Add OpenQASM 3 exporter for circuits #343
Conversation
Codecov Report❌ Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## master #343 +/- ##
==========================================
+ Coverage 79.52% 79.68% +0.15%
==========================================
Files 41 42 +1
Lines 5793 5852 +59
==========================================
+ Hits 4607 4663 +56
- Misses 1186 1189 +3 ☔ View full report in Codecov by Sentry. 🚀 New features to boost your workflow:
|
This commit introduces an OpenQASM 3 exporter for Graphix circuits. The functionality was originally proposed in TeamGraphix#245 but has not yet been merged. The added tests verify that a round-trip through the OpenQASM 3 representation preserves Graphix circuits, using the graphix-qasm3-parser plugin. This plugin is therefore added as a `requirements-dev.txt` dependency. CI is updated so that `pip install .` can detect the current version number of Graphix, instead of the default `0.1`: to do so, the whole history and the tags should be available. We removed the "-e" option from CI because it is useless in the CI context. In the long term the `qasm3_exporter` module will also host the pattern exporter, but that feature is intentionally omitted from this PR to keep the change focused.
thierry-martinez
force-pushed
the
qasm3_exporter
branch
from
a9d2236 to
1fda549
Compare
| if instruction.kind == InstructionKind.M: | ||
| return f"b[{instruction.target}] = measure q[{instruction.target}]" |
There was a problem hiding this comment.
If I understood correctly, the measure primitive projects in the graphix.instruction.M has attributes plane and angle. Shouldn't we rotate the measurement basis (i.e., cast measurement instructions into M[plane=Plane.XZ, angle=0] commands) before calling the qasm exporter ?
There was a problem hiding this comment.
Agree here also. It appears you have to define alternative measurement bases as subroutines in openQASM (see https://openqasm.com/language/subroutines.html) since they aren't included by default.
There was a problem hiding this comment.
You're right, the emitted instruction only works for M[plane=Plane.XZ, angle=0]. In a076142, I added a check that rejects any other measurements. I think we should provide functions to rewrite circuits to a particular set of instructions (in this case, measurements in a fixed form), and the OpenQASM exporter should not try to rewrite the circuit itself.
There was a problem hiding this comment.
It seems that the workaround in @emlynsg's link amounts to adding extra instructions, which can be done at the level of the circuit (and the pattern), so I agree that it's best to separate the rewrite from the OpenQASM exporter (so that it's a mere translator which doesn't change the depth of the object).
I propose that we open an issue once this PR is merged to we don't forget
There was a problem hiding this comment.
I just opened #352 to discuss about this.
matulni
left a comment
matulni
left a comment
There was a problem hiding this comment.
Thanks, the refactor is very nice. LGTM!
4 participants
This commit introduces an OpenQASM 3 exporter for Graphix circuits. The functionality was originally proposed in #245 but has not yet been merged.
The added tests verify that a round-trip through the OpenQASM 3 representation preserves Graphix circuits, using the graphix-qasm3-parser plugin. This plugin is therefore added as a
requirements-dev.txtdependency.CI is updated so that
pip install .can detect the current version number of Graphix, instead of the default0.1: to do so, the whole history and the tags should be available. We removed the "-e" option from CI because it is useless in the CI context.In the long term the
qasm3_exportermodule will also host the pattern exporter, but that feature is intentionally omitted from this PR to keep the change focused.