Skip to content

Fix bug in transformed scan variables #6573

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 1 commit into from
Mar 7, 2023
Merged

Fix bug in transformed scan variables #6573

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion pymc/logprob/transforms.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -237,7 +237,7 @@ def transform_scan_values(fgraph: FunctionGraph, node: Node) -> Optional[List[No
return None

transforms = [
values_to_transforms.get(rv_map_feature.original_values[value], None)
values_to_transforms.get(rv_map_feature.original_values[value_var], None)
for value_var in value_vars
]

Expand Down
7 changes: 6 additions & 1 deletion tests/logprob/test_transforms.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -914,8 +914,13 @@ def test_scan_transform():
init = at.random.beta(1, 1, name="init")
init_vv = init.clone()

def scan_step(prev_innov):
next_innov = at.random.beta(prev_innov * 10, (1 - prev_innov) * 10)
update = {next_innov.owner.inputs[0]: next_innov.owner.outputs[0]}
return next_innov, update
Copy link
Member Author

@ricardoV94 ricardoV94 Mar 7, 2023 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Adding more (non-probability related) outputs to scan showcases the bug.

Updates are a common output in real scan uses, so this is both more realistic and covers the fixed bug


innov, _ = scan(
fn=lambda prev_innov: at.random.beta(prev_innov * 10, (1 - prev_innov) * 10),
fn=scan_step,
outputs_info=[init],
n_steps=4,
)
Expand Down