OpenQasm support for ParameterizedEvolution

[devx5f]

Feature details

Would be great to be able to see if the ParameterizedEvolution function can be supported for openqasm support.

@qml.qnode(dev)
def test_circuit(params, duration):
    qml.evolve(H)(params, t=duration)
    return qml.expval(qml.PauliZ(wire))

Currently running into the following error:

qsam_string = qml.to_openqasm(test_circuit)([0], 1)
ValueError: Operator ParametrizedEvolution not supported with to_openqasm and does not provide a decomposition.

Implementation

No response

How important would you say this feature is?

2: Somewhat important. Needed this quarter.

Additional information

No response

[CatalinaAlbornoz]

Hi @devx5f, thanks for requesting this feature!

At a first glance this looks useful but not super easy to implement. Based on this I'd say it will be hard to get it in very soon. Nevertheless, I'll share this with our team to see what they can do.

If you can, please keep sharing your ideas for feature requests! It helps us improve PennyLane based on your feedback ❤️

[devx5f]

Understand! Thank you for taking a look at it to see what the team can do.

[rosspeili]

Can I take a dive into this? @devx5f

[devx5f]

that would be great!

[albi3ro]

So I do see that openqasm has an openpulse grammar, but I'm don't immediately see how we would map our version of ParametrizedEvolution would map to that.

To avoid confusion before anyone starts any work, for the operation:

f1 = lambda p, t: jnp.sin(p * t)
H = f1 * qml.Y(0)

ev = qml.evolve(H)
op = ev([1.2], t=[0,10])

what would that look like in openqasm? The openqasm spec doesn't seem to quite be doing the same thing.

[albi3ro]

Also, @devx5f , mind telling us a little about how you are using PennyLane's pulse programming tools? That might help us find a workaround or scope the development of the feature.

[devx5f]

Hello, thanks for taking a look at this!

The pulse programming is used for driving rabi oscillations for a given qubit