HybriD-GM: A Framework for Quantum Computing Simulation Targeted to Hybrid Parallel Architectures

Area of Science:

• Quantum Computing
• High-Performance Computing
• Computer Architecture

Background:

• Existing quantum computing simulators face challenges in efficiently utilizing hybrid CPU and GPU architectures.
• Scalability and performance optimization are critical for complex quantum simulations.

Purpose of the Study:

• To introduce the HybriD-GM model for advanced quantum computing simulations.
• To enhance the D-GM environment for efficient parallel execution on hybrid systems.
• To optimize hardware resource management in distributed quantum computations.

Main Methods:

• Developed the HybriD-GM model, integrating CPU and GPU resources.
• Extended the D-GM environment for parallel quantum computing simulations.
• Implemented projection operator management and coalescing memory access patterns.
• Organized distributed computations using tree data structures for granularity control.

Main Results:

• Achieved significant performance improvements in Shor's and Grover's algorithm simulations using HybriD-GM.
• Demonstrated superior performance compared to the previous D-GM version.
• Showcased advantages over other quantum simulators like LIQUi|⟩ and ProjectQ.

Conclusions:

• The HybriD-GM model offers substantial performance gains for quantum simulations on hybrid architectures.
• Effective management of hardware resources and computational granularity is key to optimizing quantum computing performance.
• HybriD-GM represents a significant advancement in simulating quantum algorithms efficiently.