Skip to content

[MRG] Gracefully handle cost = 0 in generic_conditional_gradient #505

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 6 commits into from
Oct 13, 2023
Merged

[MRG] Gracefully handle cost = 0 in generic_conditional_gradient #505

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
6 commits
Select commit Hold shift + click to select a range
e0a539e
Correctly handle cost = 0.
kachayev Aug 18, 2023
984f982
Merge branch 'master' into fix-line-search-zero-cost
kachayev Aug 25, 2023
47d681a
Merge branch 'master' into fix-line-search-zero-cost
kachayev Sep 1, 2023
96b3902
Merge branch 'master' into fix-line-search-zero-cost
rflamary Sep 6, 2023
25c7d88
Merge branch 'master' into fix-line-search-zero-cost
cedricvincentcuaz Oct 10, 2023
441a3a7
Merge branch 'master' into fix-line-search-zero-cost
rflamary Oct 11, 2023
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion ot/optim.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -297,7 +297,7 @@ def cost(G):
loop = 0

abs_delta_cost_G = abs(cost_G - old_cost_G)
relative_delta_cost_G = abs_delta_cost_G / abs(cost_G)
relative_delta_cost_G = abs_delta_cost_G / abs(cost_G) if cost_G != 0. else np.nan
if relative_delta_cost_G < stopThr or abs_delta_cost_G < stopThr2:
loop = 0

Expand Down
10 changes: 7 additions & 3 deletions test/test_gromov.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -8,10 +8,12 @@
# License: MIT License

import numpy as np
import pytest
import warnings

import ot
from ot.backend import NumpyBackend
from ot.backend import torch, tf
import pytest


def test_gromov(nx):
Expand Down Expand Up @@ -146,8 +148,10 @@ def test_gromov_dtype_device(nx):

C1b, C2b, pb, qb, G0b = nx.from_numpy(C1, C2, p, q, G0, type_as=tp)

Gb = ot.gromov.gromov_wasserstein(C1b, C2b, pb, qb, 'square_loss', G0=G0b, verbose=True)
gw_valb = ot.gromov.gromov_wasserstein2(C1b, C2b, pb, qb, 'kl_loss', armijo=True, G0=G0b, log=False)
with warnings.catch_warnings():
warnings.filterwarnings('error')
Gb = ot.gromov.gromov_wasserstein(C1b, C2b, pb, qb, 'square_loss', G0=G0b, verbose=True)
gw_valb = ot.gromov.gromov_wasserstein2(C1b, C2b, pb, qb, 'kl_loss', armijo=True, G0=G0b, log=False)

nx.assert_same_dtype_device(C1b, Gb)
nx.assert_same_dtype_device(C1b, gw_valb)
Expand Down