Automatic transformation of quantum circuits into Pulser code #230
-
Probably a crazy, or even meaningless, idea but... In https://arxiv.org/abs/2104.15044, the guided walkthrough (§3.2) shows how to turn a typical quantum circuit into Pulser code. Some gates can be "Pulsered" in a straightforward way. Some others (like CNOT) need to be transformed first, in a form compatible with neutral-atom devices. Would it be possible and interesting to add a feature that inputs any circuit encoded with the most common tools (Cirq, Qiskit, QuTip...) and transform it automatically into a Registrer/Sequence/Pulses set ? The result would certainly depend on the chosen device specifications. But we could imagine an intermediate representation that would be (as) device-agnostic (as possible), and can later be applied to a given device. It would be some kind of superlayer to the existing features. Note: Some tools already manipulate circuits created by others. For instance, the brilliant Mitiq has support for Braket, Cirq, PyQuil and Qiskit. |
Beta Was this translation helpful? Give feedback.
Replies: 1 comment 1 reply
-
It's a great idea, and one that we have considered already at Pasqal. Actually, you can already find a frontend for Pasqal devices inside Cirq. We have been working on an intermediate representation that would take the quantum circuit from the frontend all the way to a Pulser sequence for some time now, but this work remains internal at this point. It's worth mentioning that it is not just a naive translator of gates to pulses (although it can be that, too), but rather a series of routines that transform the given quantum circuit in a way that tries to leverage the neutral atom devices' architecture as much as possible. So, to answer your question, it would indeed be possible and interesting to have such a feature, so much so that we are already developing it 😁 Thanks for the suggestion, it's good to know we are in sync! |
Beta Was this translation helpful? Give feedback.
It's a great idea, and one that we have considered already at Pasqal. Actually, you can already find a frontend for Pasqal devices inside Cirq. We have been working on an intermediate representation that would take the quantum circuit from the frontend all the way to a Pulser sequence for some time now, but this work remains internal at this point.
It's worth mentioning that it is not just a naive translator of gates to pulses (although it can be that, too), but rather a series of routines that transform the given quantum circuit in a way that tries to leverage the neutral atom devices' architecture as much as possible.
So, to answer your question, it would indeed be possible and interest…