Incorporate custom shot noise model with the reconstruction of probability distributions

[e-eight]

We are trying to incorporate custom shot noise models with circuit knitting. Particularly we are trying to see the effects of shot noise when we reconstruct probability distributions.

With the legacy CutQC implementation we could do that with the cloud based QASM simulator. But now that the cloud based QASM simulator has been deprecated, is there any other way to incorporate custom shot noise models with the legacy CutQC implementation?

Alternatively we can use the support circuit_knitting.cutting module for this purpose. But is there an implementation of reconstruct_full_distribution in circuit_knitting.cutting or any guidelines for how to implement that? Any help is appreciated.

[garrison]

But is there an implementation of reconstruct_full_distribution in circuit_knitting.cutting or any guidelines for how to implement that?

There is a draft of such support in #428, but it has a bit different behavior: it allows generating independent samples from the full distribution rather than constructing the full distribution in memory and therefore avoids the Kronecker product and the associated exponential classical reconstruction cost.

[garrison]

We are trying to incorporate custom shot noise models with circuit knitting. Particularly we are trying to see the effects of shot noise when we reconstruct probability distributions.

Is using Qiskit Aer an option?

[e-eight]

Thanks for the Qiskit Aer suggestion! We have managed to use Qiskit Aer primitives with Cut QC for this purpose.

Looking into #428 to see if we can use something from there with the new module.