sintan (problem 3.4.5)

Average Accuracy: 49.5% → 100.0%

Time: 22.1s

Precision: binary64

Cost: 20488

Math

\[\frac{x - \sin x}{x - \tan x} \]

↓

\[\begin{array}{l} t_0 := \tan x - x\\ \mathbf{if}\;x \leq -0.098:\\ \;\;\;\;\frac{x - \sin x}{x - \tan x}\\ \mathbf{elif}\;x \leq 0.1:\\ \;\;\;\;\left(0.225 \cdot {x}^{2} + \left(-0.009642857142857142 \cdot {x}^{4} + 0.00024107142857142857 \cdot {x}^{6}\right)\right) + -0.5\\ \mathbf{else}:\\ \;\;\;\;\frac{\sin x}{t_0} - \frac{x}{t_0}\\ \end{array} \]

FPCore

(FPCore (x) :precision binary64 (/ (- x (sin x)) (- x (tan x))))

↓

(FPCore (x)
 :precision binary64
 (let* ((t_0 (- (tan x) x)))
   (if (<= x -0.098)
     (/ (- x (sin x)) (- x (tan x)))
     (if (<= x 0.1)
       (+
        (+
         (* 0.225 (pow x 2.0))
         (+
          (* -0.009642857142857142 (pow x 4.0))
          (* 0.00024107142857142857 (pow x 6.0))))
        -0.5)
       (- (/ (sin x) t_0) (/ x t_0))))))

C

double code(double x) {
	return (x - sin(x)) / (x - tan(x));
}

↓

double code(double x) {
	double t_0 = tan(x) - x;
	double tmp;
	if (x <= -0.098) {
		tmp = (x - sin(x)) / (x - tan(x));
	} else if (x <= 0.1) {
		tmp = ((0.225 * pow(x, 2.0)) + ((-0.009642857142857142 * pow(x, 4.0)) + (0.00024107142857142857 * pow(x, 6.0)))) + -0.5;
	} else {
		tmp = (sin(x) / t_0) - (x / t_0);
	}
	return tmp;
}

Fortran

real(8) function code(x)
    real(8), intent (in) :: x
    code = (x - sin(x)) / (x - tan(x))
end function

↓

real(8) function code(x)
    real(8), intent (in) :: x
    real(8) :: t_0
    real(8) :: tmp
    t_0 = tan(x) - x
    if (x <= (-0.098d0)) then
        tmp = (x - sin(x)) / (x - tan(x))
    else if (x <= 0.1d0) then
        tmp = ((0.225d0 * (x ** 2.0d0)) + (((-0.009642857142857142d0) * (x ** 4.0d0)) + (0.00024107142857142857d0 * (x ** 6.0d0)))) + (-0.5d0)
    else
        tmp = (sin(x) / t_0) - (x / t_0)
    end if
    code = tmp
end function

Java

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

↓

public static double code(double x) {
	double t_0 = Math.tan(x) - x;
	double tmp;
	if (x <= -0.098) {
		tmp = (x - Math.sin(x)) / (x - Math.tan(x));
	} else if (x <= 0.1) {
		tmp = ((0.225 * Math.pow(x, 2.0)) + ((-0.009642857142857142 * Math.pow(x, 4.0)) + (0.00024107142857142857 * Math.pow(x, 6.0)))) + -0.5;
	} else {
		tmp = (Math.sin(x) / t_0) - (x / t_0);
	}
	return tmp;
}

Python

def code(x):
	return (x - math.sin(x)) / (x - math.tan(x))

↓

def code(x):
	t_0 = math.tan(x) - x
	tmp = 0
	if x <= -0.098:
		tmp = (x - math.sin(x)) / (x - math.tan(x))
	elif x <= 0.1:
		tmp = ((0.225 * math.pow(x, 2.0)) + ((-0.009642857142857142 * math.pow(x, 4.0)) + (0.00024107142857142857 * math.pow(x, 6.0)))) + -0.5
	else:
		tmp = (math.sin(x) / t_0) - (x / t_0)
	return tmp

Julia

function code(x)
	return Float64(Float64(x - sin(x)) / Float64(x - tan(x)))
end

↓

function code(x)
	t_0 = Float64(tan(x) - x)
	tmp = 0.0
	if (x <= -0.098)
		tmp = Float64(Float64(x - sin(x)) / Float64(x - tan(x)));
	elseif (x <= 0.1)
		tmp = Float64(Float64(Float64(0.225 * (x ^ 2.0)) + Float64(Float64(-0.009642857142857142 * (x ^ 4.0)) + Float64(0.00024107142857142857 * (x ^ 6.0)))) + -0.5);
	else
		tmp = Float64(Float64(sin(x) / t_0) - Float64(x / t_0));
	end
	return tmp
end

MATLAB

function tmp = code(x)
	tmp = (x - sin(x)) / (x - tan(x));
end

↓

function tmp_2 = code(x)
	t_0 = tan(x) - x;
	tmp = 0.0;
	if (x <= -0.098)
		tmp = (x - sin(x)) / (x - tan(x));
	elseif (x <= 0.1)
		tmp = ((0.225 * (x ^ 2.0)) + ((-0.009642857142857142 * (x ^ 4.0)) + (0.00024107142857142857 * (x ^ 6.0)))) + -0.5;
	else
		tmp = (sin(x) / t_0) - (x / t_0);
	end
	tmp_2 = tmp;
end

Wolfram

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

↓

code[x_] := Block[{t$95$0 = N[(N[Tan[x], $MachinePrecision] - x), $MachinePrecision]}, If[LessEqual[x, -0.098], N[(N[(x - N[Sin[x], $MachinePrecision]), $MachinePrecision] / N[(x - N[Tan[x], $MachinePrecision]), $MachinePrecision]), $MachinePrecision], If[LessEqual[x, 0.1], N[(N[(N[(0.225 * N[Power[x, 2.0], $MachinePrecision]), $MachinePrecision] + N[(N[(-0.009642857142857142 * N[Power[x, 4.0], $MachinePrecision]), $MachinePrecision] + N[(0.00024107142857142857 * N[Power[x, 6.0], $MachinePrecision]), $MachinePrecision]), $MachinePrecision]), $MachinePrecision] + -0.5), $MachinePrecision], N[(N[(N[Sin[x], $MachinePrecision] / t$95$0), $MachinePrecision] - N[(x / t$95$0), $MachinePrecision]), $MachinePrecision]]]]

Derivation

1. Split input into 3 regimes

2. if x < -0.098000000000000004

   1. Initial program 100.0%

      \[\frac{x - \sin x}{x - \tan x} \]

   if -0.098000000000000004 < x < 0.10000000000000001

   1. Initial program 1.3%

      \[\frac{x - \sin x}{x - \tan x} \]

   2. Simplified 1.3%

      \[\leadsto \color{blue}{\frac{\sin x - x}{\tan x - x}} \]
      Proof

      [Start] 1.3	\[ \frac{x - \sin x}{x - \tan x} \]
      sub-neg [=>] 1.3	\[ \frac{\color{blue}{x + \left(-\sin x\right)}}{x - \tan x} \]
      +-commutative [=>] 1.3	\[ \frac{\color{blue}{\left(-\sin x\right) + x}}{x - \tan x} \]
      neg-sub0 [=>] 1.3	\[ \frac{\color{blue}{\left(0 - \sin x\right)} + x}{x - \tan x} \]
      associate-+l- [=>] 1.3	\[ \frac{\color{blue}{0 - \left(\sin x - x\right)}}{x - \tan x} \]
      sub0-neg [=>] 1.3	\[ \frac{\color{blue}{-\left(\sin x - x\right)}}{x - \tan x} \]
      neg-mul-1 [=>] 1.3	\[ \frac{\color{blue}{-1 \cdot \left(\sin x - x\right)}}{x - \tan x} \]
      sub-neg [=>] 1.3	\[ \frac{-1 \cdot \left(\sin x - x\right)}{\color{blue}{x + \left(-\tan x\right)}} \]
      +-commutative [=>] 1.3	\[ \frac{-1 \cdot \left(\sin x - x\right)}{\color{blue}{\left(-\tan x\right) + x}} \]
      neg-sub0 [=>] 1.3	\[ \frac{-1 \cdot \left(\sin x - x\right)}{\color{blue}{\left(0 - \tan x\right)} + x} \]
      associate-+l- [=>] 1.3	\[ \frac{-1 \cdot \left(\sin x - x\right)}{\color{blue}{0 - \left(\tan x - x\right)}} \]
      sub0-neg [=>] 1.3	\[ \frac{-1 \cdot \left(\sin x - x\right)}{\color{blue}{-\left(\tan x - x\right)}} \]
      neg-mul-1 [=>] 1.3	\[ \frac{-1 \cdot \left(\sin x - x\right)}{\color{blue}{-1 \cdot \left(\tan x - x\right)}} \]
      times-frac [=>] 1.3	\[ \color{blue}{\frac{-1}{-1} \cdot \frac{\sin x - x}{\tan x - x}} \]
      metadata-eval [=>] 1.3	\[ \color{blue}{1} \cdot \frac{\sin x - x}{\tan x - x} \]
      *-lft-identity [=>] 1.3	\[ \color{blue}{\frac{\sin x - x}{\tan x - x}} \]

   3. Taylor expanded in x around 0 100.0%

      \[\leadsto \color{blue}{\left(0.225 \cdot {x}^{2} + \left(-0.009642857142857142 \cdot {x}^{4} + 0.00024107142857142857 \cdot {x}^{6}\right)\right) - 0.5} \]

   if 0.10000000000000001 < x

   1. Initial program 100.0%

      \[\frac{x - \sin x}{x - \tan x} \]

   2. Simplified 100.0%

      \[\leadsto \color{blue}{\frac{\sin x - x}{\tan x - x}} \]
      Proof

      [Start] 100.0	\[ \frac{x - \sin x}{x - \tan x} \]
      sub-neg [=>] 100.0	\[ \frac{\color{blue}{x + \left(-\sin x\right)}}{x - \tan x} \]
      +-commutative [=>] 100.0	\[ \frac{\color{blue}{\left(-\sin x\right) + x}}{x - \tan x} \]
      neg-sub0 [=>] 100.0	\[ \frac{\color{blue}{\left(0 - \sin x\right)} + x}{x - \tan x} \]
      associate-+l- [=>] 100.0	\[ \frac{\color{blue}{0 - \left(\sin x - x\right)}}{x - \tan x} \]
      sub0-neg [=>] 100.0	\[ \frac{\color{blue}{-\left(\sin x - x\right)}}{x - \tan x} \]
      neg-mul-1 [=>] 100.0	\[ \frac{\color{blue}{-1 \cdot \left(\sin x - x\right)}}{x - \tan x} \]
      sub-neg [=>] 100.0	\[ \frac{-1 \cdot \left(\sin x - x\right)}{\color{blue}{x + \left(-\tan x\right)}} \]
      +-commutative [=>] 100.0	\[ \frac{-1 \cdot \left(\sin x - x\right)}{\color{blue}{\left(-\tan x\right) + x}} \]
      neg-sub0 [=>] 100.0	\[ \frac{-1 \cdot \left(\sin x - x\right)}{\color{blue}{\left(0 - \tan x\right)} + x} \]
      associate-+l- [=>] 100.0	\[ \frac{-1 \cdot \left(\sin x - x\right)}{\color{blue}{0 - \left(\tan x - x\right)}} \]
      sub0-neg [=>] 100.0	\[ \frac{-1 \cdot \left(\sin x - x\right)}{\color{blue}{-\left(\tan x - x\right)}} \]
      neg-mul-1 [=>] 100.0	\[ \frac{-1 \cdot \left(\sin x - x\right)}{\color{blue}{-1 \cdot \left(\tan x - x\right)}} \]
      times-frac [=>] 100.0	\[ \color{blue}{\frac{-1}{-1} \cdot \frac{\sin x - x}{\tan x - x}} \]
      metadata-eval [=>] 100.0	\[ \color{blue}{1} \cdot \frac{\sin x - x}{\tan x - x} \]
      *-lft-identity [=>] 100.0	\[ \color{blue}{\frac{\sin x - x}{\tan x - x}} \]

   3. Applied egg-rr 100.0%

      \[\leadsto \color{blue}{\frac{\sin x}{\tan x - x} - \frac{x}{\tan x - x}} \]
      Proof

      [Start] 100.0	\[ \frac{\sin x - x}{\tan x - x} \]
      div-sub [=>] 100.0	\[ \color{blue}{\frac{\sin x}{\tan x - x} - \frac{x}{\tan x - x}} \]

3. Recombined 3 regimes into one program.

4. Final simplification 100.0%

   \[\leadsto \begin{array}{l} \mathbf{if}\;x \leq -0.098:\\ \;\;\;\;\frac{x - \sin x}{x - \tan x}\\ \mathbf{elif}\;x \leq 0.1:\\ \;\;\;\;\left(0.225 \cdot {x}^{2} + \left(-0.009642857142857142 \cdot {x}^{4} + 0.00024107142857142857 \cdot {x}^{6}\right)\right) + -0.5\\ \mathbf{else}:\\ \;\;\;\;\frac{\sin x}{\tan x - x} - \frac{x}{\tan x - x}\\ \end{array} \]

Alternatives

Alternative 1
Accuracy	100.0%
Cost	20168

\[\begin{array}{l} t_0 := \tan x - x\\ \mathbf{if}\;x \leq -0.026:\\ \;\;\;\;\frac{x - \sin x}{x - \tan x}\\ \mathbf{elif}\;x \leq 0.029:\\ \;\;\;\;\left(x \cdot x\right) \cdot \left(0.225 + -0.009642857142857142 \cdot \left(x \cdot x\right)\right) + -0.5\\ \mathbf{else}:\\ \;\;\;\;\frac{\sin x}{t_0} - \frac{x}{t_0}\\ \end{array} \]

Alternative 2
Accuracy	100.0%
Cost	13513

\[\begin{array}{l} \mathbf{if}\;x \leq -0.026 \lor \neg \left(x \leq 0.031\right):\\ \;\;\;\;\frac{x - \sin x}{x - \tan x}\\ \mathbf{else}:\\ \;\;\;\;\left(x \cdot x\right) \cdot \left(0.225 + -0.009642857142857142 \cdot \left(x \cdot x\right)\right) + -0.5\\ \end{array} \]

Alternative 3
Accuracy	99.1%
Cost	7236

\[\begin{array}{l} t_0 := \tan x - x\\ \mathbf{if}\;x \leq -3.1:\\ \;\;\;\;\frac{\frac{3}{x}}{x} - \frac{x}{t_0}\\ \mathbf{elif}\;x \leq 1.55:\\ \;\;\;\;\left(x \cdot x\right) \cdot \left(0.225 + -0.009642857142857142 \cdot \left(x \cdot x\right)\right) + -0.5\\ \mathbf{else}:\\ \;\;\;\;\frac{-x}{t_0}\\ \end{array} \]

Alternative 4
Accuracy	99.1%
Cost	7049

\[\begin{array}{l} \mathbf{if}\;x \leq -2.8 \lor \neg \left(x \leq 1.55\right):\\ \;\;\;\;\frac{-x}{\tan x - x}\\ \mathbf{else}:\\ \;\;\;\;\left(x \cdot x\right) \cdot \left(0.225 + -0.009642857142857142 \cdot \left(x \cdot x\right)\right) + -0.5\\ \end{array} \]

Alternative 5
Accuracy	99.1%
Cost	1096

\[\begin{array}{l} \mathbf{if}\;x \leq -3.1:\\ \;\;\;\;\frac{\frac{3}{x}}{x} + 1\\ \mathbf{elif}\;x \leq 1.55:\\ \;\;\;\;\left(x \cdot x\right) \cdot \left(0.225 + -0.009642857142857142 \cdot \left(x \cdot x\right)\right) + -0.5\\ \mathbf{else}:\\ \;\;\;\;1\\ \end{array} \]

Alternative 6
Accuracy	99.0%
Cost	713

\[\begin{array}{l} \mathbf{if}\;x \leq -2.9 \lor \neg \left(x \leq 2.8\right):\\ \;\;\;\;\frac{\frac{3}{x}}{x} + 1\\ \mathbf{else}:\\ \;\;\;\;-0.5 + x \cdot \left(x \cdot 0.225\right)\\ \end{array} \]

Alternative 7
Accuracy	99.1%
Cost	712

\[\begin{array}{l} \mathbf{if}\;x \leq -2.6:\\ \;\;\;\;1\\ \mathbf{elif}\;x \leq 2.6:\\ \;\;\;\;-0.5 + x \cdot \left(x \cdot 0.225\right)\\ \mathbf{else}:\\ \;\;\;\;1\\ \end{array} \]

Alternative 8
Accuracy	98.8%
Cost	328

\[\begin{array}{l} \mathbf{if}\;x \leq -1.58:\\ \;\;\;\;1\\ \mathbf{elif}\;x \leq 1.55:\\ \;\;\;\;-0.5\\ \mathbf{else}:\\ \;\;\;\;1\\ \end{array} \]

Alternative 9
Accuracy	51.6%
Cost	64

\[-0.5 \]

Reproduce

herbie shell --seed 2023131 
(FPCore (x)
  :name "sintan (problem 3.4.5)"
  :precision binary64
  (/ (- x (sin x)) (- x (tan x))))