Numeric.SpecFunctions:invErfc from math-functions-0.1.5.2, B

Average Error: 0.1 → 0.1

Time: 7.9s

Precision: binary64

Cost: 14144

Math

\[0.70711 \cdot \left(\frac{2.30753 + x \cdot 0.27061}{1 + x \cdot \left(0.99229 + x \cdot 0.04481\right)} - x\right) \]

↓

\[-1 + \left(x \cdot -0.70711 + e^{\mathsf{log1p}\left(\frac{-1.6316775383 + x \cdot -0.1913510371}{-1 - x \cdot \left(x \cdot 0.04481 + 0.99229\right)}\right)}\right) \]

FPCore

(FPCore (x)
 :precision binary64
 (*
  0.70711
  (- (/ (+ 2.30753 (* x 0.27061)) (+ 1.0 (* x (+ 0.99229 (* x 0.04481))))) x)))

↓

(FPCore (x)
 :precision binary64
 (+
  -1.0
  (+
   (* x -0.70711)
   (exp
    (log1p
     (/
      (+ -1.6316775383 (* x -0.1913510371))
      (- -1.0 (* x (+ (* x 0.04481) 0.99229)))))))))

C

double code(double x) {
	return 0.70711 * (((2.30753 + (x * 0.27061)) / (1.0 + (x * (0.99229 + (x * 0.04481))))) - x);
}

↓

double code(double x) {
	return -1.0 + ((x * -0.70711) + exp(log1p(((-1.6316775383 + (x * -0.1913510371)) / (-1.0 - (x * ((x * 0.04481) + 0.99229)))))));
}

Java

public static double code(double x) {
	return 0.70711 * (((2.30753 + (x * 0.27061)) / (1.0 + (x * (0.99229 + (x * 0.04481))))) - x);
}

↓

public static double code(double x) {
	return -1.0 + ((x * -0.70711) + Math.exp(Math.log1p(((-1.6316775383 + (x * -0.1913510371)) / (-1.0 - (x * ((x * 0.04481) + 0.99229)))))));
}

Python

def code(x):
	return 0.70711 * (((2.30753 + (x * 0.27061)) / (1.0 + (x * (0.99229 + (x * 0.04481))))) - x)

↓

def code(x):
	return -1.0 + ((x * -0.70711) + math.exp(math.log1p(((-1.6316775383 + (x * -0.1913510371)) / (-1.0 - (x * ((x * 0.04481) + 0.99229)))))))

Julia

function code(x)
	return Float64(0.70711 * Float64(Float64(Float64(2.30753 + Float64(x * 0.27061)) / Float64(1.0 + Float64(x * Float64(0.99229 + Float64(x * 0.04481))))) - x))
end

↓

function code(x)
	return Float64(-1.0 + Float64(Float64(x * -0.70711) + exp(log1p(Float64(Float64(-1.6316775383 + Float64(x * -0.1913510371)) / Float64(-1.0 - Float64(x * Float64(Float64(x * 0.04481) + 0.99229))))))))
end

Wolfram

code[x_] := N[(0.70711 * N[(N[(N[(2.30753 + N[(x * 0.27061), $MachinePrecision]), $MachinePrecision] / N[(1.0 + N[(x * N[(0.99229 + N[(x * 0.04481), $MachinePrecision]), $MachinePrecision]), $MachinePrecision]), $MachinePrecision]), $MachinePrecision] - x), $MachinePrecision]), $MachinePrecision]

↓

code[x_] := N[(-1.0 + N[(N[(x * -0.70711), $MachinePrecision] + N[Exp[N[Log[1 + N[(N[(-1.6316775383 + N[(x * -0.1913510371), $MachinePrecision]), $MachinePrecision] / N[(-1.0 - N[(x * N[(N[(x * 0.04481), $MachinePrecision] + 0.99229), $MachinePrecision]), $MachinePrecision]), $MachinePrecision]), $MachinePrecision]], $MachinePrecision]], $MachinePrecision]), $MachinePrecision]), $MachinePrecision]

Derivation

1. Initial program 0.1

   \[0.70711 \cdot \left(\frac{2.30753 + x \cdot 0.27061}{1 + x \cdot \left(0.99229 + x \cdot 0.04481\right)} - x\right) \]

2. Simplified 0.1

   \[\leadsto \color{blue}{\mathsf{fma}\left(x, -0.70711, \frac{\mathsf{fma}\left(x, 0.1913510371, 1.6316775383\right)}{\mathsf{fma}\left(x, \mathsf{fma}\left(x, 0.04481, 0.99229\right), 1\right)}\right)} \]
   Proof

   [Start] 0.1	\[ 0.70711 \cdot \left(\frac{2.30753 + x \cdot 0.27061}{1 + x \cdot \left(0.99229 + x \cdot 0.04481\right)} - x\right) \]
   sub-neg [=>] 0.1	\[ 0.70711 \cdot \color{blue}{\left(\frac{2.30753 + x \cdot 0.27061}{1 + x \cdot \left(0.99229 + x \cdot 0.04481\right)} + \left(-x\right)\right)} \]
   distribute-lft-in [=>] 0.1	\[ \color{blue}{0.70711 \cdot \frac{2.30753 + x \cdot 0.27061}{1 + x \cdot \left(0.99229 + x \cdot 0.04481\right)} + 0.70711 \cdot \left(-x\right)} \]
   +-commutative [=>] 0.1	\[ \color{blue}{0.70711 \cdot \left(-x\right) + 0.70711 \cdot \frac{2.30753 + x \cdot 0.27061}{1 + x \cdot \left(0.99229 + x \cdot 0.04481\right)}} \]
   neg-mul-1 [=>] 0.1	\[ 0.70711 \cdot \color{blue}{\left(-1 \cdot x\right)} + 0.70711 \cdot \frac{2.30753 + x \cdot 0.27061}{1 + x \cdot \left(0.99229 + x \cdot 0.04481\right)} \]
   associate-*r* [=>] 0.1	\[ \color{blue}{\left(0.70711 \cdot -1\right) \cdot x} + 0.70711 \cdot \frac{2.30753 + x \cdot 0.27061}{1 + x \cdot \left(0.99229 + x \cdot 0.04481\right)} \]
   *-commutative [=>] 0.1	\[ \color{blue}{x \cdot \left(0.70711 \cdot -1\right)} + 0.70711 \cdot \frac{2.30753 + x \cdot 0.27061}{1 + x \cdot \left(0.99229 + x \cdot 0.04481\right)} \]
   *-commutative [<=] 0.1	\[ x \cdot \left(0.70711 \cdot -1\right) + \color{blue}{\frac{2.30753 + x \cdot 0.27061}{1 + x \cdot \left(0.99229 + x \cdot 0.04481\right)} \cdot 0.70711} \]
   fma-def [=>] 0.1	\[ \color{blue}{\mathsf{fma}\left(x, 0.70711 \cdot -1, \frac{2.30753 + x \cdot 0.27061}{1 + x \cdot \left(0.99229 + x \cdot 0.04481\right)} \cdot 0.70711\right)} \]
   metadata-eval [=>] 0.1	\[ \mathsf{fma}\left(x, \color{blue}{-0.70711}, \frac{2.30753 + x \cdot 0.27061}{1 + x \cdot \left(0.99229 + x \cdot 0.04481\right)} \cdot 0.70711\right) \]
   associate-*l/ [=>] 0.1	\[ \mathsf{fma}\left(x, -0.70711, \color{blue}{\frac{\left(2.30753 + x \cdot 0.27061\right) \cdot 0.70711}{1 + x \cdot \left(0.99229 + x \cdot 0.04481\right)}}\right) \]
   *-commutative [<=] 0.1	\[ \mathsf{fma}\left(x, -0.70711, \frac{\color{blue}{0.70711 \cdot \left(2.30753 + x \cdot 0.27061\right)}}{1 + x \cdot \left(0.99229 + x \cdot 0.04481\right)}\right) \]
   +-commutative [=>] 0.1	\[ \mathsf{fma}\left(x, -0.70711, \frac{0.70711 \cdot \color{blue}{\left(x \cdot 0.27061 + 2.30753\right)}}{1 + x \cdot \left(0.99229 + x \cdot 0.04481\right)}\right) \]
   distribute-lft-in [=>] 0.1	\[ \mathsf{fma}\left(x, -0.70711, \frac{\color{blue}{0.70711 \cdot \left(x \cdot 0.27061\right) + 0.70711 \cdot 2.30753}}{1 + x \cdot \left(0.99229 + x \cdot 0.04481\right)}\right) \]
   associate-*r* [=>] 0.1	\[ \mathsf{fma}\left(x, -0.70711, \frac{\color{blue}{\left(0.70711 \cdot x\right) \cdot 0.27061} + 0.70711 \cdot 2.30753}{1 + x \cdot \left(0.99229 + x \cdot 0.04481\right)}\right) \]
   *-commutative [<=] 0.1	\[ \mathsf{fma}\left(x, -0.70711, \frac{\color{blue}{\left(x \cdot 0.70711\right)} \cdot 0.27061 + 0.70711 \cdot 2.30753}{1 + x \cdot \left(0.99229 + x \cdot 0.04481\right)}\right) \]
   associate-*l* [=>] 0.1	\[ \mathsf{fma}\left(x, -0.70711, \frac{\color{blue}{x \cdot \left(0.70711 \cdot 0.27061\right)} + 0.70711 \cdot 2.30753}{1 + x \cdot \left(0.99229 + x \cdot 0.04481\right)}\right) \]
   fma-def [=>] 0.1	\[ \mathsf{fma}\left(x, -0.70711, \frac{\color{blue}{\mathsf{fma}\left(x, 0.70711 \cdot 0.27061, 0.70711 \cdot 2.30753\right)}}{1 + x \cdot \left(0.99229 + x \cdot 0.04481\right)}\right) \]
   metadata-eval [=>] 0.1	\[ \mathsf{fma}\left(x, -0.70711, \frac{\mathsf{fma}\left(x, \color{blue}{0.1913510371}, 0.70711 \cdot 2.30753\right)}{1 + x \cdot \left(0.99229 + x \cdot 0.04481\right)}\right) \]
   metadata-eval [=>] 0.1	\[ \mathsf{fma}\left(x, -0.70711, \frac{\mathsf{fma}\left(x, 0.1913510371, \color{blue}{1.6316775383}\right)}{1 + x \cdot \left(0.99229 + x \cdot 0.04481\right)}\right) \]

3. Applied egg-rr 0.1

   \[\leadsto \color{blue}{\left(x \cdot -0.70711 + e^{\mathsf{log1p}\left(\frac{\mathsf{fma}\left(x, 0.1913510371, 1.6316775383\right)}{\mathsf{fma}\left(x, \mathsf{fma}\left(x, 0.04481, 0.99229\right), 1\right)}\right)}\right) - 1} \]

4. Applied egg-rr 0.1

   \[\leadsto \left(x \cdot -0.70711 + e^{\mathsf{log1p}\left(\color{blue}{-\frac{\mathsf{fma}\left(x, 0.1913510371, 1.6316775383\right)}{-1 - x \cdot \mathsf{fma}\left(x, 0.04481, 0.99229\right)}}\right)}\right) - 1 \]

5. Simplified 0.1

   \[\leadsto \left(x \cdot -0.70711 + e^{\mathsf{log1p}\left(\color{blue}{\frac{-1.6316775383 + x \cdot -0.1913510371}{-1 - x \cdot \mathsf{fma}\left(x, 0.04481, 0.99229\right)}}\right)}\right) - 1 \]
   Proof

   [Start] 0.1	\[ \left(x \cdot -0.70711 + e^{\mathsf{log1p}\left(-\frac{\mathsf{fma}\left(x, 0.1913510371, 1.6316775383\right)}{-1 - x \cdot \mathsf{fma}\left(x, 0.04481, 0.99229\right)}\right)}\right) - 1 \]
   distribute-neg-frac [=>] 0.1	\[ \left(x \cdot -0.70711 + e^{\mathsf{log1p}\left(\color{blue}{\frac{-\mathsf{fma}\left(x, 0.1913510371, 1.6316775383\right)}{-1 - x \cdot \mathsf{fma}\left(x, 0.04481, 0.99229\right)}}\right)}\right) - 1 \]
   fma-def [<=] 0.1	\[ \left(x \cdot -0.70711 + e^{\mathsf{log1p}\left(\frac{-\color{blue}{\left(x \cdot 0.1913510371 + 1.6316775383\right)}}{-1 - x \cdot \mathsf{fma}\left(x, 0.04481, 0.99229\right)}\right)}\right) - 1 \]
   +-commutative [=>] 0.1	\[ \left(x \cdot -0.70711 + e^{\mathsf{log1p}\left(\frac{-\color{blue}{\left(1.6316775383 + x \cdot 0.1913510371\right)}}{-1 - x \cdot \mathsf{fma}\left(x, 0.04481, 0.99229\right)}\right)}\right) - 1 \]
   distribute-neg-in [=>] 0.1	\[ \left(x \cdot -0.70711 + e^{\mathsf{log1p}\left(\frac{\color{blue}{\left(-1.6316775383\right) + \left(-x \cdot 0.1913510371\right)}}{-1 - x \cdot \mathsf{fma}\left(x, 0.04481, 0.99229\right)}\right)}\right) - 1 \]
   metadata-eval [=>] 0.1	\[ \left(x \cdot -0.70711 + e^{\mathsf{log1p}\left(\frac{\color{blue}{-1.6316775383} + \left(-x \cdot 0.1913510371\right)}{-1 - x \cdot \mathsf{fma}\left(x, 0.04481, 0.99229\right)}\right)}\right) - 1 \]
   distribute-rgt-neg-in [=>] 0.1	\[ \left(x \cdot -0.70711 + e^{\mathsf{log1p}\left(\frac{-1.6316775383 + \color{blue}{x \cdot \left(-0.1913510371\right)}}{-1 - x \cdot \mathsf{fma}\left(x, 0.04481, 0.99229\right)}\right)}\right) - 1 \]
   metadata-eval [=>] 0.1	\[ \left(x \cdot -0.70711 + e^{\mathsf{log1p}\left(\frac{-1.6316775383 + x \cdot \color{blue}{-0.1913510371}}{-1 - x \cdot \mathsf{fma}\left(x, 0.04481, 0.99229\right)}\right)}\right) - 1 \]

6. Applied egg-rr 0.1

   \[\leadsto \left(x \cdot -0.70711 + e^{\mathsf{log1p}\left(\frac{-1.6316775383 + x \cdot -0.1913510371}{-1 - x \cdot \color{blue}{\left(x \cdot 0.04481 + 0.99229\right)}}\right)}\right) - 1 \]

7. Final simplification 0.1

   \[\leadsto -1 + \left(x \cdot -0.70711 + e^{\mathsf{log1p}\left(\frac{-1.6316775383 + x \cdot -0.1913510371}{-1 - x \cdot \left(x \cdot 0.04481 + 0.99229\right)}\right)}\right) \]

Alternatives

Alternative 1
Error	0.1
Cost	7488

\[\mathsf{fma}\left(x, -0.70711, \frac{-1.6316775383 + x \cdot -0.1913510371}{-1 - x \cdot \left(x \cdot 0.04481 + 0.99229\right)}\right) \]

Alternative 2
Error	0.1
Cost	1600

\[0.70711 \cdot \left(\frac{2.30753 + x \cdot 0.27061}{1 + \left(\left(1 + x \cdot 0.99229\right) + \left(-1 + x \cdot \left(x \cdot 0.04481\right)\right)\right)} - x\right) \]

Alternative 3
Error	0.1
Cost	1216

\[0.70711 \cdot \left(\frac{2.30753 + x \cdot 0.27061}{1 + x \cdot \left(x \cdot 0.04481 + 0.99229\right)} - x\right) \]

Alternative 4
Error	1.0
Cost	960

\[0.70711 \cdot \left(\frac{2.30753 + x \cdot 0.27061}{1 + x \cdot 0.99229} - x\right) \]

Alternative 5
Error	1.2
Cost	456

\[\begin{array}{l} \mathbf{if}\;x \leq -3.4:\\ \;\;\;\;x \cdot -0.70711\\ \mathbf{elif}\;x \leq 1.2:\\ \;\;\;\;1.6316775383\\ \mathbf{else}:\\ \;\;\;\;x \cdot -0.70711\\ \end{array} \]

Alternative 6
Error	1.5
Cost	320

\[0.70711 \cdot \left(2.30753 - x\right) \]

Alternative 7
Error	31.8
Cost	64

\[1.6316775383 \]

Reproduce

herbie shell --seed 2023018 
(FPCore (x)
  :name "Numeric.SpecFunctions:invErfc from math-functions-0.1.5.2, B"
  :precision binary64
  (* 0.70711 (- (/ (+ 2.30753 (* x 0.27061)) (+ 1.0 (* x (+ 0.99229 (* x 0.04481))))) x)))