Programmiererforum

Code: Select all

scipy.special.logsumexp

Code: Select all

numpy.logaddexp.reduce

Code: Select all

# test_logsumexp.py

from functools import partial
import timeit
import numpy as np
from scipy.special import logsumexp

def logsumexp_manual(a, axis=-1):
max_vals = np.max(a, axis=axis)
exp_terms = np.exp(a - max_vals[..., np.newaxis])
sum_exp_terms = np.sum(exp_terms, axis=axis)
return max_vals + np.log(sum_exp_terms)

# Validity check
arr = np.random.rand(10_000, 30, 3)
scipy_result = logsumexp(arr, axis=-1)
numpy_result = np.logaddexp.reduce(arr, axis=-1)
manual_result = logsumexp_manual(arr)
np.testing.assert_allclose(scipy_result, manual_result)
np.testing.assert_allclose(scipy_result, numpy_result)

n_loops = 10
setup = ("import numpy as np; "
"from scipy.special import logsumexp, softmax; "
"arr = np.random.rand(10_000, 30, 3)")

total_scipy = timeit.timeit(
"logsumexp(arr, axis=-1)",
setup=setup,
number=n_loops,
)

total_numpy = timeit.timeit(
"np.logaddexp.reduce(arr, axis=-1)",
setup=setup,
number=n_loops,
)

logsumexp_partial = partial(logsumexp_manual, a=arr)
total_manual = timeit.timeit(
logsumexp_partial,
setup=setup,
number=n_loops,
)

print(f"Scipy  logsumexp:  {total_scipy  / n_loops:.6f} seconds per loop")
print(f"Numpy  logaddexp:  {total_numpy  / n_loops:.6f} seconds per loop")
print(f"Manual logsumexp:  {total_manual / n_loops:.6f} seconds per loop")
smallest = min(total_scipy, total_numpy, total_manual)
< /code>
 Konsolenausgabe < /h1>
Scipy  logsumexp:   0.070824 seconds per loop
Numpy  logaddexp:   0.044226 seconds per loop
Manual logsumexp:   0.031754 seconds per loop