Merge branch 'main' into dependabot/pip/ase-3.29.0

Author: lbluque

.github/workflows/build_docs.yml

@@ -42,6 +42,12 @@ jobs:
             github.event.pull_request.head.sha ||
             github.sha
           }}
+        # Required since actions/checkout began refusing fork PR code in
+        # workflow_run jobs. Safe here: fork PRs from all external
+        # contributors require maintainer approval before this trusted,
+        # secret-holding run can execute (repo Actions setting).
+        # See https://gh.io/securely-using-pull_request_target
+        allow-unsafe-pr-checkout: true
         # Optional: get full history if needed
         fetch-depth: 0
 
@@ -99,4 +105,4 @@ jobs:
       if: always()
       uses: ./.github/actions/multi-trigger-cleanup
       with:
-          status-context: '${{ github.job }}'
+          status-context: '${{ github.job }}'

.github/workflows/test.yml

@@ -72,6 +72,12 @@ jobs:
         uses: actions/checkout@v7
         with:
           ref: ${{ env.HEAD_SHA }}
+          # Required since actions/checkout began refusing fork PR code in
+          # workflow_run jobs. Safe here: fork PRs from all external
+          # contributors require maintainer approval before this trusted,
+          # secret-holding run can execute (repo Actions setting).
+          # See https://gh.io/securely-using-pull_request_target
+          allow-unsafe-pr-checkout: true
           # Optional: get full history if needed
           fetch-depth: 0
 
@@ -141,7 +147,7 @@ jobs:
 
       - if: ${{ matrix.python_version == '3.13' }}
         name: codecov-coverage
-        uses: codecov/codecov-action@v6
+        uses: codecov/codecov-action@v7
         with:
           fail_ci_if_error: false
           files: ./coverage.xml
@@ -183,6 +189,9 @@ jobs:
         uses: actions/checkout@v7
         with:
           ref: ${{ env.HEAD_SHA }}
+          # See the checkout step in the `test` job for why this is set
+          # and why it is safe (external-contributor approval gate).
+          allow-unsafe-pr-checkout: true
           fetch-depth: 0
 
       - name: Setup
@@ -290,6 +299,12 @@ jobs:
         uses: actions/checkout@v7
         with:
           ref: ${{ env.HEAD_SHA }}
+          # Required since actions/checkout began refusing fork PR code in
+          # workflow_run jobs. Safe here: fork PRs from all external
+          # contributors require maintainer approval before this trusted,
+          # secret-holding run can execute (repo Actions setting).
+          # See https://gh.io/securely-using-pull_request_target
+          allow-unsafe-pr-checkout: true
           # Optional: get full history if needed
           fetch-depth: 0
 
@@ -369,6 +384,9 @@ jobs:
         uses: actions/checkout@v7
         with:
           ref: ${{ env.HEAD_SHA }}
+          # See the checkout step in the `test` job for why this is set
+          # and why it is safe (external-contributor approval gate).
+          allow-unsafe-pr-checkout: true
           fetch-depth: 0
 
       - name: Setup
@@ -446,6 +464,12 @@ jobs:
         uses: actions/checkout@v7
         with:
           ref: ${{ env.HEAD_SHA }}
+          # Required since actions/checkout began refusing fork PR code in
+          # workflow_run jobs. Safe here: fork PRs from all external
+          # contributors require maintainer approval before this trusted,
+          # secret-holding run can execute (repo Actions setting).
+          # See https://gh.io/securely-using-pull_request_target
+          allow-unsafe-pr-checkout: true
           # Optional: get full history if needed
           fetch-depth: 0
 

src/fairchem/applications/fastcsp/README.md

@@ -204,9 +204,11 @@ FastCSP uses YAML configuration files to control all workflow parameters. Exampl
   `vars.{Z, spg_distribution_type, num_structures_per_spg, read_z_from_file,
   read_spg_from_file}`) and SLURM block
 - `pre_relaxation_filter`: Pre-ML deduplication
-  (`assign_groups`, `remove_duplicates`, `ltol`/`stol`/`angle_tol`,
+  (`assign_groups`, `remove_duplicates`, `remove_problematic`, `ltol`/`stol`/`angle_tol`,
   `bin_by_conf`/`bin_by_z`/`bin_by_spg`, `density_bin_size`, `density_tol`,
-  `apply_niggli_filter`, `npartitions`)
+  `apply_niggli_filter`, `npartitions`). Set `remove_problematic: true` to drop structures whose
+  generation-time validity flags (`correct_z`, `molecule_matches_reference`) are False before
+  relaxation.
 - `relax`: ML relaxation settings
   (`calculator`, `optimizer`, `fmax`, `max_steps`, `fix_symmetry`,
   `relax_cell`, `write_traj`, `traj_interval`) and SLURM block

src/fairchem/applications/fastcsp/core/README.md

@@ -87,6 +87,7 @@ genarris:
 pre_relaxation_filter:
   assign_groups: true        # run dedup blocker
   remove_duplicates: true    # drop all-but-one per group (rep = closest-to-median density)
+  remove_problematic: false  # drop structures whose generation-time validity flags are False (default: false)
   ltol: 0.3                  # StructureMatcher tolerances (looser than post-relax)
   stol: 0.4
   angle_tol: 5

src/fairchem/applications/fastcsp/core/configs/example_config.yaml

@@ -47,6 +47,8 @@ pre_relaxation_filter:
   assign_groups: true        # run dedup blocker + assign group_index. Default false
   remove_duplicates: true    # drop all-but-one per group (representative = closest
                              # to the group-median density_generated). Default false
+  remove_problematic: false  # Remove structures that failed generation-time validity
+                             # checks (correct_z, molecule_matches_reference)
   # StructureMatcher tolerances (looser than post-relax)
   ltol: 0.3                  # lattice tolerance.  Default 0.2
   stol: 0.4                  # site tolerance (Å). Default 0.3

src/fairchem/applications/fastcsp/core/utils/deduplicate.py

@@ -13,15 +13,18 @@
 import os
 from collections import defaultdict
 from functools import partial
+from typing import TYPE_CHECKING
 
 import numpy as np
-import pandas as pd  # Noqa: TCH002
 from fairchem.applications.fastcsp.core.utils.logging import get_central_logger
 from fairchem.applications.fastcsp.core.utils.structure import get_structure_group
 from p_tqdm import p_map
 from pymatgen.analysis.structure_matcher import StructureMatcher
 from tqdm import tqdm
 
+if TYPE_CHECKING:
+    import pandas as pd
+
 
 def process_structure_group(
     group_data,