Resource-limited optimization

See the Japan report dictionary for every output column, annualization rule, and failure value.

This example combines state-region resource potential, arable land, banned non-economic building groups, and an era-5 technology cap. It uses selected Japanese states to demonstrate a reusable GDP optimization under autarky for employment caps from one million to one hundred million.

employment_cap	annual_gdp	employment	gdp_per_capita	construction_cost	arable_land_consumption
1000000	42703087.87793275	999999.9999999998	42.70308787793276	170660.96630680584	0.0
2000000	73526769.38362674	1999999.9999999998	36.76338469181338	547375.6423712291	0.0
3000000	103472467.60647747	2999999.9999999995	34.4908225354925	1005828.0850718003	0.0
4000000	133361640.9064289	4000000.0	33.340410226607226	1392806.377528916	0.0
5000000	163196505.01788625	4999999.999999999	32.63930100357726	1817967.0034367465	0.0

Download artifacts/analyses/optimization/japan/employment_sweep.csv · View examples/optimize_jap.py

uv run python -m examples.optimize_jap

How the constraints compose

flowchart LR
    regions[Selected Japanese states] --> resources[Resource building limits]
    regions --> land[Arable land cap]
    resources --> scenario[Scenario]
    land --> scenario
    employment[Employment cap sweep] --> scenario
    scenario --> solver[NominalOptimizer]
    solver --> results[GDP and building levels]

The important implementation pattern is that state-region helpers return ordinary limits consumed by Scenario; the optimizer does not need a country-specific code path. See the optimization guide and optimization API.