(Float64(Float64(Float64((t ^ 3.0) / Float64(l * l)) * sin(k)) * tan(k)) * Float64(Float64(1.0 + (Float64(k / t) ^ 2.0)) - 1.0)))
end

MATLAB

function tmp = code(t, l, k)
	tmp = 2.0 / (((((t ^ 3.0) / (l * l)) * sin(k)) * tan(k)) * ((1.0 + ((k / t) ^ 2.0)) - 1.0));
end

Wolfram

code[t_, l_, k_] := N[(2.0 / N[(N[(N[(N[(N[Power[t, 3.0], $MachinePrecision] / N[(l * l), $MachinePrecision]), $MachinePrecision] * N[Sin[k], $MachinePrecision]), $MachinePrecision] * N[Tan[k], $MachinePrecision]), $MachinePrecision] * N[(N[(1.0 + N[Power[N[(k / t), $MachinePrecision], 2.0], $MachinePrecision]), $MachinePrecision] - 1.0), $MachinePrecision]), $MachinePrecision]), $MachinePrecision]

Alternative 1: 94.2% accurate, 1.3× speedup

Math

\[\begin{array}{l} k_m = \left|k\right| \\ \begin{array}{l} \mathbf{if}\;k\_m \leq 1.2 \cdot 10^{-39}:\\ \;\;\;\;\frac{\left(\left(\frac{\frac{\mathsf{fma}\left(0.3333333333333333, \left(k\_m \cdot k\_m\right) \cdot \ell, \ell\right)}{k\_m}}{k\_m} \cdot \cos k\_m\right) \cdot \frac{\ell}{k\_m}\right) \cdot 2}{k\_m \cdot t}\\ \mathbf{else}:\\ \;\;\;\;2 \cdot \frac{\left(\frac{\cos k\_m}{k\_m} \cdot \frac{\ell}{t}\right) \cdot \frac{\ell}{{\sin k\_m}^{2}}}{k\_m}\\ \end{array} \end{array} \]

FPCore

k_m = (fabs.f64 k)
(FPCore (t l k_m)
 :precision binary64
 (if (<= k_m 1.2e-39)
   (/
    (*
     (*
      (*
       (/ (/ (fma 0.3333333333333333 (* (* k_m k_m) l) l) k_m) k_m)
       (cos k_m))
      (/ l k_m))
     2.0)
    (* k_m t))
   (*
    2.0
    (/ (* (* (/ (cos k_m) k_m) (/ l t)) (/ l (pow (sin k_m) 2.0))) k_m))))

C

k_m = fabs(k);
double code(double t, double l, double k_m) {
	double tmp;
	if (k_m <= 1.2e-39) {
		tmp = (((((fma(0.3333333333333333, ((k_m * k_m) * l), l) / k_m) / k_m) * cos(k_m)) * (l / k_m)) * 2.0) / (k_m * t);
	} else {
		tmp = 2.0 * ((((cos(k_m) / k_m) * (l / t)) * (l / pow(sin(k_m), 2.0))) / k_m);
	}
	return tmp;
}

Julia

k_m = abs(k)
function code(t, l, k_m)
	tmp = 0.0
	if (k_m <= 1.2e-39)
		tmp = Float64(Float64(Float64(Float64(Float64(Float64(fma(0.3333333333333333, Float64(Float64(k_m * k_m) * l), l) / k_m) / k_m) * cos(k_m)) * Float64(l / k_m)) * 2.0) / Float64(k_m * t));
	else
		tmp = Float64(2.0 * Float64(Float64(Float64(Float64(cos(k_m) / k_m) * Float64(l / t)) * Float64(l / (sin(k_m) ^ 2.0))) / k_m));
	end
	return tmp
end

Wolfram

k_m = N[Abs[k], $MachinePrecision]
code[t_, l_, k$95$m_] := If[LessEqual[k$95$m, 1.2e-39], N[(N[(N[(N[(N[(N[(N[(0.3333333333333333 * N[(N[(k$95$m * k$95$m), $MachinePrecision] * l), $MachinePrecision] + l), $MachinePrecision] / k$95$m), $MachinePrecision] / k$95$m), $MachinePrecision] * N[Cos[k$95$m], $MachinePrecision]), $MachinePrecision] * N[(l / k$95$m), $MachinePrecision]), $MachinePrecision] * 2.0), $MachinePrecision] / N[(k$95$m * t), $MachinePrecision]), $MachinePrecision], N[(2.0 * N[(N[(N[(N[(N[Cos[k$95$m], $MachinePrecision] / k$95$m), $MachinePrecision] * N[(l / t), $MachinePrecision]), $MachinePrecision] * N[(l / N[Power[N[Sin[k$95$m], $MachinePrecision], 2.0], $MachinePrecision]), $MachinePrecision]), $MachinePrecision] / k$95$m), $MachinePrecision]), $MachinePrecision]]

Derivation

1. Split input into 2 regimes

2. if k < 1.20000000000000008e-39

   1. Initial program 42.2%

      \[\frac{2}{\left(\left(\frac{{t}^{3}}{\ell \cdot \ell} \cdot \sin k\right) \cdot \tan k\right) \cdot \left(\left(1 + {\left(\frac{k}{t}\right)}^{2}\right) - 1\right)} \]
   2. Add Preprocessing

   3. Taylor expanded in t around 0

      \[\leadsto \color{blue}{2 \cdot \frac{{\ell}^{2} \cdot \cos k}{{k}^{2} \cdot \left(t \cdot {\sin k}^{2}\right)}} \]
   4. Step-by-step derivation

   5. Applied rewrites 72.5%

      \[\leadsto \color{blue}{\frac{2}{\left(k \cdot k\right) \cdot t} \cdot \frac{\left(\cos k \cdot \ell\right) \cdot \ell}{{\sin k}^{2}}} \]
   6. Step-by-step derivation

   7. Applied rewrites 87.1%

      \[\leadsto \frac{2}{k} \cdot \color{blue}{\frac{\frac{\ell}{{\sin k}^{2}} \cdot \left(\cos k \cdot \ell\right)}{k \cdot t}} \]

   8. Taylor expanded in k around 0

      \[\leadsto \frac{2}{k} \cdot \frac{\frac{\ell + \frac{1}{3} \cdot \left({k}^{2} \cdot \ell\right)}{{k}^{2}} \cdot \left(\cos k \cdot \ell\right)}{k \cdot t} \]
   9. Step-by-step derivation

   10. Applied rewrites 68.9%

       \[\leadsto \frac{2}{k} \cdot \frac{\frac{\frac{\mathsf{fma}\left(\left(k \cdot k\right) \cdot \ell, 0.3333333333333333, \ell\right)}{k}}{k} \cdot \left(\cos k \cdot \ell\right)}{k \cdot t} \]
   11. Step-by-step derivation

   12. Applied rewrites 70.3%

       \[\leadsto \frac{\left(\left(\frac{\frac{\mathsf{fma}\left(0.3333333333333333, \left(k \cdot k\right) \cdot \ell, \ell\right)}{k}}{k} \cdot \cos k\right) \cdot \frac{\ell}{k}\right) \cdot 2}{\color{blue}{k \cdot t}} \]

   if 1.20000000000000008e-39 < k

   1. Initial program 29.1%

      \[\frac{2}{\left(\left(\frac{{t}^{3}}{\ell \cdot \ell} \cdot \sin k\right) \cdot \tan k\right) \cdot \left(\left(1 + {\left(\frac{k}{t}\right)}^{2}\right) - 1\right)} \]
   2. Add Preprocessing

   3. Taylor expanded in t around 0

      \[\leadsto \color{blue}{2 \cdot \frac{{\ell}^{2} \cdot \cos k}{{k}^{2} \cdot \left(t \cdot {\sin k}^{2}\right)}} \]
   4. Step-by-step derivation

   5. Applied rewrites 78.5%

      \[\leadsto \color{blue}{\frac{2}{\left(k \cdot k\right) \cdot t} \cdot \frac{\left(\cos k \cdot \ell\right) \cdot \ell}{{\sin k}^{2}}} \]
   6. Step-by-step derivation

   7. Applied rewrites 82.8%

      \[\leadsto \frac{2}{k} \cdot \color{blue}{\frac{\frac{\ell}{{\sin k}^{2}} \cdot \left(\cos k \cdot \ell\right)}{k \cdot t}} \]
   8. Step-by-step derivation

   9. Applied rewrites 92.9%

      \[\leadsto 2 \cdot \color{blue}{\frac{\frac{\frac{\cos k \cdot \ell}{t}}{k} \cdot \frac{\ell}{{\sin k}^{2}}}{k}} \]
   10. Step-by-step derivation

   11. Applied rewrites 92.9%

       \[\leadsto 2 \cdot \frac{\left(\frac{\cos k}{k} \cdot \frac{\ell}{t}\right) \cdot \frac{\ell}{{\sin k}^{2}}}{k} \]
3. Recombined 2 regimes into one program.
4. Add Preprocessing

Alternative 2: 94.2% accurate, 1.3× speedup

Math

\[\begin{array}{l} k_m = \left|k\right| \\ \begin{array}{l} \mathbf{if}\;k\_m \leq 1.2 \cdot 10^{-39}:\\ \;\;\;\;\frac{\left(\left(\frac{\frac{\mathsf{fma}\left(0.3333333333333333, \left(k\_m \cdot k\_m\right) \cdot \ell, \ell\right)}{k\_m}}{k\_m} \cdot \cos k\_m\right) \cdot \frac{\ell}{k\_m}\right) \cdot 2}{k\_m \cdot t}\\ \mathbf{else}:\\ \;\;\;\;\frac{2}{k\_m} \cdot \left(\frac{\cos k\_m \cdot \ell}{t} \cdot \frac{\ell}{{\sin k\_m}^{2} \cdot k\_m}\right)\\ \end{array} \end{array} \]

FPCore

k_m = (fabs.f64 k)
(FPCore (t l k_m)
 :precision binary64
 (if (<= k_m 1.2e-39)
   (/
    (*
     (*
      (*
       (/ (/ (fma 0.3333333333333333 (* (* k_m k_m) l) l) k_m) k_m)
       (cos k_m))
      (/ l k_m))
     2.0)
    (* k_m t))
   (*
    (/ 2.0 k_m)
    (* (/ (* (cos k_m) l) t) (/ l (* (pow (sin k_m) 2.0) k_m))))))

C

k_m = fabs(k);
double code(double t, double l, double k_m) {
	double tmp;
	if (k_m <= 1.2e-39) {
		tmp = (((((fma(0.3333333333333333, ((k_m * k_m) * l), l) / k_m) / k_m) * cos(k_m)) * (l / k_m)) * 2.0) / (k_m * t);
	} else {
		tmp = (2.0 / k_m) * (((cos(k_m) * l) / t) * (l / (pow(sin(k_m), 2.0) * k_m)));
	}
	return tmp;
}

Julia

k_m = abs(k)
function code(t, l, k_m)
	tmp = 0.0
	if (k_m <= 1.2e-39)
		tmp = Float64(Float64(Float64(Float64(Float64(Float64(fma(0.3333333333333333, Float64(Float64(k_m * k_m) * l), l) / k_m) / k_m) * cos(k_m)) * Float64(l / k_m)) * 2.0) / Float64(k_m * t));
	else
		tmp = Float64(Float64(2.0 / k_m) * Float64(Float64(Float64(cos(k_m) * l) / t) * Float64(l / Float64((sin(k_m) ^ 2.0) * k_m))));
	end
	return tmp
end

Wolfram

k_m = N[Abs[k], $MachinePrecision]
code[t_, l_, k$95$m_] := If[LessEqual[k$95$m, 1.2e-39], N[(N[(N[(N[(N[(N[(N[(0.3333333333333333 * N[(N[(k$95$m * k$95$m), $MachinePrecision] * l), $MachinePrecision] + l), $MachinePrecision] / k$95$m), $MachinePrecision] / k$95$m), $MachinePrecision] * N[Cos[k$95$m], $MachinePrecision]), $MachinePrecision] * N[(l / k$95$m), $MachinePrecision]), $MachinePrecision] * 2.0), $MachinePrecision] / N[(k$95$m * t), $MachinePrecision]), $MachinePrecision], N[(N[(2.0 / k$95$m), $MachinePrecision] * N[(N[(N[(N[Cos[k$95$m], $MachinePrecision] * l), $MachinePrecision] / t), $MachinePrecision] * N[(l / N[(N[Power[N[Sin[k$95$m], $MachinePrecision], 2.0], $MachinePrecision] * k$95$m), $MachinePrecision]), $MachinePrecision]), $MachinePrecision]), $MachinePrecision]]

Derivation

1. Split input into 2 regimes

2. if k < 1.20000000000000008e-39

   1. Initial program 42.2%

      \[\frac{2}{\left(\left(\frac{{t}^{3}}{\ell \cdot \ell} \cdot \sin k\right) \cdot \tan k\right) \cdot \left(\left(1 + {\left(\frac{k}{t}\right)}^{2}\right) - 1\right)} \]
   2. Add Preprocessing

   3. Taylor expanded in t around 0

      \[\leadsto \color{blue}{2 \cdot \frac{{\ell}^{2} \cdot \cos k}{{k}^{2} \cdot \left(t \cdot {\sin k}^{2}\right)}} \]
   4. Step-by-step derivation

   5. Applied rewrites 72.5%

      \[\leadsto \color{blue}{\frac{2}{\left(k \cdot k\right) \cdot t} \cdot \frac{\left(\cos k \cdot \ell\right) \cdot \ell}{{\sin k}^{2}}} \]
   6. Step-by-step derivation

   7. Applied rewrites 87.1%

      \[\leadsto \frac{2}{k} \cdot \color{blue}{\frac{\frac{\ell}{{\sin k}^{2}} \cdot \left(\cos k \cdot \ell\right)}{k \cdot t}} \]

   8. Taylor expanded in k around 0

      \[\leadsto \frac{2}{k} \cdot \frac{\frac{\ell + \frac{1}{3} \cdot \left({k}^{2} \cdot \ell\right)}{{k}^{2}} \cdot \left(\cos k \cdot \ell\right)}{k \cdot t} \]
   9. Step-by-step derivation

   10. Applied rewrites 68.9%

       \[\leadsto \frac{2}{k} \cdot \frac{\frac{\frac{\mathsf{fma}\left(\left(k \cdot k\right) \cdot \ell, 0.3333333333333333, \ell\right)}{k}}{k} \cdot \left(\cos k \cdot \ell\right)}{k \cdot t} \]
   11. Step-by-step derivation

   12. Applied rewrites 70.3%

       \[\leadsto \frac{\left(\left(\frac{\frac{\mathsf{fma}\left(0.3333333333333333, \left(k \cdot k\right) \cdot \ell, \ell\right)}{k}}{k} \cdot \cos k\right) \cdot \frac{\ell}{k}\right) \cdot 2}{\color{blue}{k \cdot t}} \]

   if 1.20000000000000008e-39 < k

   1. Initial program 29.1%

      \[\frac{2}{\left(\left(\frac{{t}^{3}}{\ell \cdot \ell} \cdot \sin k\right) \cdot \tan k\right) \cdot \left(\left(1 + {\left(\frac{k}{t}\right)}^{2}\right) - 1\right)} \]
   2. Add Preprocessing

   3. Taylor expanded in t around 0

      \[\leadsto \color{blue}{2 \cdot \frac{{\ell}^{2} \cdot \cos k}{{k}^{2} \cdot \left(t \cdot {\sin k}^{2}\right)}} \]
   4. Step-by-step derivation

   5. Applied rewrites 78.5%

      \[\leadsto \color{blue}{\frac{2}{\left(k \cdot k\right) \cdot t} \cdot \frac{\left(\cos k \cdot \ell\right) \cdot \ell}{{\sin k}^{2}}} \]
   6. Step-by-step derivation

   7. Applied rewrites 82.8%

      \[\leadsto \frac{2}{k} \cdot \color{blue}{\frac{\frac{\ell}{{\sin k}^{2}} \cdot \left(\cos k \cdot \ell\right)}{k \cdot t}} \]
   8. Step-by-step derivation

   9. Applied rewrites 92.8%

      \[\leadsto \frac{2}{k} \cdot \left(\frac{\cos k \cdot \ell}{t} \cdot \color{blue}{\frac{\ell}{{\sin k}^{2} \cdot k}}\right) \]
3. Recombined 2 regimes into one program.
4. Add Preprocessing

Alternative 3: 92.0% accurate, 1.3× speedup

Math

\[\begin{array}{l} k_m = \left|k\right| \\ \begin{array}{l} t_1 := \cos k\_m \cdot \ell\\ \mathbf{if}\;k\_m \leq 2.65 \cdot 10^{+73}:\\ \;\;\;\;\left(\frac{2}{\left(k\_m \cdot k\_m\right) \cdot t} \cdot t\_1\right) \cdot \frac{\ell}{{\sin k\_m}^{2}}\\ \mathbf{else}:\\ \;\;\;\;2 \cdot \frac{\frac{\frac{t\_1}{t}}{k\_m} \cdot \frac{\ell}{0.5 - 0.5 \cdot \cos \left(2 \cdot k\_m\right)}}{k\_m}\\ \end{array} \end{array} \]

FPCore

k_m = (fabs.f64 k)
(FPCore (t l k_m)
 :precision binary64
 (let* ((t_1 (* (cos k_m) l)))
   (if (<= k_m 2.65e+73)
     (* (* (/ 2.0 (* (* k_m k_m) t)) t_1) (/ l (pow (sin k_m) 2.0)))
     (*
      2.0
      (/ (* (/ (/ t_1 t) k_m) (/ l (- 0.5 (* 0.5 (cos (* 2.0 k_m)))))) k_m)))))

C

k_m = fabs(k);
double code(double t, double l, double k_m) {
	double t_1 = cos(k_m) * l;
	double tmp;
	if (k_m <= 2.65e+73) {
		tmp = ((2.0 / ((k_m * k_m) * t)) * t_1) * (l / pow(sin(k_m), 2.0));
	} else {
		tmp = 2.0 * ((((t_1 / t) / k_m) * (l / (0.5 - (0.5 * cos((2.0 * k_m)))))) / k_m);
	}
	return tmp;
}

Fortran

k_m =     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(t, l, k_m)
use fmin_fmax_functions
    real(8), intent (in) :: t
    real(8), intent (in) :: l
    real(8), intent (in) :: k_m
    real(8) :: t_1
    real(8) :: tmp
    t_1 = cos(k_m) * l
    if (k_m <= 2.65d+73) then
        tmp = ((2.0d0 / ((k_m * k_m) * t)) * t_1) * (l / (sin(k_m) ** 2.0d0))
    else
        tmp = 2.0d0 * ((((t_1 / t) / k_m) * (l / (0.5d0 - (0.5d0 * cos((2.0d0 * k_m)))))) / k_m)
    end if
    code = tmp
end function

Java

k_m = Math.abs(k);
public static double code(double t, double l, double k_m) {
	double t_1 = Math.cos(k_m) * l;
	double tmp;
	if (k_m <= 2.65e+73) {
		tmp = ((2.0 / ((k_m * k_m) * t)) * t_1) * (l / Math.pow(Math.sin(k_m), 2.0));
	} else {
		tmp = 2.0 * ((((t_1 / t) / k_m) * (l / (0.5 - (0.5 * Math.cos((2.0 * k_m)))))) / k_m);
	}
	return tmp;
}

Python

k_m = math.fabs(k)
def code(t, l, k_m):
	t_1 = math.cos(k_m) * l
	tmp = 0
	if k_m <= 2.65e+73:
		tmp = ((2.0 / ((k_m * k_m) * t)) * t_1) * (l / math.pow(math.sin(k_m), 2.0))
	else:
		tmp = 2.0 * ((((t_1 / t) / k_m) * (l / (0.5 - (0.5 * math.cos((2.0 * k_m)))))) / k_m)
	return tmp

Julia

k_m = abs(k)
function code(t, l, k_m)
	t_1 = Float64(cos(k_m) * l)
	tmp = 0.0
	if (k_m <= 2.65e+73)
		tmp = Float64(Float64(Float64(2.0 / Float64(Float64(k_m * k_m) * t)) * t_1) * Float64(l / (sin(k_m) ^ 2.0)));
	else
		tmp = Float64(2.0 * Float64(Float64(Float64(Float64(t_1 / t) / k_m) * Float64(l / Float64(0.5 - Float64(0.5 * cos(Float64(2.0 * k_m)))))) / k_m));
	end
	return tmp
end

MATLAB

k_m = abs(k);
function tmp_2 = code(t, l, k_m)
	t_1 = cos(k_m) * l;
	tmp = 0.0;
	if (k_m <= 2.65e+73)
		tmp = ((2.0 / ((k_m * k_m) * t)) * t_1) * (l / (sin(k_m) ^ 2.0));
	else
		tmp = 2.0 * ((((t_1 / t) / k_m) * (l / (0.5 - (0.5 * cos((2.0 * k_m)))))) / k_m);
	end
	tmp_2 = tmp;
end

Wolfram

k_m = N[Abs[k], $MachinePrecision]
code[t_, l_, k$95$m_] := Block[{t$95$1 = N[(N[Cos[k$95$m], $MachinePrecision] * l), $MachinePrecision]}, If[LessEqual[k$95$m, 2.65e+73], N[(N[(N[(2.0 / N[(N[(k$95$m * k$95$m), $MachinePrecision] * t), $MachinePrecision]), $MachinePrecision] * t$95$1), $MachinePrecision] * N[(l / N[Power[N[Sin[k$95$m], $MachinePrecision], 2.0], $MachinePrecision]), $MachinePrecision]), $MachinePrecision], N[(2.0 * N[(N[(N[(N[(t$95$1 / t), $MachinePrecision] / k$95$m), $MachinePrecision] * N[(l / N[(0.5 - N[(0.5 * N[Cos[N[(2.0 * k$95$m), $MachinePrecision]], $MachinePrecision]), $MachinePrecision]), $MachinePrecision]), $MachinePrecision]), $MachinePrecision] / k$95$m), $MachinePrecision]), $MachinePrecision]]]

Derivation

1. Split input into 2 regimes

2. if k < 2.64999999999999998e73

   1. Initial program 41.7%

      \[\frac{2}{\left(\left(\frac{{t}^{3}}{\ell \cdot \ell} \cdot \sin k\right) \cdot \tan k\right) \cdot \left(\left(1 + {\left(\frac{k}{t}\right)}^{2}\right) - 1\right)} \]
   2. Add Preprocessing

   3. Taylor expanded in t around 0

      \[\leadsto \color{blue}{2 \cdot \frac{{\ell}^{2} \cdot \cos k}{{k}^{2} \cdot \left(t \cdot {\sin k}^{2}\right)}} \]
   4. Step-by-step derivation

   5. Applied rewrites 74.2%

      \[\leadsto \color{blue}{\frac{2}{\left(k \cdot k\right) \cdot t} \cdot \frac{\left(\cos k \cdot \ell\right) \cdot \ell}{{\sin k}^{2}}} \]
   6. Step-by-step derivation

   7. Applied rewrites 86.3%

      \[\leadsto \left(\frac{2}{\left(k \cdot k\right) \cdot t} \cdot \left(\cos k \cdot \ell\right)\right) \cdot \color{blue}{\frac{\ell}{{\sin k}^{2}}} \]

   if 2.64999999999999998e73 < k

   1. Initial program 24.4%

      \[\frac{2}{\left(\left(\frac{{t}^{3}}{\ell \cdot \ell} \cdot \sin k\right) \cdot \tan k\right) \cdot \left(\left(1 + {\left(\frac{k}{t}\right)}^{2}\right) - 1\right)} \]
   2. Add Preprocessing

   3. Taylor expanded in t around 0

      \[\leadsto \color{blue}{2 \cdot \frac{{\ell}^{2} \cdot \cos k}{{k}^{2} \cdot \left(t \cdot {\sin k}^{2}\right)}} \]
   4. Step-by-step derivation

   5. Applied rewrites 73.8%

      \[\leadsto \color{blue}{\frac{2}{\left(k \cdot k\right) \cdot t} \cdot \frac{\left(\cos k \cdot \ell\right) \cdot \ell}{{\sin k}^{2}}} \]
   6. Step-by-step derivation

   7. Applied rewrites 82.2%

      \[\leadsto \frac{2}{k} \cdot \color{blue}{\frac{\frac{\ell}{{\sin k}^{2}} \cdot \left(\cos k \cdot \ell\right)}{k \cdot t}} \]
   8. Step-by-step derivation

   9. Applied rewrites 93.3%

      \[\leadsto 2 \cdot \color{blue}{\frac{\frac{\frac{\cos k \cdot \ell}{t}}{k} \cdot \frac{\ell}{{\sin k}^{2}}}{k}} \]
   10. Step-by-step derivation

   11. Applied rewrites 93.3%

       \[\leadsto 2 \cdot \frac{\frac{\frac{\cos k \cdot \ell}{t}}{k} \cdot \frac{\ell}{0.5 - 0.5 \cdot \cos \left(2 \cdot k\right)}}{k} \]
3. Recombined 2 regimes into one program.
4. Add Preprocessing

Alternative 4: 92.0% accurate, 1.3× speedup

Math

\[\begin{array}{l} k_m = \left|k\right| \\ \begin{array}{l} \mathbf{if}\;k\_m \leq 2.65 \cdot 10^{+73}:\\ \;\;\;\;\left(\cos k\_m \cdot \frac{\ell}{{\sin k\_m}^{2}}\right) \cdot \left(\ell \cdot \frac{2}{\left(k\_m \cdot k\_m\right) \cdot t}\right)\\ \mathbf{else}:\\ \;\;\;\;2 \cdot \frac{\frac{\frac{\cos k\_m \cdot \ell}{t}}{k\_m} \cdot \frac{\ell}{0.5 - 0.5 \cdot \cos \left(2 \cdot k\_m\right)}}{k\_m}\\ \end{array} \end{array} \]

FPCore

k_m = (fabs.f64 k)
(FPCore (t l k_m)
 :precision binary64
 (if (<= k_m 2.65e+73)
   (* (* (cos k_m) (/ l (pow (sin k_m) 2.0))) (* l (/ 2.0 (* (* k_m k_m) t))))
   (*
    2.0
    (/
     (* (/ (/ (* (cos k_m) l) t) k_m) (/ l (- 0.5 (* 0.5 (cos (* 2.0 k_m))))))
     k_m))))

C

k_m = fabs(k);
double code(double t, double l, double k_m) {
	double tmp;
	if (k_m <= 2.65e+73) {
		tmp = (cos(k_m) * (l / pow(sin(k_m), 2.0))) * (l * (2.0 / ((k_m * k_m) * t)));
	} else {
		tmp = 2.0 * (((((cos(k_m) * l) / t) / k_m) * (l / (0.5 - (0.5 * cos((2.0 * k_m)))))) / k_m);
	}
	return tmp;
}

Fortran

k_m =     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(t, l, k_m)
use fmin_fmax_functions
    real(8), intent (in) :: t
    real(8), intent (in) :: l
    real(8), intent (in) :: k_m
    real(8) :: tmp
    if (k_m <= 2.65d+73) then
        tmp = (cos(k_m) * (l / (sin(k_m) ** 2.0d0))) * (l * (2.0d0 / ((k_m * k_m) * t)))
    else
        tmp = 2.0d0 * (((((cos(k_m) * l) / t) / k_m) * (l / (0.5d0 - (0.5d0 * cos((2.0d0 * k_m)))))) / k_m)
    end if
    code = tmp
end function

Java

k_m = Math.abs(k);
public static double code(double t, double l, double k_m) {
	double tmp;
	if (k_m <= 2.65e+73) {
		tmp = (Math.cos(k_m) * (l / Math.pow(Math.sin(k_m), 2.0))) * (l * (2.0 / ((k_m * k_m) * t)));
	} else {
		tmp = 2.0 * (((((Math.cos(k_m) * l) / t) / k_m) * (l / (0.5 - (0.5 * Math.cos((2.0 * k_m)))))) / k_m);
	}
	return tmp;
}

Python

k_m = math.fabs(k)
def code(t, l, k_m):
	tmp = 0
	if k_m <= 2.65e+73:
		tmp = (math.cos(k_m) * (l / math.pow(math.sin(k_m), 2.0))) * (l * (2.0 / ((k_m * k_m) * t)))
	else:
		tmp = 2.0 * (((((math.cos(k_m) * l) / t) / k_m) * (l / (0.5 - (0.5 * math.cos((2.0 * k_m)))))) / k_m)
	return tmp

Julia

k_m = abs(k)
function code(t, l, k_m)
	tmp = 0.0
	if (k_m <= 2.65e+73)
		tmp = Float64(Float64(cos(k_m) * Float64(l / (sin(k_m) ^ 2.0))) * Float64(l * Float64(2.0 / Float64(Float64(k_m * k_m) * t))));
	else
		tmp = Float64(2.0 * Float64(Float64(Float64(Float64(Float64(cos(k_m) * l) / t) / k_m) * Float64(l / Float64(0.5 - Float64(0.5 * cos(Float64(2.0 * k_m)))))) / k_m));
	end
	return tmp
end

MATLAB

k_m = abs(k);
function tmp_2 = code(t, l, k_m)
	tmp = 0.0;
	if (k_m <= 2.65e+73)
		tmp = (cos(k_m) * (l / (sin(k_m) ^ 2.0))) * (l * (2.0 / ((k_m * k_m) * t)));
	else
		tmp = 2.0 * (((((cos(k_m) * l) / t) / k_m) * (l / (0.5 - (0.5 * cos((2.0 * k_m)))))) / k_m);
	end
	tmp_2 = tmp;
end

Wolfram

k_m = N[Abs[k], $MachinePrecision]
code[t_, l_, k$95$m_] := If[LessEqual[k$95$m, 2.65e+73], N[(N[(N[Cos[k$95$m], $MachinePrecision] * N[(l / N[Power[N[Sin[k$95$m], $MachinePrecision], 2.0], $MachinePrecision]), $MachinePrecision]), $MachinePrecision] * N[(l * N[(2.0 / N[(N[(k$95$m * k$95$m), $MachinePrecision] * t), $MachinePrecision]), $MachinePrecision]), $MachinePrecision]), $MachinePrecision], N[(2.0 * N[(N[(N[(N[(N[(N[Cos[k$95$m], $MachinePrecision] * l), $MachinePrecision] / t), $MachinePrecision] / k$95$m), $MachinePrecision] * N[(l / N[(0.5 - N[(0.5 * N[Cos[N[(2.0 * k$95$m), $MachinePrecision]], $MachinePrecision]), $MachinePrecision]), $MachinePrecision]), $MachinePrecision]), $MachinePrecision] / k$95$m), $MachinePrecision]), $MachinePrecision]]

Derivation

1. Split input into 2 regimes

2. if k < 2.64999999999999998e73

   1. Initial program 41.7%

      \[\frac{2}{\left(\left(\frac{{t}^{3}}{\ell \cdot \ell} \cdot \sin k\right) \cdot \tan k\right) \cdot \left(\left(1 + {\left(\frac{k}{t}\right)}^{2}\right) - 1\right)} \]
   2. Add Preprocessing

   3. Taylor expanded in t around 0

      \[\leadsto \color{blue}{2 \cdot \frac{{\ell}^{2} \cdot \cos k}{{k}^{2} \cdot \left(t \cdot {\sin k}^{2}\right)}} \]
   4. Step-by-step derivation

   5. Applied rewrites 74.2%

      \[\leadsto \color{blue}{\frac{2}{\left(k \cdot k\right) \cdot t} \cdot \frac{\left(\cos k \cdot \ell\right) \cdot \ell}{{\sin k}^{2}}} \]
   6. Step-by-step derivation

   7. Applied rewrites 86.7%

      \[\leadsto \frac{2}{k} \cdot \color{blue}{\frac{\frac{\ell}{{\sin k}^{2}} \cdot \left(\cos k \cdot \ell\right)}{k \cdot t}} \]
   8. Step-by-step derivation

   9. Applied rewrites 86.3%

      \[\leadsto \left(\cos k \cdot \frac{\ell}{{\sin k}^{2}}\right) \cdot \color{blue}{\left(\ell \cdot \frac{2}{\left(k \cdot k\right) \cdot t}\right)} \]

   if 2.64999999999999998e73 < k

   1. Initial program 24.4%

      \[\frac{2}{\left(\left(\frac{{t}^{3}}{\ell \cdot \ell} \cdot \sin k\right) \cdot \tan k\right) \cdot \left(\left(1 + {\left(\frac{k}{t}\right)}^{2}\right) - 1\right)} \]
   2. Add Preprocessing

   3. Taylor expanded in t around 0

      \[\leadsto \color{blue}{2 \cdot \frac{{\ell}^{2} \cdot \cos k}{{k}^{2} \cdot \left(t \cdot {\sin k}^{2}\right)}} \]
   4. Step-by-step derivation

   5. Applied rewrites 73.8%

      \[\leadsto \color{blue}{\frac{2}{\left(k \cdot k\right) \cdot t} \cdot \frac{\left(\cos k \cdot \ell\right) \cdot \ell}{{\sin k}^{2}}} \]
   6. Step-by-step derivation

   7. Applied rewrites 82.2%

      \[\leadsto \frac{2}{k} \cdot \color{blue}{\frac{\frac{\ell}{{\sin k}^{2}} \cdot \left(\cos k \cdot \ell\right)}{k \cdot t}} \]
   8. Step-by-step derivation

   9. Applied rewrites 93.3%

      \[\leadsto 2 \cdot \color{blue}{\frac{\frac{\frac{\cos k \cdot \ell}{t}}{k} \cdot \frac{\ell}{{\sin k}^{2}}}{k}} \]
   10. Step-by-step derivation

   11. Applied rewrites 93.3%

       \[\leadsto 2 \cdot \frac{\frac{\frac{\cos k \cdot \ell}{t}}{k} \cdot \frac{\ell}{0.5 - 0.5 \cdot \cos \left(2 \cdot k\right)}}{k} \]
3. Recombined 2 regimes into one program.
4. Add Preprocessing

Alternative 5: 95.0% accurate, 1.7× speedup

Math

\[\begin{array}{l} k_m = \left|k\right| \\ \begin{array}{l} \mathbf{if}\;k\_m \leq 0.0022:\\ \;\;\;\;\frac{2}{k\_m} \cdot \left(\frac{\frac{\frac{\mathsf{fma}\left(0.3333333333333333, \left(k\_m \cdot k\_m\right) \cdot \ell, \ell\right)}{k\_m}}{k\_m} \cdot \cos k\_m}{t} \cdot \frac{\ell}{k\_m}\right)\\ \mathbf{else}:\\ \;\;\;\;2 \cdot \frac{\frac{\frac{\cos k\_m \cdot \ell}{t}}{k\_m} \cdot \frac{\ell}{0.5 - 0.5 \cdot \cos \left(2 \cdot k\_m\right)}}{k\_m}\\ \end{array} \end{array} \]

FPCore

k_m = (fabs.f64 k)
(FPCore (t l k_m)
 :precision binary64
 (if (<= k_m 0.0022)
   (*
    (/ 2.0 k_m)
    (*
     (/
      (*
       (/ (/ (fma 0.3333333333333333 (* (* k_m k_m) l) l) k_m) k_m)
       (cos k_m))
      t)
     (/ l k_m)))
   (*
    2.0
    (/
     (* (/ (/ (* (cos k_m) l) t) k_m) (/ l (- 0.5 (* 0.5 (cos (* 2.0 k_m))))))
     k_m))))

C

k_m = fabs(k);
double code(double t, double l, double k_m) {
	double tmp;
	if (k_m <= 0.0022) {
		tmp = (2.0 / k_m) * (((((fma(0.3333333333333333, ((k_m * k_m) * l), l) / k_m) / k_m) * cos(k_m)) / t) * (l / k_m));
	} else {
		tmp = 2.0 * (((((cos(k_m) * l) / t) / k_m) * (l / (0.5 - (0.5 * cos((2.0 * k_m)))))) / k_m);
	}
	return tmp;
}

Julia

k_m = abs(k)
function code(t, l, k_m)
	tmp = 0.0
	if (k_m <= 0.0022)
		tmp = Float64(Float64(2.0 / k_m) * Float64(Float64(Float64(Float64(Float64(fma(0.3333333333333333, Float64(Float64(k_m * k_m) * l), l) / k_m) / k_m) * cos(k_m)) / t) * Float64(l / k_m)));
	else
		tmp = Float64(2.0 * Float64(Float64(Float64(Float64(Float64(cos(k_m) * l) / t) / k_m) * Float64(l / Float64(0.5 - Float64(0.5 * cos(Float64(2.0 * k_m)))))) / k_m));
	end
	return tmp
end

Wolfram

k_m = N[Abs[k], $MachinePrecision]
code[t_, l_, k$95$m_] := If[LessEqual[k$95$m, 0.0022], N[(N[(2.0 / k$95$m), $MachinePrecision] * N[(N[(N[(N[(N[(N[(0.3333333333333333 * N[(N[(k$95$m * k$95$m), $MachinePrecision] * l), $MachinePrecision] + l), $MachinePrecision] / k$95$m), $MachinePrecision] / k$95$m), $MachinePrecision] * N[Cos[k$95$m], $MachinePrecision]), $MachinePrecision] / t), $MachinePrecision] * N[(l / k$95$m), $MachinePrecision]), $MachinePrecision]), $MachinePrecision], N[(2.0 * N[(N[(N[(N[(N[(N[Cos[k$95$m], $MachinePrecision] * l), $MachinePrecision] / t), $MachinePrecision] / k$95$m), $MachinePrecision] * N[(l / N[(0.5 - N[(0.5 * N[Cos[N[(2.0 * k$95$m), $MachinePrecision]], $MachinePrecision]), $MachinePrecision]), $MachinePrecision]), $MachinePrecision]), $MachinePrecision] / k$95$m), $MachinePrecision]), $MachinePrecision]]

Derivation

1. Split input into 2 regimes

2. if k < 0.00220000000000000013

   1. Initial program 43.0%

      \[\frac{2}{\left(\left(\frac{{t}^{3}}{\ell \cdot \ell} \cdot \sin k\right) \cdot \tan k\right) \cdot \left(\left(1 + {\left(\frac{k}{t}\right)}^{2}\right) - 1\right)} \]
   2. Add Preprocessing

   3. Taylor expanded in t around 0

      \[\leadsto \color{blue}{2 \cdot \frac{{\ell}^{2} \cdot \cos k}{{k}^{2} \cdot \left(t \cdot {\sin k}^{2}\right)}} \]
   4. Step-by-step derivation

   5. Applied rewrites 73.1%

      \[\leadsto \color{blue}{\frac{2}{\left(k \cdot k\right) \cdot t} \cdot \frac{\left(\cos k \cdot \ell\right) \cdot \ell}{{\sin k}^{2}}} \]
   6. Step-by-step derivation

   7. Applied rewrites 87.1%

      \[\leadsto \frac{2}{k} \cdot \color{blue}{\frac{\frac{\ell}{{\sin k}^{2}} \cdot \left(\cos k \cdot \ell\right)}{k \cdot t}} \]

   8. Taylor expanded in k around 0

      \[\leadsto \frac{2}{k} \cdot \frac{\frac{\ell + \frac{1}{3} \cdot \left({k}^{2} \cdot \ell\right)}{{k}^{2}} \cdot \left(\cos k \cdot \ell\right)}{k \cdot t} \]
   9. Step-by-step derivation

   10. Applied rewrites 69.6%

       \[\leadsto \frac{2}{k} \cdot \frac{\frac{\frac{\mathsf{fma}\left(\left(k \cdot k\right) \cdot \ell, 0.3333333333333333, \ell\right)}{k}}{k} \cdot \left(\cos k \cdot \ell\right)}{k \cdot t} \]
   11. Step-by-step derivation

   12. Applied rewrites 71.5%

       \[\leadsto \frac{2}{k} \cdot \left(\frac{\frac{\frac{\mathsf{fma}\left(0.3333333333333333, \left(k \cdot k\right) \cdot \ell, \ell\right)}{k}}{k} \cdot \cos k}{t} \cdot \color{blue}{\frac{\ell}{k}}\right) \]

   if 0.00220000000000000013 < k

   1. Initial program 24.7%

      \[\frac{2}{\left(\left(\frac{{t}^{3}}{\ell \cdot \ell} \cdot \sin k\right) \cdot \tan k\right) \cdot \left(\left(1 + {\left(\frac{k}{t}\right)}^{2}\right) - 1\right)} \]
   2. Add Preprocessing

   3. Taylor expanded in t around 0

      \[\leadsto \color{blue}{2 \cdot \frac{{\ell}^{2} \cdot \cos k}{{k}^{2} \cdot \left(t \cdot {\sin k}^{2}\right)}} \]
   4. Step-by-step derivation

   5. Applied rewrites 77.4%

      \[\leadsto \color{blue}{\frac{2}{\left(k \cdot k\right) \cdot t} \cdot \frac{\left(\cos k \cdot \ell\right) \cdot \ell}{{\sin k}^{2}}} \]
   6. Step-by-step derivation

   7. Applied rewrites 82.2%

      \[\leadsto \frac{2}{k} \cdot \color{blue}{\frac{\frac{\ell}{{\sin k}^{2}} \cdot \left(\cos k \cdot \ell\right)}{k \cdot t}} \]
   8. Step-by-step derivation

   9. Applied rewrites 92.0%

      \[\leadsto 2 \cdot \color{blue}{\frac{\frac{\frac{\cos k \cdot \ell}{t}}{k} \cdot \frac{\ell}{{\sin k}^{2}}}{k}} \]
   10. Step-by-step derivation

   11. Applied rewrites 92.0%

       \[\leadsto 2 \cdot \frac{\frac{\frac{\cos k \cdot \ell}{t}}{k} \cdot \frac{\ell}{0.5 - 0.5 \cdot \cos \left(2 \cdot k\right)}}{k} \]
3. Recombined 2 regimes into one program.
4. Add Preprocessing

Alternative 6: 86.2% accurate, 1.7× speedup

Math

\[\begin{array}{l} k_m = \left|k\right| \\ \begin{array}{l} \mathbf{if}\;k\_m \leq 0.0068:\\ \;\;\;\;\frac{2}{k\_m} \cdot \left(\frac{\frac{\frac{\mathsf{fma}\left(0.3333333333333333, \left(k\_m \cdot k\_m\right) \cdot \ell, \ell\right)}{k\_m}}{k\_m} \cdot \cos k\_m}{t} \cdot \frac{\ell}{k\_m}\right)\\ \mathbf{elif}\;k\_m \leq 1.6 \cdot 10^{+149}:\\ \;\;\;\;\frac{2}{\left(k\_m \cdot k\_m\right) \cdot t} \cdot \frac{\left(\cos k\_m \cdot \ell\right) \cdot \ell}{\mathsf{fma}\left(\cos \left(-2 \cdot k\_m\right), -0.5, 0.5\right)}\\ \mathbf{else}:\\ \;\;\;\;\frac{-0.3333333333333333}{t} \cdot \left(\frac{\ell}{k\_m} \cdot \frac{\ell}{k\_m}\right)\\ \end{array} \end{array} \]

FPCore

k_m = (fabs.f64 k)
(FPCore (t l k_m)
 :precision binary64
 (if (<= k_m 0.0068)
   (*
    (/ 2.0 k_m)
    (*
     (/
      (*
       (/ (/ (fma 0.3333333333333333 (* (* k_m k_m) l) l) k_m) k_m)
       (cos k_m))
      t)
     (/ l k_m)))
   (if (<= k_m 1.6e+149)
     (*
      (/ 2.0 (* (* k_m k_m) t))
      (/ (* (* (cos k_m) l) l) (fma (cos (* -2.0 k_m)) -0.5 0.5)))
     (* (/ -0.3333333333333333 t) (* (/ l k_m) (/ l k_m))))))

C

k_m = fabs(k);
double code(double t, double l, double k_m) {
	double tmp;
	if (k_m <= 0.0068) {
		tmp = (2.0 / k_m) * (((((fma(0.3333333333333333, ((k_m * k_m) * l), l) / k_m) / k_m) * cos(k_m)) / t) * (l / k_m));
	} else if (k_m <= 1.6e+149) {
		tmp = (2.0 / ((k_m * k_m) * t)) * (((cos(k_m) * l) * l) / fma(cos((-2.0 * k_m)), -0.5, 0.5));
	} else {
		tmp = (-0.3333333333333333 / t) * ((l / k_m) * (l / k_m));
	}
	return tmp;
}

Julia

k_m = abs(k)
function code(t, l, k_m)
	tmp = 0.0
	if (k_m <= 0.0068)
		tmp = Float64(Float64(2.0 / k_m) * Float64(Float64(Float64(Float64(Float64(fma(0.3333333333333333, Float64(Float64(k_m * k_m) * l), l) / k_m) / k_m) * cos(k_m)) / t) * Float64(l / k_m)));
	elseif (k_m <= 1.6e+149)
		tmp = Float64(Float64(2.0 / Float64(Float64(k_m * k_m) * t)) * Float64(Float64(Float64(cos(k_m) * l) * l) / fma(cos(Float64(-2.0 * k_m)), -0.5, 0.5)));
	else
		tmp = Float64(Float64(-0.3333333333333333 / t) * Float64(Float64(l / k_m) * Float64(l / k_m)));
	end
	return tmp
end

Wolfram

k_m = N[Abs[k], $MachinePrecision]
code[t_, l_, k$95$m_] := If[LessEqual[k$95$m, 0.0068], N[(N[(2.0 / k$95$m), $MachinePrecision] * N[(N[(N[(N[(N[(N[(0.3333333333333333 * N[(N[(k$95$m * k$95$m), $MachinePrecision] * l), $MachinePrecision] + l), $MachinePrecision] / k$95$m), $MachinePrecision] / k$95$m), $MachinePrecision] * N[Cos[k$95$m], $MachinePrecision]), $MachinePrecision] / t), $MachinePrecision] * N[(l / k$95$m), $MachinePrecision]), $MachinePrecision]), $MachinePrecision], If[LessEqual[k$95$m, 1.6e+149], N[(N[(2.0 / N[(N[(k$95$m * k$95$m), $MachinePrecision] * t), $MachinePrecision]), $MachinePrecision] * N[(N[(N[(N[Cos[k$95$m], $MachinePrecision] * l), $MachinePrecision] * l), $MachinePrecision] / N[(N[Cos[N[(-2.0 * k$95$m), $MachinePrecision]], $MachinePrecision] * -0.5 + 0.5), $MachinePrecision]), $MachinePrecision]), $MachinePrecision], N[(N[(-0.3333333333333333 / t), $MachinePrecision] * N[(N[(l / k$95$m), $MachinePrecision] * N[(l / k$95$m), $MachinePrecision]), $MachinePrecision]), $MachinePrecision]]]

Derivation

1. Split input into 3 regimes

2. if k < 0.00679999999999999962

   1. Initial program 43.0%

      \[\frac{2}{\left(\left(\frac{{t}^{3}}{\ell \cdot \ell} \cdot \sin k\right) \cdot \tan k\right) \cdot \left(\left(1 + {\left(\frac{k}{t}\right)}^{2}\right) - 1\right)} \]
   2. Add Preprocessing

   3. Taylor expanded in t around 0

      \[\leadsto \color{blue}{2 \cdot \frac{{\ell}^{2} \cdot \cos k}{{k}^{2} \cdot \left(t \cdot {\sin k}^{2}\right)}} \]
   4. Step-by-step derivation

   5. Applied rewrites 73.1%

      \[\leadsto \color{blue}{\frac{2}{\left(k \cdot k\right) \cdot t} \cdot \frac{\left(\cos k \cdot \ell\right) \cdot \ell}{{\sin k}^{2}}} \]
   6. Step-by-step derivation

   7. Applied rewrites 87.1%

      \[\leadsto \frac{2}{k} \cdot \color{blue}{\frac{\frac{\ell}{{\sin k}^{2}} \cdot \left(\cos k \cdot \ell\right)}{k \cdot t}} \]

   8. Taylor expanded in k around 0

      \[\leadsto \frac{2}{k} \cdot \frac{\frac{\ell + \frac{1}{3} \cdot \left({k}^{2} \cdot \ell\right)}{{k}^{2}} \cdot \left(\cos k \cdot \ell\right)}{k \cdot t} \]
   9. Step-by-step derivation

   10. Applied rewrites 69.6%

       \[\leadsto \frac{2}{k} \cdot \frac{\frac{\frac{\mathsf{fma}\left(\left(k \cdot k\right) \cdot \ell, 0.3333333333333333, \ell\right)}{k}}{k} \cdot \left(\cos k \cdot \ell\right)}{k \cdot t} \]
   11. Step-by-step derivation

   12. Applied rewrites 71.5%

       \[\leadsto \frac{2}{k} \cdot \left(\frac{\frac{\frac{\mathsf{fma}\left(0.3333333333333333, \left(k \cdot k\right) \cdot \ell, \ell\right)}{k}}{k} \cdot \cos k}{t} \cdot \color{blue}{\frac{\ell}{k}}\right) \]

   if 0.00679999999999999962 < k < 1.6000000000000001e149

   1. Initial program 23.1%

      \[\frac{2}{\left(\left(\frac{{t}^{3}}{\ell \cdot \ell} \cdot \sin k\right) \cdot \tan k\right) \cdot \left(\left(1 + {\left(\frac{k}{t}\right)}^{2}\right) - 1\right)} \]
   2. Add Preprocessing

   3. Taylor expanded in t around 0

      \[\leadsto \color{blue}{2 \cdot \frac{{\ell}^{2} \cdot \cos k}{{k}^{2} \cdot \left(t \cdot {\sin k}^{2}\right)}} \]
   4. Step-by-step derivation

   5. Applied rewrites 94.1%

      \[\leadsto \color{blue}{\frac{2}{\left(k \cdot k\right) \cdot t} \cdot \frac{\left(\cos k \cdot \ell\right) \cdot \ell}{{\sin k}^{2}}} \]
   6. Step-by-step derivation

   7. Applied rewrites 94.2%

      \[\leadsto \frac{2}{\left(k \cdot k\right) \cdot t} \cdot \frac{\left(\cos k \cdot \ell\right) \cdot \ell}{0.5 - \color{blue}{0.5 \cdot \cos \left(2 \cdot k\right)}} \]
   8. Step-by-step derivation

   9. Applied rewrites 94.2%

      \[\leadsto \frac{2}{\left(k \cdot k\right) \cdot t} \cdot \frac{\left(\cos k \cdot \ell\right) \cdot \ell}{\mathsf{fma}\left(\cos \left(-2 \cdot k\right), \color{blue}{-0.5}, 0.5\right)} \]

   if 1.6000000000000001e149 < k

   1. Initial program 26.9%

      \[\frac{2}{\left(\left(\frac{{t}^{3}}{\ell \cdot \ell} \cdot \sin k\right) \cdot \tan k\right) \cdot \left(\left(1 + {\left(\frac{k}{t}\right)}^{2}\right) - 1\right)} \]
   2. Add Preprocessing

   3. Taylor expanded in k around 0

      \[\leadsto \color{blue}{\frac{\frac{-1}{3} \cdot \frac{{k}^{2} \cdot {\ell}^{2}}{t} + 2 \cdot \frac{{\ell}^{2}}{t}}{{k}^{4}}} \]
   4. Step-by-step derivation

   5. Applied rewrites 3.8%

      \[\leadsto \color{blue}{\frac{\frac{\ell \cdot \ell}{t} \cdot \mathsf{fma}\left(k \cdot k, -0.3333333333333333, 2\right)}{{k}^{4}}} \]

   6. Taylor expanded in k around inf

      \[\leadsto \frac{-1}{3} \cdot \color{blue}{\frac{{\ell}^{2}}{{k}^{2} \cdot t}} \]
   7. Step-by-step derivation

   8. Applied rewrites 62.0%

      \[\leadsto \frac{-0.3333333333333333}{t} \cdot \color{blue}{\left(\frac{\ell}{k} \cdot \frac{\ell}{k}\right)} \]
3. Recombined 3 regimes into one program.
4. Add Preprocessing

Alternative 7: 88.9% accurate, 1.7× speedup

Math

\[\begin{array}{l} k_m = \left|k\right| \\ \begin{array}{l} \mathbf{if}\;k\_m \leq 0.0068:\\ \;\;\;\;\frac{2}{k\_m} \cdot \left(\frac{\frac{\frac{\mathsf{fma}\left(0.3333333333333333, \left(k\_m \cdot k\_m\right) \cdot \ell, \ell\right)}{k\_m}}{k\_m} \cdot \cos k\_m}{t} \cdot \frac{\ell}{k\_m}\right)\\ \mathbf{else}:\\ \;\;\;\;\frac{2}{k\_m} \cdot \frac{\left(\cos k\_m \cdot \ell\right) \cdot \ell}{\left(0.5 - 0.5 \cdot \cos \left(2 \cdot k\_m\right)\right) \cdot \left(k\_m \cdot t\right)}\\ \end{array} \end{array} \]

FPCore

k_m = (fabs.f64 k)
(FPCore (t l k_m)
 :precision binary64
 (if (<= k_m 0.0068)
   (*
    (/ 2.0 k_m)
    (*
     (/
      (*
       (/ (/ (fma 0.3333333333333333 (* (* k_m k_m) l) l) k_m) k_m)
       (cos k_m))
      t)
     (/ l k_m)))
   (*
    (/ 2.0 k_m)
    (/
     (* (* (cos k_m) l) l)
     (* (- 0.5 (* 0.5 (cos (* 2.0 k_m)))) (* k_m t))))))

C

k_m = fabs(k);
double code(double t, double l, double k_m) {
	double tmp;
	if (k_m <= 0.0068) {
		tmp = (2.0 / k_m) * (((((fma(0.3333333333333333, ((k_m * k_m) * l), l) / k_m) / k_m) * cos(k_m)) / t) * (l / k_m));
	} else {
		tmp = (2.0 / k_m) * (((cos(k_m) * l) * l) / ((0.5 - (0.5 * cos((2.0 * k_m)))) * (k_m * t)));
	}
	return tmp;
}

Julia

k_m = abs(k)
function code(t, l, k_m)
	tmp = 0.0
	if (k_m <= 0.0068)
		tmp = Float64(Float64(2.0 / k_m) * Float64(Float64(Float64(Float64(Float64(fma(0.3333333333333333, Float64(Float64(k_m * k_m) * l), l) / k_m) / k_m) * cos(k_m)) / t) * Float64(l / k_m)));
	else
		tmp = Float64(Float64(2.0 / k_m) * Float64(Float64(Float64(cos(k_m) * l) * l) / Float64(Float64(0.5 - Float64(0.5 * cos(Float64(2.0 * k_m)))) * Float64(k_m * t))));
	end
	return tmp
end

Wolfram

k_m = N[Abs[k], $MachinePrecision]
code[t_, l_, k$95$m_] := If[LessEqual[k$95$m, 0.0068], N[(N[(2.0 / k$95$m), $MachinePrecision] * N[(N[(N[(N[(N[(N[(0.3333333333333333 * N[(N[(k$95$m * k$95$m), $MachinePrecision] * l), $MachinePrecision] + l), $MachinePrecision] / k$95$m), $MachinePrecision] / k$95$m), $MachinePrecision] * N[Cos[k$95$m], $MachinePrecision]), $MachinePrecision] / t), $MachinePrecision] * N[(l / k$95$m), $MachinePrecision]), $MachinePrecision]), $MachinePrecision], N[(N[(2.0 / k$95$m), $MachinePrecision] * N[(N[(N[(N[Cos[k$95$m], $MachinePrecision] * l), $MachinePrecision] * l), $MachinePrecision] / N[(N[(0.5 - N[(0.5 * N[Cos[N[(2.0 * k$95$m), $MachinePrecision]], $MachinePrecision]), $MachinePrecision]), $MachinePrecision] * N[(k$95$m * t), $MachinePrecision]), $MachinePrecision]), $MachinePrecision]), $MachinePrecision]]

Derivation

1. Split input into 2 regimes

2. if k < 0.00679999999999999962

   1. Initial program 43.0%

      \[\frac{2}{\left(\left(\frac{{t}^{3}}{\ell \cdot \ell} \cdot \sin k\right) \cdot \tan k\right) \cdot \left(\left(1 + {\left(\frac{k}{t}\right)}^{2}\right) - 1\right)} \]
   2. Add Preprocessing

   3. Taylor expanded in t around 0

      \[\leadsto \color{blue}{2 \cdot \frac{{\ell}^{2} \cdot \cos k}{{k}^{2} \cdot \left(t \cdot {\sin k}^{2}\right)}} \]
   4. Step-by-step derivation

   5. Applied rewrites 73.1%

      \[\leadsto \color{blue}{\frac{2}{\left(k \cdot k\right) \cdot t} \cdot \frac{\left(\cos k \cdot \ell\right) \cdot \ell}{{\sin k}^{2}}} \]
   6. Step-by-step derivation

   7. Applied rewrites 87.1%

      \[\leadsto \frac{2}{k} \cdot \color{blue}{\frac{\frac{\ell}{{\sin k}^{2}} \cdot \left(\cos k \cdot \ell\right)}{k \cdot t}} \]

   8. Taylor expanded in k around 0

      \[\leadsto \frac{2}{k} \cdot \frac{\frac{\ell + \frac{1}{3} \cdot \left({k}^{2} \cdot \ell\right)}{{k}^{2}} \cdot \left(\cos k \cdot \ell\right)}{k \cdot t} \]
   9. Step-by-step derivation

   10. Applied rewrites 69.6%

       \[\leadsto \frac{2}{k} \cdot \frac{\frac{\frac{\mathsf{fma}\left(\left(k \cdot k\right) \cdot \ell, 0.3333333333333333, \ell\right)}{k}}{k} \cdot \left(\cos k \cdot \ell\right)}{k \cdot t} \]
   11. Step-by-step derivation

   12. Applied rewrites 71.5%

       \[\leadsto \frac{2}{k} \cdot \left(\frac{\frac{\frac{\mathsf{fma}\left(0.3333333333333333, \left(k \cdot k\right) \cdot \ell, \ell\right)}{k}}{k} \cdot \cos k}{t} \cdot \color{blue}{\frac{\ell}{k}}\right) \]

   if 0.00679999999999999962 < k

   1. Initial program 24.7%

      \[\frac{2}{\left(\left(\frac{{t}^{3}}{\ell \cdot \ell} \cdot \sin k\right) \cdot \tan k\right) \cdot \left(\left(1 + {\left(\frac{k}{t}\right)}^{2}\right) - 1\right)} \]
   2. Add Preprocessing

   3. Taylor expanded in t around 0

      \[\leadsto \color{blue}{2 \cdot \frac{{\ell}^{2} \cdot \cos k}{{k}^{2} \cdot \left(t \cdot {\sin k}^{2}\right)}} \]
   4. Step-by-step derivation

   5. Applied rewrites 77.4%

      \[\leadsto \color{blue}{\frac{2}{\left(k \cdot k\right) \cdot t} \cdot \frac{\left(\cos k \cdot \ell\right) \cdot \ell}{{\sin k}^{2}}} \]
   6. Step-by-step derivation

   7. Applied rewrites 82.2%

      \[\leadsto \frac{2}{k} \cdot \color{blue}{\frac{\frac{\ell}{{\sin k}^{2}} \cdot \left(\cos k \cdot \ell\right)}{k \cdot t}} \]
   8. Step-by-step derivation

   9. Applied rewrites 82.4%

      \[\leadsto \frac{2}{k} \cdot \frac{\left(\cos k \cdot \ell\right) \cdot \ell}{\color{blue}{{\sin k}^{2} \cdot \left(k \cdot t\right)}} \]
   10. Step-by-step derivation

   11. Applied rewrites 82.3%

       \[\leadsto \frac{2}{k} \cdot \frac{\left(\cos k \cdot \ell\right) \cdot \ell}{\left(0.5 - 0.5 \cdot \cos \left(2 \cdot k\right)\right) \cdot \left(\color{blue}{k} \cdot t\right)} \]
3. Recombined 2 regimes into one program.
4. Add Preprocessing

Alternative 8: 80.6% accurate, 2.4× speedup

Math

\[\begin{array}{l} k_m = \left|k\right| \\ \begin{array}{l} \mathbf{if}\;k\_m \leq 3.2 \cdot 10^{+136}:\\ \;\;\;\;\frac{2}{k\_m} \cdot \left(\frac{\frac{\frac{\mathsf{fma}\left(0.3333333333333333, \left(k\_m \cdot k\_m\right) \cdot \ell, \ell\right)}{k\_m}}{k\_m} \cdot \cos k\_m}{t} \cdot \frac{\ell}{k\_m}\right)\\ \mathbf{else}:\\ \;\;\;\;\frac{-0.3333333333333333}{t} \cdot \left(\frac{\ell}{k\_m} \cdot \frac{\ell}{k\_m}\right)\\ \end{array} \end{array} \]

FPCore

k_m = (fabs.f64 k)
(FPCore (t l k_m)
 :precision binary64
 (if (<= k_m 3.2e+136)
   (*
    (/ 2.0 k_m)
    (*
     (/
      (*
       (/ (/ (fma 0.3333333333333333 (* (* k_m k_m) l) l) k_m) k_m)
       (cos k_m))
      t)
     (/ l k_m)))
   (* (/ -0.3333333333333333 t) (* (/ l k_m) (/ l k_m)))))

C

k_m = fabs(k);
double code(double t, double l, double k_m) {
	double tmp;
	if (k_m <= 3.2e+136) {
		tmp = (2.0 / k_m) * (((((fma(0.3333333333333333, ((k_m * k_m) * l), l) / k_m) / k_m) * cos(k_m)) / t) * (l / k_m));
	} else {
		tmp = (-0.3333333333333333 / t) * ((l / k_m) * (l / k_m));
	}
	return tmp;
}

Julia

k_m = abs(k)
function code(t, l, k_m)
	tmp = 0.0
	if (k_m <= 3.2e+136)
		tmp = Float64(Float64(2.0 / k_m) * Float64(Float64(Float64(Float64(Float64(fma(0.3333333333333333, Float64(Float64(k_m * k_m) * l), l) / k_m) / k_m) * cos(k_m)) / t) * Float64(l / k_m)));
	else
		tmp = Float64(Float64(-0.3333333333333333 / t) * Float64(Float64(l / k_m) * Float64(l / k_m)));
	end
	return tmp
end

Wolfram

k_m = N[Abs[k], $MachinePrecision]
code[t_, l_, k$95$m_] := If[LessEqual[k$95$m, 3.2e+136], N[(N[(2.0 / k$95$m), $MachinePrecision] * N[(N[(N[(N[(N[(N[(0.3333333333333333 * N[(N[(k$95$m * k$95$m), $MachinePrecision] * l), $MachinePrecision] + l), $MachinePrecision] / k$95$m), $MachinePrecision] / k$95$m), $MachinePrecision] * N[Cos[k$95$m], $MachinePrecision]), $MachinePrecision] / t), $MachinePrecision] * N[(l / k$95$m), $MachinePrecision]), $MachinePrecision]), $MachinePrecision], N[(N[(-0.3333333333333333 / t), $MachinePrecision] * N[(N[(l / k$95$m), $MachinePrecision] * N[(l / k$95$m), $MachinePrecision]), $MachinePrecision]), $MachinePrecision]]

Derivation

1. Split input into 2 regimes

2. if k < 3.19999999999999988e136

   1. Initial program 40.4%

      \[\frac{2}{\left(\left(\frac{{t}^{3}}{\ell \cdot \ell} \cdot \sin k\right) \cdot \tan k\right) \cdot \left(\left(1 + {\left(\frac{k}{t}\right)}^{2}\right) - 1\right)} \]
   2. Add Preprocessing

   3. Taylor expanded in t around 0

      \[\leadsto \color{blue}{2 \cdot \frac{{\ell}^{2} \cdot \cos k}{{k}^{2} \cdot \left(t \cdot {\sin k}^{2}\right)}} \]
   4. Step-by-step derivation

   5. Applied rewrites 75.8%

      \[\leadsto \color{blue}{\frac{2}{\left(k \cdot k\right) \cdot t} \cdot \frac{\left(\cos k \cdot \ell\right) \cdot \ell}{{\sin k}^{2}}} \]
   6. Step-by-step derivation

   7. Applied rewrites 87.5%

      \[\leadsto \frac{2}{k} \cdot \color{blue}{\frac{\frac{\ell}{{\sin k}^{2}} \cdot \left(\cos k \cdot \ell\right)}{k \cdot t}} \]

   8. Taylor expanded in k around 0

      \[\leadsto \frac{2}{k} \cdot \frac{\frac{\ell + \frac{1}{3} \cdot \left({k}^{2} \cdot \ell\right)}{{k}^{2}} \cdot \left(\cos k \cdot \ell\right)}{k \cdot t} \]
   9. Step-by-step derivation

   10. Applied rewrites 69.0%

       \[\leadsto \frac{2}{k} \cdot \frac{\frac{\frac{\mathsf{fma}\left(\left(k \cdot k\right) \cdot \ell, 0.3333333333333333, \ell\right)}{k}}{k} \cdot \left(\cos k \cdot \ell\right)}{k \cdot t} \]
   11. Step-by-step derivation

   12. Applied rewrites 70.7%

       \[\leadsto \frac{2}{k} \cdot \left(\frac{\frac{\frac{\mathsf{fma}\left(0.3333333333333333, \left(k \cdot k\right) \cdot \ell, \ell\right)}{k}}{k} \cdot \cos k}{t} \cdot \color{blue}{\frac{\ell}{k}}\right) \]

   if 3.19999999999999988e136 < k

   1. Initial program 25.8%

      \[\frac{2}{\left(\left(\frac{{t}^{3}}{\ell \cdot \ell} \cdot \sin k\right) \cdot \tan k\right) \cdot \left(\left(1 + {\left(\frac{k}{t}\right)}^{2}\right) - 1\right)} \]
   2. Add Preprocessing

   3. Taylor expanded in k around 0

      \[\leadsto \color{blue}{\frac{\frac{-1}{3} \cdot \frac{{k}^{2} \cdot {\ell}^{2}}{t} + 2 \cdot \frac{{\ell}^{2}}{t}}{{k}^{4}}} \]
   4. Step-by-step derivation

   5. Applied rewrites 9.7%

      \[\leadsto \color{blue}{\frac{\frac{\ell \cdot \ell}{t} \cdot \mathsf{fma}\left(k \cdot k, -0.3333333333333333, 2\right)}{{k}^{4}}} \]

   6. Taylor expanded in k around inf

      \[\leadsto \frac{-1}{3} \cdot \color{blue}{\frac{{\ell}^{2}}{{k}^{2} \cdot t}} \]
   7. Step-by-step derivation

   8. Applied rewrites 62.3%

      \[\leadsto \frac{-0.3333333333333333}{t} \cdot \color{blue}{\left(\frac{\ell}{k} \cdot \frac{\ell}{k}\right)} \]
3. Recombined 2 regimes into one program.
4. Add Preprocessing

Alternative 9: 79.6% accurate, 2.5× speedup

Math

\[\begin{array}{l} k_m = \left|k\right| \\ \begin{array}{l} \mathbf{if}\;k\_m \leq 3.2 \cdot 10^{+136}:\\ \;\;\;\;\frac{2}{k\_m} \cdot \left(\left(\frac{\frac{\mathsf{fma}\left(0.3333333333333333, \left(k\_m \cdot k\_m\right) \cdot \ell, \ell\right)}{k\_m}}{k\_m} \cdot \cos k\_m\right) \cdot \frac{\ell}{k\_m \cdot t}\right)\\ \mathbf{else}:\\ \;\;\;\;\frac{-0.3333333333333333}{t} \cdot \left(\frac{\ell}{k\_m} \cdot \frac{\ell}{k\_m}\right)\\ \end{array} \end{array} \]

FPCore

k_m = (fabs.f64 k)
(FPCore (t l k_m)
 :precision binary64
 (if (<= k_m 3.2e+136)
   (*
    (/ 2.0 k_m)
    (*
     (* (/ (/ (fma 0.3333333333333333 (* (* k_m k_m) l) l) k_m) k_m) (cos k_m))
     (/ l (* k_m t))))
   (* (/ -0.3333333333333333 t) (* (/ l k_m) (/ l k_m)))))

C

k_m = fabs(k);
double code(double t, double l, double k_m) {
	double tmp;
	if (k_m <= 3.2e+136) {
		tmp = (2.0 / k_m) * ((((fma(0.3333333333333333, ((k_m * k_m) * l), l) / k_m) / k_m) * cos(k_m)) * (l / (k_m * t)));
	} else {
		tmp = (-0.3333333333333333 / t) * ((l / k_m) * (l / k_m));
	}
	return tmp;
}

Julia

k_m = abs(k)
function code(t, l, k_m)
	tmp = 0.0
	if (k_m <= 3.2e+136)
		tmp = Float64(Float64(2.0 / k_m) * Float64(Float64(Float64(Float64(fma(0.3333333333333333, Float64(Float64(k_m * k_m) * l), l) / k_m) / k_m) * cos(k_m)) * Float64(l / Float64(k_m * t))));
	else
		tmp = Float64(Float64(-0.3333333333333333 / t) * Float64(Float64(l / k_m) * Float64(l / k_m)));
	end
	return tmp
end

Wolfram

k_m = N[Abs[k], $MachinePrecision]
code[t_, l_, k$95$m_] := If[LessEqual[k$95$m, 3.2e+136], N[(N[(2.0 / k$95$m), $MachinePrecision] * N[(N[(N[(N[(N[(0.3333333333333333 * N[(N[(k$95$m * k$95$m), $MachinePrecision] * l), $MachinePrecision] + l), $MachinePrecision] / k$95$m), $MachinePrecision] / k$95$m), $MachinePrecision] * N[Cos[k$95$m], $MachinePrecision]), $MachinePrecision] * N[(l / N[(k$95$m * t), $MachinePrecision]), $MachinePrecision]), $MachinePrecision]), $MachinePrecision], N[(N[(-0.3333333333333333 / t), $MachinePrecision] * N[(N[(l / k$95$m), $MachinePrecision] * N[(l / k$95$m), $MachinePrecision]), $MachinePrecision]), $MachinePrecision]]

Derivation

1. Split input into 2 regimes

2. if k < 3.19999999999999988e136

   1. Initial program 40.4%

      \[\frac{2}{\left(\left(\frac{{t}^{3}}{\ell \cdot \ell} \cdot \sin k\right) \cdot \tan k\right) \cdot \left(\left(1 + {\left(\frac{k}{t}\right)}^{2}\right) - 1\right)} \]
   2. Add Preprocessing

   3. Taylor expanded in t around 0

      \[\leadsto \color{blue}{2 \cdot \frac{{\ell}^{2} \cdot \cos k}{{k}^{2} \cdot \left(t \cdot {\sin k}^{2}\right)}} \]
   4. Step-by-step derivation

   5. Applied rewrites 75.8%

      \[\leadsto \color{blue}{\frac{2}{\left(k \cdot k\right) \cdot t} \cdot \frac{\left(\cos k \cdot \ell\right) \cdot \ell}{{\sin k}^{2}}} \]
   6. Step-by-step derivation

   7. Applied rewrites 87.5%

      \[\leadsto \frac{2}{k} \cdot \color{blue}{\frac{\frac{\ell}{{\sin k}^{2}} \cdot \left(\cos k \cdot \ell\right)}{k \cdot t}} \]

   8. Taylor expanded in k around 0

      \[\leadsto \frac{2}{k} \cdot \frac{\frac{\ell + \frac{1}{3} \cdot \left({k}^{2} \cdot \ell\right)}{{k}^{2}} \cdot \left(\cos k \cdot \ell\right)}{k \cdot t} \]
   9. Step-by-step derivation

   10. Applied rewrites 69.0%

       \[\leadsto \frac{2}{k} \cdot \frac{\frac{\frac{\mathsf{fma}\left(\left(k \cdot k\right) \cdot \ell, 0.3333333333333333, \ell\right)}{k}}{k} \cdot \left(\cos k \cdot \ell\right)}{k \cdot t} \]
   11. Step-by-step derivation

   12. Applied rewrites 70.3%

       \[\leadsto \frac{2}{k} \cdot \left(\left(\frac{\frac{\mathsf{fma}\left(0.3333333333333333, \left(k \cdot k\right) \cdot \ell, \ell\right)}{k}}{k} \cdot \cos k\right) \cdot \color{blue}{\frac{\ell}{k \cdot t}}\right) \]

   if 3.19999999999999988e136 < k

   1. Initial program 25.8%

      \[\frac{2}{\left(\left(\frac{{t}^{3}}{\ell \cdot \ell} \cdot \sin k\right) \cdot \tan k\right) \cdot \left(\left(1 + {\left(\frac{k}{t}\right)}^{2}\right) - 1\right)} \]
   2. Add Preprocessing

   3. Taylor expanded in k around 0

      \[\leadsto \color{blue}{\frac{\frac{-1}{3} \cdot \frac{{k}^{2} \cdot {\ell}^{2}}{t} + 2 \cdot \frac{{\ell}^{2}}{t}}{{k}^{4}}} \]
   4. Step-by-step derivation

   5. Applied rewrites 9.7%

      \[\leadsto \color{blue}{\frac{\frac{\ell \cdot \ell}{t} \cdot \mathsf{fma}\left(k \cdot k, -0.3333333333333333, 2\right)}{{k}^{4}}} \]

   6. Taylor expanded in k around inf

      \[\leadsto \frac{-1}{3} \cdot \color{blue}{\frac{{\ell}^{2}}{{k}^{2} \cdot t}} \]
   7. Step-by-step derivation

   8. Applied rewrites 62.3%

      \[\leadsto \frac{-0.3333333333333333}{t} \cdot \color{blue}{\left(\frac{\ell}{k} \cdot \frac{\ell}{k}\right)} \]
3. Recombined 2 regimes into one program.
4. Add Preprocessing

Alternative 10: 76.8% accurate, 2.6× speedup

Math

\[\begin{array}{l} k_m = \left|k\right| \\ \begin{array}{l} \mathbf{if}\;k\_m \leq 3.8 \cdot 10^{-13}:\\ \;\;\;\;\frac{\frac{\frac{\ell}{\left(k\_m \cdot k\_m\right) \cdot t}}{k\_m}}{k\_m} \cdot \left(\ell + \ell\right)\\ \mathbf{elif}\;k\_m \leq 8.5 \cdot 10^{+205}:\\ \;\;\;\;\frac{2}{k\_m} \cdot \frac{\frac{\left(0.3333333333333333 \cdot \ell\right) \cdot k\_m}{k\_m} \cdot \left(\cos k\_m \cdot \ell\right)}{k\_m \cdot t}\\ \mathbf{else}:\\ \;\;\;\;\frac{-0.3333333333333333}{t} \cdot \left(\frac{\ell}{k\_m} \cdot \frac{\ell}{k\_m}\right)\\ \end{array} \end{array} \]

FPCore

k_m = (fabs.f64 k)
(FPCore (t l k_m)
 :precision binary64
 (if (<= k_m 3.8e-13)
   (* (/ (/ (/ l (* (* k_m k_m) t)) k_m) k_m) (+ l l))
   (if (<= k_m 8.5e+205)
     (*
      (/ 2.0 k_m)
      (/
       (* (/ (* (* 0.3333333333333333 l) k_m) k_m) (* (cos k_m) l))
       (* k_m t)))
     (* (/ -0.3333333333333333 t) (* (/ l k_m) (/ l k_m))))))

C

k_m = fabs(k);
double code(double t, double l, double k_m) {
	double tmp;
	if (k_m <= 3.8e-13) {
		tmp = (((l / ((k_m * k_m) * t)) / k_m) / k_m) * (l + l);
	} else if (k_m <= 8.5e+205) {
		tmp = (2.0 / k_m) * (((((0.3333333333333333 * l) * k_m) / k_m) * (cos(k_m) * l)) / (k_m * t));
	} else {
		tmp = (-0.3333333333333333 / t) * ((l / k_m) * (l / k_m));
	}
	return tmp;
}

Fortran

k_m =     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(t, l, k_m)
use fmin_fmax_functions
    real(8), intent (in) :: t
    real(8), intent (in) :: l
    real(8), intent (in) :: k_m
    real(8) :: tmp
    if (k_m <= 3.8d-13) then
        tmp = (((l / ((k_m * k_m) * t)) / k_m) / k_m) * (l + l)
    else if (k_m <= 8.5d+205) then
        tmp = (2.0d0 / k_m) * (((((0.3333333333333333d0 * l) * k_m) / k_m) * (cos(k_m) * l)) / (k_m * t))
    else
        tmp = ((-0.3333333333333333d0) / t) * ((l / k_m) * (l / k_m))
    end if
    code = tmp
end function

Java

k_m = Math.abs(k);
public static double code(double t, double l, double k_m) {
	double tmp;
	if (k_m <= 3.8e-13) {
		tmp = (((l / ((k_m * k_m) * t)) / k_m) / k_m) * (l + l);
	} else if (k_m <= 8.5e+205) {
		tmp = (2.0 / k_m) * (((((0.3333333333333333 * l) * k_m) / k_m) * (Math.cos(k_m) * l)) / (k_m * t));
	} else {
		tmp = (-0.3333333333333333 / t) * ((l / k_m) * (l / k_m));
	}
	return tmp;
}

Python

k_m = math.fabs(k)
def code(t, l, k_m):
	tmp = 0
	if k_m <= 3.8e-13:
		tmp = (((l / ((k_m * k_m) * t)) / k_m) / k_m) * (l + l)
	elif k_m <= 8.5e+205:
		tmp = (2.0 / k_m) * (((((0.3333333333333333 * l) * k_m) / k_m) * (math.cos(k_m) * l)) / (k_m * t))
	else:
		tmp = (-0.3333333333333333 / t) * ((l / k_m) * (l / k_m))
	return tmp

Julia

k_m = abs(k)
function code(t, l, k_m)
	tmp = 0.0
	if (k_m <= 3.8e-13)
		tmp = Float64(Float64(Float64(Float64(l / Float64(Float64(k_m * k_m) * t)) / k_m) / k_m) * Float64(l + l));
	elseif (k_m <= 8.5e+205)
		tmp = Float64(Float64(2.0 / k_m) * Float64(Float64(Float64(Float64(Float64(0.3333333333333333 * l) * k_m) / k_m) * Float64(cos(k_m) * l)) / Float64(k_m * t)));
	else
		tmp = Float64(Float64(-0.3333333333333333 / t) * Float64(Float64(l / k_m) * Float64(l / k_m)));
	end
	return tmp
end

MATLAB

k_m = abs(k);
function tmp_2 = code(t, l, k_m)
	tmp = 0.0;
	if (k_m <= 3.8e-13)
		tmp = (((l / ((k_m * k_m) * t)) / k_m) / k_m) * (l + l);
	elseif (k_m <= 8.5e+205)
		tmp = (2.0 / k_m) * (((((0.3333333333333333 * l) * k_m) / k_m) * (cos(k_m) * l)) / (k_m * t));
	else
		tmp = (-0.3333333333333333 / t) * ((l / k_m) * (l / k_m));
	end
	tmp_2 = tmp;
end

Wolfram

k_m = N[Abs[k], $MachinePrecision]
code[t_, l_, k$95$m_] := If[LessEqual[k$95$m, 3.8e-13], N[(N[(N[(N[(l / N[(N[(k$95$m * k$95$m), $MachinePrecision] * t), $MachinePrecision]), $MachinePrecision] / k$95$m), $MachinePrecision] / k$95$m), $MachinePrecision] * N[(l + l), $MachinePrecision]), $MachinePrecision], If[LessEqual[k$95$m, 8.5e+205], N[(N[(2.0 / k$95$m), $MachinePrecision] * N[(N[(N[(N[(N[(0.3333333333333333 * l), $MachinePrecision] * k$95$m), $MachinePrecision] / k$95$m), $MachinePrecision] * N[(N[Cos[k$95$m], $MachinePrecision] * l), $MachinePrecision]), $MachinePrecision] / N[(k$95$m * t), $MachinePrecision]), $MachinePrecision]), $MachinePrecision], N[(N[(-0.3333333333333333 / t), $MachinePrecision] * N[(N[(l / k$95$m), $MachinePrecision] * N[(l / k$95$m), $MachinePrecision]), $MachinePrecision]), $MachinePrecision]]]

Derivation

1. Split input into 3 regimes

2. if k < 3.8e-13

   1. Initial program 42.7%

      \[\frac{2}{\left(\left(\frac{{t}^{3}}{\ell \cdot \ell} \cdot \sin k\right) \cdot \tan k\right) \cdot \left(\left(1 + {\left(\frac{k}{t}\right)}^{2}\right) - 1\right)} \]
   2. Add Preprocessing

   3. Taylor expanded in k around 0

      \[\leadsto \color{blue}{2 \cdot \frac{{\ell}^{2}}{{k}^{4} \cdot t}} \]
   4. Step-by-step derivation

   5. Applied rewrites 75.6%

      \[\leadsto \color{blue}{\frac{\ell}{{k}^{4} \cdot t} \cdot \left(\ell + \ell\right)} \]
   6. Step-by-step derivation

   7. Applied rewrites 78.6%

      \[\leadsto \frac{\ell}{\left(\left(k \cdot k\right) \cdot t\right) \cdot \left(k \cdot k\right)} \cdot \left(\ell + \ell\right) \]
   8. Step-by-step derivation

   9. Applied rewrites 81.2%

      \[\leadsto \frac{\frac{\frac{\ell}{\left(k \cdot k\right) \cdot t}}{k}}{k} \cdot \left(\color{blue}{\ell} + \ell\right) \]

   if 3.8e-13 < k < 8.49999999999999997e205

   1. Initial program 23.6%

      \[\frac{2}{\left(\left(\frac{{t}^{3}}{\ell \cdot \ell} \cdot \sin k\right) \cdot \tan k\right) \cdot \left(\left(1 + {\left(\frac{k}{t}\right)}^{2}\right) - 1\right)} \]
   2. Add Preprocessing

   3. Taylor expanded in t around 0

      \[\leadsto \color{blue}{2 \cdot \frac{{\ell}^{2} \cdot \cos k}{{k}^{2} \cdot \left(t \cdot {\sin k}^{2}\right)}} \]
   4. Step-by-step derivation

   5. Applied rewrites 85.2%

      \[\leadsto \color{blue}{\frac{2}{\left(k \cdot k\right) \cdot t} \cdot \frac{\left(\cos k \cdot \ell\right) \cdot \ell}{{\sin k}^{2}}} \]
   6. Step-by-step derivation

   7. Applied rewrites 85.5%

      \[\leadsto \frac{2}{k} \cdot \color{blue}{\frac{\frac{\ell}{{\sin k}^{2}} \cdot \left(\cos k \cdot \ell\right)}{k \cdot t}} \]

   8. Taylor expanded in k around 0

      \[\leadsto \frac{2}{k} \cdot \frac{\frac{\ell + \frac{1}{3} \cdot \left({k}^{2} \cdot \ell\right)}{{k}^{2}} \cdot \left(\cos k \cdot \ell\right)}{k \cdot t} \]
   9. Step-by-step derivation

   10. Applied rewrites 48.5%

       \[\leadsto \frac{2}{k} \cdot \frac{\frac{\frac{\mathsf{fma}\left(\left(k \cdot k\right) \cdot \ell, 0.3333333333333333, \ell\right)}{k}}{k} \cdot \left(\cos k \cdot \ell\right)}{k \cdot t} \]

   11. Taylor expanded in k around inf

       \[\leadsto \frac{2}{k} \cdot \frac{\frac{\frac{1}{3} \cdot \left(k \cdot \ell\right)}{k} \cdot \left(\cos k \cdot \ell\right)}{k \cdot t} \]
   12. Step-by-step derivation

   13. Applied rewrites 64.5%

       \[\leadsto \frac{2}{k} \cdot \frac{\frac{\left(0.3333333333333333 \cdot \ell\right) \cdot k}{k} \cdot \left(\cos k \cdot \ell\right)}{k \cdot t} \]

   if 8.49999999999999997e205 < k

   1. Initial program 33.3%

      \[\frac{2}{\left(\left(\frac{{t}^{3}}{\ell \cdot \ell} \cdot \sin k\right) \cdot \tan k\right) \cdot \left(\left(1 + {\left(\frac{k}{t}\right)}^{2}\right) - 1\right)} \]
   2. Add Preprocessing

   3. Taylor expanded in k around 0

      \[\leadsto \color{blue}{\frac{\frac{-1}{3} \cdot \frac{{k}^{2} \cdot {\ell}^{2}}{t} + 2 \cdot \frac{{\ell}^{2}}{t}}{{k}^{4}}} \]
   4. Step-by-step derivation

   5. Applied rewrites 0.0%

      \[\leadsto \color{blue}{\frac{\frac{\ell \cdot \ell}{t} \cdot \mathsf{fma}\left(k \cdot k, -0.3333333333333333, 2\right)}{{k}^{4}}} \]

   6. Taylor expanded in k around inf

      \[\leadsto \frac{-1}{3} \cdot \color{blue}{\frac{{\ell}^{2}}{{k}^{2} \cdot t}} \]
   7. Step-by-step derivation

   8. Applied rewrites 64.5%

      \[\leadsto \frac{-0.3333333333333333}{t} \cdot \color{blue}{\left(\frac{\ell}{k} \cdot \frac{\ell}{k}\right)} \]
3. Recombined 3 regimes into one program.
4. Add Preprocessing

Alternative 11: 76.1% accurate, 2.8× speedup

Math

\[\begin{array}{l} k_m = \left|k\right| \\ \begin{array}{l} t_1 := \left(k\_m \cdot k\_m\right) \cdot t\\ \mathbf{if}\;k\_m \leq 3.15 \cdot 10^{-13}:\\ \;\;\;\;\frac{\frac{\frac{\ell}{t\_1}}{k\_m}}{k\_m} \cdot \left(\ell + \ell\right)\\ \mathbf{elif}\;k\_m \leq 3.4 \cdot 10^{+137}:\\ \;\;\;\;\frac{2}{t\_1} \cdot \frac{\left(\cos k\_m \cdot \ell\right) \cdot \ell}{k\_m \cdot k\_m}\\ \mathbf{else}:\\ \;\;\;\;\frac{-0.3333333333333333}{t} \cdot \left(\frac{\ell}{k\_m} \cdot \frac{\ell}{k\_m}\right)\\ \end{array} \end{array} \]

FPCore

k_m = (fabs.f64 k)
(FPCore (t l k_m)
 :precision binary64
 (let* ((t_1 (* (* k_m k_m) t)))
   (if (<= k_m 3.15e-13)
     (* (/ (/ (/ l t_1) k_m) k_m) (+ l l))
     (if (<= k_m 3.4e+137)
       (* (/ 2.0 t_1) (/ (* (* (cos k_m) l) l) (* k_m k_m)))
       (* (/ -0.3333333333333333 t) (* (/ l k_m) (/ l k_m)))))))

C

k_m = fabs(k);
double code(double t, double l, double k_m) {
	double t_1 = (k_m * k_m) * t;
	double tmp;
	if (k_m <= 3.15e-13) {
		tmp = (((l / t_1) / k_m) / k_m) * (l + l);
	} else if (k_m <= 3.4e+137) {
		tmp = (2.0 / t_1) * (((cos(k_m) * l) * l) / (k_m * k_m));
	} else {
		tmp = (-0.3333333333333333 / t) * ((l / k_m) * (l / k_m));
	}
	return tmp;
}

Fortran

k_m =     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(t, l, k_m)
use fmin_fmax_functions
    real(8), intent (in) :: t
    real(8), intent (in) :: l
    real(8), intent (in) :: k_m
    real(8) :: t_1
    real(8) :: tmp
    t_1 = (k_m * k_m) * t
    if (k_m <= 3.15d-13) then
        tmp = (((l / t_1) / k_m) / k_m) * (l + l)
    else if (k_m <= 3.4d+137) then
        tmp = (2.0d0 / t_1) * (((cos(k_m) * l) * l) / (k_m * k_m))
    else
        tmp = ((-0.3333333333333333d0) / t) * ((l / k_m) * (l / k_m))
    end if
    code = tmp
end function

Java

k_m = Math.abs(k);
public static double code(double t, double l, double k_m) {
	double t_1 = (k_m * k_m) * t;
	double tmp;
	if (k_m <= 3.15e-13) {
		tmp = (((l / t_1) / k_m) / k_m) * (l + l);
	} else if (k_m <= 3.4e+137) {
		tmp = (2.0 / t_1) * (((Math.cos(k_m) * l) * l) / (k_m * k_m));
	} else {
		tmp = (-0.3333333333333333 / t) * ((l / k_m) * (l / k_m));
	}
	return tmp;
}

Python

k_m = math.fabs(k)
def code(t, l, k_m):
	t_1 = (k_m * k_m) * t
	tmp = 0
	if k_m <= 3.15e-13:
		tmp = (((l / t_1) / k_m) / k_m) * (l + l)
	elif k_m <= 3.4e+137:
		tmp = (2.0 / t_1) * (((math.cos(k_m) * l) * l) / (k_m * k_m))
	else:
		tmp = (-0.3333333333333333 / t) * ((l / k_m) * (l / k_m))
	return tmp

Julia

k_m = abs(k)
function code(t, l, k_m)
	t_1 = Float64(Float64(k_m * k_m) * t)
	tmp = 0.0
	if (k_m <= 3.15e-13)
		tmp = Float64(Float64(Float64(Float64(l / t_1) / k_m) / k_m) * Float64(l + l));
	elseif (k_m <= 3.4e+137)
		tmp = Float64(Float64(2.0 / t_1) * Float64(Float64(Float64(cos(k_m) * l) * l) / Float64(k_m * k_m)));
	else
		tmp = Float64(Float64(-0.3333333333333333 / t) * Float64(Float64(l / k_m) * Float64(l / k_m)));
	end
	return tmp
end

MATLAB

k_m = abs(k);
function tmp_2 = code(t, l, k_m)
	t_1 = (k_m * k_m) * t;
	tmp = 0.0;
	if (k_m <= 3.15e-13)
		tmp = (((l / t_1) / k_m) / k_m) * (l + l);
	elseif (k_m <= 3.4e+137)
		tmp = (2.0 / t_1) * (((cos(k_m) * l) * l) / (k_m * k_m));
	else
		tmp = (-0.3333333333333333 / t) * ((l / k_m) * (l / k_m));
	end
	tmp_2 = tmp;
end

Wolfram

k_m = N[Abs[k], $MachinePrecision]
code[t_, l_, k$95$m_] := Block[{t$95$1 = N[(N[(k$95$m * k$95$m), $MachinePrecision] * t), $MachinePrecision]}, If[LessEqual[k$95$m, 3.15e-13], N[(N[(N[(N[(l / t$95$1), $MachinePrecision] / k$95$m), $MachinePrecision] / k$95$m), $MachinePrecision] * N[(l + l), $MachinePrecision]), $MachinePrecision], If[LessEqual[k$95$m, 3.4e+137], N[(N[(2.0 / t$95$1), $MachinePrecision] * N[(N[(N[(N[Cos[k$95$m], $MachinePrecision] * l), $MachinePrecision] * l), $MachinePrecision] / N[(k$95$m * k$95$m), $MachinePrecision]), $MachinePrecision]), $MachinePrecision], N[(N[(-0.3333333333333333 / t), $MachinePrecision] * N[(N[(l / k$95$m), $MachinePrecision] * N[(l / k$95$m), $MachinePrecision]), $MachinePrecision]), $MachinePrecision]]]]

Derivation

1. Split input into 3 regimes

2. if k < 3.15000000000000021e-13

   1. Initial program 42.7%

      \[\frac{2}{\left(\left(\frac{{t}^{3}}{\ell \cdot \ell} \cdot \sin k\right) \cdot \tan k\right) \cdot \left(\left(1 + {\left(\frac{k}{t}\right)}^{2}\right) - 1\right)} \]
   2. Add Preprocessing

   3. Taylor expanded in k around 0

      \[\leadsto \color{blue}{2 \cdot \frac{{\ell}^{2}}{{k}^{4} \cdot t}} \]
   4. Step-by-step derivation

   5. Applied rewrites 75.6%

      \[\leadsto \color{blue}{\frac{\ell}{{k}^{4} \cdot t} \cdot \left(\ell + \ell\right)} \]
   6. Step-by-step derivation

   7. Applied rewrites 78.6%

      \[\leadsto \frac{\ell}{\left(\left(k \cdot k\right) \cdot t\right) \cdot \left(k \cdot k\right)} \cdot \left(\ell + \ell\right) \]
   8. Step-by-step derivation

   9. Applied rewrites 81.2%

      \[\leadsto \frac{\frac{\frac{\ell}{\left(k \cdot k\right) \cdot t}}{k}}{k} \cdot \left(\color{blue}{\ell} + \ell\right) \]

   if 3.15000000000000021e-13 < k < 3.39999999999999986e137

   1. Initial program 26.1%

      \[\frac{2}{\left(\left(\frac{{t}^{3}}{\ell \cdot \ell} \cdot \sin k\right) \cdot \tan k\right) \cdot \left(\left(1 + {\left(\frac{k}{t}\right)}^{2}\right) - 1\right)} \]
   2. Add Preprocessing

   3. Taylor expanded in t around 0

      \[\leadsto \color{blue}{2 \cdot \frac{{\ell}^{2} \cdot \cos k}{{k}^{2} \cdot \left(t \cdot {\sin k}^{2}\right)}} \]
   4. Step-by-step derivation

   5. Applied rewrites 93.4%

      \[\leadsto \color{blue}{\frac{2}{\left(k \cdot k\right) \cdot t} \cdot \frac{\left(\cos k \cdot \ell\right) \cdot \ell}{{\sin k}^{2}}} \]

   6. Taylor expanded in k around 0

      \[\leadsto \frac{2}{\left(k \cdot k\right) \cdot t} \cdot \frac{\left(\cos k \cdot \ell\right) \cdot \ell}{{k}^{\color{blue}{2}}} \]
   7. Step-by-step derivation

   8. Applied rewrites 63.6%

      \[\leadsto \frac{2}{\left(k \cdot k\right) \cdot t} \cdot \frac{\left(\cos k \cdot \ell\right) \cdot \ell}{k \cdot \color{blue}{k}} \]

   if 3.39999999999999986e137 < k

   1. Initial program 25.8%

      \[\frac{2}{\left(\left(\frac{{t}^{3}}{\ell \cdot \ell} \cdot \sin k\right) \cdot \tan k\right) \cdot \left(\left(1 + {\left(\frac{k}{t}\right)}^{2}\right) - 1\right)} \]
   2. Add Preprocessing

   3. Taylor expanded in k around 0

      \[\leadsto \color{blue}{\frac{\frac{-1}{3} \cdot \frac{{k}^{2} \cdot {\ell}^{2}}{t} + 2 \cdot \frac{{\ell}^{2}}{t}}{{k}^{4}}} \]
   4. Step-by-step derivation

   5. Applied rewrites 9.7%

      \[\leadsto \color{blue}{\frac{\frac{\ell \cdot \ell}{t} \cdot \mathsf{fma}\left(k \cdot k, -0.3333333333333333, 2\right)}{{k}^{4}}} \]

   6. Taylor expanded in k around inf

      \[\leadsto \frac{-1}{3} \cdot \color{blue}{\frac{{\ell}^{2}}{{k}^{2} \cdot t}} \]
   7. Step-by-step derivation

   8. Applied rewrites 62.3%

      \[\leadsto \frac{-0.3333333333333333}{t} \cdot \color{blue}{\left(\frac{\ell}{k} \cdot \frac{\ell}{k}\right)} \]
3. Recombined 3 regimes into one program.
4. Add Preprocessing

Alternative 12: 74.3% accurate, 8.0× speedup

Math

\[\begin{array}{l} k_m = \left|k\right| \\ \begin{array}{l} \mathbf{if}\;k\_m \leq 2.2 \cdot 10^{+114}:\\ \;\;\;\;\frac{\frac{\frac{\ell}{\left(k\_m \cdot k\_m\right) \cdot t}}{k\_m}}{k\_m} \cdot \left(\ell + \ell\right)\\ \mathbf{else}:\\ \;\;\;\;\frac{-0.3333333333333333}{t} \cdot \left(\frac{\ell}{k\_m} \cdot \frac{\ell}{k\_m}\right)\\ \end{array} \end{array} \]

FPCore

k_m = (fabs.f64 k)
(FPCore (t l k_m)
 :precision binary64
 (if (<= k_m 2.2e+114)
   (* (/ (/ (/ l (* (* k_m k_m) t)) k_m) k_m) (+ l l))
   (* (/ -0.3333333333333333 t) (* (/ l k_m) (/ l k_m)))))

C

k_m = fabs(k);
double code(double t, double l, double k_m) {
	double tmp;
	if (k_m <= 2.2e+114) {
		tmp = (((l / ((k_m * k_m) * t)) / k_m) / k_m) * (l + l);
	} else {
		tmp = (-0.3333333333333333 / t) * ((l / k_m) * (l / k_m));
	}
	return tmp;
}

Fortran

k_m =     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(t, l, k_m)
use fmin_fmax_functions
    real(8), intent (in) :: t
    real(8), intent (in) :: l
    real(8), intent (in) :: k_m
    real(8) :: tmp
    if (k_m <= 2.2d+114) then
        tmp = (((l / ((k_m * k_m) * t)) / k_m) / k_m) * (l + l)
    else
        tmp = ((-0.3333333333333333d0) / t) * ((l / k_m) * (l / k_m))
    end if
    code = tmp
end function

Java

k_m = Math.abs(k);
public static double code(double t, double l, double k_m) {
	double tmp;
	if (k_m <= 2.2e+114) {
		tmp = (((l / ((k_m * k_m) * t)) / k_m) / k_m) * (l + l);
	} else {
		tmp = (-0.3333333333333333 / t) * ((l / k_m) * (l / k_m));
	}
	return tmp;
}

Python

k_m = math.fabs(k)
def code(t, l, k_m):
	tmp = 0
	if k_m <= 2.2e+114:
		tmp = (((l / ((k_m * k_m) * t)) / k_m) / k_m) * (l + l)
	else:
		tmp = (-0.3333333333333333 / t) * ((l / k_m) * (l / k_m))
	return tmp

Julia

k_m = abs(k)
function code(t, l, k_m)
	tmp = 0.0
	if (k_m <= 2.2e+114)
		tmp = Float64(Float64(Float64(Float64(l / Float64(Float64(k_m * k_m) * t)) / k_m) / k_m) * Float64(l + l));
	else
		tmp = Float64(Float64(-0.3333333333333333 / t) * Float64(Float64(l / k_m) * Float64(l / k_m)));
	end
	return tmp
end

MATLAB

k_m = abs(k);
function tmp_2 = code(t, l, k_m)
	tmp = 0.0;
	if (k_m <= 2.2e+114)
		tmp = (((l / ((k_m * k_m) * t)) / k_m) / k_m) * (l + l);
	else
		tmp = (-0.3333333333333333 / t) * ((l / k_m) * (l / k_m));
	end
	tmp_2 = tmp;
end

Wolfram

k_m = N[Abs[k], $MachinePrecision]
code[t_, l_, k$95$m_] := If[LessEqual[k$95$m, 2.2e+114], N[(N[(N[(N[(l / N[(N[(k$95$m * k$95$m), $MachinePrecision] * t), $MachinePrecision]), $MachinePrecision] / k$95$m), $MachinePrecision] / k$95$m), $MachinePrecision] * N[(l + l), $MachinePrecision]), $MachinePrecision], N[(N[(-0.3333333333333333 / t), $MachinePrecision] * N[(N[(l / k$95$m), $MachinePrecision] * N[(l / k$95$m), $MachinePrecision]), $MachinePrecision]), $MachinePrecision]]

Derivation

1. Split input into 2 regimes

2. if k < 2.2e114

   1. Initial program 40.9%

      \[\frac{2}{\left(\left(\frac{{t}^{3}}{\ell \cdot \ell} \cdot \sin k\right) \cdot \tan k\right) \cdot \left(\left(1 + {\left(\frac{k}{t}\right)}^{2}\right) - 1\right)} \]
   2. Add Preprocessing

   3. Taylor expanded in k around 0

      \[\leadsto \color{blue}{2 \cdot \frac{{\ell}^{2}}{{k}^{4} \cdot t}} \]
   4. Step-by-step derivation

   5. Applied rewrites 71.9%

      \[\leadsto \color{blue}{\frac{\ell}{{k}^{4} \cdot t} \cdot \left(\ell + \ell\right)} \]
   6. Step-by-step derivation

   7. Applied rewrites 75.0%

      \[\leadsto \frac{\ell}{\left(\left(k \cdot k\right) \cdot t\right) \cdot \left(k \cdot k\right)} \cdot \left(\ell + \ell\right) \]
   8. Step-by-step derivation

   9. Applied rewrites 77.3%

      \[\leadsto \frac{\frac{\frac{\ell}{\left(k \cdot k\right) \cdot t}}{k}}{k} \cdot \left(\color{blue}{\ell} + \ell\right) \]

   if 2.2e114 < k

   1. Initial program 25.0%

      \[\frac{2}{\left(\left(\frac{{t}^{3}}{\ell \cdot \ell} \cdot \sin k\right) \cdot \tan k\right) \cdot \left(\left(1 + {\left(\frac{k}{t}\right)}^{2}\right) - 1\right)} \]
   2. Add Preprocessing

   3. Taylor expanded in k around 0

      \[\leadsto \color{blue}{\frac{\frac{-1}{3} \cdot \frac{{k}^{2} \cdot {\ell}^{2}}{t} + 2 \cdot \frac{{\ell}^{2}}{t}}{{k}^{4}}} \]
   4. Step-by-step derivation

   5. Applied rewrites 14.0%

      \[\leadsto \color{blue}{\frac{\frac{\ell \cdot \ell}{t} \cdot \mathsf{fma}\left(k \cdot k, -0.3333333333333333, 2\right)}{{k}^{4}}} \]

   6. Taylor expanded in k around inf

      \[\leadsto \frac{-1}{3} \cdot \color{blue}{\frac{{\ell}^{2}}{{k}^{2} \cdot t}} \]
   7. Step-by-step derivation

   8. Applied rewrites 62.4%

      \[\leadsto \frac{-0.3333333333333333}{t} \cdot \color{blue}{\left(\frac{\ell}{k} \cdot \frac{\ell}{k}\right)} \]
3. Recombined 2 regimes into one program.
4. Add Preprocessing

Alternative 13: 74.2% accurate, 8.9× speedup

Math

\[\begin{array}{l} k_m = \left|k\right| \\ \begin{array}{l} \mathbf{if}\;k\_m \leq 2.2 \cdot 10^{+114}:\\ \;\;\;\;\frac{\frac{\ell}{k\_m \cdot k\_m}}{\left(k\_m \cdot k\_m\right) \cdot t} \cdot \left(\ell + \ell\right)\\ \mathbf{else}:\\ \;\;\;\;\frac{-0.3333333333333333}{t} \cdot \left(\frac{\ell}{k\_m} \cdot \frac{\ell}{k\_m}\right)\\ \end{array} \end{array} \]

FPCore

k_m = (fabs.f64 k)
(FPCore (t l k_m)
 :precision binary64
 (if (<= k_m 2.2e+114)
   (* (/ (/ l (* k_m k_m)) (* (* k_m k_m) t)) (+ l l))
   (* (/ -0.3333333333333333 t) (* (/ l k_m) (/ l k_m)))))

C

k_m = fabs(k);
double code(double t, double l, double k_m) {
	double tmp;
	if (k_m <= 2.2e+114) {
		tmp = ((l / (k_m * k_m)) / ((k_m * k_m) * t)) * (l + l);
	} else {
		tmp = (-0.3333333333333333 / t) * ((l / k_m) * (l / k_m));
	}
	return tmp;
}

Fortran

k_m =     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(t, l, k_m)
use fmin_fmax_functions
    real(8), intent (in) :: t
    real(8), intent (in) :: l
    real(8), intent (in) :: k_m
    real(8) :: tmp
    if (k_m <= 2.2d+114) then
        tmp = ((l / (k_m * k_m)) / ((k_m * k_m) * t)) * (l + l)
    else
        tmp = ((-0.3333333333333333d0) / t) * ((l / k_m) * (l / k_m))
    end if
    code = tmp
end function

Java

k_m = Math.abs(k);
public static double code(double t, double l, double k_m) {
	double tmp;
	if (k_m <= 2.2e+114) {
		tmp = ((l / (k_m * k_m)) / ((k_m * k_m) * t)) * (l + l);
	} else {
		tmp = (-0.3333333333333333 / t) * ((l / k_m) * (l / k_m));
	}
	return tmp;
}

Python

k_m = math.fabs(k)
def code(t, l, k_m):
	tmp = 0
	if k_m <= 2.2e+114:
		tmp = ((l / (k_m * k_m)) / ((k_m * k_m) * t)) * (l + l)
	else:
		tmp = (-0.3333333333333333 / t) * ((l / k_m) * (l / k_m))
	return tmp

Julia

k_m = abs(k)
function code(t, l, k_m)
	tmp = 0.0
	if (k_m <= 2.2e+114)
		tmp = Float64(Float64(Float64(l / Float64(k_m * k_m)) / Float64(Float64(k_m * k_m) * t)) * Float64(l + l));
	else
		tmp = Float64(Float64(-0.3333333333333333 / t) * Float64(Float64(l / k_m) * Float64(l / k_m)));
	end
	return tmp
end

MATLAB

k_m = abs(k);
function tmp_2 = code(t, l, k_m)
	tmp = 0.0;
	if (k_m <= 2.2e+114)
		tmp = ((l / (k_m * k_m)) / ((k_m * k_m) * t)) * (l + l);
	else
		tmp = (-0.3333333333333333 / t) * ((l / k_m) * (l / k_m));
	end
	tmp_2 = tmp;
end

Wolfram

k_m = N[Abs[k], $MachinePrecision]
code[t_, l_, k$95$m_] := If[LessEqual[k$95$m, 2.2e+114], N[(N[(N[(l / N[(k$95$m * k$95$m), $MachinePrecision]), $MachinePrecision] / N[(N[(k$95$m * k$95$m), $MachinePrecision] * t), $MachinePrecision]), $MachinePrecision] * N[(l + l), $MachinePrecision]), $MachinePrecision], N[(N[(-0.3333333333333333 / t), $MachinePrecision] * N[(N[(l / k$95$m), $MachinePrecision] * N[(l / k$95$m), $MachinePrecision]), $MachinePrecision]), $MachinePrecision]]

Derivation

1. Split input into 2 regimes

2. if k < 2.2e114

   1. Initial program 40.9%

      \[\frac{2}{\left(\left(\frac{{t}^{3}}{\ell \cdot \ell} \cdot \sin k\right) \cdot \tan k\right) \cdot \left(\left(1 + {\left(\frac{k}{t}\right)}^{2}\right) - 1\right)} \]
   2. Add Preprocessing

   3. Taylor expanded in k around 0

      \[\leadsto \color{blue}{2 \cdot \frac{{\ell}^{2}}{{k}^{4} \cdot t}} \]
   4. Step-by-step derivation

   5. Applied rewrites 71.9%

      \[\leadsto \color{blue}{\frac{\ell}{{k}^{4} \cdot t} \cdot \left(\ell + \ell\right)} \]
   6. Step-by-step derivation

   7. Applied rewrites 75.0%

      \[\leadsto \frac{\ell}{\left(\left(k \cdot k\right) \cdot t\right) \cdot \left(k \cdot k\right)} \cdot \left(\ell + \ell\right) \]
   8. Step-by-step derivation

   9. Applied rewrites 77.2%

      \[\leadsto \frac{\frac{\ell}{k \cdot k}}{\left(k \cdot k\right) \cdot t} \cdot \left(\color{blue}{\ell} + \ell\right) \]

   if 2.2e114 < k

   1. Initial program 25.0%

      \[\frac{2}{\left(\left(\frac{{t}^{3}}{\ell \cdot \ell} \cdot \sin k\right) \cdot \tan k\right) \cdot \left(\left(1 + {\left(\frac{k}{t}\right)}^{2}\right) - 1\right)} \]
   2. Add Preprocessing

   3. Taylor expanded in k around 0

      \[\leadsto \color{blue}{\frac{\frac{-1}{3} \cdot \frac{{k}^{2} \cdot {\ell}^{2}}{t} + 2 \cdot \frac{{\ell}^{2}}{t}}{{k}^{4}}} \]
   4. Step-by-step derivation

   5. Applied rewrites 14.0%

      \[\leadsto \color{blue}{\frac{\frac{\ell \cdot \ell}{t} \cdot \mathsf{fma}\left(k \cdot k, -0.3333333333333333, 2\right)}{{k}^{4}}} \]

   6. Taylor expanded in k around inf

      \[\leadsto \frac{-1}{3} \cdot \color{blue}{\frac{{\ell}^{2}}{{k}^{2} \cdot t}} \]
   7. Step-by-step derivation

   8. Applied rewrites 62.4%

      \[\leadsto \frac{-0.3333333333333333}{t} \cdot \color{blue}{\left(\frac{\ell}{k} \cdot \frac{\ell}{k}\right)} \]
3. Recombined 2 regimes into one program.
4. Add Preprocessing

Alternative 14: 72.2% accurate, 9.2× speedup

Math

\[\begin{array}{l} k_m = \left|k\right| \\ \begin{array}{l} \mathbf{if}\;k\_m \leq 2.2 \cdot 10^{+114}:\\ \;\;\;\;\frac{\ell}{k\_m \cdot \left(\left(k\_m \cdot t\right) \cdot \left(k\_m \cdot k\_m\right)\right)} \cdot \left(\ell + \ell\right)\\ \mathbf{else}:\\ \;\;\;\;\frac{-0.3333333333333333}{t} \cdot \left(\frac{\ell}{k\_m} \cdot \frac{\ell}{k\_m}\right)\\ \end{array} \end{array} \]

FPCore

k_m = (fabs.f64 k)
(FPCore (t l k_m)
 :precision binary64
 (if (<= k_m 2.2e+114)
   (* (/ l (* k_m (* (* k_m t) (* k_m k_m)))) (+ l l))
   (* (/ -0.3333333333333333 t) (* (/ l k_m) (/ l k_m)))))

C

k_m = fabs(k);
double code(double t, double l, double k_m) {
	double tmp;
	if (k_m <= 2.2e+114) {
		tmp = (l / (k_m * ((k_m * t) * (k_m * k_m)))) * (l + l);
	} else {
		tmp = (-0.3333333333333333 / t) * ((l / k_m) * (l / k_m));
	}
	return tmp;
}

Fortran

k_m =     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(t, l, k_m)
use fmin_fmax_functions
    real(8), intent (in) :: t
    real(8), intent (in) :: l
    real(8), intent (in) :: k_m
    real(8) :: tmp
    if (k_m <= 2.2d+114) then
        tmp = (l / (k_m * ((k_m * t) * (k_m * k_m)))) * (l + l)
    else
        tmp = ((-0.3333333333333333d0) / t) * ((l / k_m) * (l / k_m))
    end if
    code = tmp
end function

Java

k_m = Math.abs(k);
public static double code(double t, double l, double k_m) {
	double tmp;
	if (k_m <= 2.2e+114) {
		tmp = (l / (k_m * ((k_m * t) * (k_m * k_m)))) * (l + l);
	} else {
		tmp = (-0.3333333333333333 / t) * ((l / k_m) * (l / k_m));
	}
	return tmp;
}

Python

k_m = math.fabs(k)
def code(t, l, k_m):
	tmp = 0
	if k_m <= 2.2e+114:
		tmp = (l / (k_m * ((k_m * t) * (k_m * k_m)))) * (l + l)
	else:
		tmp = (-0.3333333333333333 / t) * ((l / k_m) * (l / k_m))
	return tmp

Julia

k_m = abs(k)
function code(t, l, k_m)
	tmp = 0.0
	if (k_m <= 2.2e+114)
		tmp = Float64(Float64(l / Float64(k_m * Float64(Float64(k_m * t) * Float64(k_m * k_m)))) * Float64(l + l));
	else
		tmp = Float64(Float64(-0.3333333333333333 / t) * Float64(Float64(l / k_m) * Float64(l / k_m)));
	end
	return tmp
end

MATLAB

k_m = abs(k);
function tmp_2 = code(t, l, k_m)
	tmp = 0.0;
	if (k_m <= 2.2e+114)
		tmp = (l / (k_m * ((k_m * t) * (k_m * k_m)))) * (l + l);
	else
		tmp = (-0.3333333333333333 / t) * ((l / k_m) * (l / k_m));
	end
	tmp_2 = tmp;
end

Wolfram

k_m = N[Abs[k], $MachinePrecision]
code[t_, l_, k$95$m_] := If[LessEqual[k$95$m, 2.2e+114], N[(N[(l / N[(k$95$m * N[(N[(k$95$m * t), $MachinePrecision] * N[(k$95$m * k$95$m), $MachinePrecision]), $MachinePrecision]), $MachinePrecision]), $MachinePrecision] * N[(l + l), $MachinePrecision]), $MachinePrecision], N[(N[(-0.3333333333333333 / t), $MachinePrecision] * N[(N[(l / k$95$m), $MachinePrecision] * N[(l / k$95$m), $MachinePrecision]), $MachinePrecision]), $MachinePrecision]]

Derivation

1. Split input into 2 regimes

2. if k < 2.2e114

   1. Initial program 40.9%

      \[\frac{2}{\left(\left(\frac{{t}^{3}}{\ell \cdot \ell} \cdot \sin k\right) \cdot \tan k\right) \cdot \left(\left(1 + {\left(\frac{k}{t}\right)}^{2}\right) - 1\right)} \]
   2. Add Preprocessing

   3. Taylor expanded in k around 0

      \[\leadsto \color{blue}{2 \cdot \frac{{\ell}^{2}}{{k}^{4} \cdot t}} \]
   4. Step-by-step derivation

   5. Applied rewrites 71.9%

      \[\leadsto \color{blue}{\frac{\ell}{{k}^{4} \cdot t} \cdot \left(\ell + \ell\right)} \]
   6. Step-by-step derivation

   7. Applied rewrites 75.0%

      \[\leadsto \frac{\ell}{\left(\left(k \cdot k\right) \cdot t\right) \cdot \left(k \cdot k\right)} \cdot \left(\ell + \ell\right) \]
   8. Step-by-step derivation

   9. Applied rewrites 75.0%

      \[\leadsto \frac{\ell}{k \cdot \left(\left(k \cdot t\right) \cdot \left(k \cdot k\right)\right)} \cdot \left(\ell + \ell\right) \]

   if 2.2e114 < k

   1. Initial program 25.0%

      \[\frac{2}{\left(\left(\frac{{t}^{3}}{\ell \cdot \ell} \cdot \sin k\right) \cdot \tan k\right) \cdot \left(\left(1 + {\left(\frac{k}{t}\right)}^{2}\right) - 1\right)} \]
   2. Add Preprocessing

   3. Taylor expanded in k around 0

      \[\leadsto \color{blue}{\frac{\frac{-1}{3} \cdot \frac{{k}^{2} \cdot {\ell}^{2}}{t} + 2 \cdot \frac{{\ell}^{2}}{t}}{{k}^{4}}} \]
   4. Step-by-step derivation

   5. Applied rewrites 14.0%

      \[\leadsto \color{blue}{\frac{\frac{\ell \cdot \ell}{t} \cdot \mathsf{fma}\left(k \cdot k, -0.3333333333333333, 2\right)}{{k}^{4}}} \]

   6. Taylor expanded in k around inf

      \[\leadsto \frac{-1}{3} \cdot \color{blue}{\frac{{\ell}^{2}}{{k}^{2} \cdot t}} \]
   7. Step-by-step derivation

   8. Applied rewrites 62.4%

      \[\leadsto \frac{-0.3333333333333333}{t} \cdot \color{blue}{\left(\frac{\ell}{k} \cdot \frac{\ell}{k}\right)} \]
3. Recombined 2 regimes into one program.
4. Add Preprocessing

Alternative 15: 71.2% accurate, 11.6× speedup

Math

\[\begin{array}{l} k_m = \left|k\right| \\ \frac{\ell}{k\_m \cdot \left(\left(k\_m \cdot t\right) \cdot \left(k\_m \cdot k\_m\right)\right)} \cdot \left(\ell + \ell\right) \end{array} \]

FPCore

k_m = (fabs.f64 k)
(FPCore (t l k_m)
 :precision binary64
 (* (/ l (* k_m (* (* k_m t) (* k_m k_m)))) (+ l l)))

C

k_m = fabs(k);
double code(double t, double l, double k_m) {
	return (l / (k_m * ((k_m * t) * (k_m * k_m)))) * (l + l);
}

Fortran

k_m =     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(t, l, k_m)
use fmin_fmax_functions
    real(8), intent (in) :: t
    real(8), intent (in) :: l
    real(8), intent (in) :: k_m
    code = (l / (k_m * ((k_m * t) * (k_m * k_m)))) * (l + l)
end function

Java

k_m = Math.abs(k);
public static double code(double t, double l, double k_m) {
	return (l / (k_m * ((k_m * t) * (k_m * k_m)))) * (l + l);
}

Python

k_m = math.fabs(k)
def code(t, l, k_m):
	return (l / (k_m * ((k_m * t) * (k_m * k_m)))) * (l + l)

Julia

k_m = abs(k)
function code(t, l, k_m)
	return Float64(Float64(l / Float64(k_m * Float64(Float64(k_m * t) * Float64(k_m * k_m)))) * Float64(l + l))
end

MATLAB

k_m = abs(k);
function tmp = code(t, l, k_m)
	tmp = (l / (k_m * ((k_m * t) * (k_m * k_m)))) * (l + l);
end

Wolfram

k_m = N[Abs[k], $MachinePrecision]
code[t_, l_, k$95$m_] := N[(N[(l / N[(k$95$m * N[(N[(k$95$m * t), $MachinePrecision] * N[(k$95$m * k$95$m), $MachinePrecision]), $MachinePrecision]), $MachinePrecision]), $MachinePrecision] * N[(l + l), $MachinePrecision]), $MachinePrecision]

Derivation

1. Initial program 38.7%

   \[\frac{2}{\left(\left(\frac{{t}^{3}}{\ell \cdot \ell} \cdot \sin k\right) \cdot \tan k\right) \cdot \left(\left(1 + {\left(\frac{k}{t}\right)}^{2}\right) - 1\right)} \]
2. Add Preprocessing

3. Taylor expanded in k around 0

   \[\leadsto \color{blue}{2 \cdot \frac{{\ell}^{2}}{{k}^{4} \cdot t}} \]
4. Step-by-step derivation

5. Applied rewrites 69.5%

   \[\leadsto \color{blue}{\frac{\ell}{{k}^{4} \cdot t} \cdot \left(\ell + \ell\right)} \]
6. Step-by-step derivation

7. Applied rewrites 72.1%

   \[\leadsto \frac{\ell}{\left(\left(k \cdot k\right) \cdot t\right) \cdot \left(k \cdot k\right)} \cdot \left(\ell + \ell\right) \]
8. Step-by-step derivation

9. Applied rewrites 72.2%

   \[\leadsto \frac{\ell}{k \cdot \left(\left(k \cdot t\right) \cdot \left(k \cdot k\right)\right)} \cdot \left(\ell + \ell\right) \]
10. Add Preprocessing

Alternative 16: 55.9% accurate, 14.4× speedup

Math

\[\begin{array}{l} k_m = \left|k\right| \\ \frac{2}{\left(\left(\left(k\_m \cdot k\_m\right) \cdot t\right) \cdot k\_m\right) \cdot k\_m} \end{array} \]

FPCore

k_m = (fabs.f64 k)
(FPCore (t l k_m)
 :precision binary64
 (/ 2.0 (* (* (* (* k_m k_m) t) k_m) k_m)))

C

k_m = fabs(k);
double code(double t, double l, double k_m) {
	return 2.0 / ((((k_m * k_m) * t) * k_m) * k_m);
}

Fortran

k_m =     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(t, l, k_m)
use fmin_fmax_functions
    real(8), intent (in) :: t
    real(8), intent (in) :: l
    real(8), intent (in) :: k_m
    code = 2.0d0 / ((((k_m * k_m) * t) * k_m) * k_m)
end function

Java

k_m = Math.abs(k);
public static double code(double t, double l, double k_m) {
	return 2.0 / ((((k_m * k_m) * t) * k_m) * k_m);
}

Python

k_m = math.fabs(k)
def code(t, l, k_m):
	return 2.0 / ((((k_m * k_m) * t) * k_m) * k_m)

Julia

k_m = abs(k)
function code(t, l, k_m)
	return Float64(2.0 / Float64(Float64(Float64(Float64(k_m * k_m) * t) * k_m) * k_m))
end

MATLAB

k_m = abs(k);
function tmp = code(t, l, k_m)
	tmp = 2.0 / ((((k_m * k_m) * t) * k_m) * k_m);
end

Wolfram

k_m = N[Abs[k], $MachinePrecision]
code[t_, l_, k$95$m_] := N[(2.0 / N[(N[(N[(N[(k$95$m * k$95$m), $MachinePrecision] * t), $MachinePrecision] * k$95$m), $MachinePrecision] * k$95$m), $MachinePrecision]), $MachinePrecision]

Derivation

1. Initial program 38.7%

   \[\frac{2}{\left(\left(\frac{{t}^{3}}{\ell \cdot \ell} \cdot \sin k\right) \cdot \tan k\right) \cdot \left(\left(1 + {\left(\frac{k}{t}\right)}^{2}\right) - 1\right)} \]
2. Add Preprocessing

3. Taylor expanded in k around 0

   \[\leadsto \color{blue}{2 \cdot \frac{{\ell}^{2}}{{k}^{4} \cdot t}} \]
4. Step-by-step derivation

5. Applied rewrites 69.5%

   \[\leadsto \color{blue}{\frac{\ell}{{k}^{4} \cdot t} \cdot \left(\ell + \ell\right)} \]
6. Step-by-step derivation

7. Applied rewrites 72.1%

   \[\leadsto \frac{\ell}{\left(\left(k \cdot k\right) \cdot t\right) \cdot \left(k \cdot k\right)} \cdot \left(\ell + \ell\right) \]
8. Step-by-step derivation

9. Applied rewrites 55.8%

   \[\leadsto \frac{\frac{2}{\left(k \cdot k\right) \cdot t}}{\color{blue}{k \cdot k}} \]
10. Step-by-step derivation

11. Applied rewrites 56.2%

    \[\leadsto \frac{2}{\color{blue}{\left(\left(\left(k \cdot k\right) \cdot t\right) \cdot k\right) \cdot k}} \]
12. Add Preprocessing

Reproduce

herbie shell --seed 2025021 
(FPCore (t l k)
  :name "Toniolo and Linder, Equation (10-)"
  :precision binary64
  (/ 2.0 (* (* (* (/ (pow t 3.0) (* l l)) (sin k)) (tan k)) (- (+ 1.0 (pow (/ k t) 2.0)) 1.0))))