symmetry log of sum of exp

Average Accuracy: 54.1% → 98.4%

Time: 15.4s

Precision: binary64

Cost: 19648

• Unsound rule application detected in e-graph. Results may not be sound.

• could not determine a ground truth

  1. a = 4.4547743634403135e+30
  2. b = -1.736456706245406e-189

\[ \begin{array}{c}[a, b] = \mathsf{sort}([a, b])\\ \end{array} \]

Math

\[\log \left(e^{a} + e^{b}\right) \]

↓

\[\mathsf{log1p}\left(e^{a}\right) + \frac{b}{e^{a} + 1} \]

FPCore

(FPCore (a b) :precision binary64 (log (+ (exp a) (exp b))))

↓

(FPCore (a b) :precision binary64 (+ (log1p (exp a)) (/ b (+ (exp a) 1.0))))

C

double code(double a, double b) {
	return log((exp(a) + exp(b)));
}

↓

double code(double a, double b) {
	return log1p(exp(a)) + (b / (exp(a) + 1.0));
}

Java

public static double code(double a, double b) {
	return Math.log((Math.exp(a) + Math.exp(b)));
}

↓

public static double code(double a, double b) {
	return Math.log1p(Math.exp(a)) + (b / (Math.exp(a) + 1.0));
}

Python

def code(a, b):
	return math.log((math.exp(a) + math.exp(b)))

↓

def code(a, b):
	return math.log1p(math.exp(a)) + (b / (math.exp(a) + 1.0))

Julia

function code(a, b)
	return log(Float64(exp(a) + exp(b)))
end

↓

function code(a, b)
	return Float64(log1p(exp(a)) + Float64(b / Float64(exp(a) + 1.0)))
end

Wolfram

code[a_, b_] := N[Log[N[(N[Exp[a], $MachinePrecision] + N[Exp[b], $MachinePrecision]), $MachinePrecision]], $MachinePrecision]

↓

code[a_, b_] := N[(N[Log[1 + N[Exp[a], $MachinePrecision]], $MachinePrecision] + N[(b / N[(N[Exp[a], $MachinePrecision] + 1.0), $MachinePrecision]), $MachinePrecision]), $MachinePrecision]

Derivation

1. Initial program 54.1%

   \[\log \left(e^{a} + e^{b}\right) \]

2. Taylor expanded in b around 0 73.8%

   \[\leadsto \color{blue}{\log \left(1 + e^{a}\right) + \frac{b}{1 + e^{a}}} \]

3. Simplified 74.2%

   \[\leadsto \color{blue}{\mathsf{log1p}\left(e^{a}\right) + \frac{b}{1 + e^{a}}} \]
   Step-by-step derivation

   [Start] 73.8	\[ \log \left(1 + e^{a}\right) + \frac{b}{1 + e^{a}} \]
   log1p-def [=>] 74.2	\[ \color{blue}{\mathsf{log1p}\left(e^{a}\right)} + \frac{b}{1 + e^{a}} \]

4. Final simplification 74.2%

   \[\leadsto \mathsf{log1p}\left(e^{a}\right) + \frac{b}{e^{a} + 1} \]

Alternatives

Alternative 1
Accuracy	96.7%
Cost	25928

\[\begin{array}{l} \mathbf{if}\;e^{b} \leq 1.005:\\ \;\;\;\;\mathsf{log1p}\left(e^{a} + b\right)\\ \mathbf{elif}\;e^{b} \leq 5 \cdot 10^{+219}:\\ \;\;\;\;\mathsf{log1p}\left(e^{b}\right)\\ \mathbf{else}:\\ \;\;\;\;b \cdot 0.5 + \log 2\\ \end{array} \]

Alternative 2
Accuracy	97.1%
Cost	19524

\[\begin{array}{l} \mathbf{if}\;b \leq 1.15 \cdot 10^{-8}:\\ \;\;\;\;\mathsf{log1p}\left(e^{a} + b\right)\\ \mathbf{else}:\\ \;\;\;\;\log \left(e^{a} + e^{b}\right)\\ \end{array} \]

Alternative 3
Accuracy	57.8%
Cost	19396

\[\begin{array}{l} \mathbf{if}\;e^{a} \leq 0:\\ \;\;\;\;b \cdot 0.5\\ \mathbf{else}:\\ \;\;\;\;\mathsf{log1p}\left(e^{a}\right)\\ \end{array} \]

Alternative 4
Accuracy	57.6%
Cost	19396

\[\begin{array}{l} \mathbf{if}\;e^{a} \leq 0:\\ \;\;\;\;b \cdot 0.5\\ \mathbf{else}:\\ \;\;\;\;\mathsf{log1p}\left(e^{b}\right)\\ \end{array} \]

Alternative 5
Accuracy	98.1%
Cost	19392

\[\mathsf{log1p}\left(e^{a} + \mathsf{expm1}\left(b\right)\right) \]

Alternative 6
Accuracy	11.9%
Cost	7108

\[\begin{array}{l} \mathbf{if}\;a \leq 1.4:\\ \;\;\;\;b \cdot 0.5\\ \mathbf{else}:\\ \;\;\;\;\frac{a}{b + 2} + \log \left(b + 2\right)\\ \end{array} \]

Alternative 7
Accuracy	11.9%
Cost	6852

\[\begin{array}{l} \mathbf{if}\;a \leq 1:\\ \;\;\;\;b \cdot 0.5\\ \mathbf{else}:\\ \;\;\;\;\log \left(b + \left(a + 2\right)\right)\\ \end{array} \]

Alternative 8
Accuracy	11.9%
Cost	6724

\[\begin{array}{l} \mathbf{if}\;a \leq 86:\\ \;\;\;\;b \cdot 0.5\\ \mathbf{else}:\\ \;\;\;\;\log \left(b + 2\right)\\ \end{array} \]

Alternative 9
Accuracy	11.9%
Cost	6596

\[\begin{array}{l} \mathbf{if}\;a \leq 105:\\ \;\;\;\;b \cdot 0.5\\ \mathbf{else}:\\ \;\;\;\;\log 2\\ \end{array} \]

Alternative 10
Accuracy	11.9%
Cost	192

\[b \cdot 0.5 \]

Reproduce

herbie shell --seed 2023159 
(FPCore (a b)
  :name "symmetry log of sum of exp"
  :precision binary64
  (log (+ (exp a) (exp b))))