UnitaryHACK2024 Challenge to create custom optimizer/transpiler for Qiskit to Native Gates.
- Understand the Main Objective:
- The main goal is to write a custom optimizer or transpiler that outperforms Qiskit's built-in optimization for native gates.
- Identify Potential Blockers:
- Recognize that understanding the Qiskit transpiler might be challenging and necessary for this project.
- Ensure the custom IonQ (or trapped-ion) method is better optimized than Qiskit's transmon-qubit optimizer, using all-to-all connectivity.
- Break Down the Project:
- Consider splitting the project into smaller tasks:
- Overriding Qiskit's optimizer.
- Performing basic matrix optimizations.
- Implementing trapped-ion optimizations.
- Consider splitting the project into smaller tasks:
- Phrase the Issue Clearly:
- Describe the issue as you would for a backlog issue, focusing on engaging newcomers. Make it clear and detailed to facilitate understanding and contribution.
- Team Review:
- Have team members like Spencer, Vadim, and Jon review the issue descriptions to ensure clarity and completeness.
- Specific Project Recommendations:
- Define performance metrics: Investigate and demonstrate the performance of the optimizer across a range of common algorithms (e.g., AQ benchmarking suite).
- The optimizer should aim to use the most efficient circuits (e.g., prefer ZZ gates over MS gates if they are easier to implement).
- Focus initially on Qiskit, with potential for future porting to Pennylane and Cirq.