Diagrams.TwoD.Segment.Bernstein:evaluateBernstein from diagrams-lib-1.3.0.3

Average Error: 10.3 → 0.5

Time: 2.8s

Precision: binary64

Math

\[\frac{x \cdot \left(\left(y - z\right) + 1\right)}{z} \]

↓

\[\begin{array}{l} t_0 := \frac{x \cdot \left(\left(y - z\right) + 1\right)}{z}\\ \mathbf{if}\;t_0 \leq -0.0033332639133751216:\\ \;\;\;\;\left(\frac{y}{\frac{z}{x}} + \frac{x}{z}\right) - x\\ \mathbf{elif}\;t_0 \leq 1.7476870974454313 \cdot 10^{-157}:\\ \;\;\;\;\mathsf{fma}\left(\frac{y}{z}, x, \frac{x}{z}\right) - x\\ \mathbf{else}:\\ \;\;\;\;\left(\frac{x}{z} + y \cdot \frac{x}{z}\right) - x\\ \end{array} \]

FPCore

(FPCore (x y z) :precision binary64 (/ (* x (+ (- y z) 1.0)) z))

↓

(FPCore (x y z)
 :precision binary64
 (let* ((t_0 (/ (* x (+ (- y z) 1.0)) z)))
   (if (<= t_0 -0.0033332639133751216)
     (- (+ (/ y (/ z x)) (/ x z)) x)
     (if (<= t_0 1.7476870974454313e-157)
       (- (fma (/ y z) x (/ x z)) x)
       (- (+ (/ x z) (* y (/ x z))) x)))))

C

double code(double x, double y, double z) {
	return (x * ((y - z) + 1.0)) / z;
}

↓

double code(double x, double y, double z) {
	double t_0 = (x * ((y - z) + 1.0)) / z;
	double tmp;
	if (t_0 <= -0.0033332639133751216) {
		tmp = ((y / (z / x)) + (x / z)) - x;
	} else if (t_0 <= 1.7476870974454313e-157) {
		tmp = fma((y / z), x, (x / z)) - x;
	} else {
		tmp = ((x / z) + (y * (x / z))) - x;
	}
	return tmp;
}

Julia

function code(x, y, z)
	return Float64(Float64(x * Float64(Float64(y - z) + 1.0)) / z)
end

↓

function code(x, y, z)
	t_0 = Float64(Float64(x * Float64(Float64(y - z) + 1.0)) / z)
	tmp = 0.0
	if (t_0 <= -0.0033332639133751216)
		tmp = Float64(Float64(Float64(y / Float64(z / x)) + Float64(x / z)) - x);
	elseif (t_0 <= 1.7476870974454313e-157)
		tmp = Float64(fma(Float64(y / z), x, Float64(x / z)) - x);
	else
		tmp = Float64(Float64(Float64(x / z) + Float64(y * Float64(x / z))) - x);
	end
	return tmp
end

Wolfram

code[x_, y_, z_] := N[(N[(x * N[(N[(y - z), $MachinePrecision] + 1.0), $MachinePrecision]), $MachinePrecision] / z), $MachinePrecision]

↓

code[x_, y_, z_] := Block[{t$95$0 = N[(N[(x * N[(N[(y - z), $MachinePrecision] + 1.0), $MachinePrecision]), $MachinePrecision] / z), $MachinePrecision]}, If[LessEqual[t$95$0, -0.0033332639133751216], N[(N[(N[(y / N[(z / x), $MachinePrecision]), $MachinePrecision] + N[(x / z), $MachinePrecision]), $MachinePrecision] - x), $MachinePrecision], If[LessEqual[t$95$0, 1.7476870974454313e-157], N[(N[(N[(y / z), $MachinePrecision] * x + N[(x / z), $MachinePrecision]), $MachinePrecision] - x), $MachinePrecision], N[(N[(N[(x / z), $MachinePrecision] + N[(y * N[(x / z), $MachinePrecision]), $MachinePrecision]), $MachinePrecision] - x), $MachinePrecision]]]]

Error

Bits error versus x

Bits error versus y

Bits error versus z

Target

Original	10.3
Target	0.5
Herbie	0.5

\[\begin{array}{l} \mathbf{if}\;x < -2.71483106713436 \cdot 10^{-162}:\\ \;\;\;\;\left(1 + y\right) \cdot \frac{x}{z} - x\\ \mathbf{elif}\;x < 3.874108816439546 \cdot 10^{-197}:\\ \;\;\;\;\left(x \cdot \left(\left(y - z\right) + 1\right)\right) \cdot \frac{1}{z}\\ \mathbf{else}:\\ \;\;\;\;\left(1 + y\right) \cdot \frac{x}{z} - x\\ \end{array} \]

Derivation

1. Split input into 3 regimes

2. if (/.f64 (*.f64 x (+.f64 (-.f64 y z) 1)) z) < -0.0033332639133751216

   1. Initial program 16.3

      \[\frac{x \cdot \left(\left(y - z\right) + 1\right)}{z} \]

   2. Simplified 5.3

      \[\leadsto \color{blue}{\frac{\mathsf{fma}\left(x, y, x\right)}{z} - x} \]

   3. Taylor expanded in y around 0 5.3

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

   4. Applied egg-rr 0.1

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

   5. Applied egg-rr 0.1

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

   if -0.0033332639133751216 < (/.f64 (*.f64 x (+.f64 (-.f64 y z) 1)) z) < 1.74768709744543126e-157

   1. Initial program 0.2

      \[\frac{x \cdot \left(\left(y - z\right) + 1\right)}{z} \]

   2. Simplified 0.1

      \[\leadsto \color{blue}{\frac{\mathsf{fma}\left(x, y, x\right)}{z} - x} \]

   3. Taylor expanded in y around 0 0.1

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

   4. Applied egg-rr 0.0

      \[\leadsto \color{blue}{\mathsf{fma}\left(\frac{y}{z}, x, \frac{x}{z}\right)} - x \]

   if 1.74768709744543126e-157 < (/.f64 (*.f64 x (+.f64 (-.f64 y z) 1)) z)

   1. Initial program 12.5

      \[\frac{x \cdot \left(\left(y - z\right) + 1\right)}{z} \]

   2. Simplified 4.4

      \[\leadsto \color{blue}{\frac{\mathsf{fma}\left(x, y, x\right)}{z} - x} \]

   3. Taylor expanded in y around 0 4.4

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

   4. Applied egg-rr 1.1

      \[\leadsto \left(\color{blue}{y \cdot \frac{x}{z}} + \frac{x}{z}\right) - x \]
3. Recombined 3 regimes into one program.

4. Final simplification 0.5

   \[\leadsto \begin{array}{l} \mathbf{if}\;\frac{x \cdot \left(\left(y - z\right) + 1\right)}{z} \leq -0.0033332639133751216:\\ \;\;\;\;\left(\frac{y}{\frac{z}{x}} + \frac{x}{z}\right) - x\\ \mathbf{elif}\;\frac{x \cdot \left(\left(y - z\right) + 1\right)}{z} \leq 1.7476870974454313 \cdot 10^{-157}:\\ \;\;\;\;\mathsf{fma}\left(\frac{y}{z}, x, \frac{x}{z}\right) - x\\ \mathbf{else}:\\ \;\;\;\;\left(\frac{x}{z} + y \cdot \frac{x}{z}\right) - x\\ \end{array} \]

Reproduce

herbie shell --seed 2022151 
(FPCore (x y z)
  :name "Diagrams.TwoD.Segment.Bernstein:evaluateBernstein from diagrams-lib-1.3.0.3"
  :precision binary64

  :herbie-target
  (if (< x -2.71483106713436e-162) (- (* (+ 1.0 y) (/ x z)) x) (if (< x 3.874108816439546e-197) (* (* x (+ (- y z) 1.0)) (/ 1.0 z)) (- (* (+ 1.0 y) (/ x z)) x)))

  (/ (* x (+ (- y z) 1.0)) z))