[DISCUSSION] `edge` arg in shift may reference a parameter

[FabianHofmann]

What can be improved?

I am not sure whether there is a work-around for this and whether latex conversion might be too difficult, but I am filing to not lose track of it.

We need some way to insert a varying edge in the shift function. Like,

shift(status, over=snapshot, offset=1, edge=status_initial)
shift(soc,    over=snapshot, offset=1, edge=soc_initial)

It inherits the dim rule the other three value keyword args already have. the parameter must not span the dimension being translated.

Version

v0.0.0

[FBumann]

Might be obsolete and more readable if we do this with cased expressions #2 and #70

Like this

expressions:
  previous_soc:
    foreach: [snapshot, storage]
    cases:
      cyclic_boundary:
        when: "cyclic and position(snapshot) == 0"
        expression: shift(soc, over=snapshot, offset=1, edge='wrap')
      boundary:
        when: "position(snapshot) == 0"
        expression: soc_initial
      interior:
        expression: shift(soc, over=snapshot, offset=1)

constraints:
  storage_balance:
    foreach: [snapshot, storage]
    expression: soc == previous_soc * (1 - loss) + charge - discharge

[FBumann]

We might still add this, but not urgent i think.

[FabianHofmann]

We might still add this, but not urgent i think.

agree!

[FBumann]

Note

The following content was generated by AI.

Closed by #300. Link delay lands as rung 16, but not through this issue's shape: edge=<parameter> gives a per-link number, and wrap is a quoted keyword, never a value — so a parameter could never have carried the wrap/lose choice. That choice is a branch, carried by the shipped cases: (#168): a cases: on cyclic_delay picks edge='wrap' or edge=0 over shift(offset=delay).

Reproduces PyPSA where snapshot_weightings.generators is uniform. The weighted-duration case (a resample that varies along the snapshot axis, above shift's positional reach) is split out to #299.