Taking generators as an example, in order to enable "scalar" attributes (e.g. capital_cost) to vary over investment periods, one has to loop over the model investment periods and add a unique generator with the capital_cost value for that year as a float and apply the build_year:

capital_cost_by_period = {2025: 10.0, 2030: 20.0}
for period in capital_cost_by_period:
    n.add("Generator", name=f"my-generator-{period}", capital_cost=capital_cost_by_period[period], build_year=period)

This is a somewhat clunky user experience because "build_year" implies something not actually being built until that year, but it's being used for other situations like variable costs.

Could it be possible instead for the API to allow passing a pandas series indexed against the investment periods for these "scalar" attributes?

capital_cost_by_period = pd.Series([10.0, 20.0], index=pd.Index([2025, 2030], name="period"))  # index name matches multiindex level name for investment periods
n.add("Generator", name="my-generator", capital_cost=capital_cost_by_period)

In the case where you are using n.add with its implicit multi-add feature, you would pass a dataframe with the same index but columns with names matching each generator name. I suspect there may be ambiguities with the existing multi-add functionality to deal with here...