We opensourced the code used to write Ansatz Optimization using Simulated Annealing in Variational Quantum Algorithms for the TSP paper.
This work was presented at IEEE QAI 2025 conference.
Abstract
In this paper we present a Variational Quantum Algorithm (VQA) for solving the Traveling Salesman Problem (TSP) that requires only $O(n log n)$ qubits and an ansatz whose topology evolves via Simulated Annealing (SA). Instead of fixing the circuit in advance, the optimizer dynamically adds, removes, or rearranges rotation gates and entanglement gates to balance exploration and exploitation. On synthetic instances with 5 −7 cities (7 −13 qubits, 21 −39 parameters), the algorithm identifies the optimal tour in almost all its runs within a few hundred iterations, demonstrating that co-optimization of encoding and circuit structure can efficiently address combinatorial problems.
