Closed Batch21 closed 2 weeks ago
switching the units to milliseconds simplifies the code a bit and IMO the precision of using nanoseconds was probably unnecessary. What do you think?
I've removed the unused error.
I don't think the reference to nanoseconds in generate_timesteps_from_frequency
can be removed completely as the polars Duration
uses them internally to store a proportion of the stochastics. We could convert to milliseconds when creating the new datetime, but this seem a bit unnecessary:
+ TimeDelta::milliseconds(duration.nanoseconds() / NANOSECS_IN_MILLISECOND)
Main change is to switch to using milliseconds as the main unit for comparison and resampling. polars is also updated to v0.40.0