Disney BSSRDF, sample scattering profile, lower

Average Accuracy: 61.1% → 99.4%

Time: 12.1s

Precision: binary32

Cost: 3392

\[\left(0 \leq s \land s \leq 256\right) \land \left(2.328306437 \cdot 10^{-10} \leq u \land u \leq 0.25\right)\]

Math

\[s \cdot \log \left(\frac{1}{1 - 4 \cdot u}\right) \]

↓

\[\mathsf{log1p}\left(u \cdot -4\right) \cdot \left(-s\right) \]

FPCore

(FPCore (s u) :precision binary32 (* s (log (/ 1.0 (- 1.0 (* 4.0 u))))))

↓

(FPCore (s u) :precision binary32 (* (log1p (* u -4.0)) (- s)))

C

float code(float s, float u) {
	return s * logf((1.0f / (1.0f - (4.0f * u))));
}

↓

float code(float s, float u) {
	return log1pf((u * -4.0f)) * -s;
}

Julia

function code(s, u)
	return Float32(s * log(Float32(Float32(1.0) / Float32(Float32(1.0) - Float32(Float32(4.0) * u)))))
end

↓

function code(s, u)
	return Float32(log1p(Float32(u * Float32(-4.0))) * Float32(-s))
end

Derivation

1. Initial program 58.5%

   \[s \cdot \log \left(\frac{1}{1 - 4 \cdot u}\right) \]

2. Simplified 99.5%

   \[\leadsto \color{blue}{\mathsf{log1p}\left(u \cdot -4\right) \cdot \left(-s\right)} \]
   Step-by-step derivation

   [Start] 58.5	\[ s \cdot \log \left(\frac{1}{1 - 4 \cdot u}\right) \]
   *-commutative [=>] 58.5	\[ \color{blue}{\log \left(\frac{1}{1 - 4 \cdot u}\right) \cdot s} \]
   log-rec [=>] 61.0	\[ \color{blue}{\left(-\log \left(1 - 4 \cdot u\right)\right)} \cdot s \]
   distribute-lft-neg-out [=>] 61.0	\[ \color{blue}{-\log \left(1 - 4 \cdot u\right) \cdot s} \]
   distribute-rgt-neg-in [=>] 61.0	\[ \color{blue}{\log \left(1 - 4 \cdot u\right) \cdot \left(-s\right)} \]
   sub-neg [=>] 61.0	\[ \log \color{blue}{\left(1 + \left(-4 \cdot u\right)\right)} \cdot \left(-s\right) \]
   log1p-def [=>] 99.5	\[ \color{blue}{\mathsf{log1p}\left(-4 \cdot u\right)} \cdot \left(-s\right) \]
   *-commutative [=>] 99.5	\[ \mathsf{log1p}\left(-\color{blue}{u \cdot 4}\right) \cdot \left(-s\right) \]
   distribute-rgt-neg-in [=>] 99.5	\[ \mathsf{log1p}\left(\color{blue}{u \cdot \left(-4\right)}\right) \cdot \left(-s\right) \]
   metadata-eval [=>] 99.5	\[ \mathsf{log1p}\left(u \cdot \color{blue}{-4}\right) \cdot \left(-s\right) \]

3. Final simplification 99.5%

   \[\leadsto \mathsf{log1p}\left(u \cdot -4\right) \cdot \left(-s\right) \]

Alternatives

Alternative 1
Accuracy	86.9%
Cost	352

\[s \cdot \left(u \cdot 4 + 8 \cdot \left(u \cdot u\right)\right) \]

Alternative 2
Accuracy	86.7%
Cost	288

\[s \cdot \left(u \cdot \left(4 + u \cdot 8\right)\right) \]

Alternative 3
Accuracy	73.8%
Cost	160

\[4 \cdot \left(u \cdot s\right) \]

Alternative 4
Accuracy	74.0%
Cost	160

\[s \cdot \left(u \cdot 4\right) \]

Reproduce

herbie shell --seed 2023157 
(FPCore (s u)
  :name "Disney BSSRDF, sample scattering profile, lower"
  :precision binary32
  :pre (and (and (<= 0.0 s) (<= s 256.0)) (and (<= 2.328306437e-10 u) (<= u 0.25)))
  (* s (log (/ 1.0 (- 1.0 (* 4.0 u))))))