Implements a reverse_map() method for JordanWignerMapper().#1344
Implements a reverse_map() method for JordanWignerMapper().#1344MarcoBarroca wants to merge 21 commits into
Conversation
MarcoBarroca
requested review from
ftroisi,
matteoacrossi,
mrossinek,
robertodr and
woodsp-ibm
as code owners
|
|
|
|
MarcoBarroca
changed the title
|
I see there are failures which are not of this change but due to a newly released pylint version - I created #1346 as this needs fixing outside of this PR to get CI running again and passing on the code that is already in main here. |
|
Sure, I'll fix the mypy error by pushing the complex inside FermionicOp. |
|
Included the pylint fixes in this PR. Hope that's'okay. Should fix #1346 |
Would it be too much to ask if you could do them in a separate PR. I think we will need to get the same fixes onto the stable branch so would want to backport the PR. Given this PR introduces a new feature its not something we would backport - that is more for critical fixes where we can do a bug fix release to address things. Of course once such a separate PR was merged, and the branch here updated, it would all pass. |
|
I'll let this run so we can check things anyway if you like. (Once you have a PR that has been merged here it will not longer need a maintainers approval to run - its just on the first PR) |
MarcoBarroca
force-pushed
the
main
branch
2 times, most recently
from
be27527 to
4c15ca7
Compare
Coverage Report for CI Build 26298873383Coverage increased (+0.03%) to 86.548%Details
Uncovered Changes
Coverage RegressionsNo coverage regressions found. Coverage Stats
💛 - Coveralls |
| return total_fermionic_op.normal_order() | ||
|
|
||
| @staticmethod | ||
| def invert_pauli_terms(sparse_pauli_op: SparsePauliOp) -> SparsePauliOp: |
There was a problem hiding this comment.
I am thinking this should be private method - ie something we do not want as part of supported public API. (If it was it would havte to have full docstring stating args, return etc. I think this is more intended as some private utility right?
There was a problem hiding this comment.
Right now I think I’m the only one asking for this. I don’t see why we shouldn’t have this on the public API though. If Open Fermion can do this and thinks it’s relevant why shouldn’t Qiskit Nature?
There was a problem hiding this comment.
We are talking just about the invert_pauli_terms method right - that was all I was meaning as being private not the whole reverse map method
There was a problem hiding this comment.
Oh! my bad. Sure, I can turn that into a private method. It's just a utility for the reverse_map().
|
|
||
| @classmethod | ||
| def reverse_map(cls, qubit_op: SparsePauliOp) -> FermionicOp: | ||
| """Maps a qubit operator ``SparsePauliOp`` back into the second |
There was a problem hiding this comment.
I have a question around this. This seems to take qubit operator and maps it back regardless of what it really is. I get if I this is given a JW mapped Fermionic Op as a qubit operator it will reverse that. What happens to some other operator - does it always work and produce something even if its meaningless? Is there any check possible? Should we add any note here about this?
There was a problem hiding this comment.
It will work for any operator that is a SparsePauliOp and write that in terms of fermionic operators. But if a user tries to reverse an operator that was mapped with a different mapper(I.e. ParityMapper()) they will just get different ones.
It can also take any qubit operator and turn into a fermionic Op even if it doesn’t make sense, so it will always give a result.
Maybe a note making it clear that this works for JW? I wouldn’t know how to check if an qubit operator was transformed via Jordan Wigner or something else.
There was a problem hiding this comment.
A note sounds like the best way then - that says it users responsibility to provide an operator that was mapped with JW to get a meaningful outcome as this does the reverse mapping for any operator on the given assumption it was done with a JW mapping in the first place - or however you want to phrase it,.
…f61.yaml Co-authored-by: Steve Wood <40241007+woodsp-ibm@users.noreply.github.com>
…f61.yaml Co-authored-by: Steve Wood <40241007+woodsp-ibm@users.noreply.github.com>
|
Sorry my bad in the suggestion, if you did not correct it was
should have been
|
…f61.yaml Co-authored-by: Steve Wood <40241007+woodsp-ibm@users.noreply.github.com>
|
@woodsp-ibm Let me know if you think the note is okay when you can. |
Co-authored-by: Steve Wood <40241007+woodsp-ibm@users.noreply.github.com>
|
@mrossinek Can I get you to take a look at this too. |
|
Hey, just wanted an update here. Is there anything I should do about this still? |
|
The build errors with mypy are fixed by #1362, which is still open. As far as progressing this hopefully @mrossinek can take a look. |
robertodr
requested review from
Panadestein,
fpietra and
ludmilaasb
as code owners
|
Saw that this repo is seeing some activity recently. Let me know if you need anything from me to get this merged. |
5 participants
Summary
This implements a reverse_map() method in the JordanWignerMapper(). It allows users to go back from qubit operators into fermionic operators. Completes issue #255.
Also included a unit test.
Details and comments
Instead of using the implementation linked in issue #255 I modified it to use the FermionicOp object for the mapping. Seemed like it would be easier for anyone to understand the code. Both implementations are equivalent.