[FBumann]

Prompt: Lets scan the pysa files about those missing values. And judge which should get 0, 1 or absent. As a stacked PR

Note

The following content was generated by AI.

Of the 73 PyPSA parameters that declare missing:, 61 change from 0 to absent. For each of them, every reading gives the same rows either way. Four keep 0, eight keep their reading, and none needs 1. Stacked on #894.

The judgement, by how the files read each parameter

Parameters	Read as	0 and absent	Now
the 18 *_set schedules and builds (Generator_p_set, Line_s_nom_set, StorageUnit_state_of_charge_set, …)	a bare name in the where: of the row that fixes the value, and that row's right side	a bare number in a where: is false at a missing row under both, and a real 0.0 is a row under both, so the rows are the same, and the missing value is never read	absent
the 16 weights of the global constraints (*_primary_energy_weight, *_operational_limit_weight, *_tech_capacity_weight, *_volume_weight, *_expansion_cost_weight) and the 5 cycle and outage weights (Line_cycle_weight, Transformer_cycle_weight, Transformer_phase_shift_weight, the two *_BODF)	a factor of a summand in a sum, or a coefficient under a where: that names it	absent drops the summand and keeps the row; 0 keeps a term with no coefficient, which specsolve does not build	absent
the 12 ramp limits, Line_v_ang_max, Transformer_v_ang_max, Carrier_max_growth ("no value means no limit")	a bare name in a where: or a when:, and a value under it	false at a missing row under both, so the value is never read	absent
the 3 *_maintenance_duration	only in assumptions, under where: *_maintainable	a comparison is false at a missing row under both, so a maintainable unit without one fails the assumption either way	absent
CVaR_alpha	the divisor 1 - CVaR_alpha, under where: CVaR_omega	read only where data prep writes it beside CVaR_omega	absent
the 3 *_big_m	a coefficient in the com-ext rows, under where: *_com_ext	0 would force p <= 0 at a missing row, and absent would drop the row. Neither is right: data prep writes one for each committable, extendable unit (get_committable_big_m_values)	absent, with an assumption holds: *_big_m where *_com_ext
the 3 *_p_init	status_initial * p_init, the output carried into the first snapshot	a unit that came in off ramps from 0 * p_init, and absent would drop that row. PyPSA's default p_init is 0	0
CVaR_omega	a factor of the cost expression (1 - CVaR_omega) * … + CVaR_omega * CVaR	absent would drop that expression wherever no risk preference is set. 0 is risk-neutral, PyPSA's default	0
GlobalConstraint_counts_snapshot	a bool flag	false is a snapshot the row does not count	false
the 7 *_nom_mod	a divisor	0 is refused at load as a divisor	absent, as in #894

No parameter here takes 1. The values a PyPSA default of 1 would fill, such as efficiency and p_max_pu, declare no missing:. They stay refused, because data prep writes PyPSA's default into every row.

Verification

Method. A script listed every place a fragment reads each of the 73 parameters: the entry, the field and the text. I judged each parameter against the table "What a missing row reads as" in absence.md, from the position of each reading. Then a second script rewrote the 61 missing: 0 lines, in the fragments and in examples/pypsa.yaml alike.

Gates on 6c6afb9, which merges #894's head 2d7355f, and so main with #898, #899, #900, #901, #904, #907, #909 and #911:

• pixi run lint passes.
• pixi run test passes: 3077 passed, 1 skipped. test_the_fragments_merge_to_the_one_file holds, so the fragments and the one file agree.
• pixi run docs-build passes.
• compile-tex was not run, because the container cannot download the TeX bundle. CI runs it.
• The merge had no conflict. Every generator was re-run on it. Only four gallery pages changed: pypsa.md and the three *_commitment.md pages. In those, the big-M assumptions print their condition's indices sorted by name, as fix(language): two files that declare their dimensions in different orders load to the same program #911 prints every derived frame.

Not checked. I did not build the PyPSA ladder in specsolve with this mathspec, because specsolve pins a mathspec release. The claim that the rows stay the same rests on the table in absence.md, and on specsolve already keeping a missing: 0 table sparse.

Not done

• efficiency, p_max_pu and the other parameters with a PyPSA default stay refused, which brynpickering's review on feat(language)!: a missing row reads as the value the file names, and missing: neutral is gone #894 asks about.
• No mathspec test fails without the three big-M assumptions. specsolve checks an assumption against the data, so the test for them belongs in its PyPSA ladder.

🤖 Generated with Claude Code

https://claude.ai/code/session_01A2crTbivTGt2WzGpTebdmf

[read-the-docs-community]

Documentation build overview

📚 mathspec | 🛠️ Build #35288864 | 📁 Comparing 6c6afb9 against latest (18207f1)

  🔍 Preview build  

33 files changed · ± 33 modified

± Modified

• reference/changelog/index.html
• reference/glossary/index.html
• reference/language/index.html
• reference/notation/index.html
• reference/program/index.html
• reference/reading/index.html
• examples/pypsa/carrier/index.html
• examples/pypsa/cost/index.html
• examples/pypsa/generator/index.html
• examples/pypsa/generator_commitment/index.html
• and 23 more...

[FabianHofmann]

good catch @FBumann