default.qubit returns the new DefaultQubit device
#4436
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This reverts commit de4e5fa.
albi3ro
reviewed
Aug 11, 2023
albi3ro
reviewed
Aug 11, 2023
albi3ro
reviewed
Aug 11, 2023
…ing" This reverts commit 840f30a.
# Conflicts: # tests/ops/qubit/test_special_unitary.py # tests/optimize/test_optimize_shot_adaptive.py
albi3ro
reviewed
Sep 19, 2023
albi3ro
reviewed
Sep 20, 2023
albi3ro
approved these changes
Sep 20, 2023
There was a problem hiding this comment.
Fantastic work 🎉 So excited to have this finally here. 🚀
For the code-cov issues, I am willing to override, but let's mark them down to investigate after this PR goes through.
For the qcut thing, I am still rather curious what is happening, but indulging my curiosity is not worth holding up any of the work we need to do. It can also be marked as "Further investigation for when we have time". If we ever actually have time for something like that.
lillian542
approved these changes
Sep 20, 2023
mudit2812
pushed a commit
that referenced
this pull request
Sep 20, 2023
Note: kindly ignore the codecov failure, it seems totally unrelated to
this PR.
**Context:**
It's time for `qml.device("default.qubit")` to return the new device
(look, no wires!).
**Description of the Change:**
- 99% of this PR is changing `dev.batch_execute` to `dev.execute`
- I am sorry about all the pylint changes - it’s from before I silenced
them
- I couldn’t keep track but there were some little things that we
asserted didn’t work, but now they do! Like custom wires with some
measurement processes
- `test_state.py` used to try to test different dtypes, but the returned
state should now _always_ be complex, so that's what I test for.
- One stupid qcut torch test fails on CI only… idk why. it just says
`UNSUPPORTED DTYPE` (that's it because it's jit-compiled)
- I changed `math.size` for tensorflow because autoray [changed
it](jcmgray/autoray@58f02d5)
in 0.6.4 (I was still on 0.6.3 locally), but we actually want to use the
base `tf.size(x)`
Test formatting changes:
- Duplicated tests in `test_hadamard_gradient.py` for legacy devices
- Any test that used `default.qubit.some_interface` has been
parametrized to also test with `default.qubit` to make sure
interface-specific stuff also works with DQ2
- test_qnode.py uses legacy because it’s deeply steeped in legacy
- test_vqe.py uses the new device but the results are nonsense. I
commented the actual expected results out
- Pulse and qchem tests are being fixed in other PRs
- test_return_types.py uses legacy - test_return_types_dq2.py is a
copy-paste with DQ2
- The things in `test_tensor_measurements.py` just felt unnecessary and
too hard to move over
**Benefits:**
We use the new device!
**Possible Drawbacks:**
3 things that have changed afaik:
- Non-hermitian observables fail hard now, still needs investigation
- `expval(SparseHamiltonian)` used to fail nicely, now it raises a
`DiagGatesUndefined` error
- `qml.Snapshot()` instances are manually spotted and skipped while
doing `adjoint_jacobian` (when `diff_method="adjoint"`)
---------
Co-authored-by: Christina Lee <christina@xanadu.ai>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Note: kindly ignore the codecov failure, it seems totally unrelated to this PR.
Context:
It's time for
qml.device("default.qubit")to return the new device (look, no wires!).Description of the Change:
dev.batch_executetodev.executetest_state.pyused to try to test different dtypes, but the returned state should now always be complex, so that's what I test for.UNSUPPORTED DTYPE(that's it because it's jit-compiled)math.sizefor tensorflow because autoray changed it in 0.6.4 (I was still on 0.6.3 locally), but we actually want to use the basetf.size(x)Test formatting changes:
test_hadamard_gradient.pyfor legacy devicesdefault.qubit.some_interfacehas been parametrized to also test withdefault.qubitto make sure interface-specific stuff also works with DQ2test_tensor_measurements.pyjust felt unnecessary and too hard to move overBenefits:
We use the new device!
Possible Drawbacks:
3 things that have changed afaik:
expval(SparseHamiltonian)used to fail nicely, now it raises aDiagGatesUndefinederrorqml.Snapshot()instances are manually spotted and skipped while doingadjoint_jacobian(whendiff_method="adjoint")