Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Advanced Scheduling Strategies for Distributed Quantum Computing Jobs

Gongyu Ni, Davide Ferrari, Lester Ho, Michele Amoretti

TL;DR

A range of scheduling strategies are proposed, simulated, and evaluated, including heuristics that prioritize resource maximization for QPU utilization, node selection based on heterogeneous network connectivity, asynchronous node release upon job completion, and a scheduling strategy based on reinforcement learning with proximal policy optimization.

Abstract

Scaling the number of qubits available across multiple quantum devices is an active area of research within distributed quantum computing (DQC). This includes quantum circuit compilation and execution management on multiple quantum devices in the network. The latter aspect is very challenging because, while reducing the makespan of job batches remains a relevant objective, novel quantum-specific constraints must be considered, including QPU utilization, non-local gate density, and the latency associated with queued DQC jobs. In this work, a range of scheduling strategies is proposed, simulated, and evaluated, including heuristics that prioritize resource maximization for QPU utilization, node selection based on heterogeneous network connectivity, asynchronous node release upon job completion, and a scheduling strategy based on reinforcement learning with proximal policy optimization. These approaches are benchmarked against traditional FIFO and LIST schedulers under varying DQC job types and network conditions for the allocation of DQC jobs to devices within a network.

Advanced Scheduling Strategies for Distributed Quantum Computing Jobs

TL;DR

A range of scheduling strategies are proposed, simulated, and evaluated, including heuristics that prioritize resource maximization for QPU utilization, node selection based on heterogeneous network connectivity, asynchronous node release upon job completion, and a scheduling strategy based on reinforcement learning with proximal policy optimization.

Abstract

Scaling the number of qubits available across multiple quantum devices is an active area of research within distributed quantum computing (DQC). This includes quantum circuit compilation and execution management on multiple quantum devices in the network. The latter aspect is very challenging because, while reducing the makespan of job batches remains a relevant objective, novel quantum-specific constraints must be considered, including QPU utilization, non-local gate density, and the latency associated with queued DQC jobs. In this work, a range of scheduling strategies is proposed, simulated, and evaluated, including heuristics that prioritize resource maximization for QPU utilization, node selection based on heterogeneous network connectivity, asynchronous node release upon job completion, and a scheduling strategy based on reinforcement learning with proximal policy optimization. These approaches are benchmarked against traditional FIFO and LIST schedulers under varying DQC job types and network conditions for the allocation of DQC jobs to devices within a network.
Paper Structure (50 sections, 38 equations, 10 figures, 9 tables, 4 algorithms)

This paper contains 50 sections, 38 equations, 10 figures, 9 tables, 4 algorithms.

Table of Contents

  1. Introduction
  2. Related Work
  3. DQC Framework
  4. DQC Workflow
  5. Network Model
  6. Heterogeneous links
  7. Synchronized Network Simulation
  8. Performance Metrics
  9. Makespan
  10. QPU Utilization
  11. Non-Local Gate Density
  12. System Execution-Latency Performance
  13. Fairness Among Jobs
  14. Scheduling Methods
  15. Resource-Prioritize Scheduler
  16. ...and 35 more sections

Figures (10)

  • Figure 1: DQC workflow, schedule the sequence of parallelized compiled sub-circuits.
  • Figure 2: Fully connected QPU network with heterogeneous links.
  • Figure 3: The PPO model contains: environment, state, action, reward and loss function.
  • Figure 4: $5$-qubit GHZ state preparation. The first qubit is put into superposition via a Hadamard gate, and subsequent qubits are entangled via controlled-NOT operations.
  • Figure 5: $5$-qubit graph state preparation. Each qubit is initialized in the $|+\rangle$ state via a Hadamard gate, and controlled-Z (CZ) gates is introduced according to the edges of the graph.
  • ...and 5 more figures