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

Percentage Accurate: 77.2% → 99.8%

Time: 4.4s

Alternatives: 9

Speedup: 2.7×

Specification

Math

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

FPCore

(FPCore (x)
  :precision binary64
  :pre TRUE
  (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)
use fmin_fmax_functions
    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]]

ReFlow

f(x):
	x in [-inf, +inf]
code: THEORY
BEGIN
f(x: real): real =
	LET t_0 = (sin((x * (5e-1)))) IN
	((((8) / (3)) * t_0) * t_0) / (sin(x))
END code

Initial Program: 77.2% accurate, 1.0× speedup

Math

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

FPCore

(FPCore (x)
  :precision binary64
  :pre TRUE
  (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)
use fmin_fmax_functions
    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]]

ReFlow

f(x):
	x in [-inf, +inf]
code: THEORY
BEGIN
f(x: real): real =
	LET t_0 = (sin((x * (5e-1)))) IN
	((((8) / (3)) * t_0) * t_0) / (sin(x))
END code

Alternative 1: 99.8% accurate, 2.7× speedup

Math

\[\frac{\tan \left(-0.5 \cdot x\right)}{-0.75} \]

FPCore

(FPCore (x)
  :precision binary64
  :pre TRUE
  (/ (tan (* -0.5 x)) -0.75))

C

double code(double x) {
	return tan((-0.5 * x)) / -0.75;
}

Fortran

real(8) function code(x)
use fmin_fmax_functions
    real(8), intent (in) :: x
    code = tan(((-0.5d0) * x)) / (-0.75d0)
end function

Java

public static double code(double x) {
	return Math.tan((-0.5 * x)) / -0.75;
}

Python

def code(x):
	return math.tan((-0.5 * x)) / -0.75

Julia

function code(x)
	return Float64(tan(Float64(-0.5 * x)) / -0.75)
end

MATLAB

function tmp = code(x)
	tmp = tan((-0.5 * x)) / -0.75;
end

Wolfram

code[x_] := N[(N[Tan[N[(-0.5 * x), $MachinePrecision]], $MachinePrecision] / -0.75), $MachinePrecision]

ReFlow

f(x):
	x in [-inf, +inf]
code: THEORY
BEGIN
f(x: real): real =
	(tan(((-5e-1) * x))) / (-75e-2)
END code

Derivation

1. Initial program 77.2%

   \[\frac{\left(\frac{8}{3} \cdot \sin \left(x \cdot 0.5\right)\right) \cdot \sin \left(x \cdot 0.5\right)}{\sin x} \]
2. Step-by-step derivation

3. Applied rewrites 53.2%

   \[\leadsto \frac{1}{\frac{\sin x}{\left(1 - \cos x\right) \cdot 1.3333333333333333}} \]
4. Step-by-step derivation

5. Applied rewrites 53.3%

   \[\leadsto \frac{1}{\sin x \cdot \frac{0.75}{1 - \cos x}} \]

7. Applied rewrites 99.3%

   \[\leadsto \frac{1.3333333333333333}{\frac{1}{\tan \left(0.5 \cdot x\right)}} \]
8. Step-by-step derivation

9. Applied rewrites 99.8%

   \[\leadsto \frac{\tan \left(-0.5 \cdot x\right)}{-0.75} \]
10. Add Preprocessing

Alternative 2: 99.4% accurate, 2.7× speedup

Math

\[\tan \left(0.5 \cdot x\right) \cdot 1.3333333333333333 \]

FPCore

(FPCore (x)
  :precision binary64
  :pre TRUE
  (* (tan (* 0.5 x)) 1.3333333333333333))

C

double code(double x) {
	return tan((0.5 * x)) * 1.3333333333333333;
}

Fortran

real(8) function code(x)
use fmin_fmax_functions
    real(8), intent (in) :: x
    code = tan((0.5d0 * x)) * 1.3333333333333333d0
end function

Java

public static double code(double x) {
	return Math.tan((0.5 * x)) * 1.3333333333333333;
}

Python

def code(x):
	return math.tan((0.5 * x)) * 1.3333333333333333

Julia

function code(x)
	return Float64(tan(Float64(0.5 * x)) * 1.3333333333333333)
end

MATLAB

function tmp = code(x)
	tmp = tan((0.5 * x)) * 1.3333333333333333;
end

Wolfram

code[x_] := N[(N[Tan[N[(0.5 * x), $MachinePrecision]], $MachinePrecision] * 1.3333333333333333), $MachinePrecision]

ReFlow

f(x):
	x in [-inf, +inf]
code: THEORY
BEGIN
f(x: real): real =
	(tan(((5e-1) * x))) * (13333333333333332593184650249895639717578887939453125e-52)
END code

Derivation

1. Initial program 77.2%

   \[\frac{\left(\frac{8}{3} \cdot \sin \left(x \cdot 0.5\right)\right) \cdot \sin \left(x \cdot 0.5\right)}{\sin x} \]
2. Step-by-step derivation

3. Applied rewrites 53.3%

   \[\leadsto \frac{\left(1 - \cos x\right) \cdot 1.3333333333333333}{\sin x} \]
4. Step-by-step derivation

5. Applied rewrites 53.2%

   \[\leadsto \frac{1}{\frac{\sin x}{1 - \cos x}} \cdot 1.3333333333333333 \]
6. Step-by-step derivation

7. Applied rewrites 99.4%

   \[\leadsto \tan \left(0.5 \cdot x\right) \cdot 1.3333333333333333 \]
8. Add Preprocessing

Alternative 3: 58.7% accurate, 2.9× speedup

Math

\[\frac{\sin x}{2} \cdot 1.3333333333333333 \]

FPCore

(FPCore (x)
  :precision binary64
  :pre TRUE
  (* (/ (sin x) 2.0) 1.3333333333333333))

C

double code(double x) {
	return (sin(x) / 2.0) * 1.3333333333333333;
}

Fortran

real(8) function code(x)
use fmin_fmax_functions
    real(8), intent (in) :: x
    code = (sin(x) / 2.0d0) * 1.3333333333333333d0
end function

Java

public static double code(double x) {
	return (Math.sin(x) / 2.0) * 1.3333333333333333;
}

Python

def code(x):
	return (math.sin(x) / 2.0) * 1.3333333333333333

Julia

function code(x)
	return Float64(Float64(sin(x) / 2.0) * 1.3333333333333333)
end

MATLAB

function tmp = code(x)
	tmp = (sin(x) / 2.0) * 1.3333333333333333;
end

Wolfram

code[x_] := N[(N[(N[Sin[x], $MachinePrecision] / 2.0), $MachinePrecision] * 1.3333333333333333), $MachinePrecision]

ReFlow

f(x):
	x in [-inf, +inf]
code: THEORY
BEGIN
f(x: real): real =
	((sin(x)) / (2)) * (13333333333333332593184650249895639717578887939453125e-52)
END code

Derivation

1. Initial program 77.2%

   \[\frac{\left(\frac{8}{3} \cdot \sin \left(x \cdot 0.5\right)\right) \cdot \sin \left(x \cdot 0.5\right)}{\sin x} \]
2. Step-by-step derivation

3. Applied rewrites 53.3%

   \[\leadsto \frac{\left(1 - \cos x\right) \cdot 1.3333333333333333}{\sin x} \]
4. Step-by-step derivation

5. Applied rewrites 53.2%

   \[\leadsto \frac{1}{\frac{\sin x}{1 - \cos x}} \cdot 1.3333333333333333 \]
6. Step-by-step derivation

7. Applied rewrites 99.0%

   \[\leadsto \frac{\sin x}{1 + \cos x} \cdot 1.3333333333333333 \]

8. Taylor expanded in x around 0

   \[\leadsto \frac{\sin x}{2} \cdot 1.3333333333333333 \]
9. Step-by-step derivation

10. Applied rewrites 58.7%

    \[\leadsto \frac{\sin x}{2} \cdot 1.3333333333333333 \]
11. Add Preprocessing

Alternative 4: 51.0% accurate, 3.3× speedup

Math

\[\frac{1}{\frac{1}{\frac{x}{\mathsf{fma}\left(x \cdot x, \mathsf{fma}\left(x \cdot x, \mathsf{fma}\left(x \cdot x, -4.96031746031746 \cdot 10^{-5}, -0.0020833333333333333\right), -0.125\right), 1.5\right)}}} \]

FPCore

(FPCore (x)
  :precision binary64
  :pre TRUE
  (/
 1.0
 (/
  1.0
  (/
   x
   (fma
    (* x x)
    (fma
     (* x x)
     (fma (* x x) -4.96031746031746e-5 -0.0020833333333333333)
     -0.125)
    1.5)))))

C

double code(double x) {
	return 1.0 / (1.0 / (x / fma((x * x), fma((x * x), fma((x * x), -4.96031746031746e-5, -0.0020833333333333333), -0.125), 1.5)));
}

Julia

function code(x)
	return Float64(1.0 / Float64(1.0 / Float64(x / fma(Float64(x * x), fma(Float64(x * x), fma(Float64(x * x), -4.96031746031746e-5, -0.0020833333333333333), -0.125), 1.5))))
end

Wolfram

code[x_] := N[(1.0 / N[(1.0 / N[(x / N[(N[(x * x), $MachinePrecision] * N[(N[(x * x), $MachinePrecision] * N[(N[(x * x), $MachinePrecision] * -4.96031746031746e-5 + -0.0020833333333333333), $MachinePrecision] + -0.125), $MachinePrecision] + 1.5), $MachinePrecision]), $MachinePrecision]), $MachinePrecision]), $MachinePrecision]

ReFlow

f(x):
	x in [-inf, +inf]
code: THEORY
BEGIN
f(x: real): real =
	(1) / ((1) / (x / (((x * x) * (((x * x) * (((x * x) * (-49603174603174603131579278869622839920339174568653106689453125e-66)) + (-20833333333333333044212754003865484264679253101348876953125e-61))) + (-125e-3))) + (15e-1))))
END code

Derivation

1. Initial program 77.2%

   \[\frac{\left(\frac{8}{3} \cdot \sin \left(x \cdot 0.5\right)\right) \cdot \sin \left(x \cdot 0.5\right)}{\sin x} \]
2. Step-by-step derivation

3. Applied rewrites 53.2%

   \[\leadsto \frac{1}{\frac{\sin x}{\left(1 - \cos x\right) \cdot 1.3333333333333333}} \]

4. Taylor expanded in x around 0

   \[\leadsto \frac{1}{\frac{\frac{3}{2} + {x}^{2} \cdot \left({x}^{2} \cdot \left(\frac{-1}{20160} \cdot {x}^{2} - \frac{1}{480}\right) - \frac{1}{8}\right)}{x}} \]
5. Step-by-step derivation

6. Applied rewrites 50.8%

   \[\leadsto \frac{1}{\frac{1.5 + {x}^{2} \cdot \left({x}^{2} \cdot \left(-4.96031746031746 \cdot 10^{-5} \cdot {x}^{2} - 0.0020833333333333333\right) - 0.125\right)}{x}} \]
7. Step-by-step derivation

8. Applied rewrites 50.9%

   \[\leadsto \frac{1}{\frac{1}{\frac{x}{\mathsf{fma}\left(x \cdot x, \mathsf{fma}\left(x \cdot x, \mathsf{fma}\left(x \cdot x, -4.96031746031746 \cdot 10^{-5}, -0.0020833333333333333\right), -0.125\right), 1.5\right)}}} \]
9. Add Preprocessing

Alternative 5: 50.9% accurate, 8.9× speedup

Math

\[\frac{1}{\mathsf{fma}\left(-0.125, x, \frac{1.5}{x}\right)} \]

FPCore

(FPCore (x)
  :precision binary64
  :pre TRUE
  (/ 1.0 (fma -0.125 x (/ 1.5 x))))

C

double code(double x) {
	return 1.0 / fma(-0.125, x, (1.5 / x));
}

Julia

function code(x)
	return Float64(1.0 / fma(-0.125, x, Float64(1.5 / x)))
end

Wolfram

code[x_] := N[(1.0 / N[(-0.125 * x + N[(1.5 / x), $MachinePrecision]), $MachinePrecision]), $MachinePrecision]

ReFlow

f(x):
	x in [-inf, +inf]
code: THEORY
BEGIN
f(x: real): real =
	(1) / (((-125e-3) * x) + ((15e-1) / x))
END code

Derivation

1. Initial program 77.2%

   \[\frac{\left(\frac{8}{3} \cdot \sin \left(x \cdot 0.5\right)\right) \cdot \sin \left(x \cdot 0.5\right)}{\sin x} \]
2. Step-by-step derivation

3. Applied rewrites 53.2%

   \[\leadsto \frac{1}{\frac{\sin x}{\left(1 - \cos x\right) \cdot 1.3333333333333333}} \]

4. Taylor expanded in x around 0

   \[\leadsto \frac{1}{\frac{\frac{3}{2} + \frac{-1}{8} \cdot {x}^{2}}{x}} \]
5. Step-by-step derivation

6. Applied rewrites 51.0%

   \[\leadsto \frac{1}{\frac{1.5 + -0.125 \cdot {x}^{2}}{x}} \]
7. Step-by-step derivation

8. Applied rewrites 51.0%

   \[\leadsto \frac{1}{\frac{\mathsf{fma}\left(x \cdot x, -0.125, 1.5\right)}{x}} \]
9. Step-by-step derivation

10. Applied rewrites 51.0%

    \[\leadsto \frac{1}{\mathsf{fma}\left(-0.125, x, \frac{1.5}{x}\right)} \]
11. Add Preprocessing

Alternative 6: 50.5% accurate, 9.8× speedup

Math

\[\frac{\frac{1}{\frac{2}{x}}}{0.75} \]

FPCore

(FPCore (x)
  :precision binary64
  :pre TRUE
  (/ (/ 1.0 (/ 2.0 x)) 0.75))

C

double code(double x) {
	return (1.0 / (2.0 / x)) / 0.75;
}

Fortran

real(8) function code(x)
use fmin_fmax_functions
    real(8), intent (in) :: x
    code = (1.0d0 / (2.0d0 / x)) / 0.75d0
end function

Java

public static double code(double x) {
	return (1.0 / (2.0 / x)) / 0.75;
}

Python

def code(x):
	return (1.0 / (2.0 / x)) / 0.75

Julia

function code(x)
	return Float64(Float64(1.0 / Float64(2.0 / x)) / 0.75)
end

MATLAB

function tmp = code(x)
	tmp = (1.0 / (2.0 / x)) / 0.75;
end

Wolfram

code[x_] := N[(N[(1.0 / N[(2.0 / x), $MachinePrecision]), $MachinePrecision] / 0.75), $MachinePrecision]

ReFlow

f(x):
	x in [-inf, +inf]
code: THEORY
BEGIN
f(x: real): real =
	((1) / ((2) / x)) / (75e-2)
END code

Derivation

1. Initial program 77.2%

   \[\frac{\left(\frac{8}{3} \cdot \sin \left(x \cdot 0.5\right)\right) \cdot \sin \left(x \cdot 0.5\right)}{\sin x} \]
2. Step-by-step derivation

3. Applied rewrites 53.3%

   \[\leadsto \frac{\left(1 - \cos x\right) \cdot 1.3333333333333333}{\sin x} \]
4. Step-by-step derivation

5. Applied rewrites 53.2%

   \[\leadsto \frac{1.3333333333333333}{\frac{\sin x}{1 - \cos x}} \]

6. Taylor expanded in x around 0

   \[\leadsto \frac{1.3333333333333333}{\frac{2}{x}} \]
7. Step-by-step derivation

8. Applied rewrites 50.3%

   \[\leadsto \frac{1.3333333333333333}{\frac{2}{x}} \]
9. Step-by-step derivation

10. Applied rewrites 50.5%

    \[\leadsto \frac{\frac{1}{\frac{2}{x}}}{0.75} \]
11. Add Preprocessing

Alternative 7: 50.5% accurate, 9.8× speedup

Math

\[\frac{1}{\frac{1}{\frac{x}{1.5}}} \]

FPCore

(FPCore (x)
  :precision binary64
  :pre TRUE
  (/ 1.0 (/ 1.0 (/ x 1.5))))

C

double code(double x) {
	return 1.0 / (1.0 / (x / 1.5));
}

Fortran

real(8) function code(x)
use fmin_fmax_functions
    real(8), intent (in) :: x
    code = 1.0d0 / (1.0d0 / (x / 1.5d0))
end function

Java

public static double code(double x) {
	return 1.0 / (1.0 / (x / 1.5));
}

Python

def code(x):
	return 1.0 / (1.0 / (x / 1.5))

Julia

function code(x)
	return Float64(1.0 / Float64(1.0 / Float64(x / 1.5)))
end

MATLAB

function tmp = code(x)
	tmp = 1.0 / (1.0 / (x / 1.5));
end

Wolfram

code[x_] := N[(1.0 / N[(1.0 / N[(x / 1.5), $MachinePrecision]), $MachinePrecision]), $MachinePrecision]

ReFlow

f(x):
	x in [-inf, +inf]
code: THEORY
BEGIN
f(x: real): real =
	(1) / ((1) / (x / (15e-1)))
END code

Derivation

1. Initial program 77.2%

   \[\frac{\left(\frac{8}{3} \cdot \sin \left(x \cdot 0.5\right)\right) \cdot \sin \left(x \cdot 0.5\right)}{\sin x} \]
2. Step-by-step derivation

3. Applied rewrites 53.2%

   \[\leadsto \frac{1}{\frac{\sin x}{\left(1 - \cos x\right) \cdot 1.3333333333333333}} \]

4. Taylor expanded in x around 0

   \[\leadsto \frac{1}{\frac{\frac{3}{2}}{x}} \]
5. Step-by-step derivation

6. Applied rewrites 50.4%

   \[\leadsto \frac{1}{\frac{1.5}{x}} \]
7. Step-by-step derivation

8. Applied rewrites 50.5%

   \[\leadsto \frac{1}{\frac{1}{\frac{x}{1.5}}} \]
9. Add Preprocessing

Alternative 8: 50.4% accurate, 14.1× speedup

Math

\[\frac{1}{\frac{1.5}{x}} \]

FPCore

(FPCore (x)
  :precision binary64
  :pre TRUE
  (/ 1.0 (/ 1.5 x)))

C

double code(double x) {
	return 1.0 / (1.5 / x);
}

Fortran

real(8) function code(x)
use fmin_fmax_functions
    real(8), intent (in) :: x
    code = 1.0d0 / (1.5d0 / x)
end function

Java

public static double code(double x) {
	return 1.0 / (1.5 / x);
}

Python

def code(x):
	return 1.0 / (1.5 / x)

Julia

function code(x)
	return Float64(1.0 / Float64(1.5 / x))
end

MATLAB

function tmp = code(x)
	tmp = 1.0 / (1.5 / x);
end

Wolfram

code[x_] := N[(1.0 / N[(1.5 / x), $MachinePrecision]), $MachinePrecision]

ReFlow

f(x):
	x in [-inf, +inf]
code: THEORY
BEGIN
f(x: real): real =
	(1) / ((15e-1) / x)
END code

Derivation

1. Initial program 77.2%

   \[\frac{\left(\frac{8}{3} \cdot \sin \left(x \cdot 0.5\right)\right) \cdot \sin \left(x \cdot 0.5\right)}{\sin x} \]
2. Step-by-step derivation

3. Applied rewrites 53.2%

   \[\leadsto \frac{1}{\frac{\sin x}{\left(1 - \cos x\right) \cdot 1.3333333333333333}} \]

4. Taylor expanded in x around 0

   \[\leadsto \frac{1}{\frac{\frac{3}{2}}{x}} \]
5. Step-by-step derivation

6. Applied rewrites 50.4%

   \[\leadsto \frac{1}{\frac{1.5}{x}} \]
7. Add Preprocessing

Alternative 9: 50.4% accurate, 30.3× speedup

Math

\[0.6666666666666666 \cdot x \]

FPCore

(FPCore (x)
  :precision binary64
  :pre TRUE
  (* 0.6666666666666666 x))

C

double code(double x) {
	return 0.6666666666666666 * x;
}

Fortran

real(8) function code(x)
use fmin_fmax_functions
    real(8), intent (in) :: x
    code = 0.6666666666666666d0 * x
end function

Java

public static double code(double x) {
	return 0.6666666666666666 * x;
}

Python

def code(x):
	return 0.6666666666666666 * x

Julia

function code(x)
	return Float64(0.6666666666666666 * x)
end

MATLAB

function tmp = code(x)
	tmp = 0.6666666666666666 * x;
end

Wolfram

code[x_] := N[(0.6666666666666666 * x), $MachinePrecision]

ReFlow

f(x):
	x in [-inf, +inf]
code: THEORY
BEGIN
f(x: real): real =
	(66666666666666662965923251249478198587894439697265625e-53) * x
END code

Derivation

1. Initial program 77.2%

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

2. Taylor expanded in x around 0

   \[\leadsto \frac{2}{3} \cdot x \]
3. Step-by-step derivation

4. Applied rewrites 50.4%

   \[\leadsto 0.6666666666666666 \cdot x \]
5. Add Preprocessing

Reproduce

herbie shell --seed 2026092 
(FPCore (x)
  :name "Graphics.Rasterific.Svg.PathConverter:segmentToBezier from rasterific-svg-0.2.3.1, A"
  :precision binary64
  (/ (* (* (/ 8.0 3.0) (sin (* x 0.5))) (sin (* x 0.5))) (sin x)))