Graphics.Rasterific.Svg.PathConverter:segmentToBezier from rasterific-svg-0.2.3.1, A

Percentage Accurate: 76.8% → 80.0%

Time: 1.7s

Alternatives: 4

Speedup: 0.3×

Specification

\[\mathsf{TRUE}\left(\right)\]

Math

\[\begin{array}{l} \\ \begin{array}{l} t_0 := \sin \left(x \cdot 0.5\right)\\ \frac{\left(\frac{8}{3} \cdot t\_0\right) \cdot t\_0}{\sin x} \end{array} \end{array} \]

FPCore

(FPCore (x)
 :precision binary64
 (let* ((t_0 (sin (* x 0.5)))) (/ (* (* (/ 8.0 3.0) t_0) t_0) (sin x))))

C

double code(double x) {
	double t_0 = sin((x * 0.5));
	return (((8.0 / 3.0) * t_0) * t_0) / sin(x);
}

Fortran

real(8) function code(x)
    real(8), intent (in) :: x
    real(8) :: t_0
    t_0 = sin((x * 0.5d0))
    code = (((8.0d0 / 3.0d0) * t_0) * t_0) / sin(x)
end function

Java

public static double code(double x) {
	double t_0 = Math.sin((x * 0.5));
	return (((8.0 / 3.0) * t_0) * t_0) / Math.sin(x);
}

Python

def code(x):
	t_0 = math.sin((x * 0.5))
	return (((8.0 / 3.0) * t_0) * t_0) / math.sin(x)

Julia

function code(x)
	t_0 = sin(Float64(x * 0.5))
	return Float64(Float64(Float64(Float64(8.0 / 3.0) * t_0) * t_0) / sin(x))
end

MATLAB

function tmp = code(x)
	t_0 = sin((x * 0.5));
	tmp = (((8.0 / 3.0) * t_0) * t_0) / sin(x);
end

Wolfram

code[x_] := Block[{t$95$0 = N[Sin[N[(x * 0.5), $MachinePrecision]], $MachinePrecision]}, N[(N[(N[(N[(8.0 / 3.0), $MachinePrecision] * t$95$0), $MachinePrecision] * t$95$0), $MachinePrecision] / N[Sin[x], $MachinePrecision]), $MachinePrecision]]

Initial Program: 76.8% accurate, 1.0× speedup

Math

\[\begin{array}{l} \\ \begin{array}{l} t_0 := \sin \left(x \cdot 0.5\right)\\ \frac{\left(\frac{8}{3} \cdot t\_0\right) \cdot t\_0}{\sin x} \end{array} \end{array} \]

FPCore

(FPCore (x)
 :precision binary64
 (let* ((t_0 (sin (* x 0.5)))) (/ (* (* (/ 8.0 3.0) t_0) t_0) (sin x))))

C

double code(double x) {
	double t_0 = sin((x * 0.5));
	return (((8.0 / 3.0) * t_0) * t_0) / sin(x);
}

Fortran

real(8) function code(x)
    real(8), intent (in) :: x
    real(8) :: t_0
    t_0 = sin((x * 0.5d0))
    code = (((8.0d0 / 3.0d0) * t_0) * t_0) / sin(x)
end function

Java

public static double code(double x) {
	double t_0 = Math.sin((x * 0.5));
	return (((8.0 / 3.0) * t_0) * t_0) / Math.sin(x);
}

Python

def code(x):
	t_0 = math.sin((x * 0.5))
	return (((8.0 / 3.0) * t_0) * t_0) / math.sin(x)

Julia

function code(x)
	t_0 = sin(Float64(x * 0.5))
	return Float64(Float64(Float64(Float64(8.0 / 3.0) * t_0) * t_0) / sin(x))
end

MATLAB

function tmp = code(x)
	t_0 = sin((x * 0.5));
	tmp = (((8.0 / 3.0) * t_0) * t_0) / sin(x);
end

Wolfram

code[x_] := Block[{t$95$0 = N[Sin[N[(x * 0.5), $MachinePrecision]], $MachinePrecision]}, N[(N[(N[(N[(8.0 / 3.0), $MachinePrecision] * t$95$0), $MachinePrecision] * t$95$0), $MachinePrecision] / N[Sin[x], $MachinePrecision]), $MachinePrecision]]

Alternative 1: 80.0% accurate, 0.3× speedup

Math

\[\begin{array}{l} \\ \begin{array}{l} t_0 := \sin \left(x \cdot 0.5\right)\\ t_1 := \frac{\left(\frac{8}{3} \cdot t\_0\right) \cdot t\_0}{\sin x}\\ \mathbf{if}\;t\_1 \leq -4 \cdot 10^{-159}:\\ \;\;\;\;t\_1\\ \mathbf{elif}\;t\_1 \leq 0:\\ \;\;\;\;x \cdot 0.5\\ \mathbf{else}:\\ \;\;\;\;t\_1\\ \end{array} \end{array} \]

FPCore

(FPCore (x)
 :precision binary64
 (let* ((t_0 (sin (* x 0.5))) (t_1 (/ (* (* (/ 8.0 3.0) t_0) t_0) (sin x))))
   (if (<= t_1 -4e-159) t_1 (if (<= t_1 0.0) (* x 0.5) t_1))))

C

double code(double x) {
	double t_0 = sin((x * 0.5));
	double t_1 = (((8.0 / 3.0) * t_0) * t_0) / sin(x);
	double tmp;
	if (t_1 <= -4e-159) {
		tmp = t_1;
	} else if (t_1 <= 0.0) {
		tmp = x * 0.5;
	} else {
		tmp = t_1;
	}
	return tmp;
}

Fortran

real(8) function code(x)
    real(8), intent (in) :: x
    real(8) :: t_0
    real(8) :: t_1
    real(8) :: tmp
    t_0 = sin((x * 0.5d0))
    t_1 = (((8.0d0 / 3.0d0) * t_0) * t_0) / sin(x)
    if (t_1 <= (-4d-159)) then
        tmp = t_1
    else if (t_1 <= 0.0d0) then
        tmp = x * 0.5d0
    else
        tmp = t_1
    end if
    code = tmp
end function

Java

public static double code(double x) {
	double t_0 = Math.sin((x * 0.5));
	double t_1 = (((8.0 / 3.0) * t_0) * t_0) / Math.sin(x);
	double tmp;
	if (t_1 <= -4e-159) {
		tmp = t_1;
	} else if (t_1 <= 0.0) {
		tmp = x * 0.5;
	} else {
		tmp = t_1;
	}
	return tmp;
}

Python

def code(x):
	t_0 = math.sin((x * 0.5))
	t_1 = (((8.0 / 3.0) * t_0) * t_0) / math.sin(x)
	tmp = 0
	if t_1 <= -4e-159:
		tmp = t_1
	elif t_1 <= 0.0:
		tmp = x * 0.5
	else:
		tmp = t_1
	return tmp

Julia

function code(x)
	t_0 = sin(Float64(x * 0.5))
	t_1 = Float64(Float64(Float64(Float64(8.0 / 3.0) * t_0) * t_0) / sin(x))
	tmp = 0.0
	if (t_1 <= -4e-159)
		tmp = t_1;
	elseif (t_1 <= 0.0)
		tmp = Float64(x * 0.5);
	else
		tmp = t_1;
	end
	return tmp
end

MATLAB

function tmp_2 = code(x)
	t_0 = sin((x * 0.5));
	t_1 = (((8.0 / 3.0) * t_0) * t_0) / sin(x);
	tmp = 0.0;
	if (t_1 <= -4e-159)
		tmp = t_1;
	elseif (t_1 <= 0.0)
		tmp = x * 0.5;
	else
		tmp = t_1;
	end
	tmp_2 = tmp;
end

Wolfram

code[x_] := Block[{t$95$0 = N[Sin[N[(x * 0.5), $MachinePrecision]], $MachinePrecision]}, Block[{t$95$1 = N[(N[(N[(N[(8.0 / 3.0), $MachinePrecision] * t$95$0), $MachinePrecision] * t$95$0), $MachinePrecision] / N[Sin[x], $MachinePrecision]), $MachinePrecision]}, If[LessEqual[t$95$1, -4e-159], t$95$1, If[LessEqual[t$95$1, 0.0], N[(x * 0.5), $MachinePrecision], t$95$1]]]]

Derivation

1. Split input into 2 regimes

2. if (/.f64 (*.f64 (*.f64 (/.f64 #s(literal 8 binary64) #s(literal 3 binary64)) (sin.f64 (*.f64 x #s(literal 1/2 binary64)))) (sin.f64 (*.f64 x #s(literal 1/2 binary64)))) (sin.f64 x)) < -3.99999999999999995e-159 or -0.0 < (/.f64 (*.f64 (*.f64 (/.f64 #s(literal 8 binary64) #s(literal 3 binary64)) (sin.f64 (*.f64 x #s(literal 1/2 binary64)))) (sin.f64 (*.f64 x #s(literal 1/2 binary64)))) (sin.f64 x))

   1. Initial program 98.7%

      \[\frac{\left(\frac{8}{3} \cdot \sin \left(x \cdot 0.5\right)\right) \cdot \sin \left(x \cdot 0.5\right)}{\sin x} \]
   2. Add Preprocessing

   if -3.99999999999999995e-159 < (/.f64 (*.f64 (*.f64 (/.f64 #s(literal 8 binary64) #s(literal 3 binary64)) (sin.f64 (*.f64 x #s(literal 1/2 binary64)))) (sin.f64 (*.f64 x #s(literal 1/2 binary64)))) (sin.f64 x)) < -0.0

   1. Initial program 7.1%

      \[\frac{\left(\frac{8}{3} \cdot \sin \left(x \cdot 0.5\right)\right) \cdot \sin \left(x \cdot 0.5\right)}{\sin x} \]
   2. Add Preprocessing

   3. Taylor expanded in x around 0

      \[\leadsto \color{blue}{x \cdot \left(\frac{2}{3} + {x}^{2} \cdot \left(\frac{1}{18} + \frac{1}{180} \cdot {x}^{2}\right)\right)} \]

   5. Applied rewrites 18.8%

      \[\leadsto \color{blue}{\frac{8}{3} \cdot \sin \left(x \cdot 0.5\right)} \]

   6. Taylor expanded in x around 0

      \[\leadsto x \cdot \color{blue}{\left(\frac{4}{3} + {x}^{2} \cdot \left(\frac{1}{1440} \cdot {x}^{2} - \frac{1}{18}\right)\right)} \]

   8. Applied rewrites 20.7%

      \[\leadsto x \cdot \color{blue}{0.5} \]
3. Recombined 2 regimes into one program.
4. Add Preprocessing

Alternative 2: 15.6% accurate, 1.0× speedup

Math

\[\begin{array}{l} \\ \begin{array}{l} t_0 := \sin \left(x \cdot 0.5\right)\\ \mathbf{if}\;\frac{\left(\frac{8}{3} \cdot t\_0\right) \cdot t\_0}{\sin x} \leq 5 \cdot 10^{-8}:\\ \;\;\;\;x \cdot 0.5\\ \mathbf{else}:\\ \;\;\;\;\frac{8}{3}\\ \end{array} \end{array} \]

FPCore

(FPCore (x)
 :precision binary64
 (let* ((t_0 (sin (* x 0.5))))
   (if (<= (/ (* (* (/ 8.0 3.0) t_0) t_0) (sin x)) 5e-8)
     (* x 0.5)
     (/ 8.0 3.0))))

C

double code(double x) {
	double t_0 = sin((x * 0.5));
	double tmp;
	if (((((8.0 / 3.0) * t_0) * t_0) / sin(x)) <= 5e-8) {
		tmp = x * 0.5;
	} else {
		tmp = 8.0 / 3.0;
	}
	return tmp;
}

Fortran

real(8) function code(x)
    real(8), intent (in) :: x
    real(8) :: t_0
    real(8) :: tmp
    t_0 = sin((x * 0.5d0))
    if (((((8.0d0 / 3.0d0) * t_0) * t_0) / sin(x)) <= 5d-8) then
        tmp = x * 0.5d0
    else
        tmp = 8.0d0 / 3.0d0
    end if
    code = tmp
end function

Java

public static double code(double x) {
	double t_0 = Math.sin((x * 0.5));
	double tmp;
	if (((((8.0 / 3.0) * t_0) * t_0) / Math.sin(x)) <= 5e-8) {
		tmp = x * 0.5;
	} else {
		tmp = 8.0 / 3.0;
	}
	return tmp;
}

Python

def code(x):
	t_0 = math.sin((x * 0.5))
	tmp = 0
	if ((((8.0 / 3.0) * t_0) * t_0) / math.sin(x)) <= 5e-8:
		tmp = x * 0.5
	else:
		tmp = 8.0 / 3.0
	return tmp

Julia

function code(x)
	t_0 = sin(Float64(x * 0.5))
	tmp = 0.0
	if (Float64(Float64(Float64(Float64(8.0 / 3.0) * t_0) * t_0) / sin(x)) <= 5e-8)
		tmp = Float64(x * 0.5);
	else
		tmp = Float64(8.0 / 3.0);
	end
	return tmp
end

MATLAB

function tmp_2 = code(x)
	t_0 = sin((x * 0.5));
	tmp = 0.0;
	if (((((8.0 / 3.0) * t_0) * t_0) / sin(x)) <= 5e-8)
		tmp = x * 0.5;
	else
		tmp = 8.0 / 3.0;
	end
	tmp_2 = tmp;
end

Wolfram

code[x_] := Block[{t$95$0 = N[Sin[N[(x * 0.5), $MachinePrecision]], $MachinePrecision]}, If[LessEqual[N[(N[(N[(N[(8.0 / 3.0), $MachinePrecision] * t$95$0), $MachinePrecision] * t$95$0), $MachinePrecision] / N[Sin[x], $MachinePrecision]), $MachinePrecision], 5e-8], N[(x * 0.5), $MachinePrecision], N[(8.0 / 3.0), $MachinePrecision]]]

Derivation

1. Split input into 2 regimes

2. if (/.f64 (*.f64 (*.f64 (/.f64 #s(literal 8 binary64) #s(literal 3 binary64)) (sin.f64 (*.f64 x #s(literal 1/2 binary64)))) (sin.f64 (*.f64 x #s(literal 1/2 binary64)))) (sin.f64 x)) < 4.9999999999999998e-8

   1. Initial program 70.8%

      \[\frac{\left(\frac{8}{3} \cdot \sin \left(x \cdot 0.5\right)\right) \cdot \sin \left(x \cdot 0.5\right)}{\sin x} \]
   2. Add Preprocessing

   3. Taylor expanded in x around 0

      \[\leadsto \color{blue}{x \cdot \left(\frac{2}{3} + {x}^{2} \cdot \left(\frac{1}{18} + \frac{1}{180} \cdot {x}^{2}\right)\right)} \]

   5. Applied rewrites 16.0%

      \[\leadsto \color{blue}{\frac{8}{3} \cdot \sin \left(x \cdot 0.5\right)} \]

   6. Taylor expanded in x around 0

      \[\leadsto x \cdot \color{blue}{\left(\frac{4}{3} + {x}^{2} \cdot \left(\frac{1}{1440} \cdot {x}^{2} - \frac{1}{18}\right)\right)} \]

   8. Applied rewrites 14.4%

      \[\leadsto x \cdot \color{blue}{0.5} \]

   if 4.9999999999999998e-8 < (/.f64 (*.f64 (*.f64 (/.f64 #s(literal 8 binary64) #s(literal 3 binary64)) (sin.f64 (*.f64 x #s(literal 1/2 binary64)))) (sin.f64 (*.f64 x #s(literal 1/2 binary64)))) (sin.f64 x))

   1. Initial program 98.9%

      \[\frac{\left(\frac{8}{3} \cdot \sin \left(x \cdot 0.5\right)\right) \cdot \sin \left(x \cdot 0.5\right)}{\sin x} \]
   2. Add Preprocessing

   3. Taylor expanded in x around 0

      \[\leadsto \color{blue}{x \cdot \left(\frac{2}{3} + {x}^{2} \cdot \left(\frac{1}{18} + {x}^{2} \cdot \left(\frac{1}{180} + \frac{17}{30240} \cdot {x}^{2}\right)\right)\right)} \]

   5. Applied rewrites 18.2%

      \[\leadsto \color{blue}{\frac{8}{3}} \]
3. Recombined 2 regimes into one program.
4. Add Preprocessing

Alternative 3: 19.7% accurate, 3.4× speedup

Math

\[\begin{array}{l} \\ \sin x \end{array} \]

FPCore

(FPCore (x) :precision binary64 (sin x))

C

double code(double x) {
	return sin(x);
}

Fortran

real(8) function code(x)
    real(8), intent (in) :: x
    code = sin(x)
end function

Java

public static double code(double x) {
	return Math.sin(x);
}

Python

def code(x):
	return math.sin(x)

Julia

function code(x)
	return sin(x)
end

MATLAB

function tmp = code(x)
	tmp = sin(x);
end

Wolfram

code[x_] := N[Sin[x], $MachinePrecision]

Derivation

1. Initial program 77.3%

   \[\frac{\left(\frac{8}{3} \cdot \sin \left(x \cdot 0.5\right)\right) \cdot \sin \left(x \cdot 0.5\right)}{\sin x} \]
2. Add Preprocessing

3. Taylor expanded in x around 0

   \[\leadsto \color{blue}{x \cdot \left(\frac{2}{3} + {x}^{2} \cdot \left(\frac{1}{18} + \frac{1}{180} \cdot {x}^{2}\right)\right)} \]

5. Applied rewrites 15.1%

   \[\leadsto \color{blue}{\frac{8}{3} \cdot \sin \left(x \cdot 0.5\right)} \]

6. Taylor expanded in x around 0

   \[\leadsto x \cdot \color{blue}{\left(\frac{4}{3} + \frac{-1}{18} \cdot {x}^{2}\right)} \]

8. Applied rewrites 19.8%

   \[\leadsto \sin x \]
9. Add Preprocessing

Alternative 4: 11.8% accurate, 57.2× speedup

Math

\[\begin{array}{l} \\ x \cdot 0.5 \end{array} \]

FPCore

(FPCore (x) :precision binary64 (* x 0.5))

C

double code(double x) {
	return x * 0.5;
}

Fortran

real(8) function code(x)
    real(8), intent (in) :: x
    code = x * 0.5d0
end function

Java

public static double code(double x) {
	return x * 0.5;
}

Python

def code(x):
	return x * 0.5

Julia

function code(x)
	return Float64(x * 0.5)
end

MATLAB

function tmp = code(x)
	tmp = x * 0.5;
end

Wolfram

code[x_] := N[(x * 0.5), $MachinePrecision]

Derivation

1. Initial program 77.3%

   \[\frac{\left(\frac{8}{3} \cdot \sin \left(x \cdot 0.5\right)\right) \cdot \sin \left(x \cdot 0.5\right)}{\sin x} \]
2. Add Preprocessing

3. Taylor expanded in x around 0

   \[\leadsto \color{blue}{x \cdot \left(\frac{2}{3} + {x}^{2} \cdot \left(\frac{1}{18} + \frac{1}{180} \cdot {x}^{2}\right)\right)} \]

5. Applied rewrites 15.1%

   \[\leadsto \color{blue}{\frac{8}{3} \cdot \sin \left(x \cdot 0.5\right)} \]

6. Taylor expanded in x around 0

   \[\leadsto x \cdot \color{blue}{\left(\frac{4}{3} + {x}^{2} \cdot \left(\frac{1}{1440} \cdot {x}^{2} - \frac{1}{18}\right)\right)} \]

8. Applied rewrites 12.2%

   \[\leadsto x \cdot \color{blue}{0.5} \]
9. Add Preprocessing

Developer Target 1: 99.5% accurate, 1.0× speedup

Math

\[\begin{array}{l} \\ \begin{array}{l} t_0 := \sin \left(x \cdot 0.5\right)\\ \frac{\frac{8 \cdot t\_0}{3}}{\frac{\sin x}{t\_0}} \end{array} \end{array} \]

FPCore

(FPCore (x)
 :precision binary64
 (let* ((t_0 (sin (* x 0.5)))) (/ (/ (* 8.0 t_0) 3.0) (/ (sin x) t_0))))

C

double code(double x) {
	double t_0 = sin((x * 0.5));
	return ((8.0 * t_0) / 3.0) / (sin(x) / t_0);
}

Fortran

real(8) function code(x)
    real(8), intent (in) :: x
    real(8) :: t_0
    t_0 = sin((x * 0.5d0))
    code = ((8.0d0 * t_0) / 3.0d0) / (sin(x) / t_0)
end function

Java

public static double code(double x) {
	double t_0 = Math.sin((x * 0.5));
	return ((8.0 * t_0) / 3.0) / (Math.sin(x) / t_0);
}

Python

def code(x):
	t_0 = math.sin((x * 0.5))
	return ((8.0 * t_0) / 3.0) / (math.sin(x) / t_0)

Julia

function code(x)
	t_0 = sin(Float64(x * 0.5))
	return Float64(Float64(Float64(8.0 * t_0) / 3.0) / Float64(sin(x) / t_0))
end

MATLAB

function tmp = code(x)
	t_0 = sin((x * 0.5));
	tmp = ((8.0 * t_0) / 3.0) / (sin(x) / t_0);
end

Wolfram

code[x_] := Block[{t$95$0 = N[Sin[N[(x * 0.5), $MachinePrecision]], $MachinePrecision]}, N[(N[(N[(8.0 * t$95$0), $MachinePrecision] / 3.0), $MachinePrecision] / N[(N[Sin[x], $MachinePrecision] / t$95$0), $MachinePrecision]), $MachinePrecision]]

Reproduce

herbie shell --seed 2024321 
(FPCore (x)
  :name "Graphics.Rasterific.Svg.PathConverter:segmentToBezier from rasterific-svg-0.2.3.1, A"
  :precision binary64
  :pre (TRUE)

  :alt
  (! :herbie-platform default (/ (/ (* 8 (sin (* x 1/2))) 3) (/ (sin x) (sin (* x 1/2)))))

  (/ (* (* (/ 8.0 3.0) (sin (* x 0.5))) (sin (* x 0.5))) (sin x)))