Thank you for opening a new pull request.

Before your PR can be merged it will first need to pass continuous integration tests and be reviewed. Sometimes the review process can be slow, so please be patient.

While you're waiting, please feel free to review other open PRs. While only a subset of people are authorized to approve pull requests for merging, everyone is encouraged to review open pull requests. Doing reviews helps reduce the burden on the core team and helps make the project's code better for everyone.

One or more of the the following people are requested to review this:

• @Qiskit/terra-core

All committers have signed the CLA.

Sure, the runtime is roughly halved:

The code seems to have some issues with the test runner though. E.g. the tests

• test.python.quantum_info.operators.test_operator.TestOperator.test_from_circuit_constructor_reverse_embedded_layout
• test.python.quantum_info.operators.test_operator.TestOperator.test_circuit_init
  fail on my local machine if I start the tests with
  tox -epy37
  however, they pass if the tests are run individually e.g.
  tox -epy37 -- -n test.python.quantum_info.operators.test_operator.TestOperator.test_circuit_init
  Can you have a look @mtreinish, @chriseclectic ?

Thanks @Sebastian-Brandhofer. I don't fully understand the reason for test failures above, but calling np.reshape as in

op1 = np.reshape(node1.op.to_matrix(), (2, 2) * len(qarg1))

copies the array, while setting .shape directly as in

op1 = node1.op.to_matrix()
op1.shape = (2, 2) * len(qarg1)

modifies the array in place. Which means that your code may inadvertently modify matrices for already existing gates. Which might explain why the test failures are only exhibited when running multiple tests and not just the last problematic test.

It's also interesting that executing the following code:

a = _get_identity(4)
print(f"{a.shape = }")

b = a
b.shape = (2, 2, 2, 2)
print(f"{b.shape = }")

c = _get_identity(4)
print(f"{c.shape = }")

gives the output:

a.shape = (4, 4)
b.shape = (2, 2, 2, 2)
c.shape = (2, 2, 2, 2)

That is, the first call to _get_identity(4) returns the matrix with shape (4, 4), but the next call to _get_identity(4) returns the matrix with shape (2, 2, 2, 2) because of lrucache and b.shape = ..... Though this is something that you may want to have here.

And a bit of an off-topic question. What did you use to produce such wonderful graphics above? I have recently started using SnakeViz, but the graphics above look better. Additionally, are your runtimes with profiler enabled or disabled?

Thanks @alexanderivrii, that is probably the reason for the test fails. I expected to_matrix to return a copy. The runtimes are with a profiler disabled (and also without your code) and the graphs are called flame graphs ( https://github.com/baverman/flameprof ) which you can generate out of cProfile's output. :-)

New runtime using np.reshape instead of .shape

Pull Request Test Coverage Report for Build 2415450922

• 8 of 8 (100.0%) changed or added relevant lines in 1 file are covered.
• 31 unchanged lines in 5 files lost coverage.
• Overall coverage increased (+0.02%) to 84.393%

Totals
Change from base Build 2349198842:	0.02%
Covered Lines:	54565
Relevant Lines:	64656

💛 - Coveralls