[challenge]: Extreme-efficiency VQE numerical simulation

[fliingelephant]

Released by

Shi-Xin Zhang, Institute of Physics, CAS

Contact email

shixinzhang@iphy.ac.cn

Method

Quantum Circuit Simulation

Challenge issue

Extreme-efficiency VQE numerical simulation

Goal

Build an iteratable, auto-evaluatable harness framework that uses semi-automated algorithm–engineering co-innovation to achieve a VQE computation scheme surpassing the current tensorcircuit-ng official baseline in both space efficiency and time efficiency (including first-time JIT compilation efficiency and subsequent runtime execution efficiency).

More ambitiously, the harness framework should automatically and reliably help users optimize TensorCircuit-NG-based code scripts for performance.

Optimization dimensions (reference pool)

Algorithm level:

• Tensor network contraction path search
• Slicing search

Engineering level:

• scan control flow
• AOT compilation
• Operator fusion and customization
• Automatic differentiation checkpointing
• Memory offloading
• Mixed precision