Blackbox Hamiltonian Learning from Quantum Measurements

This project is being conducted as my graduation project (CMPE491) at Boğaziçi University under the supervision of Asst. Prof. İnci Meliha Baytaş from February 2025 to June 2025.

Project Overview

This project developed deep learning models for blackbox Hamiltonian learning from minimal quantum measurement data. The focus was on predicting both system dimension and Hamiltonian parameters jointly using multi-output neural networks, as well as 1D CNNs, RNNs, and Transformers. I simulated open quantum systems via Julia and QuantumOptics.jl for dataset generation, creating a comprehensive framework for quantum system parameter estimation.

Research Objectives

• Joint Parameter Prediction: Develop models that can simultaneously predict quantum system dimensions and Hamiltonian parameters
• Minimal Data Requirements: Design efficient algorithms that work with limited measurement data
• Architecture Comparison: Evaluate the performance of different neural network architectures for this task
• Open Quantum Systems: Simulate realistic quantum systems with decoherence and noise