pi-thon 3.14 fixes by mimakaev · Pull Request #75 · open2c/polychrom

mimakaev · 2025-10-26T20:28:09Z

Description

This PR contains fixes for the python 3.14 switching from fork to forkserver multiprocessing method, addresses some deprecation warnings, and reformats things with black.

Changed CI to include 3.14 and two latest versions of MacOS
black and isort on the codebase
Switched contactmaps to be compatible with spawn and forkserver multiprocessing (latest default in 3.14)
removed confusing test for an old starting conformation generator
Addressed all flake8 issues and all valid VSCode warnings such as possibly unbound variables
Deprecated some unnecessary functions, like streaming ndarray agg
redid imports from simtk.unit due to dynamic module nature (pylance errors)

The PR also revisits topology preserving simulation, and adds a forcekit and a tested example (using claude fable model).

PR Checklist

[] apply "black" to the whole codebase (black .)
[] apply isort to the codebase (isort .)
[] fun flake8 and try to resolve all the issues (work in progress!)

Copilot

Pull Request Overview

This PR updates the polychrom codebase for Python 3.14 compatibility, focusing on multiprocessing method changes (fork → forkserver), addressing deprecation warnings, and applying code formatting with black/isort.

Key Changes:

Updated CI/CD to test against Python 3.14 and recent macOS versions
Refactored contactmaps module for spawn/forkserver multiprocessing compatibility (lambda functions → named functions)
Fixed NumPy deprecation (np.in1d → np.isin) and improved type checking patterns
Applied consistent code formatting across the codebase

Reviewed Changes

Copilot reviewed 24 out of 25 changed files in this pull request and generated 6 comments.

Show a summary per file

File	Description
`.github/workflows/pythonapp.yml`	Added Python 3.14 and macOS testing matrix
`polychrom/contactmaps.py`	Refactored for multiprocessing compatibility: moved lambdas to module-level functions, added default function parameters
`tests/test_contactmaps.py`	Replaced lambda with named `load_function` for spawn/forkserver compatibility; added module-level test data
`polychrom/forces.py`	Updated cylindrical_confinement logic and type hints; contains a bug with bottom=0 handling
`polychrom/forcekits.py`	Added explicit error for unsupported nonbonded forces (potential breaking change)
`tests/test_polymer_analyses.py`	Fixed NumPy deprecation: `np.in1d` → `np.isin`
`tests/test_io.py`	Updated to use new `load_URI` API instead of deprecated `fetch_block`
`polychrom/polymer_analyses.py`	Added TYPE_CHECKING pattern for better Cython import handling
`polychrom/param_units.py`	Reformatted docstrings with black (r""" prefix)
`polychrom/hdf5_format.py`	Changed `type()` to `isinstance()` for better practices
`polychrom/legacy/forces.py`	Reordered imports and changed to `isinstance()`
Various files	Import reordering (isort), spacing fixes, removed extra blank lines (black formatting)
`tests/test_configurations.py`	Removed entire test file for old starting conformation generator
Examples	Import reordering and formatting improvements

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Copilot · 2025-11-11T16:59:27Z

+    bottom = 0 if bottom is True else bottom

-    if bottom is not None:
+    if bottom:


The logic change here introduces a bug. When bottom=0 is passed, the condition if bottom: (line 734) will evaluate to False because 0 is falsy in Python, causing the code to use the wrong branch.

The original code correctly used if bottom is not None: which would handle bottom=0 properly. The fix should be:

bottom = 0 if bottom is True else bottom if bottom is not None: # ... existing code for bottom confinement

Suggested change

if bottom:

if bottom is not None:

Copilot · 2025-11-11T16:59:28Z

@@ -1 +1 @@
-from polychrom.polymerutils import *  # noqa: F403
+from polychrom.polymerutils import *  # noqa: F403


Import pollutes the enclosing namespace, as the imported module polychrom.polymerutils does not define 'all'.

Suggested change

from polychrom.polymerutils import * # noqa: F403

# TODO: Replace the following line with explicit imports of only the required names.

# For example: from polychrom.polymerutils import foo, bar

# from polychrom.polymerutils import * # noqa: F403

mimakaev · 2025-11-11T17:25:38Z

unused imports and other things will be addressed in a next PR

Copilot

Pull request overview

Copilot reviewed 25 out of 26 changed files in this pull request and generated 5 comments.

Comments suppressed due to low confidence (1)

polychrom/param_units.py:189

guess_bondWiggleDistance doesn’t use self but is now an instance method, which is a silent breaking change for any code calling it as a class/static utility (e.g. SimulationParams.guess_bondWiggleDistance(...)). Consider making it a @staticmethod again (or accepting both calling styles) to preserve the previous API.

    def guess_bondWiggleDistance(self, L0, b, mean_linker_length, a=None):
        """Return bond wiggle distance based on the amount of DNA per bead (L0), the
        Kuhn length (b) in basepairs, and the mean linker length in basepairs, and the
        expected radius of a monomer in nanometers (a)."""
        L0_nm = L0 / (1 + 146 / mean_linker_length) * 0.34
        b_nm = b / (1 + 146 / mean_linker_length) * 0.34
        if a is None:
            a = b_nm
        return np.sqrt(2 * L0_nm * b_nm / 3) / a

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Copilot · 2026-02-18T20:07:44Z

    for file in files:
        try:
            h5py.File(file, "r")
        except Exception:


list_URIs opens each HDF5 file with h5py.File(file, "r") but never closes it. Use a context manager to ensure file handles are released (or explicitly close) to avoid leaking descriptors during large directory scans.

Copilot · 2026-02-18T20:07:44Z

+import openmm  # if this fails, update openmm



simulation.py now hard-imports openmm with no fallback to simtk.openmm. Since OpenMM isn’t listed in requirements.txt and other modules still support the legacy import path, this can break environments that only have simtk.openmm installed. Consider restoring the previous try/except import fallback (or updating installation requirements consistently across the project).

Suggested change

import openmm # if this fails, update openmm

try:

import openmm # preferred OpenMM namespace

except ImportError: # fall back to legacy OpenMM package layout

from simtk import openmm # type: ignore[no-redef]

Copilot · 2026-02-18T20:07:44Z

+    except ImportError:
+        warnings.warn(
+            "C++ simplification module not available. " "Please compile the Cython extensions.", RuntimeWarning
+        )
+        return data


This only catches ImportError, but if _polymer_math failed to import at module import time (the top-level try/except does pass), calling _polymer_math.simplifyPolymer(...) will raise NameError, not ImportError, and won’t be handled. Consider setting _polymer_math = None when import fails and/or catching NameError here so the intended warning + fallback path works.

Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>

…ogy code - hdf5_format: write strings/lists-of-strings as original Python objects (vlen UTF-8), not numpy S/U conversions - fixes TypeError crash on lists of strings and the 64KB fixed-length attribute limit on older HDF5 - hdf5_format: continue_trajectory now compares block numbers, not array positions - no longer destroys blocks in non-contiguous trajectories; re-buffers all collateral blocks from deleted files - hdf5_format: list_URIs(read_error=False) warns when skipping unreadable files instead of silently truncating the trajectory - _polymer_math C: fix stale-N bug in _simplifyCpp (sweeps 2+ read garbage tail points and lost the ring-closure edge; output now contains only real points and topology is verified preserved against an independent Alexander-invariant implementation for 3_1/4_1/5_1/7_1 and lattice unknots) - _polymer_math C: getLinkingNumber returned -2x the true linking number (signed crossing sum, never halved, inverted sign); now matches the Gauss integral exactly (Hopf = +-1) - polymer_analyses: simplifyPolymer no longer swallows its own missing- extension ImportError (was silently returning unsimplified data); single _require_polymer_math() helper shared by all three topology functions - polymer_analyses: restore ndarray_groupby_aggregate and streaming_ndarray_agg (public documented API, deleted without deprecation) - contactmaps: exceptionsToIgnore=None no longer crashes with TypeError(tuple(None)) in findN/iterators; binnedContactMap default fixed - forcekits: raise ValueError when except_bonds is requested but the nonbonded force supports neither exceptions nor exclusions (was print-and-continue with silently wrong physics) - simulation: do_block docstring no longer promises a steps default that does not exist - CLAUDE.md: fix doBlock -> do_block, correct polymerutils.load() claims - tests: regression tests for string roundtrip, gapped continue_trajectory; polymer_math tests now assert exact invariants (|lk|=1 for Hopf, stick number bound for trefoil, no garbage rows) instead of codifying the bugs Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

- forcekits: add grosberg_polymer_chains - the Kremer-Grest/Halverson force set (FENE + WCA + bending) preconfigured correctly, most importantly with except_bonds=False: the polymer_chains default of excluding bonded pairs from the nonbonded force silently breaks FENE (its minimum is at r=0; the 0.97 sigma bond length comes from the WCA balance). Docstring carries the validated numerics (dt <= 82 fs langevinMiddle, collision_rate 0.079/ps, fixed timestep only, warmup protocol). No trunc parameter on purpose: truncated repulsion + FENE collapses and NaNs. - forces: correct grosberg_* docstrings - grosberg_polymer_bonds claimed a built-in repulsion it does not have; grosberg_angle claimed k=1.5 "maximizes entanglement length" when it reduces Ne to ~28 (that being the point); grosberg_repulsive_force trunc guidance replaced with measured crossing rates (trunc=3 is leaky, only trunc=None preserves topology). - polymer_analyses: add alexander_invariants(ring) - exact knot detection: |Alexander(-1)| and odd |Alexander(-2)| via integer Bareiss determinants on a generic projection, with degenerate-projection rejection, clearance-scaled perturbation, and two-projection agreement. Unknot=(1,1), trefoil=(3,7), 4_1=(5,11), 5_1=(5,31). Uses the (fixed) C simplifyPolymer for pre-reduction. Tested against the knot table and grow_cubic unknots. - examples/topologyPreservingRingMelt: executed notebook showcasing the whole workflow: unentangled ring-melt construction, parameter table with validated timestep/friction limits, simulation via the forcekit, knot + periodic-image-aware linking verification, and a crossable (trunc=1.5, harmonic bonds) positive control demonstrating detection. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

review-notebook-app · 2026-08-08T23:23:36Z

Check out this pull request on

See visual diffs & provide feedback on Jupyter Notebooks.

Powered by ReviewNB

mimakaev added 3 commits October 26, 2025 11:54

contactmaps 3.14 compatible; black/isort

b7f3c69

contactmaps multiprocessing friendly

a164043

legacy code README

8735dc3

mimakaev requested a review from Copilot November 11, 2025 16:56

Copilot started reviewing on behalf of mimakaev November 11, 2025 16:56 View session

Copilot finished reviewing on behalf of mimakaev November 11, 2025 16:58

Copilot AI reviewed Nov 11, 2025

View reviewed changes

test and type hints

c941c06

mimakaev force-pushed the mi/3.14_fixes branch from 74ce2eb to c941c06 Compare February 18, 2026 18:51

mimakaev added 6 commits February 18, 2026 13:52

black isort

5f34fc8

change compression ops to be faster

4547dd0

Merge branch 'master' into mi/3.14_fixes

a441dba

deprecate old code

dff3fa5

address vscode warnings

3831b64

docstrings to raw strings

2d8ac8e

mimakaev requested a review from Copilot February 18, 2026 20:00

Copilot started reviewing on behalf of mimakaev February 18, 2026 20:01 View session

Copilot AI reviewed Feb 18, 2026

View reviewed changes

mimakaev and others added 6 commits February 18, 2026 12:08

last updates

7740357

Update tests/test_starting_conformations.py

48404d8

Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>

Update tests/test_hdf5_format.py

6acac51

Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>

fix in list_URIs

06af7ca

mimakaev merged commit 1e9ac0f into master Aug 9, 2026
4 checks passed

Navigation Menu

Uh oh!

pi-thon 3.14 fixes - #75

pi-thon 3.14 fixes#75
mimakaev merged 16 commits into
masterfrom
mi/3.14_fixes

mimakaev commented Oct 26, 2025 •

edited

Loading

Uh oh!

Copilot AI left a comment

Uh oh!

Copilot AI Nov 11, 2025

Uh oh!

Uh oh!

Copilot AI Nov 11, 2025

Uh oh!

Uh oh!

Uh oh!

Uh oh!

mimakaev commented Nov 11, 2025

Uh oh!

Copilot AI left a comment

Uh oh!

Uh oh!

Uh oh!

Copilot AI Feb 18, 2026

Uh oh!

Copilot AI Feb 18, 2026

Uh oh!

Copilot AI Feb 18, 2026

Uh oh!

review-notebook-app Bot commented Aug 8, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

		@@ -1 +1 @@
		from polychrom.polymerutils import * # noqa: F403
		from polychrom.polymerutils import * # noqa: F403

-from polychrom.polymerutils import *  # noqa: F403
+# TODO: Replace the following line with explicit imports of only the required names.
+# For example: from polychrom.polymerutils import foo, bar
+# from polychrom.polymerutils import *  # noqa: F403

-import openmm  # if this fails, update openmm
+try:
+    import openmm  # preferred OpenMM namespace
+except ImportError:  # fall back to legacy OpenMM package layout
+    from simtk import openmm  # type: ignore[no-redef]

Uh oh!

Conversation

mimakaev commented Oct 26, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Description

PR Checklist

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull Request Overview

Reviewed Changes

Uh oh!

Copilot AI Nov 11, 2025

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Copilot AI Nov 11, 2025

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

mimakaev commented Nov 11, 2025

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

Uh oh!

Uh oh!

Copilot AI Feb 18, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Feb 18, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Feb 18, 2026

Choose a reason for hiding this comment

Uh oh!

review-notebook-app Bot commented Aug 8, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

mimakaev commented Oct 26, 2025 •

edited

Loading