Skip to content

Add support for simplifications based on symmetry#160

Merged
apozas merged 63 commits intodevelfrom
pedro_symmetries
Apr 21, 2025
Merged

Add support for simplifications based on symmetry#160
apozas merged 63 commits intodevelfrom
pedro_symmetries

Conversation

@apozas
Copy link
Collaborator

@apozas apozas commented Apr 21, 2025

This PR implements the (optional) exploitation of symmetries in the target distribution for variable reduction. The most important additions are InflationProblem.add_symmetries() to add custom symmetries expressed as permutations of the lexicographical order, and symmetry_utils for helper functions. Documentation and tests are ready.

apozas and others added 30 commits June 3, 2024 09:02
Merge pull request #136 from ecboghiu/devel
Merge pull request #138 from ecboghiu/update-website
Update intro to be the same as in the documentation
Fix print of scenario information for hybrid scenarios
Update the documentation website to v2.0.1
Update the documentation website to v2.0.1
We see that the error stems from [A^{1,1}, B^{2,2}] which factorizes being indistinguishable from [A^{1,1}, A^{2,2}], despite this not following directly from symmetry. Perhaps a feature, not a bug.
…h are equal only after accounting for factorization.
…ful symmetries. Currently experience a bug in sdp_utils that should have been fixed...
…d outcome-relabelling symmetry assumptions (which essentially define a unique nonclassical distribution)
…DP even when outcome relabelling symmetries are present
…y array which is actual less efficient for iteration.

Fixed issue caused by prior commit #0317f943a562f2ec7669a16030d02b2c584ac86f
Elie Wolfe and others added 28 commits December 2, 2024 21:39
…tors coming from outcome relabelling (and private setting relabelling)
This simplifies the code and reduces points of failure
In this way we just compute it once. It is useful if we consider creating a single InflationProblem and recycling it for different symmetries
This is a function that is pretty standalone, and does not need to be inside InflationProblem
This has made me realize that lexorder is different in InflationLP and InflationSDP. We may want to correct this
@apozas apozas merged commit bce9e79 into devel Apr 21, 2025
0 of 3 checks passed
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