Skip to content
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.

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

Add output_symbols arg to codegen as performance mitigation #361

Closed

Conversation

gareth-cross
Copy link
Contributor

For complex expressions, the calculation of Codegen.output_symbols can grow quite large (encroaching on a minute or more). In some cases, the value of output_symbols can be specified externally. For example:

  • The user may have a list of output_symbols readily available at the callsite (frequently it matches the input symbol set exactly).
  • The output_symbols can be determined from another Codegen object, for example in most (though not all) cases the following can work:
cg = Codegen(...)
cg_factor = cg.with_linearization(output_symbols=cg.output_symbols) # avoid recomputation

Ideally free_symbols would be faster, but this is a mitigation/workaround that can offer worthwhile savings in some cases.

This change adds an optional argument to the various constructors of Codegen so that the user can optionally provide the symbols themselves, and avoid an expensive computation.

Copy link
Contributor

@chao-qu-skydio chao-qu-skydio left a comment

Choose a reason for hiding this comment

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

lgtm

# Convert to Matrix before calling free_symbols because it's much faster to call once
return sf.S(sf.Matrix(codegen_util.flat_symbols_from_values(self.outputs)).mat).free_symbols
@staticmethod
def _compute_output_symbols(output_values: Values) -> T.Set[sf.Symbol]:
Copy link
Contributor

Choose a reason for hiding this comment

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

Prefer to add back the comments stating that this is slow.

@@ -159,6 +164,12 @@ def __init__(
self.inputs = inputs
self.outputs = outputs

if output_symbols is None:
# compute the output symbols from the expressions:
Copy link
Contributor

Choose a reason for hiding this comment

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

Why do you end sentences in ":"? 🦀

@@ -828,6 +838,7 @@ def with_jacobians(
include_results: bool = True,
name: str = None,
sparse_jacobians: bool = False,
output_symbols: T.Optional[T.Set[sf.Symbol]] = None,
Copy link
Member

Choose a reason for hiding this comment

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

I think this one still needs to be forwarded to the Codegen constructor below?

@gareth-cross gareth-cross closed this by deleting the head repository Aug 23, 2024
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants