feat: add density matrix simulation via qiskit Aer #380

CalMacCQ · 2024-08-22T14:54:28Z

Description

Adding a new AerDensityMatrixBackend class. Unlike other density matrix simulators available through the pytket extensions this emulator should support an optional NoiseModel in the constructor.

There's a decent amount of duplication from the AerBackend implementation for BackendInfo. Maybe there's a smarter and less error prone way to do it.

Related issues

closes #231

Checklist

I have performed a self-review of my code.
I have commented hard-to-understand parts of my code.
I have made corresponding changes to the public API documentation.
I have added tests that prove my fix is effective or that my feature works.
I have updated the changelog with any user-facing changes.

sjdilkes · 2024-08-23T12:14:11Z

docs/changelog.rst

@@ -1,9 +1,13 @@
 Changelog
 ~~~~~~~~~

+.. currentmodule:: pytket.extensions.qiskit


Is this deliberate? I'm unfamiliar with the syntax

Yeah, basically it means that the refernces to classes and functions don't need to be fully qualified. The link is also generated

i.e.

:py:class:`IBMQBackend`

vs

:py:class:`~pytket.extensions.qiskit.IBMQBackend`

sjdilkes · 2024-08-23T12:15:55Z

pytket/extensions/qiskit/backends/aer.py

+    """
+    Backend for running simulations on the Qiskit Aer density matrix simulator.
+
+    :param noise_model: Noise model to apply during simulation. Defaults to None.


Should these be defined here or in the init? I'm not sure what our standard is

Seems to me that we document this in the class docstring rather than the __init__ method. I don't have a strong prefernce either way.

sjdilkes · 2024-08-23T12:16:35Z

pytket/extensions/qiskit/backends/aer.py

+    _supports_density_matrix = True
+    _supports_state = False
+    _memory = False
+    _noise_model = None


type information or all these attributes would be helpful

added in 1d5fd8c

sjdilkes · 2024-08-23T12:17:09Z

pytket/extensions/qiskit/backends/aer.py

+        super().__init__()
+        self._qiskit_backend = qiskit_aer_backend(self._qiskit_backend_name)
+        self._noise_model = noise_model
+        gate_set = _tket_gate_set_from_qiskit_backend(self._qiskit_backend).union(


type for this would be great

added in 1d5fd8c

sjdilkes · 2024-08-23T12:18:27Z

pytket/extensions/qiskit/backends/aer.py

+            name=type(self).__name__,
+            device_name=self._qiskit_backend_name,
+            version=__extension_version__,
+            architecture=FullyConnected(n_qubits),


Should this switch between characterisation.architecture and FullyConnected? Though maybe using an architecture with a density matrix is not likely behaviour anyway?

I think you may be right. See 64a3ad8

sjdilkes · 2024-08-23T12:18:55Z

pytket/extensions/qiskit/backends/aer.py

+            gate_set=_tket_gate_set_from_qiskit_backend(self._qiskit_backend),
+            supports_midcircuit_measurement=True,
+            supports_reset=True,
+            supports_fast_feedforward=True,


This is definitely true for the density matrix simulator?

I believe so. I have a test case for this here

sjdilkes · 2024-08-23T12:19:27Z

pytket/extensions/qiskit/result_convert.py

    return BackendResult(
        c_bits=c_bits,
        q_bits=q_bits,
        shots=shots,
        counts=counts,
        state=state,
        unitary=unitary,
+        density_matrix=density_matrix,


BackendResult already has a density_matrix attribute?

I'm just passing it in as an arg when defining the BackendResult though right?

cqc-alec · 2024-09-03T12:37:54Z

tests/backend_test.py

+        result1.get_density_matrix(), np.outer(output_state, output_state.conj())
+    )
+    # Example with resets and conditional gates
+    # Deterministic if input is a computational basis state


Don't quite understand this comment. What exactly is deterministic?

So despite the fact that the circuit contains measures and resets it prepares a state deterministically assuming that it starts from the $|0\rangle^{\otimes n}$ state.

updated the comment.

cqc-alec · 2024-09-03T12:38:56Z

tests/backend_test.py

+    assert noisy_density_sim.valid_circuit(circ)
+
+    result = noisy_density_sim.run_circuit(circ)
+    assert result.get_density_matrix().shape == (8, 8)


Can we also check that the density matrix differs from what we get with a non-noisy simulation?

The way I've done this is to check the purity $\gamma$ of the state represented by the density matrix $\rho$

$$ \gamma = \text{Tr}(\rho^2) $$

If $\gamma =1$, the state is pure. If $\gamma < 1$ the state is mixed. I've verified in 6d00b85 that the density matrix given by the noisy simulator is a mixed state and the noisless a pure state.

Another option would be to calcluate the expectation value of some observable $E$ and verify that noise changes the value of $\langle E \rangle$.

cqc-alec · 2024-09-03T12:40:46Z

pytket/extensions/qiskit/backends/aer.py

+                if self.supports_state or self.supports_density_matrix:
                    qc.save_state()


Shouldn't we be using https://qiskit.github.io/qiskit-aer/stubs/qiskit_aer.library.save_density_matrix.html ?

Yes, the QuantumCircuit.save_state gave equivalent answers but I think you're right. Its more correct to save the density matrix.

Thanks

cqc-alec

Couple of small comments.

cqc-alec · 2024-09-04T12:11:26Z

tests/backend_test.py

+    noisy_dm = result.get_density_matrix()
+    assert noisy_dm.shape == (8, 8)
+    # Check purity to verify mixed state
+    assert np.trace(noisy_dm**2).real < 1


Could we change this to "< 0.99" or something so that we rule out the possibility of rounding errors causing the test to pass when it shouldn't?

Thanks, done.

cqc-alec · 2024-09-04T12:12:24Z

tests/backend_test.py

+    noiseless_dm2 = result2.get_density_matrix()
+    assert np.allclose(noiseless_dm2, np.outer(statevector, statevector.conj()))
+    # Check purity to verify that we have a pure state
+    assert np.trace(noiseless_dm2**2).real == 1


Better to use np.isclose() for floating-point comparison, to account for rounding error.

Good point, done in ff91d0a

CalMacCQ added 5 commits August 15, 2024 14:35

add AerDensityMatrixBackend class

607f639

update documentation

7c73f2d

add AerDensityMatrixBackend

35fd4f0

add test for the noiseless case

611f5ae

add backend to API docs

8b6c4b8

CalMacCQ requested a review from cqc-melf as a code owner August 22, 2024 14:54

remove duplicate documentation

166333b

CalMacCQ marked this pull request as draft August 22, 2024 14:55

CalMacCQ added 6 commits August 22, 2024 15:57

Merge branch 'main' into feat/Aer_density

67456b7

add changelog entry

853ca8e

add docstring for density matrix backend

4e8e969

slight simplification to process_circuits

a6ede1f

make noise_model arg conistent with AerBackend

36953f9

add test for noisy density matrix simulation

7a62e76

CalMacCQ requested a review from yao-cqc August 23, 2024 11:28

CalMacCQ marked this pull request as ready for review August 23, 2024 11:29

CalMacCQ marked this pull request as draft August 23, 2024 11:36

CalMacCQ added 2 commits August 23, 2024 12:47

attempt to add NoiseModel to BackendInfo

152d569

clean up test file

04db9ed

CalMacCQ requested review from sjdilkes and removed request for yao-cqc August 23, 2024 11:57

CalMacCQ added 2 commits August 23, 2024 12:58

remove print

4b9d421

remove redundant assignment

efbeed7

sjdilkes requested changes Aug 23, 2024

View reviewed changes

CalMacCQ added 4 commits August 30, 2024 13:59

typing improvements

1d5fd8c

set architecture in BackendInfo

64a3ad8

test some architecture properties

3dcdd81

minor simplification to test circuit

6f5cc47

CalMacCQ requested a review from sjdilkes August 30, 2024 14:08

CalMacCQ marked this pull request as ready for review August 30, 2024 14:13

CalMacCQ requested review from cqc-alec and sjdilkes and removed request for sjdilkes and cqc-alec September 3, 2024 09:17

cqc-alec reviewed Sep 3, 2024

View reviewed changes

CalMacCQ added 5 commits September 3, 2024 17:42

update comment

b9e2b0d

use save_density_matrix method

10078ab

add test of the purity

6d00b85

clean up test file

8ef8332

additonal purity check in density matrix test

01f1fdd

CalMacCQ requested a review from cqc-alec September 4, 2024 12:07

cqc-alec reviewed Sep 4, 2024

View reviewed changes

adjust purity check to avoid rounding error

ff91d0a

CalMacCQ requested a review from cqc-alec September 4, 2024 12:37

cqc-alec approved these changes Sep 4, 2024

View reviewed changes

CalMacCQ requested a review from sjdilkes September 4, 2024 12:54

Merge branch 'main' into feat/Aer_density

6f1e99a

sjdilkes approved these changes Sep 5, 2024

View reviewed changes

CalMacCQ merged commit 368df3e into main Sep 5, 2024
6 checks passed

CalMacCQ deleted the feat/Aer_density branch September 6, 2024 14:50

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

feat: add density matrix simulation via qiskit Aer #380

feat: add density matrix simulation via qiskit Aer #380

CalMacCQ commented Aug 22, 2024 •

edited

Loading

sjdilkes Aug 23, 2024

CalMacCQ Aug 30, 2024

sjdilkes Aug 23, 2024

CalMacCQ Aug 30, 2024

sjdilkes Aug 23, 2024

CalMacCQ Aug 30, 2024

sjdilkes Aug 23, 2024

CalMacCQ Aug 30, 2024

sjdilkes Aug 23, 2024

CalMacCQ Aug 30, 2024

sjdilkes Aug 23, 2024

CalMacCQ Aug 30, 2024

sjdilkes Aug 23, 2024

CalMacCQ Aug 30, 2024

cqc-alec Sep 3, 2024

CalMacCQ Sep 3, 2024 •

edited

Loading

cqc-alec Sep 3, 2024

CalMacCQ Sep 3, 2024 •

edited

Loading

cqc-alec Sep 3, 2024

CalMacCQ Sep 3, 2024

cqc-alec left a comment

cqc-alec Sep 4, 2024

CalMacCQ Sep 4, 2024

cqc-alec Sep 4, 2024

CalMacCQ Sep 4, 2024

		if self.supports_state or self.supports_density_matrix:
		qc.save_state()

feat: add density matrix simulation via qiskit Aer #380

feat: add density matrix simulation via qiskit Aer #380

Conversation

CalMacCQ commented Aug 22, 2024 • edited Loading

Description

Related issues

Checklist

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

CalMacCQ Sep 3, 2024 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

CalMacCQ Sep 3, 2024 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

cqc-alec left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

CalMacCQ commented Aug 22, 2024 •

edited

Loading

CalMacCQ Sep 3, 2024 •

edited

Loading

CalMacCQ Sep 3, 2024 •

edited

Loading