Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Clifford Assisted Optimal Pass Selection for Quantum Transpilation

Siddharth Dangwal, Gokul Subramanian Ravi, Lennart Maximilian Seifert, Poulami Das, James Sud, Frederic T. Chong

TL;DR

This work addresses the challenge of selecting an optimal subset of transpiler passes to maximize fidelity on NISQ devices. It introduces COMPASS, which uses Clifford-dummy circuits to efficiently predict how pass combinations interact, enabling circuit-specific pass selection. To scale, it presents E-COMPASS, a divide-and-conquer approach with approximate fidelity estimation that reduces search and measurement overhead while preserving predictive power. Empirical results on IBM devices show COMPASS achieving up to ~$96.9\%$ of the maximum possible fidelity and E-COMPASS delivering substantial gains with significantly lower overhead, highlighting a practical path toward more reliable quantum compilation in the NISQ era.

Abstract

The fidelity of quantum programs in the NISQ era is limited by high levels of device noise. To increase the fidelity of quantum programs running on NISQ devices, a variety of optimizations have been proposed. These include mapping passes, routing passes, scheduling methods and standalone optimisations which are usually incorporated into a transpiler as passes. Popular transpilers such as those proposed by Qiskit, Cirq and Cambridge Quantum Computing make use of these extensively. However, choosing the right set of transpiler passes and the right configuration for each pass is a challenging problem. Transpilers often make critical decisions using heuristics since the ideal choices are impossible to identify without knowing the target application outcome. Further, the transpiler also makes simplifying assumptions about device noise that often do not hold in the real world. As a result, we often see effects where the fidelity of a target application decreases despite using state-of-the-art optimisations. To overcome this challenge, we propose OPTRAN, a framework for Choosing an Optimal Pass Set for Quantum Transpilation. OPTRAN uses classically simulable quantum circuits composed entirely of Clifford gates, that resemble the target application, to estimate how different passes interact with each other in the context of the target application. OPTRAN then uses this information to choose the optimal combination of passes that maximizes the target application's fidelity when run on the actual device. Our experiments on IBM machines show that OPTRAN improves fidelity by 87.66% of the maximum possible limit over the baseline used by IBM Qiskit. We also propose low-cost variants of OPTRAN, called OPTRAN-E-3 and OPTRAN-E-1 that improve fidelity by 78.33% and 76.66% of the maximum permissible limit over the baseline at a 58.33% and 69.44% reduction in cost compared to OPTRAN respectively.

Clifford Assisted Optimal Pass Selection for Quantum Transpilation

TL;DR

This work addresses the challenge of selecting an optimal subset of transpiler passes to maximize fidelity on NISQ devices. It introduces COMPASS, which uses Clifford-dummy circuits to efficiently predict how pass combinations interact, enabling circuit-specific pass selection. To scale, it presents E-COMPASS, a divide-and-conquer approach with approximate fidelity estimation that reduces search and measurement overhead while preserving predictive power. Empirical results on IBM devices show COMPASS achieving up to ~ of the maximum possible fidelity and E-COMPASS delivering substantial gains with significantly lower overhead, highlighting a practical path toward more reliable quantum compilation in the NISQ era.

Abstract

The fidelity of quantum programs in the NISQ era is limited by high levels of device noise. To increase the fidelity of quantum programs running on NISQ devices, a variety of optimizations have been proposed. These include mapping passes, routing passes, scheduling methods and standalone optimisations which are usually incorporated into a transpiler as passes. Popular transpilers such as those proposed by Qiskit, Cirq and Cambridge Quantum Computing make use of these extensively. However, choosing the right set of transpiler passes and the right configuration for each pass is a challenging problem. Transpilers often make critical decisions using heuristics since the ideal choices are impossible to identify without knowing the target application outcome. Further, the transpiler also makes simplifying assumptions about device noise that often do not hold in the real world. As a result, we often see effects where the fidelity of a target application decreases despite using state-of-the-art optimisations. To overcome this challenge, we propose OPTRAN, a framework for Choosing an Optimal Pass Set for Quantum Transpilation. OPTRAN uses classically simulable quantum circuits composed entirely of Clifford gates, that resemble the target application, to estimate how different passes interact with each other in the context of the target application. OPTRAN then uses this information to choose the optimal combination of passes that maximizes the target application's fidelity when run on the actual device. Our experiments on IBM machines show that OPTRAN improves fidelity by 87.66% of the maximum possible limit over the baseline used by IBM Qiskit. We also propose low-cost variants of OPTRAN, called OPTRAN-E-3 and OPTRAN-E-1 that improve fidelity by 78.33% and 76.66% of the maximum permissible limit over the baseline at a 58.33% and 69.44% reduction in cost compared to OPTRAN respectively.
Paper Structure (32 sections, 12 figures, 3 tables)

This paper contains 32 sections, 12 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Background
  3. Quantum Errors
  4. Primer on Quantum Transpilation
  5. Transpiler Passes for Error Mitigation
  6. Challenges In Optimal Pass Set Selection
  7. Optimal Pass Set Varies with Circuit, Device, and Time
  8. Passes Interact Both Constructively and Destructively
  9. Compiler Passes Rarely Model Operational Crosstalk
  10. Existing Fidelity Estimation Tools are Insufficient
  11. COMPASS: Design
  12. Insights: CNOT Structure Dictates Pass Performance, Leverage Clifford Dummy Circuits
  13. COMPASS: Description
  14. Why do Clifford Dummy Circuits Work?
  15. Efficient COMPASS (E-COMPASS)
  16. ...and 17 more sections

Figures (12)

  • Figure 1: (a) A quantum circuit- it must be before execution on hardware (b) All possible pass combinations that can be used to compile (assume three passes - Passes 1, 2, and 3) (c) Each combination creates a different executable. Each schedule is impacted by hardware errors differently. (d) The combination that corresponds to the highest fidelity (marked B) may be a subset of all "optimizations". It is hard to predict this optimal set using analytical tools like estimating the Expected Success Probability (ESP) of executing a compiled schedule. The performance of the combination predicted by ESP (marked A) is sub-optimal.
  • Figure 2: The IBM-Qiskit transpiler pipeline. The details of each pass are discussed in Section\ref{['subsec:transpilation']}. The transpiler takes an abstract quantum circuit as input and gives an optimized physical quantum circuit as output. The first three steps of the pipeline are compulsory. The physical circuit optimizations are optional.
  • Figure 3: Variation of optimal mapping-scheduling-routing pass set for four circuits on IBM Hanoi and IBMQ Guadalupe on 2 Sep 2022 and 8 Sep 2022. The best pass combination in each subplot is highlighted with a yellow circle.
  • Figure 4: Frequency of different optimal M-S-R combinations, when computed over 4 device backends, 13 days and 13 circuits.
  • Figure 5: (a) Pass Interactions for 7 qubit cnxh7 benchmark on IBMQ-Toronto (b) Increase in error rate of the (6,7) CNOT due to crosstalk from other CNOT links in the circuit. We see a substantial increase in error due to crosstalk between CNOT gates that are 3-5 hops away (c) Fidelity trends for device execution, noise model and ESP for the ADDER4 benchmark. The X-axis represents different transpiler pass combinations.
  • ...and 7 more figures