Mismatch between qaoa_training_pipeline and qiskit_aer MPS-based simulators. #23
Description
Environment
- Qiskit version: 2.1.0
- Python version: python3.13
- Operating system: MacOS
What is happening?
The qiskit_aer-based MPS simulator and the quimb-based implementation give, for a given bond dimension, different results. This should be either better understood, or fixed.
FYI @eggerdj
How can we reproduce the issue?
See notebook attached to the discussion.
What should happen?
The MPS simulators should give, for the same bond dimension, the same energy.
Any suggestions?
No response