bug366, discussion (missed optimization)

Percentage Accurate: 54.0% → 99.2%

Time: 5.6s

Alternatives: 3

Speedup: 107.0×

Specification

Math

\[\begin{array}{l} \\ \sqrt{a \cdot a - b \cdot b} \end{array} \]

FPCore

(FPCore (a b) :precision binary64 (sqrt (- (* a a) (* b b))))

C

double code(double a, double b) {
	return sqrt(((a * a) - (b * b)));
}

Fortran

real(8) function code(a, b)
    real(8), intent (in) :: a
    real(8), intent (in) :: b
    code = sqrt(((a * a) - (b * b)))
end function

Java

public static double code(double a, double b) {
	return Math.sqrt(((a * a) - (b * b)));
}

Python

def code(a, b):
	return math.sqrt(((a * a) - (b * b)))

Julia

function code(a, b)
	return sqrt(Float64(Float64(a * a) - Float64(b * b)))
end

MATLAB

function tmp = code(a, b)
	tmp = sqrt(((a * a) - (b * b)));
end

Wolfram

code[a_, b_] := N[Sqrt[N[(N[(a * a), $MachinePrecision] - N[(b * b), $MachinePrecision]), $MachinePrecision]], $MachinePrecision]

Initial Program: 54.0% accurate, 1.0× speedup

Math

\[\begin{array}{l} \\ \sqrt{a \cdot a - b \cdot b} \end{array} \]

FPCore

(FPCore (a b) :precision binary64 (sqrt (- (* a a) (* b b))))

C

double code(double a, double b) {
	return sqrt(((a * a) - (b * b)));
}

Fortran

real(8) function code(a, b)
    real(8), intent (in) :: a
    real(8), intent (in) :: b
    code = sqrt(((a * a) - (b * b)))
end function

Java

public static double code(double a, double b) {
	return Math.sqrt(((a * a) - (b * b)));
}

Python

def code(a, b):
	return math.sqrt(((a * a) - (b * b)))

Julia

function code(a, b)
	return sqrt(Float64(Float64(a * a) - Float64(b * b)))
end

MATLAB

function tmp = code(a, b)
	tmp = sqrt(((a * a) - (b * b)));
end

Wolfram

code[a_, b_] := N[Sqrt[N[(N[(a * a), $MachinePrecision] - N[(b * b), $MachinePrecision]), $MachinePrecision]], $MachinePrecision]

Alternative 1: 99.2% accurate, 0.5× speedup

Math

\[\begin{array}{l} a_m = \left|a\right| \\ \sqrt{b + a\_m} \cdot \sqrt{a\_m - b} \end{array} \]

FPCore

a_m = (fabs.f64 a)
(FPCore (a_m b) :precision binary64 (* (sqrt (+ b a_m)) (sqrt (- a_m b))))

C

a_m = fabs(a);
double code(double a_m, double b) {
	return sqrt((b + a_m)) * sqrt((a_m - b));
}

Fortran

a_m = abs(a)
real(8) function code(a_m, b)
    real(8), intent (in) :: a_m
    real(8), intent (in) :: b
    code = sqrt((b + a_m)) * sqrt((a_m - b))
end function

Java

a_m = Math.abs(a);
public static double code(double a_m, double b) {
	return Math.sqrt((b + a_m)) * Math.sqrt((a_m - b));
}

Python

a_m = math.fabs(a)
def code(a_m, b):
	return math.sqrt((b + a_m)) * math.sqrt((a_m - b))

Julia

a_m = abs(a)
function code(a_m, b)
	return Float64(sqrt(Float64(b + a_m)) * sqrt(Float64(a_m - b)))
end

MATLAB

a_m = abs(a);
function tmp = code(a_m, b)
	tmp = sqrt((b + a_m)) * sqrt((a_m - b));
end

Wolfram

a_m = N[Abs[a], $MachinePrecision]
code[a$95$m_, b_] := N[(N[Sqrt[N[(b + a$95$m), $MachinePrecision]], $MachinePrecision] * N[Sqrt[N[(a$95$m - b), $MachinePrecision]], $MachinePrecision]), $MachinePrecision]

Derivation

1. Initial program 54.4%

   \[\sqrt{a \cdot a - b \cdot b} \]
2. Add Preprocessing
3. Step-by-step derivation

   1. pow1/2 54.4%

      \[\leadsto \color{blue}{{\left(a \cdot a - b \cdot b\right)}^{0.5}} \]

   2. difference-of-squares 55.0%

      \[\leadsto {\color{blue}{\left(\left(a + b\right) \cdot \left(a - b\right)\right)}}^{0.5} \]

   3. unpow-prod-down 50.0%

      \[\leadsto \color{blue}{{\left(a + b\right)}^{0.5} \cdot {\left(a - b\right)}^{0.5}} \]

4. Applied egg-rr 50.0%

   \[\leadsto \color{blue}{{\left(a + b\right)}^{0.5} \cdot {\left(a - b\right)}^{0.5}} \]
5. Step-by-step derivation

   1. unpow1/2 50.0%

      \[\leadsto \color{blue}{\sqrt{a + b}} \cdot {\left(a - b\right)}^{0.5} \]

   2. unpow1/2 50.0%

      \[\leadsto \sqrt{a + b} \cdot \color{blue}{\sqrt{a - b}} \]

   3. +-commutative 50.0%

      \[\leadsto \sqrt{\color{blue}{b + a}} \cdot \sqrt{a - b} \]

6. Simplified 50.0%

   \[\leadsto \color{blue}{\sqrt{b + a} \cdot \sqrt{a - b}} \]
7. Add Preprocessing

Alternative 2: 99.6% accurate, 11.9× speedup

Math

\[\begin{array}{l} a_m = \left|a\right| \\ a\_m + b \cdot \left(\frac{b}{a\_m} \cdot -0.5\right) \end{array} \]

FPCore

a_m = (fabs.f64 a)
(FPCore (a_m b) :precision binary64 (+ a_m (* b (* (/ b a_m) -0.5))))

C

a_m = fabs(a);
double code(double a_m, double b) {
	return a_m + (b * ((b / a_m) * -0.5));
}

Fortran

a_m = abs(a)
real(8) function code(a_m, b)
    real(8), intent (in) :: a_m
    real(8), intent (in) :: b
    code = a_m + (b * ((b / a_m) * (-0.5d0)))
end function

Java

a_m = Math.abs(a);
public static double code(double a_m, double b) {
	return a_m + (b * ((b / a_m) * -0.5));
}

Python

a_m = math.fabs(a)
def code(a_m, b):
	return a_m + (b * ((b / a_m) * -0.5))

Julia

a_m = abs(a)
function code(a_m, b)
	return Float64(a_m + Float64(b * Float64(Float64(b / a_m) * -0.5)))
end

MATLAB

a_m = abs(a);
function tmp = code(a_m, b)
	tmp = a_m + (b * ((b / a_m) * -0.5));
end

Wolfram

a_m = N[Abs[a], $MachinePrecision]
code[a$95$m_, b_] := N[(a$95$m + N[(b * N[(N[(b / a$95$m), $MachinePrecision] * -0.5), $MachinePrecision]), $MachinePrecision]), $MachinePrecision]

Derivation

1. Initial program 54.4%

   \[\sqrt{a \cdot a - b \cdot b} \]
2. Add Preprocessing
3. Step-by-step derivation

   1. pow1/2 54.4%

      \[\leadsto \color{blue}{{\left(a \cdot a - b \cdot b\right)}^{0.5}} \]

   2. difference-of-squares 55.0%

      \[\leadsto {\color{blue}{\left(\left(a + b\right) \cdot \left(a - b\right)\right)}}^{0.5} \]

   3. unpow-prod-down 50.0%

      \[\leadsto \color{blue}{{\left(a + b\right)}^{0.5} \cdot {\left(a - b\right)}^{0.5}} \]

4. Applied egg-rr 50.0%

   \[\leadsto \color{blue}{{\left(a + b\right)}^{0.5} \cdot {\left(a - b\right)}^{0.5}} \]
5. Step-by-step derivation

   1. unpow1/2 50.0%

      \[\leadsto \color{blue}{\sqrt{a + b}} \cdot {\left(a - b\right)}^{0.5} \]

   2. unpow1/2 50.0%

      \[\leadsto \sqrt{a + b} \cdot \color{blue}{\sqrt{a - b}} \]

   3. +-commutative 50.0%

      \[\leadsto \sqrt{\color{blue}{b + a}} \cdot \sqrt{a - b} \]

6. Simplified 50.0%

   \[\leadsto \color{blue}{\sqrt{b + a} \cdot \sqrt{a - b}} \]

7. Taylor expanded in b around 0 49.2%

   \[\leadsto \color{blue}{a + b \cdot \left(-0.5 \cdot \frac{b \cdot \left(1 + 0.25 \cdot \frac{{\left(a + -1 \cdot a\right)}^{2}}{{a}^{2}}\right)}{a} + 0.5 \cdot \frac{a + -1 \cdot a}{a}\right)} \]

9. Applied egg-rr 50.8%

   \[\leadsto a + b \cdot \color{blue}{\left(\frac{b}{a} \cdot -0.5\right)} \]
10. Add Preprocessing

Alternative 3: 99.0% accurate, 107.0× speedup

Math

\[\begin{array}{l} a_m = \left|a\right| \\ a\_m \end{array} \]

FPCore

a_m = (fabs.f64 a)
(FPCore (a_m b) :precision binary64 a_m)

C

a_m = fabs(a);
double code(double a_m, double b) {
	return a_m;
}

Fortran

a_m = abs(a)
real(8) function code(a_m, b)
    real(8), intent (in) :: a_m
    real(8), intent (in) :: b
    code = a_m
end function

Java

a_m = Math.abs(a);
public static double code(double a_m, double b) {
	return a_m;
}

Python

a_m = math.fabs(a)
def code(a_m, b):
	return a_m

Julia

a_m = abs(a)
function code(a_m, b)
	return a_m
end

MATLAB

a_m = abs(a);
function tmp = code(a_m, b)
	tmp = a_m;
end

Wolfram

a_m = N[Abs[a], $MachinePrecision]
code[a$95$m_, b_] := a$95$m

Derivation

1. Initial program 54.4%

   \[\sqrt{a \cdot a - b \cdot b} \]
2. Add Preprocessing

3. Taylor expanded in a around inf 50.5%

   \[\leadsto \color{blue}{a} \]
4. Add Preprocessing

Developer Target 1: 99.2% accurate, 0.2× speedup

Math

\[\begin{array}{l} \\ \sqrt{\left|a\right| + \left|b\right|} \cdot \sqrt{\left|a\right| - \left|b\right|} \end{array} \]

FPCore

(FPCore (a b)
 :precision binary64
 (* (sqrt (+ (fabs a) (fabs b))) (sqrt (- (fabs a) (fabs b)))))

C

double code(double a, double b) {
	return sqrt((fabs(a) + fabs(b))) * sqrt((fabs(a) - fabs(b)));
}

Fortran

real(8) function code(a, b)
    real(8), intent (in) :: a
    real(8), intent (in) :: b
    code = sqrt((abs(a) + abs(b))) * sqrt((abs(a) - abs(b)))
end function

Java

public static double code(double a, double b) {
	return Math.sqrt((Math.abs(a) + Math.abs(b))) * Math.sqrt((Math.abs(a) - Math.abs(b)));
}

Python

def code(a, b):
	return math.sqrt((math.fabs(a) + math.fabs(b))) * math.sqrt((math.fabs(a) - math.fabs(b)))

Julia

function code(a, b)
	return Float64(sqrt(Float64(abs(a) + abs(b))) * sqrt(Float64(abs(a) - abs(b))))
end

MATLAB

function tmp = code(a, b)
	tmp = sqrt((abs(a) + abs(b))) * sqrt((abs(a) - abs(b)));
end

Wolfram

code[a_, b_] := N[(N[Sqrt[N[(N[Abs[a], $MachinePrecision] + N[Abs[b], $MachinePrecision]), $MachinePrecision]], $MachinePrecision] * N[Sqrt[N[(N[Abs[a], $MachinePrecision] - N[Abs[b], $MachinePrecision]), $MachinePrecision]], $MachinePrecision]), $MachinePrecision]

Reproduce

herbie shell --seed 2024123 
(FPCore (a b)
  :name "bug366, discussion (missed optimization)"
  :precision binary64

  :alt
  (! :herbie-platform default (let* ((fa (fabs a)) (fb (fabs b))) (* (sqrt (+ fa fb)) (sqrt (- fa fb)))))

  (sqrt (- (* a a) (* b b))))