qiskit-advocate/qamp-fall-21

Integrate `AdaptVQE` with Qiskit Terra's gradients

Opened this issue ยท 7 comments

Description

The goal of this project, is to refactor the implementation of AdaptVQE (in Qiskit Nature) to make use of the native gradient functionality included in Qiskit Terra.

History:

The adaptive VQE was originally implemented in Qiskit Aqua as the outcome of the Qiskit Hackathon Europe 2019, link. Since then, it has already been refactored multiple times during the migration to the GroundStateEigensolver interface and later transition to Qiskit Nature.

As it stands today, it still performs the gradient computation manually, see here, instead of leveraging Qiskit Terra's gradient functionality released as part of Terra 0.17.0

Link to the original issue: qiskit-community/qiskit-nature#123

Mentor/s

Co-mentoring by Max Rossmannek (@mrossinek), PhD Student, Qiskit Nature developer and Dariusz Lasecki (@dlasecki), Qiskit Applications software engineer

Type of participant

Requirements:

  • familiarity with Qiskit
  • proficiency in Python
  • familiarity with Git/Github

Nice to have:

  • knowledge of VQE
  • knowledge of Qiskit Terra's operator gradients

Number of participants

1-2 people

Deliverable

The outcome of this project should be a PR to Qiskit Nature which refactors the AdaptVQE eigensolver to leverage Qiskit Terra's operator gradients.

@dlasecki @fs1132429 Can you please comment in the issue so that I can assign you?

@dlasecki @fs1132429 Can you please comment in the issue so that I can assign you?

Ok

yeah ๐Ÿ‘

GIF-211110_153322

checkpoint 2.pdf

Here is the work update and a GIF of H2 unittest iteratively producing computed gradients for checkpoint 2.

Here is my final presentation slides ๐Ÿ™‚
adapt_vqe_final.pdf