-
Notifications
You must be signed in to change notification settings - Fork 6
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
fix: Check for rewrite composition in badger #255
Conversation
This is a short-term quick-but-incomplete fix for the rewrite composition generating loops in the circuit. We now check right after the composition whether we have invalidated the circuit. Still, it is theoretically possible for an invalid chain of rewrites to end up producing a valid (but not equivalent) circuit. There is a discussion open for discussing more general solutions to this problem.
Codecov ReportAttention:
Additional details and impacted files@@ Coverage Diff @@
## main #255 +/- ##
==========================================
- Coverage 82.92% 82.71% -0.21%
==========================================
Files 34 34
Lines 4246 4247 +1
Branches 4147 4148 +1
==========================================
- Hits 3521 3513 -8
- Misses 537 547 +10
+ Partials 188 187 -1 ☔ View full report in Codecov by Sentry. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Clever. Would it be difficult to add a test case to exercise this failure mode (which we'd of course hope to remove in due course)?
I added a test, but left it ignored-by-default since loading the rewrite set takes a while (#257). |
This is a short-term quick-but-incomplete fix for #239. We now check right after the composition whether we have invalidated the circuit by making a loop. We leverage that circuit hashing already does a toposort traversal, adding an error when a loop is found and catching it. Still, it is theoretically possible for an invalid chain of rewrites to end up producing a valid (but not equivalent) circuit. There is a discussion open for discussing more general solutions to this problem #242.
## 🤖 New release * `tket2`: 0.1.0-alpha.1 <details><summary><i><b>Changelog</b></i></summary><p> ## `tket2` <blockquote> ## [0.0.0-alpha.1](https://github.com/CQCL/tket2/releases/tag/tket2-v0.0.0-alpha.1) - 2024-05-24 ### Bug Fixes - Check for rewrite composition in badger ([#255](#255)) - induced cycles in depth optimisation ([#264](#264)) - Encode opaque symbolic constants ([#273](#273)) - Correctly detect custom ops by name ([#281](#281)) - Track input linear units in `Command` ([#310](#310)) - Don't convert tket2 measurements into tket1 ops ([#331](#331)) ### Documentation - Expand the main module and README docs ([#298](#298)) ### New Features - add angle type to tket2 extension ([#231](#231)) - bindings for circuit cost and hash ([#252](#252)) - Implement `PyErr` conversion locally in `tket2-py` ([#258](#258)) - Add a "progress timeout" to badger ([#259](#259)) - [**breaking**] Add lexicographic cost ([#270](#270)) - rewrite tracing ([#267](#267)) - Move pre/post rewrite cost to the RewriteStrategy API ([#276](#276)) - [**breaking**] Lexicographic cost fn ([#277](#277)) - Return rewrite strategies as a generator ([#275](#275)) - add qalloc, qfree, reset ops ([#284](#284)) - [**breaking**] Support any ops in portmatching ([#293](#293)) - Add `PatternMatch::nodes` and `subcircuit` + matching example ([#299](#299)) - Use `IncomingPort` and `OutgoingPort` instead of `Port` where possible. ([#296](#296)) - expose Tk2Op name ([#307](#307)) ### Refactor - Move tket2 code to a workspace member ([#210](#210)) - Restructure the python code ([#211](#211)) - s/taso/badger/ ([#228](#228)) - Move python bindings from `tket2` to `tket2-py` ([#235](#235)) - rename t2op ([#256](#256)) ### Testing - Add small parallel badger test ([#237](#237)) - fix non-deterministic badger test ([#245](#245)) </blockquote> </p></details> --- This PR was generated with [release-plz](https://github.com/MarcoIeni/release-plz/). --------- Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com> Co-authored-by: Agustin Borgna <[email protected]>
This is a short-term quick-but-incomplete fix for #239.
We now check right after the composition whether we have invalidated the circuit by making a loop.
We leverage that circuit hashing already does a toposort traversal, adding an error when a loop is found and catching it.
Still, it is theoretically possible for an invalid chain of rewrites to end up producing a valid (but not equivalent) circuit. There is a discussion open for discussing more general solutions to this problem #242.