easy fixes

[shimwell]

Easy Fixes

1. Spelling Typos (9 occurrences)

"begining" → "beginning"

• openmc/data/function.py:171

"Accomodate" → "Accommodate" (8 occurrences)

• openmc/mgxs/mgxs.py:1167, 1280, 2338, 4668, 5756, 6146
• openmc/mgxs/mdgxs.py:1386, 2354

2. Anti-pattern Type Checks

Using isinstance(x, type(None)) instead of is None

• openmc/deplete/stepresult.py:583

Current:

if isinstance(op_results.k, type(None)):

Better:

if op_results.k is None:

Using type() comparison instead of isinstance()

• openmc/tallies.py:3359

Current:

if type(self_filter) == filter_type:

Better:

if isinstance(self_filter, filter_type):

3. Overly Broad Exception Handling

Two locations catch generic Exception which could be more specific:

• openmc/config.py:205
• openmc/data/multipole.py:1116

[shimwell]

Quick wins

helpers.py: DirectReactionRateHelper.get_material_rates currently loops over product(nuc_index, rx_index) and assigns one by one. If the tally order is (nuclide, reaction) (it usually is), you can reshape once and do a single slice assignment (or np.reshape + np.ix_) to eliminate the Python loop.
helpers.py: FluxCollapseHelper.get_material_rates uses list.index inside nested loops for direct‑tally nuclides/reactions. Precompute dicts {score: idx} and {nuclide: idx} once (per call or cached) to avoid O(n^2) index lookups.
Likely wins, still small changes

chain.py: add_redox_term densifies the matrix (toarray()) and grows redox_change with np.append inside a loop. Replace with sparse math: redox_change = matrix.T.dot(os) (or os @ matrix) and avoid toarray() unless strictly needed. This removes O(n^2) dense work and repeated reallocations.
chain.py: form_rr_term and form_ext_source_term call get_components and get_external_rate for every nuclide. Precompute the component set and any per‑component rates once per (material, timestep) and then just lookup inside the loop.
Minor but free

stepresult.py: get_material sorts self.index_nuc every call. Cache a sorted list of nuclide names when allocate() is called and reuse it.

Bigger impact if you run many steps

pool.py: deplete creates a multiprocessing.Pool each call. If you have many depletion steps, pool startup dominates. A small refactor to keep a persistent pool per run (or allow the integrator to own a pool) can save a lot of overhead.
If you want, I can pick 1–2 of these and implement them with minimal behavior change.