Doubts on multiprocessing

PennyLane Help
1

Hello everyone,

I have been doing some tests with multiprocessing and dask libraries and they ended up having issues with pennylane queue or getting executions mixed. I was checking the forum and I only saw a couple of old posts from 2021 that mention the use of QNodeCollection, which I can’t find in current documentation.

The only thing I could find in documentation was on DefaultQubit ( [

Accelerate calculations with multiprocessing

](qml.devices.default_qubit.DefaultQubit — PennyLane 0.37.0 documentation). Is there any more information regarding this topic?

Thanks and have a good day,

Albert

2

Hi @Albert_Nieto,

Are you looking to do multithreading? Or batching observables? Or parameter broadcasting?

Eg: use OpenMP to change the number of threads? Or limit the number of observables that get sent to a specific GPU (in case you’re using may)? Or run the same circuit many times but changing the parameters in an efficient way?

If you can share more details I can help you find the way of achieving this. Also if you can share details on the system you’re using (Linux, Mac, GPUs, which ones, …) then I can see what options are available.