Noise Mitigation Toolbox
The noise-mitigation toolbox is being developed by the Unitary Fund. This will be an open-source tool that will provide interfaces and routines for the noise mitigation methods developed under this program. It will have interfaces into multiple cloud-based QPU platforms.
Improvements to Error-mitigation techniques:
- Gate-level noise scaling for zero-noise extrapolation. Pulse access not required.
- Adaptive learning for zero-noise extrapolation. Reduces runtime and improves accuracy of error-mitigation.
- Integration of non-Markovian noise models
Unitary Fund and Standford
Programming Language Based Noise Mitigation
This thrust will explore how ideas from probabilistic and approximate computing can be leveraged for NISQ computing.
Investigate and Adapt Classical Program Optimizing Transformation to Quantum Programs. And make them error-aware.
- Active measures to mitigate error-effects in programs without fault-tolerant qubits.
- A Domain-Specific Language (DSL) + Machine Learning (ML) approach
- Formulate the error semantics and error-handling primitives, their possible way of combination, but leaving room for flexibility.
- Write a DSL sketch program and figure out the rest by program synthesis + reinforcement learning on real quantum data/devices.
- A way to include domain-knowledge + machine learning.
- Generality, portability, verifiability, and modularity.
Maryland, Stanford, and Unitary Fund