Scaling Portfolio Diversification with Quantum Circuit Cutting Techniques

TL;DR

This paper introduces QuantCut, a framework that enables large quantum circuit execution by decomposing gates, demonstrated on a 71-qubit QAOA for portfolio diversification, showing promising results in simulation and real-world scenarios.

Contribution

We develop QuantCut, an automatic circuit-cutting framework that allows efficient execution of large quantum circuits, specifically applied to portfolio diversification problems.

Findings

QuantCut enables decomposition of large quantum circuits into manageable sub-circuits.

The framework improves performance in noise simulations for Max-Cut problems.

Applied to a 71-qubit QAOA, QuantCut shows competitive results in financial optimization.

Abstract

Quantum Approximate Optimization Algorithms (QAOA) have demonstrated a strong potential in addressing graph-based optimization problems. However, the execution of large-scale quantum circuits remains constrained by the limitations of current quantum hardware. In this work, we introduce QuantCut, an automatic framework for circuit cutting that enables efficient execution of large quantum circuits by decomposing entangling two-qubit gates into manageable sub-circuits. Specifically, we focus on gate-cutting techniques. We apply QuantCut to a 71-qubit QAOA circuit ansatz for portfolio diversification in the S&P 500 stock market, aiming to maximize asset diversification. Our approach iteratively optimizes the expectation value while leveraging circuit-cutting strategies to reduce the qubit register size. To validate our framework, we first conduct experiments on a toy model using quantum noise simulations for the Max-Cut problem, analyzing performance improvements with an increasing number of layers. Subsequently, we extend our methodology to a real-world financial optimization scenario, showing competitive results. The results suggest that QuantCut effectively facilitates large-scale quantum computations with circuit-cutting technologies.