Basic TTN simulation - Githubissues

CQCL / pytket-cutensornet

cuTensorNet Python API extensions for pytket quantum SDK

Apache License 2.0

8 stars 1 forks source link

Basic TTN simulation #41

Closed PabloAndresCQ closed 5 months ago

PabloAndresCQ commented 8 months ago

This PR includes:

A new simulation algorithm TTNxGate currently only supporting:
- exact and approximate chi-bound simulation.
- some basic TTN state querying functions (vdot, get_amplitude, get_statevector)
A new abstract class TNState that encapsulates both MPS and TTN.
Updated docs
Updated the MPI examples for the new API.
Updated the MPS Jupyter notebook for the new API.

TODO

[x] Merge the three pending feature/ttn_* PRs into this one
[x] Cleanup the bonds_from_q0_to_ancestor code and sanity checks after merging the above
[x] Check the code again to look for weird artefacts created by merging the PRs
[x] Run tests locally
[ ] Add stuff to changelog (under "Unreleased")
[ ] Check that the generated docs look good
[x] Change the name of tnstate module to structured
[ ] Write a new Jupyter notebook for TTN

PabloAndresCQ commented 7 months ago

I still need to update the contents of the examples/ folder to use the slightly changed API (essentially, some name changes).

I will not write anything about TTNxGate just yet in the Jupyter Notebook. I want to improve it up a bit first in future PRs.

PabloAndresCQ commented 7 months ago

Folder examples/ updated. Ready to merge after final approval.

erinaldiq commented 6 months ago

Can this be merged? It would be great to start using it.

PabloAndresCQ commented 6 months ago

Hi @erinaldiq, I intended to do some changes to this (by creating other PRs and merging to this one) before I merge to develop. The reason being that, at the moment, there are multiple efficiency bottlenecks that I have not had time to sit down and fix, so the user experience won't be particularly good. I'll come back strong after the new year and work on this for the first couple of weeks. I also intend to create a Jupyter notebook example showing how to use this feature. Also, notice the API has changed slightly!

If you'd still want to use this version of TTN, feel free to pull this branch (feature/ttn) and install it locally via pip install .. I'd be happy to assist you.

PabloAndresCQ commented 5 months ago

We agreed that the tasks that remain (see below) will be done on a separate PR.

[ ] Add stuff to changelog (under "Unreleased")
[ ] Check that the generated docs look good
[ ] Write a new Jupyter notebook for TTN
[ ] Add a NetworkX option for _get_qubit_partition to not depend on KaHypar installation

erinaldiq commented 5 months ago

Thanks @PabloAndresCQ ! Let me know if you need help with the example and/or testing the docs 👍🏼