ENH: Allow adjust=False when times is provided

[tserrao]

• [ x ] closes #54328 (partially)
• [ x ] Tests added and passed if fixing a bug or adding a new feature
• [ x ] All code checks passed.
• [ x ] Added type annotations to new arguments/methods/functions.
• [ x ] Added an entry in the latest doc/source/whatsnew/v3.0.0.rst

This change enables EWMA to be calculated recursively ("infinite history", i.e. adjust=False) for irregular-interval time series. Previously new_wt was held constant in the ewm run loop, but now new_wt is updated each step through the series under the narrow condition that adjust=False and com=1.

There may be a more elegant expression that will generalize to com values different than one, but using 1 - old_wt works nicely if halflife is the only provided decay parameter since we are guaranteed com=1. The additional parameter restrictions to force this condition seem acceptable since there is already some restriction on decay parameters.

Exactly one of com, span, halflife, or alpha must be provided if times is not provided. If times is provided, halflife and one of com, span or alpha may be provided.

FWIW I confirmed that results match polars and @azmyrajab's nice test function:

ema_no_adjust = ema_test(vals, dt_seconds, half_life=half_life_seconds)[-1] ema_no_adjust_pl = ema_polars(df["val"].rename_axis(index="ts").rename("val"), half_life, by="ts").select("val")[-1].item() print(ema_no_adjust) print(ema_no_adjust_pl) print(df.ewm(halflife=half_life, times=id, adjust=False).mean().iloc[-1]['val']) 0.2062994740159002 0.2062994740159002 0.2062994740159002 

[tserrao]

pre-commit is finally green but a lot of the actions are failing with the same message:

Error: Failed to CreateArtifact: Received non-retryable error: Failed request: (409) Conflict: an artifact with this name already exists on the workflow run

Seeing lots of PRs get these errors. Maybe a recent GHA setup change is breaking some tests?

[tserrao]

CI all green :)

[mroeschke]

Thanks @tserrao