symmetry log of sum of exp

Percentage Accurate: 54.3% → 98.4%

Time: 8.8s

Alternatives: 9

Speedup: 1.2×

• could not determine a ground truth

  1. a = 5.7628694798851157e-129
  2. b = 1.3168576548067078e+103

Specification

Math

\[\begin{array}{l} \\ \log \left(e^{a} + e^{b}\right) \end{array} \]

FPCore

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

C

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

Fortran

module fmin_fmax_functions
    implicit none
    private
    public fmax
    public fmin

    interface fmax
        module procedure fmax88
        module procedure fmax44
        module procedure fmax84
        module procedure fmax48
    end interface
    interface fmin
        module procedure fmin88
        module procedure fmin44
        module procedure fmin84
        module procedure fmin48
    end interface
contains
    real(8) function fmax88(x, y) result (res)
        real(8), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(x, max(x, y), y /= y), x /= x)
    end function
    real(4) function fmax44(x, y) result (res)
        real(4), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(y, merge(x, max(x, y), y /= y), x /= x)
    end function
    real(8) function fmax84(x, y) result(res)
        real(8), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(dble(y), merge(x, max(x, dble(y)), y /= y), x /= x)
    end function
    real(8) function fmax48(x, y) result(res)
        real(4), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(dble(x), max(dble(x), y), y /= y), x /= x)
    end function
    real(8) function fmin88(x, y) result (res)
        real(8), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(x, min(x, y), y /= y), x /= x)
    end function
    real(4) function fmin44(x, y) result (res)
        real(4), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(y, merge(x, min(x, y), y /= y), x /= x)
    end function
    real(8) function fmin84(x, y) result(res)
        real(8), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(dble(y), merge(x, min(x, dble(y)), y /= y), x /= x)
    end function
    real(8) function fmin48(x, y) result(res)
        real(4), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(dble(x), min(dble(x), y), y /= y), x /= x)
    end function
end module

real(8) function code(a, b)
use fmin_fmax_functions
    real(8), intent (in) :: a
    real(8), intent (in) :: b
    code = log((exp(a) + exp(b)))
end function

Java

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

Python

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

Julia

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

MATLAB

function tmp = code(a, b)
	tmp = log((exp(a) + exp(b)));
end

Wolfram

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

Initial Program: 54.3% accurate, 1.0× speedup

Math

\[\begin{array}{l} \\ \log \left(e^{a} + e^{b}\right) \end{array} \]

FPCore

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

C

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

Fortran

module fmin_fmax_functions
    implicit none
    private
    public fmax
    public fmin

    interface fmax
        module procedure fmax88
        module procedure fmax44
        module procedure fmax84
        module procedure fmax48
    end interface
    interface fmin
        module procedure fmin88
        module procedure fmin44
        module procedure fmin84
        module procedure fmin48
    end interface
contains
    real(8) function fmax88(x, y) result (res)
        real(8), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(x, max(x, y), y /= y), x /= x)
    end function
    real(4) function fmax44(x, y) result (res)
        real(4), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(y, merge(x, max(x, y), y /= y), x /= x)
    end function
    real(8) function fmax84(x, y) result(res)
        real(8), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(dble(y), merge(x, max(x, dble(y)), y /= y), x /= x)
    end function
    real(8) function fmax48(x, y) result(res)
        real(4), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(dble(x), max(dble(x), y), y /= y), x /= x)
    end function
    real(8) function fmin88(x, y) result (res)
        real(8), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(x, min(x, y), y /= y), x /= x)
    end function
    real(4) function fmin44(x, y) result (res)
        real(4), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(y, merge(x, min(x, y), y /= y), x /= x)
    end function
    real(8) function fmin84(x, y) result(res)
        real(8), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(dble(y), merge(x, min(x, dble(y)), y /= y), x /= x)
    end function
    real(8) function fmin48(x, y) result(res)
        real(4), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(dble(x), min(dble(x), y), y /= y), x /= x)
    end function
end module

real(8) function code(a, b)
use fmin_fmax_functions
    real(8), intent (in) :: a
    real(8), intent (in) :: b
    code = log((exp(a) + exp(b)))
end function

Java

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

Python

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

Julia

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

MATLAB

function tmp = code(a, b)
	tmp = log((exp(a) + exp(b)));
end

Wolfram

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

Alternative 1: 98.4% accurate, 0.7× speedup

Math

\[\begin{array}{l} [a, b] = \mathsf{sort}([a, b])\\ \\ \frac{b}{1 + e^{a}} + \mathsf{log1p}\left(e^{a}\right) \end{array} \]

FPCore

NOTE: a and b should be sorted in increasing order before calling this function.
(FPCore (a b) :precision binary64 (+ (/ b (+ 1.0 (exp a))) (log1p (exp a))))

C

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

Java

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

Python

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

Julia

a, b = sort([a, b])
function code(a, b)
	return Float64(Float64(b / Float64(1.0 + exp(a))) + log1p(exp(a)))
end

Wolfram

NOTE: a and b should be sorted in increasing order before calling this function.
code[a_, b_] := N[(N[(b / N[(1.0 + N[Exp[a], $MachinePrecision]), $MachinePrecision]), $MachinePrecision] + N[Log[1 + N[Exp[a], $MachinePrecision]], $MachinePrecision]), $MachinePrecision]

Derivation

1. Initial program 54.3%

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

2. Taylor expanded in b around 0

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

   1. +-commutative N/A

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

   2. lower-+.f64 N/A

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

   3. lower-/.f64 N/A

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

   4. lower-+.f64 N/A

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

   5. lift-exp.f64 N/A

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

   6. lower-log.f64 N/A

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

   7. lower-+.f64 N/A

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

   8. lift-exp.f64 98.2

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

4. Applied rewrites 98.2%

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

   1. lift-log.f64 N/A

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

   2. lift-+.f64 N/A

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

   3. lift-exp.f64 N/A

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

   4. lower-log1p.f64 N/A

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

   5. lift-exp.f64 98.4

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

6. Applied rewrites 98.4%

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

Alternative 2: 98.2% accurate, 0.8× speedup

Math

\[\begin{array}{l} [a, b] = \mathsf{sort}([a, b])\\ \\ \begin{array}{l} t_0 := 1 + e^{a}\\ \frac{b}{t\_0} + \log t\_0 \end{array} \end{array} \]

FPCore

NOTE: a and b should be sorted in increasing order before calling this function.
(FPCore (a b)
 :precision binary64
 (let* ((t_0 (+ 1.0 (exp a)))) (+ (/ b t_0) (log t_0))))

C

assert(a < b);
double code(double a, double b) {
	double t_0 = 1.0 + exp(a);
	return (b / t_0) + log(t_0);
}

Fortran

NOTE: a and b should be sorted in increasing order before calling this function.
module fmin_fmax_functions
    implicit none
    private
    public fmax
    public fmin

    interface fmax
        module procedure fmax88
        module procedure fmax44
        module procedure fmax84
        module procedure fmax48
    end interface
    interface fmin
        module procedure fmin88
        module procedure fmin44
        module procedure fmin84
        module procedure fmin48
    end interface
contains
    real(8) function fmax88(x, y) result (res)
        real(8), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(x, max(x, y), y /= y), x /= x)
    end function
    real(4) function fmax44(x, y) result (res)
        real(4), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(y, merge(x, max(x, y), y /= y), x /= x)
    end function
    real(8) function fmax84(x, y) result(res)
        real(8), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(dble(y), merge(x, max(x, dble(y)), y /= y), x /= x)
    end function
    real(8) function fmax48(x, y) result(res)
        real(4), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(dble(x), max(dble(x), y), y /= y), x /= x)
    end function
    real(8) function fmin88(x, y) result (res)
        real(8), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(x, min(x, y), y /= y), x /= x)
    end function
    real(4) function fmin44(x, y) result (res)
        real(4), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(y, merge(x, min(x, y), y /= y), x /= x)
    end function
    real(8) function fmin84(x, y) result(res)
        real(8), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(dble(y), merge(x, min(x, dble(y)), y /= y), x /= x)
    end function
    real(8) function fmin48(x, y) result(res)
        real(4), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(dble(x), min(dble(x), y), y /= y), x /= x)
    end function
end module

real(8) function code(a, b)
use fmin_fmax_functions
    real(8), intent (in) :: a
    real(8), intent (in) :: b
    real(8) :: t_0
    t_0 = 1.0d0 + exp(a)
    code = (b / t_0) + log(t_0)
end function

Java

assert a < b;
public static double code(double a, double b) {
	double t_0 = 1.0 + Math.exp(a);
	return (b / t_0) + Math.log(t_0);
}

Python

[a, b] = sort([a, b])
def code(a, b):
	t_0 = 1.0 + math.exp(a)
	return (b / t_0) + math.log(t_0)

Julia

a, b = sort([a, b])
function code(a, b)
	t_0 = Float64(1.0 + exp(a))
	return Float64(Float64(b / t_0) + log(t_0))
end

MATLAB

a, b = num2cell(sort([a, b])){:}
function tmp = code(a, b)
	t_0 = 1.0 + exp(a);
	tmp = (b / t_0) + log(t_0);
end

Wolfram

NOTE: a and b should be sorted in increasing order before calling this function.
code[a_, b_] := Block[{t$95$0 = N[(1.0 + N[Exp[a], $MachinePrecision]), $MachinePrecision]}, N[(N[(b / t$95$0), $MachinePrecision] + N[Log[t$95$0], $MachinePrecision]), $MachinePrecision]]

Derivation

1. Initial program 54.3%

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

2. Taylor expanded in b around 0

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

   1. +-commutative N/A

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

   2. lower-+.f64 N/A

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

   3. lower-/.f64 N/A

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

   4. lower-+.f64 N/A

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

   5. lift-exp.f64 N/A

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

   6. lower-log.f64 N/A

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

   7. lower-+.f64 N/A

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

   8. lift-exp.f64 98.2

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

4. Applied rewrites 98.2%

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

Alternative 3: 58.6% accurate, 1.1× speedup

Math

\[\begin{array}{l} [a, b] = \mathsf{sort}([a, b])\\ \\ 0.5 \cdot b + \mathsf{log1p}\left(e^{a}\right) \end{array} \]

FPCore

NOTE: a and b should be sorted in increasing order before calling this function.
(FPCore (a b) :precision binary64 (+ (* 0.5 b) (log1p (exp a))))

C

assert(a < b);
double code(double a, double b) {
	return (0.5 * b) + log1p(exp(a));
}

Java

assert a < b;
public static double code(double a, double b) {
	return (0.5 * b) + Math.log1p(Math.exp(a));
}

Python

[a, b] = sort([a, b])
def code(a, b):
	return (0.5 * b) + math.log1p(math.exp(a))

Julia

a, b = sort([a, b])
function code(a, b)
	return Float64(Float64(0.5 * b) + log1p(exp(a)))
end

Wolfram

NOTE: a and b should be sorted in increasing order before calling this function.
code[a_, b_] := N[(N[(0.5 * b), $MachinePrecision] + N[Log[1 + N[Exp[a], $MachinePrecision]], $MachinePrecision]), $MachinePrecision]

Derivation

1. Initial program 54.3%

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

2. Taylor expanded in b around 0

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

   1. +-commutative N/A

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

   2. lower-+.f64 N/A

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

   3. lower-/.f64 N/A

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

   4. lower-+.f64 N/A

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

   5. lift-exp.f64 N/A

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

   6. lower-log.f64 N/A

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

   7. lower-+.f64 N/A

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

   8. lift-exp.f64 98.2

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

4. Applied rewrites 98.2%

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

   1. lift-log.f64 N/A

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

   2. lift-+.f64 N/A

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

   3. lift-exp.f64 N/A

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

   4. lower-log1p.f64 N/A

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

   5. lift-exp.f64 98.4

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

6. Applied rewrites 98.4%

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

7. Taylor expanded in a around 0

   \[\leadsto \frac{1}{2} \cdot b + \mathsf{log1p}\left(\color{blue}{e^{a}}\right) \]
8. Step-by-step derivation

   1. lower-*.f64 58.6

      \[\leadsto 0.5 \cdot b + \mathsf{log1p}\left(e^{a}\right) \]

9. Applied rewrites 58.6%

   \[\leadsto 0.5 \cdot b + \mathsf{log1p}\left(\color{blue}{e^{a}}\right) \]
10. Add Preprocessing

Alternative 4: 58.4% accurate, 1.2× speedup

Math

\[\begin{array}{l} [a, b] = \mathsf{sort}([a, b])\\ \\ 0.5 \cdot b + \log \left(1 + e^{a}\right) \end{array} \]

FPCore

NOTE: a and b should be sorted in increasing order before calling this function.
(FPCore (a b) :precision binary64 (+ (* 0.5 b) (log (+ 1.0 (exp a)))))

C

assert(a < b);
double code(double a, double b) {
	return (0.5 * b) + log((1.0 + exp(a)));
}

Fortran

NOTE: a and b should be sorted in increasing order before calling this function.
module fmin_fmax_functions
    implicit none
    private
    public fmax
    public fmin

    interface fmax
        module procedure fmax88
        module procedure fmax44
        module procedure fmax84
        module procedure fmax48
    end interface
    interface fmin
        module procedure fmin88
        module procedure fmin44
        module procedure fmin84
        module procedure fmin48
    end interface
contains
    real(8) function fmax88(x, y) result (res)
        real(8), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(x, max(x, y), y /= y), x /= x)
    end function
    real(4) function fmax44(x, y) result (res)
        real(4), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(y, merge(x, max(x, y), y /= y), x /= x)
    end function
    real(8) function fmax84(x, y) result(res)
        real(8), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(dble(y), merge(x, max(x, dble(y)), y /= y), x /= x)
    end function
    real(8) function fmax48(x, y) result(res)
        real(4), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(dble(x), max(dble(x), y), y /= y), x /= x)
    end function
    real(8) function fmin88(x, y) result (res)
        real(8), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(x, min(x, y), y /= y), x /= x)
    end function
    real(4) function fmin44(x, y) result (res)
        real(4), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(y, merge(x, min(x, y), y /= y), x /= x)
    end function
    real(8) function fmin84(x, y) result(res)
        real(8), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(dble(y), merge(x, min(x, dble(y)), y /= y), x /= x)
    end function
    real(8) function fmin48(x, y) result(res)
        real(4), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(dble(x), min(dble(x), y), y /= y), x /= x)
    end function
end module

real(8) function code(a, b)
use fmin_fmax_functions
    real(8), intent (in) :: a
    real(8), intent (in) :: b
    code = (0.5d0 * b) + log((1.0d0 + exp(a)))
end function

Java

assert a < b;
public static double code(double a, double b) {
	return (0.5 * b) + Math.log((1.0 + Math.exp(a)));
}

Python

[a, b] = sort([a, b])
def code(a, b):
	return (0.5 * b) + math.log((1.0 + math.exp(a)))

Julia

a, b = sort([a, b])
function code(a, b)
	return Float64(Float64(0.5 * b) + log(Float64(1.0 + exp(a))))
end

MATLAB

a, b = num2cell(sort([a, b])){:}
function tmp = code(a, b)
	tmp = (0.5 * b) + log((1.0 + exp(a)));
end

Wolfram

NOTE: a and b should be sorted in increasing order before calling this function.
code[a_, b_] := N[(N[(0.5 * b), $MachinePrecision] + N[Log[N[(1.0 + N[Exp[a], $MachinePrecision]), $MachinePrecision]], $MachinePrecision]), $MachinePrecision]

Derivation

1. Initial program 54.3%

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

2. Taylor expanded in b around 0

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

   1. +-commutative N/A

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

   2. lower-+.f64 N/A

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

   3. lower-/.f64 N/A

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

   4. lower-+.f64 N/A

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

   5. lift-exp.f64 N/A

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

   6. lower-log.f64 N/A

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

   7. lower-+.f64 N/A

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

   8. lift-exp.f64 98.2

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

4. Applied rewrites 98.2%

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

5. Taylor expanded in a around 0

   \[\leadsto \frac{1}{2} \cdot b + \log \color{blue}{\left(1 + e^{a}\right)} \]
6. Step-by-step derivation

   1. lower-*.f64 58.4

      \[\leadsto 0.5 \cdot b + \log \left(1 + \color{blue}{e^{a}}\right) \]

7. Applied rewrites 58.4%

   \[\leadsto 0.5 \cdot b + \log \color{blue}{\left(1 + e^{a}\right)} \]
8. Add Preprocessing

Alternative 5: 52.6% accurate, 1.3× speedup

Math

\[\begin{array}{l} [a, b] = \mathsf{sort}([a, b])\\ \\ \log \left(e^{a} + \left(1 + b\right)\right) \end{array} \]

FPCore

NOTE: a and b should be sorted in increasing order before calling this function.
(FPCore (a b) :precision binary64 (log (+ (exp a) (+ 1.0 b))))

C

assert(a < b);
double code(double a, double b) {
	return log((exp(a) + (1.0 + b)));
}

Fortran

NOTE: a and b should be sorted in increasing order before calling this function.
module fmin_fmax_functions
    implicit none
    private
    public fmax
    public fmin

    interface fmax
        module procedure fmax88
        module procedure fmax44
        module procedure fmax84
        module procedure fmax48
    end interface
    interface fmin
        module procedure fmin88
        module procedure fmin44
        module procedure fmin84
        module procedure fmin48
    end interface
contains
    real(8) function fmax88(x, y) result (res)
        real(8), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(x, max(x, y), y /= y), x /= x)
    end function
    real(4) function fmax44(x, y) result (res)
        real(4), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(y, merge(x, max(x, y), y /= y), x /= x)
    end function
    real(8) function fmax84(x, y) result(res)
        real(8), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(dble(y), merge(x, max(x, dble(y)), y /= y), x /= x)
    end function
    real(8) function fmax48(x, y) result(res)
        real(4), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(dble(x), max(dble(x), y), y /= y), x /= x)
    end function
    real(8) function fmin88(x, y) result (res)
        real(8), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(x, min(x, y), y /= y), x /= x)
    end function
    real(4) function fmin44(x, y) result (res)
        real(4), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(y, merge(x, min(x, y), y /= y), x /= x)
    end function
    real(8) function fmin84(x, y) result(res)
        real(8), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(dble(y), merge(x, min(x, dble(y)), y /= y), x /= x)
    end function
    real(8) function fmin48(x, y) result(res)
        real(4), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(dble(x), min(dble(x), y), y /= y), x /= x)
    end function
end module

real(8) function code(a, b)
use fmin_fmax_functions
    real(8), intent (in) :: a
    real(8), intent (in) :: b
    code = log((exp(a) + (1.0d0 + b)))
end function

Java

assert a < b;
public static double code(double a, double b) {
	return Math.log((Math.exp(a) + (1.0 + b)));
}

Python

[a, b] = sort([a, b])
def code(a, b):
	return math.log((math.exp(a) + (1.0 + b)))

Julia

a, b = sort([a, b])
function code(a, b)
	return log(Float64(exp(a) + Float64(1.0 + b)))
end

MATLAB

a, b = num2cell(sort([a, b])){:}
function tmp = code(a, b)
	tmp = log((exp(a) + (1.0 + b)));
end

Wolfram

NOTE: a and b should be sorted in increasing order before calling this function.
code[a_, b_] := N[Log[N[(N[Exp[a], $MachinePrecision] + N[(1.0 + b), $MachinePrecision]), $MachinePrecision]], $MachinePrecision]

Derivation

1. Initial program 54.3%

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

2. Taylor expanded in b around 0

   \[\leadsto \log \left(e^{a} + \color{blue}{\left(1 + b\right)}\right) \]
3. Step-by-step derivation

   1. lower-+.f64 52.6

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

4. Applied rewrites 52.6%

   \[\leadsto \log \left(e^{a} + \color{blue}{\left(1 + b\right)}\right) \]
5. Add Preprocessing

Alternative 6: 51.2% accurate, 1.5× speedup

Math

\[\begin{array}{l} [a, b] = \mathsf{sort}([a, b])\\ \\ \log \left(1 + e^{a}\right) \end{array} \]

FPCore

NOTE: a and b should be sorted in increasing order before calling this function.
(FPCore (a b) :precision binary64 (log (+ 1.0 (exp a))))

C

assert(a < b);
double code(double a, double b) {
	return log((1.0 + exp(a)));
}

Fortran

NOTE: a and b should be sorted in increasing order before calling this function.
module fmin_fmax_functions
    implicit none
    private
    public fmax
    public fmin

    interface fmax
        module procedure fmax88
        module procedure fmax44
        module procedure fmax84
        module procedure fmax48
    end interface
    interface fmin
        module procedure fmin88
        module procedure fmin44
        module procedure fmin84
        module procedure fmin48
    end interface
contains
    real(8) function fmax88(x, y) result (res)
        real(8), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(x, max(x, y), y /= y), x /= x)
    end function
    real(4) function fmax44(x, y) result (res)
        real(4), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(y, merge(x, max(x, y), y /= y), x /= x)
    end function
    real(8) function fmax84(x, y) result(res)
        real(8), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(dble(y), merge(x, max(x, dble(y)), y /= y), x /= x)
    end function
    real(8) function fmax48(x, y) result(res)
        real(4), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(dble(x), max(dble(x), y), y /= y), x /= x)
    end function
    real(8) function fmin88(x, y) result (res)
        real(8), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(x, min(x, y), y /= y), x /= x)
    end function
    real(4) function fmin44(x, y) result (res)
        real(4), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(y, merge(x, min(x, y), y /= y), x /= x)
    end function
    real(8) function fmin84(x, y) result(res)
        real(8), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(dble(y), merge(x, min(x, dble(y)), y /= y), x /= x)
    end function
    real(8) function fmin48(x, y) result(res)
        real(4), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(dble(x), min(dble(x), y), y /= y), x /= x)
    end function
end module

real(8) function code(a, b)
use fmin_fmax_functions
    real(8), intent (in) :: a
    real(8), intent (in) :: b
    code = log((1.0d0 + exp(a)))
end function

Java

assert a < b;
public static double code(double a, double b) {
	return Math.log((1.0 + Math.exp(a)));
}

Python

[a, b] = sort([a, b])
def code(a, b):
	return math.log((1.0 + math.exp(a)))

Julia

a, b = sort([a, b])
function code(a, b)
	return log(Float64(1.0 + exp(a)))
end

MATLAB

a, b = num2cell(sort([a, b])){:}
function tmp = code(a, b)
	tmp = log((1.0 + exp(a)));
end

Wolfram

NOTE: a and b should be sorted in increasing order before calling this function.
code[a_, b_] := N[Log[N[(1.0 + N[Exp[a], $MachinePrecision]), $MachinePrecision]], $MachinePrecision]

Derivation

1. Initial program 54.3%

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

2. Taylor expanded in b around 0

   \[\leadsto \log \color{blue}{\left(1 + e^{a}\right)} \]
3. Step-by-step derivation

   1. lower-+.f64 N/A

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

   2. lift-exp.f64 51.2

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

4. Applied rewrites 51.2%

   \[\leadsto \log \color{blue}{\left(1 + e^{a}\right)} \]
5. Add Preprocessing

Alternative 7: 50.1% accurate, 1.6× speedup

Math

\[\begin{array}{l} [a, b] = \mathsf{sort}([a, b])\\ \\ \frac{a}{2 + b} + \log \left(2 + b\right) \end{array} \]

FPCore

NOTE: a and b should be sorted in increasing order before calling this function.
(FPCore (a b) :precision binary64 (+ (/ a (+ 2.0 b)) (log (+ 2.0 b))))

C

assert(a < b);
double code(double a, double b) {
	return (a / (2.0 + b)) + log((2.0 + b));
}

Fortran

NOTE: a and b should be sorted in increasing order before calling this function.
module fmin_fmax_functions
    implicit none
    private
    public fmax
    public fmin

    interface fmax
        module procedure fmax88
        module procedure fmax44
        module procedure fmax84
        module procedure fmax48
    end interface
    interface fmin
        module procedure fmin88
        module procedure fmin44
        module procedure fmin84
        module procedure fmin48
    end interface
contains
    real(8) function fmax88(x, y) result (res)
        real(8), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(x, max(x, y), y /= y), x /= x)
    end function
    real(4) function fmax44(x, y) result (res)
        real(4), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(y, merge(x, max(x, y), y /= y), x /= x)
    end function
    real(8) function fmax84(x, y) result(res)
        real(8), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(dble(y), merge(x, max(x, dble(y)), y /= y), x /= x)
    end function
    real(8) function fmax48(x, y) result(res)
        real(4), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(dble(x), max(dble(x), y), y /= y), x /= x)
    end function
    real(8) function fmin88(x, y) result (res)
        real(8), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(x, min(x, y), y /= y), x /= x)
    end function
    real(4) function fmin44(x, y) result (res)
        real(4), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(y, merge(x, min(x, y), y /= y), x /= x)
    end function
    real(8) function fmin84(x, y) result(res)
        real(8), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(dble(y), merge(x, min(x, dble(y)), y /= y), x /= x)
    end function
    real(8) function fmin48(x, y) result(res)
        real(4), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(dble(x), min(dble(x), y), y /= y), x /= x)
    end function
end module

real(8) function code(a, b)
use fmin_fmax_functions
    real(8), intent (in) :: a
    real(8), intent (in) :: b
    code = (a / (2.0d0 + b)) + log((2.0d0 + b))
end function

Java

assert a < b;
public static double code(double a, double b) {
	return (a / (2.0 + b)) + Math.log((2.0 + b));
}

Python

[a, b] = sort([a, b])
def code(a, b):
	return (a / (2.0 + b)) + math.log((2.0 + b))

Julia

a, b = sort([a, b])
function code(a, b)
	return Float64(Float64(a / Float64(2.0 + b)) + log(Float64(2.0 + b)))
end

MATLAB

a, b = num2cell(sort([a, b])){:}
function tmp = code(a, b)
	tmp = (a / (2.0 + b)) + log((2.0 + b));
end

Wolfram

NOTE: a and b should be sorted in increasing order before calling this function.
code[a_, b_] := N[(N[(a / N[(2.0 + b), $MachinePrecision]), $MachinePrecision] + N[Log[N[(2.0 + b), $MachinePrecision]], $MachinePrecision]), $MachinePrecision]

Derivation

1. Initial program 54.3%

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

2. Taylor expanded in a around 0

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

   1. +-commutative N/A

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

   2. lower-+.f64 N/A

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

   3. lower-/.f64 N/A

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

   4. lower-+.f64 N/A

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

   5. lift-exp.f64 N/A

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

   6. lower-log.f64 N/A

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

   7. lower-+.f64 N/A

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

   8. lift-exp.f64 50.5

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

4. Applied rewrites 50.5%

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

5. Taylor expanded in b around 0

   \[\leadsto \frac{a}{2 + b} + \log \left(1 + \color{blue}{e^{b}}\right) \]
6. Step-by-step derivation

   1. lower-+.f64 50.4

      \[\leadsto \frac{a}{2 + b} + \log \left(1 + e^{b}\right) \]

7. Applied rewrites 50.4%

   \[\leadsto \frac{a}{2 + b} + \log \left(1 + \color{blue}{e^{b}}\right) \]

8. Taylor expanded in b around 0

   \[\leadsto \frac{a}{2 + b} + \log \left(2 + b\right) \]
9. Step-by-step derivation

   1. lower-+.f64 49.6

      \[\leadsto \frac{a}{2 + b} + \log \left(2 + b\right) \]

10. Applied rewrites 49.6%

    \[\leadsto \frac{a}{2 + b} + \log \left(2 + b\right) \]
11. Add Preprocessing

Alternative 8: 49.6% accurate, 2.5× speedup

Math

\[\begin{array}{l} [a, b] = \mathsf{sort}([a, b])\\ \\ \mathsf{fma}\left(0.5, b, \log 2\right) \end{array} \]

FPCore

NOTE: a and b should be sorted in increasing order before calling this function.
(FPCore (a b) :precision binary64 (fma 0.5 b (log 2.0)))

C

assert(a < b);
double code(double a, double b) {
	return fma(0.5, b, log(2.0));
}

Julia

a, b = sort([a, b])
function code(a, b)
	return fma(0.5, b, log(2.0))
end

Wolfram

NOTE: a and b should be sorted in increasing order before calling this function.
code[a_, b_] := N[(0.5 * b + N[Log[2.0], $MachinePrecision]), $MachinePrecision]

Derivation

1. Initial program 54.3%

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

2. Taylor expanded in b around 0

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

   1. +-commutative N/A

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

   2. lower-+.f64 N/A

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

   3. lower-/.f64 N/A

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

   4. lower-+.f64 N/A

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

   5. lift-exp.f64 N/A

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

   6. lower-log.f64 N/A

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

   7. lower-+.f64 N/A

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

   8. lift-exp.f64 98.2

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

4. Applied rewrites 98.2%

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

5. Taylor expanded in a around 0

   \[\leadsto \log 2 + \color{blue}{\frac{1}{2} \cdot b} \]
6. Step-by-step derivation

   1. +-commutative N/A

      \[\leadsto \frac{1}{2} \cdot b + \log 2 \]

   2. lower-fma.f64 N/A

      \[\leadsto \mathsf{fma}\left(\frac{1}{2}, b, \log 2\right) \]

   3. lower-log.f64 50.1

      \[\leadsto \mathsf{fma}\left(0.5, b, \log 2\right) \]

7. Applied rewrites 50.1%

   \[\leadsto \mathsf{fma}\left(0.5, \color{blue}{b}, \log 2\right) \]
8. Add Preprocessing

Alternative 9: 49.3% accurate, 4.4× speedup

Math

\[\begin{array}{l} [a, b] = \mathsf{sort}([a, b])\\ \\ \log 2 \end{array} \]

FPCore

NOTE: a and b should be sorted in increasing order before calling this function.
(FPCore (a b) :precision binary64 (log 2.0))

C

assert(a < b);
double code(double a, double b) {
	return log(2.0);
}

Fortran

NOTE: a and b should be sorted in increasing order before calling this function.
module fmin_fmax_functions
    implicit none
    private
    public fmax
    public fmin

    interface fmax
        module procedure fmax88
        module procedure fmax44
        module procedure fmax84
        module procedure fmax48
    end interface
    interface fmin
        module procedure fmin88
        module procedure fmin44
        module procedure fmin84
        module procedure fmin48
    end interface
contains
    real(8) function fmax88(x, y) result (res)
        real(8), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(x, max(x, y), y /= y), x /= x)
    end function
    real(4) function fmax44(x, y) result (res)
        real(4), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(y, merge(x, max(x, y), y /= y), x /= x)
    end function
    real(8) function fmax84(x, y) result(res)
        real(8), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(dble(y), merge(x, max(x, dble(y)), y /= y), x /= x)
    end function
    real(8) function fmax48(x, y) result(res)
        real(4), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(dble(x), max(dble(x), y), y /= y), x /= x)
    end function
    real(8) function fmin88(x, y) result (res)
        real(8), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(x, min(x, y), y /= y), x /= x)
    end function
    real(4) function fmin44(x, y) result (res)
        real(4), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(y, merge(x, min(x, y), y /= y), x /= x)
    end function
    real(8) function fmin84(x, y) result(res)
        real(8), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(dble(y), merge(x, min(x, dble(y)), y /= y), x /= x)
    end function
    real(8) function fmin48(x, y) result(res)
        real(4), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(dble(x), min(dble(x), y), y /= y), x /= x)
    end function
end module

real(8) function code(a, b)
use fmin_fmax_functions
    real(8), intent (in) :: a
    real(8), intent (in) :: b
    code = log(2.0d0)
end function

Java

assert a < b;
public static double code(double a, double b) {
	return Math.log(2.0);
}

Python

[a, b] = sort([a, b])
def code(a, b):
	return math.log(2.0)

Julia

a, b = sort([a, b])
function code(a, b)
	return log(2.0)
end

MATLAB

a, b = num2cell(sort([a, b])){:}
function tmp = code(a, b)
	tmp = log(2.0);
end

Wolfram

NOTE: a and b should be sorted in increasing order before calling this function.
code[a_, b_] := N[Log[2.0], $MachinePrecision]

Derivation

1. Initial program 54.3%

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

2. Taylor expanded in b around 0

   \[\leadsto \log \color{blue}{\left(1 + e^{a}\right)} \]
3. Step-by-step derivation

   1. lower-+.f64 N/A

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

   2. lift-exp.f64 51.2

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

4. Applied rewrites 51.2%

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

5. Taylor expanded in a around 0

   \[\leadsto \log 2 \]
6. Step-by-step derivation

7. Applied rewrites 49.3%

   \[\leadsto \log 2 \]
8. Add Preprocessing

Reproduce

herbie shell --seed 2025139 
(FPCore (a b)
  :name "symmetry log of sum of exp"
  :precision binary64
  (log (+ (exp a) (exp b))))