Skip to content

Fix v2 gap - #23

Open
jinzhta wants to merge 18 commits into
mir-group:fix_v2_gapfrom
jinzhta:fix_v2_gap
Open

Fix v2 gap#23
jinzhta wants to merge 18 commits into
mir-group:fix_v2_gapfrom
jinzhta:fix_v2_gap

Conversation

@jinzhta

@jinzhta jinzhta commented Oct 26, 2025

Copy link
Copy Markdown
Collaborator

[This is the version I used for training/running scider calculations (date: 2025-10-10 16:19)]

[This is the version I used for training scider models with sl-nst features (date: 2025-10-25 23:59)]

kylebystrom and others added 18 commits June 10, 2025 04:05
…n fill_feat_ to avoid training paused (seen in execution_train_2025-04-22-02-42-39)
…tements. This version works for training complete band-gap scider model (trained with molecules and solids data), see execution_train_2025-05-16-05-38-54; But different from misc_fixes_and_misc branch commit ab4295a because this commit uses dbaseline[orb] += np.sum(da[s] * drho_tmp) in MOLGP2 instead of dbaseline[orb] += np.sum(da[s] * drho_tmp * wt)
….py to avoid 'NLDF exponent is too large' error when running cider dft calculation
… libxc_baselines.c and add arguments in different functions in Kernel related functions to pass omega all the way to libxc) and re-cmake
…camke again with new CMakeLists.txt (compared with fix_sl_feat branch) and link ciderpress to my local libxc-6.2.2 instead of downloading libxc-7.0.0
… [2,18) in function get_psetup_func_counts in ciderpress.gpaw.atom_utils.py
…sl feature (originally omega feature only works for sl-NPA), this change is companied with changes in orchard.scripts.train_mol_nes.py
…patiable with NST sl feature since alpha + s^2 = tau/tau_0
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants