fix: preserve per-period sizes in transform.fix_sizes() for multi-period models

[FBumann]

Fixes #695

Problem

flow_system.transform.fix_sizes() crashed with

ValueError: can only convert an array of size 1 to a Python scalar

on any multi-period model, because it collapsed every solution size to a Python scalar via float(sizes[size_var].item()). In multi-period models, investment sizes carry a period dimension (one size per period unless linked_periods is used), so .item() fails. The same applies to models with a scenario dimension.

This broke the documented two-stage workflow (size on a reduced model → fix_sizes() → dispatch at full resolution) for multi-period systems, e.g. the multi-period clustering notebook.

Fix

Keep sizes that have period/scenario dimensions as DataArrays so each coordinate retains its own fixed size; only collapse 0-d arrays to plain floats. InvestParameters.fixed_size already accepts per-period/per-scenario data (it is fitted with dims=['period', 'scenario']), so no further changes are needed downstream.

Verification

• New regression test test_fix_sizes_preserves_per_period_sizes in tests/test_math/test_multi_period.py: a 2-period sizing problem with per-period demand peaks (50 / 80) is solved, sizes are fixed, and the dispatch stage must reproduce the same per-period sizes [50, 80] and objective (1700) — including through NetCDF save/reload roundtrips via the optimize fixture.
• tests/test_math/test_multi_period.py: 33/33 pass.
• Scalar dict input (fix_sizes({'Boiler(heat)': 100})) still works as before.
• Pre-existing failures in tests/test_math/test_flow_invest.py (invest retirement / non-mandatory tests) occur on main without this change and are unrelated.

🤖 Generated with Claude Code

Summary by CodeRabbit

• Bug Fixes

  • Improved multi-period and multi-scenario optimization to correctly preserve period and scenario-specific investment sizes during sizing operations, preventing loss of granular optimization results.

• Tests

  • Added regression test verifying that per-period investment sizes and objectives are maintained across optimization stages in multi-period workflows.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro

Run ID: 18ef7878-af69-44f8-98a3-8c1d83b91808

📥 Commits

Reviewing files that changed from the base of the PR and between a825371 and 36fc283.

📒 Files selected for processing (2)

• flixopt/transform_accessor.py
• tests/test_math/test_multi_period.py

---

📝 Walkthrough

Walkthrough

TransformAccessor.fix_sizes() is updated to preserve period and scenario dimensions in size arrays instead of collapsing them to Python scalars. The docstring clarifies this behavior, the implementation conditionally handles scalar vs. non-scalar inputs, and a regression test validates per-period size preservation in a two-stage multi-period optimization workflow.

Changes

fix_sizes() Per-Period Size Preservation

Layer / File(s)	Summary
fix_sizes() documentation and implementation flixopt/transform_accessor.py	Docstring updated to clarify that period/scenario dimensions in input sizes are preserved. Implementation changes from always collapsing sizes via .item() to conditionally: scalars are converted to float, non-scalar arrays are passed through as fixed_value objects.
Multi-period regression test for fix_sizes() tests/test_math/test_multi_period.py	New test test_fix_sizes_preserves_per_period_sizes validates a two-stage workflow: stage 1 sizes/optimizes a multi-period model with period-dependent demand, then stage 2 applies fix_sizes() and re-optimizes dispatch, asserting per-period sizes and objective are preserved.

Estimated code review effort

🎯 3 (Moderate) | ⏱️ ~20 minutes

Possibly related issues

• [BUG] transform.fix_sizes() fails on multi-period models #695: Addresses the same bug in TransformAccessor.fix_sizes() by removing .item()/float(...) collapsing and preserving per-period/scenario arrays instead of reducing them to Python scalars.

Poem

A rabbit hops through periods bright,
With sizes saved, dimension-wise,
No collapse to scalar's plight,
Each phase gets its own size prize,
Multi-stage dreams take their flight! 🐰✨

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title clearly and specifically summarizes the main change: fixing preserve per-period sizes in the fix_sizes() function for multi-period models, which directly matches the core change in the pull request.
Description check	✅ Passed	The PR description comprehensively covers the problem, fix, and verification sections with specific examples and test results, though it does not strictly follow the template's checkbox format.
Docstring Coverage	✅ Passed	Docstring coverage is 100.00% which is sufficient. The required threshold is 80.00%.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

📝 Generate docstrings

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch fix/fix-sizes-multi-period

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}