ENH: add trip scheduling choice explicit chunking#1041
Open
m-richards wants to merge 1 commit into
Open
Conversation
m-richards
commented
Mar 10, 2026
| # check that tours with no inbound stops have zero inbound duration | ||
| assert out_tours[tsc.IB_DURATION].mask(in_tours[tsc.HAS_IB_STOPS], 0).sum() == 0 | ||
|
|
||
| # confirm explicit chunking is supported and doesn't affect results |
Author
There was a problem hiding this comment.
I can convert this to a parametrised test in the same way as test_agg_accessibility_orig_land_use, I'm not sure if that's preferred or not
m-richards
force-pushed
the
matt/trip_scheduling_choice_explicit_chunking
branch
from
d26da46 to
9aec46f
Compare
Contributor
There was a problem hiding this comment.
Pull request overview
This PR wires ActivitySim’s explicit chunking mechanism into the trip_scheduling_choice model step to reduce memory peaks (per #1040), and adds a regression test intended to confirm that explicit chunking is accepted and does not change results.
Changes:
- Pass
explicit_chunk_size=model_settings.explicit_chunkintochunk.adaptive_chunked_chooserswithinrun_trip_scheduling_choice. - Add
explicit_chunktoTripSchedulingChoiceSettingsso it can be configured via YAML. - Add a unit test intended to validate chunking support and result stability.
Reviewed changes
Copilot reviewed 2 out of 2 changed files in this pull request and generated 1 comment.
| File | Description |
|---|---|
activitysim/abm/models/trip_scheduling_choice.py |
Plumbs explicit_chunk through to adaptive_chunked_choosers and exposes the setting via the component settings model. |
activitysim/abm/test/test_misc/test_trip_scheduling_choice.py |
Adds a test asserting equality between unchunked and “explicit chunked” runs (but currently doesn’t actually exercise explicit chunking). |
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
Comment on lines
+290
to
+308
| # confirm explicit chunking is supported and doesn't affect results | ||
| model_settings_explicit_chunk = tsc.TripSchedulingChoiceSettings( | ||
| **{ | ||
| "SPEC": "placeholder.csv", | ||
| "compute_settings": { | ||
| "protect_columns": ["origin", "destination", "schedule_id"] | ||
| }, | ||
| } | ||
| ) | ||
| out_tours_chunked = tsc.run_trip_scheduling_choice( | ||
| state, | ||
| model_spec, | ||
| tours, | ||
| skims, | ||
| locals_dict, | ||
| trace_label="PyTest Trip Scheduling", | ||
| model_settings=model_settings_explicit_chunk, | ||
| ) | ||
| assert_frame_equal(out_tours, out_tours_chunked) |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Closes #1040
Based on discussion in #1040 - this model is still being used in the DTP activitysim implementation and with certain other tweaks on their activitysim fork, this step will out of memory on 128GB hardware.
This is pretty simple to wire up. I've added a basic test which shows the parameter is accepted (but it would also pass if I hadn't wired up chunk.adaptive_chunked_choosers, so it's not ideal)
I'm not quite sure what the testing philosophy is at the moment - I've read #1038 and the google doc, not exactly sure what's pragmatic with a small change like this - given there aren't a host of other tests testing for explicit chunking.