function_benchmarking.py

import timeit
from mpmath import mp, mpf
import pandas as pd

LIST_OF_FUNCTION_STRINGS = [' acosh(x)',  ' acosh(x**2)',  ' acos(x)',  ' acot(x)',  ' acot(x**2)',  ' acoth(x)',  ' acsc(x)',  ' acsc(x**2)',  ' acsch(x)',  ' agm(x, x**2)',  ' airyai(x)',  ' airybi(1/x)',  ' asec(x)',  ' asech(x)',  ' asin(x)',  ' asinh(x)',  ' atan(x)',  ' atan2(x, x)',  ' atanh(x)',  ' barnesg(1/x)', ' besseli(x, 1/x)',  ' besselk(x, 1/x)',  ' bessely(x, 1/x)',  ' beta(x, x)',  ' cbrt(x**0.1)',  ' chebyt(x, x**2)',  ' chebyu(x, x**2)',  ' chi(x)',  ' clcos(x, x)',  ' clsin(x, x)',  ' cos(x)',  ' cos(x**2)',  ' cosh(x)',  ' cot(x)',  ' csch(x)',  ' degrees(x)',  ' ellipk(x)',  ' elliprc(x, x)',  'ellipf(x,-x)',  'ellipe(x,-x)', ' elliprg(x, x, x)',  ' erfc(x)',  ' exp(x)',  ' exp(x**2)',  ' expm1(x)',  ' gamma(x)',  ' hankel1(x, 2)',  ' hankel2(x, 2)',  ' harmonic(x)',  ' harmonic(x**2)',  ' hermite(x, x)',  ' hypot(x,x**2)',  ' jacobi(x, x, x, x)',  ' laguerre(x, 1/x, 1/x, maxterms=10**10)',  ' lambertw(x)',  ' lambertw(x**2)',  ' legendre(x, x)',  ' ln(x)',  ' ln(x**2)',  ' log(x)',  ' log(x**2)',  ' log10(x)',  ' loggamma(x**2)',  ' powm1(x,x)',  ' power(x, 3)',  ' power(x, 4)',  ' power(x, 5)',  ' power(x, 6)',  ' power(x, 7)',  ' power(x, 8)',  ' power(x, 9)',  ' power(x, 10)',  ' power(x, 11)',  ' power(x, 12)',  ' psi(x, x**2)',  ' radians(x)',  ' riemannr(x)',  ' root(x,3)',  ' root(x,4)',  ' root(x,5)',  ' root(x,6)',  ' root(x,7)',  ' root(x,8)',  ' root(x,9)',  ' root(x,10)',  ' root(x,11)',  ' root(x,12)',  ' scorerhi(x)',  ' sec(x)',  ' sech(x)',  ' shi(x**2)',  ' shi(x)',  ' si(x**2)',  ' si(x)',  ' siegeltheta(x)',  ' sin(x)',  ' sin(x**2)',  ' sinc(x)',  ' sinh(x)' ,  ' tanh(x)']

def benchmark_mpmath_functions( precisions, num_iterations):
    results = []
    for func_name in LIST_OF_FUNCTION_STRINGS:
        for precision in precisions:
            print(f"Current function: {func_name} | Precision: {precision}")
            mp.dps = precision
            setup_code = f"from mpmath import mp, mpf; mp.dps = {precision}; x = mpf(1) / mpf(10**{10})"
            test_code = f"mp.{func_name.strip()}"
            execution_time = timeit.timeit(stmt=test_code, setup=setup_code, number=num_iterations)
            computations_per_second = num_iterations / execution_time
            results.append([func_name.strip(), precision, computations_per_second])

    df = pd.DataFrame(results, columns=["Function Name", "Precision", "Computations per Second"])
    return df

# Set the precision levels to benchmark
precisions = [16, 32, 64, 128, 256, 512] # 1024, 2056, 4096, 8192]

# Set the number of iterations for each benchmark
num_iterations = 5

# Run the benchmarks
results_df = benchmark_mpmath_functions( precisions, num_iterations)
# Print the results DataFrame
print(results_df)

results_df.to_csv('mpmath_benchmark_results.csv', index=False)