search

DiffEqML / torchdyn

A PyTorch library entirely dedicated to neural differential equations, implicit models and related numerical methods
https://torchdyn.org
Apache License 2.0
1.33k stars 124 forks source link

Error control #165

Closed Zymrael closed 1 year ago

Zymrael commented 1 year ago

Resolves #164.

Slightly faster than before (left, torchdyn, right torchdiffeq for test_sensitivity.py configurations):

2022-08-22 06:38:59 [    INFO] Fwd times: 0.065, 0.008 (test_sensitivity.py:58)
2022-08-22 06:38:59 [    INFO] Bwd times: 0.231, 0.330 (test_sensitivity.py:70)
2022-08-22 06:39:00 [    INFO] Fwd times: 0.063, 0.008 (test_sensitivity.py:58)
2022-08-22 06:39:01 [    INFO] Bwd times: 0.283, 0.327 (test_sensitivity.py:70)
2022-08-22 06:39:01 [    INFO] Fwd times: 0.063, 0.008 (test_sensitivity.py:58)
2022-08-22 06:39:02 [    INFO] Bwd times: 0.222, 0.327 (test_sensitivity.py:70)
2022-08-22 06:39:02 [    INFO] Fwd times: 0.063, 0.008 (test_sensitivity.py:58)
2022-08-22 06:39:03 [    INFO] Bwd times: 0.284, 0.326 (test_sensitivity.py:70)
2022-08-22 06:39:04 [    INFO] Fwd times: 0.063, 0.009 (test_sensitivity.py:58)
2022-08-22 06:39:04 [    INFO] Bwd times: 0.221, 0.327 (test_sensitivity.py:70)
2022-08-22 06:39:05 [    INFO] Fwd times: 0.063, 0.009 (test_sensitivity.py:58)
2022-08-22 06:39:05 [    INFO] Bwd times: 0.282, 0.326 (test_sensitivity.py:70)
2022-08-22 06:39:06 [    INFO] Fwd times: 0.063, 0.011 (test_sensitivity.py:58)
2022-08-22 06:39:07 [    INFO] Bwd times: 0.221, 0.326 (test_sensitivity.py:70)
2022-08-22 06:39:07 [    INFO] Fwd times: 0.063, 0.010 (test_sensitivity.py:58)
2022-08-22 06:39:08 [    INFO] Bwd times: 0.283, 0.327 (test_sensitivity.py:70)