Quantum-Chiplet: A VLSI-Like Methodology for Hierarchical Quantum Design and Verification

In this work, we introduce “Quantum-Chiplet”, a novel design flow for quantum circuits via a VLSI-like methodology. A large-scale quantum problem is hierarchically decomposed into several functions, or “chiplets,” and designed one by one. Each “chiplet” is initially represented with a behavior model without the need for quantum gate-level details, allowing for greater flexibility in designing arbitrary digital logic under any qubit count within a quantum environment. Focusing on behavior-level design, we also shorten the simulation time by more than an order of magnitude, as demonstrated by comparing the simulation time of QuantumChiplet to the conventional quantum simulator, Qiskit. To synthesize the design into real quantum gates, we leverage the conventional VLSI tool–Design Compiler for the automatic generation of quantum-gate-level design and automatically translate it into code interpretable by conventional quantum simulators. Moreover, Quantum-Chiplet utilizes matrices and the Python NumPy library for simulation. So even if suitable quantum gates are hard to find in existing libraries, simulations can still be done. These advancements, from bottom to top, pave the way for efficiently simulating and scaling up quantum circuits, and offer insight into electronic design automation for next-generation fault-tolerant quantum computing.