Research projects

Tackling the worlds most pressing challenges

In the last decade, machine learning has positively impacted numerous areas of society, particularly in chemistry and innovative materials. Applications include discovering eco-friendly chemical processes, creating efficient materials for energy, and advancing drug discovery. However, these challenges are computationally demanding due to the complexity of simulating quantum systems on classical computers. The integration of machine learning with quantum simulation and quantum computing now offers unprecedented opportunities to address these complex problems, surpassing the capabilities of even the largest supercomputers.

Open positions

Project 1. Machine learning and quantum simulations for quantum state generation

We are looking for candidates to join our research project focused on realizing versatile multiqubit Hamiltonians in Rydberg arrays. The objectives include using multifrequency coupling fields for machine-assisted generation and optimization of strongly-correlated quantum states, with applications in quantum sensing and chemistry simulations. This position offers an exciting opportunity to contribute to cutting-edge research at the intersection of experimental quantum physics, machine learning, and numerical methods.

Tasks:

  1. Upgrade the Strasbourg quantum simulator to a modular Python-based control environment with a remote interface and integrate machine learning algorithms.
  2. Experimentally implement a recent theoretical proposal using programmable multifrequency coupling fields for continuous- and discrete-time quantum dynamics.
  3. Implement hybrid quantum-classical variational methods for steering quantum dynamics to strongly-correlated states, capitalizing on parallel quantum operations in the Rydberg platform.

Scientific Training Goals:

  1. Acquire expert knowledge in the principles and operation of Rydberg atom-based quantum simulators.
  2. Develop a deep understanding of machine learning techniques, programming, and data management best practices.
  3. Gain proficiency in state-of-the-art numerical methods and hybrid quantum-classical techniques for strongly-correlated quantum systems and dynamics.

Contact: Shannon Whitlock (whitlock at unistra.fr)

Project 2. Machine learning for numerical simulations and quantum state tomography for near-term quantum devices

We are seeking candidates for a research project focusing on the machine-learning assisted acceleration of Markov Chain-based Quantum Monte Carlo (MCQMC) methods for strongly correlated systems. Additionally, we aim to develop approximate tomography methods using machine learning techniques to characterize quantum states. This position offers an exciting opportunity to contribute to the advancement of quantum Monte Carlo methods and the application of machine learning techniques in quantum state characterization.

Tasks:

  1. Develop strategies employing neural networks to accelerate Markov chain Monte Carlo methods, facilitating efficient global updates in path-integral and perturbation-series representations.
  2. Train artificial neural networks using measurement data from numerical simulations and quantum computers, enabling the determination of properties of underlying quantum states.

Scientific Training Goals:

  1. Acquire expert knowledge in Quantum Monte Carlo approaches for strongly interacting bosonic and fermionic systems.
  2. Gain a deep understanding of machine learning, neural network techniques, and programming using existing open-source frameworks and libraries.
  3. Develop expertise in benchmarking and characterization techniques for quantum machines.

Contact: Guido Pupillo (pupillo at unistra.fr)

Project 3. Qubit decoherence simulation on a PARSEC machine

We are looking for candidates to contribute to a research project with the following objectives: Understanding how to simulate the instantaneous exploration of probability space by entangled quantum particles as they decohere following an observation, and Implementing the algorithm on a PARSEC massively parallel emergent machine and evaluating its performance compared to parallelization on a standard supercomputer. This position offers an exciting opportunity to contribute to advancing quantum computing simulation methodologies and exploring their performance on emergent parallel computing architectures.

Tasks:

  1. Upgrade the EASEA software platform, utilizing PARSEC machines, to efficiently perform massively parallel quantum qubits decoherence simulations.
  2. Use the developed method to simulate and experiment with quantum computing on a massively parallel synchronous/asynchronous system.
  3. Compare the results with theoretical and actual outcomes obtained on a real Quantum Computer, considering both quality and quantity.

Scientific Training Goals:

  1. Acquire knowledge on efficiently discretizing a quantum search space based on the energy received by entangled qubits.
  2. Understand how to perform global or local searches in the discretized hypervolumes.
  3. Develop expertise in massively parallel synchronous and asynchronous stochastic optimization within the defined hypervolumes to approach the actual results of the observation of entangled qubits.

Contact: Aline Deruyver / Pierre Collet (aline.deruyver at unistra.fr)

Project 4. Machine learning interface for the MIMIQ quantum emulator

This research project will explore quantum machine learning on state-of-the-art emulators for quantum circuits. A fundamental goal is to analyze the applicability of quantum neural network algorithms and to determine their usefulness for practical tasks. The project will analyze resource requirements and fundamentally explore new ways for obtaining a quantum advantage in machine learning. The quantum emulation will simulate execution of perfect and noisy quantum circuits using matrix product and tensor network concepts. For this we are seeking a highly motivated candidate skilled in both analytical and numerical quantum many-body theory concepts.

Tasks:

  1. Collaborate with other fellows to develop quantum machine learning methods on quantum circuits, utilizing exact state-vector simulation and Matrix Product State (MPS) methods.
  2. Deploy the Quantum Machine Learning (QML) ideas on the MIMIQ quantum emulator.

Scientific Training Goals:

  1. Acquire expert knowledge in matrix product state/tensor network algorithms, QML, entanglement as resource for quantum computing and QML.
  2. Develop expert skills in quantum software design and development.

Contact: Johannes Schachenmayer (schachenmayer at unistra.fr)

Project 5. Molecular qudits for quantum machine learning

We are seeking candidates to contribute to our research project focused on the synthesis, implementation, and utilization of qudits in quantum computing. The primary objectives are threefold: (i) obtain a significant reduction of the number of quantum operations. (ii) Improvement of the circuit depth in the realization of quantum gates, such as the Toffoli gate. (iii) Application of qudit-based gates for optimization tasks in Quantum Machine Learning. This position offers an exciting opportunity to contribute to the advancement of quantum computing by exploring the potential advantages of qudits and their application in quantum algorithms, particularly in the field of Quantum Machine Learning.

Tasks:

  1. Produce enlarged Hilbert space qudits.
  2. Implement, read-out, and steer the enlarged Hilbert space of qudits within quantum devices.
  3. Perform qubit-to-qudit mapping and compare it to a standard realization of quantum algorithms to highlight the potential advantages of qudits.

Scientific Training Goals:

  1. Design appropriate molecular qudits.
  2. Implement qudits in Quantum Machine Learning algorithms.
  3. Program qudit-based quantum algorithms.

Contact: Mario Ruben (mario.ruben at kit.edu)

Project 6. Tensor network machine learning

We are looking for candidates to contribute to our research project with the following two objectives: Development of a tensor network machine learning suite. Application of the developed suite to relevant problems in Rydberg atoms quantum technology physics.

Tasks:

  1. Integrate tensor network methods into machine learning codes.
  2. Investigate the application of tensor network machine learning to Rydberg quantum simulators and quantum computers.

Scientific Training Goals:

  1. Acquire knowledge of tensor networks and their application to machine learning tasks.
  2. Develop an optimizer for Rydberg quantum computers and compare its performance with standard machine learning methods.

Contact: Simone Montangero (simone.montangero at unipd.it and matilde.cassin at unipd.it)

Project 7. Machine learning for developing better error mitigation schemes and gate decompositions for quantum algorithms

We are seeking candidates to contribute to our research project with the following objectives: Scaling up currently available error mitigation schemes through the application of machine learning techniques and developing machine-learning assisted gate decomposition methods. This position offers an exciting opportunity to contribute to the advancement of error mitigation schemes and gate decomposition methods in quantum computing, leveraging machine learning techniques for practical applications in variational quantum circuits and near-term quantum devices.

Tasks:

  1. Develop neural network-based techniques for learning readout errors, testing them on prototype quantum processors.
  2. Mitigate noise detectors using a neural network description of readout errors, applying it for variational quantum algorithms.
  3. Optimize gate decompositions using reinforcement learning (RL).
  4. Tailor RL-based gate decompositions for near-term devices, considering native gates and device specifications.
  5. Apply developed gate-decomposition methods for near-term applications, such as variational quantum circuits.

Scientific Training Goals:

  1. Gain an overview of current quantum hardware, noise models, and error mitigation schemes.
  2. Acquire a deep understanding of both neural network methods and reinforcement learning techniques.
  3. Acquire expert knowledge on gate decompositions with different native gate sets and on variational quantum algorithms.

Contact: Zoltan Zimboras (zimboras.zoltan at wigner.hu)

Project 8. Machine learning for simulation in quantum materials discovery

We are seeking candidates to contribute to our research project with the following objectives. Firstly, to develop Machine Learning universal models capable of predicting specific quantum mechanical properties of molecules, solids, and quantum materials with maximum data efficiency. Secondly, to apply these new Machine Learning models to quantum materials discovery. This position offers an exciting opportunity to contribute to the development of efficient Machine Learning models for quantum materials discovery and gain expertise in cutting-edge computational methods and technologies.

Tasks:

  1. Construct databases and descriptors for supervised learning for use in material discovery with specified sets of electronic structure methods.
  2. Validate and test Machine Learning potentials.
  3. Enable the fast and precise prediction of new materials with optimized required electronic properties.

Scientific Training Goals:

  1. Gain knowledge and expertise in high-throughput ab-initio methods and technologies.
  2. Gain knowledge and expertise in the development of Machine Learning interatomic potentials and density functionals.
  3. Gain knowledge and expertise on a variety of electronic structure computational methods, both wave function-based and density-based, complemented by Machine Learning, to predict molecular properties.
  4. Use and develop cloud-based material databases.

Contact: Guido Goldoni (guido.goldoni at unimore.it)

Project 9. Quantum algorithms, including quantum machine learning, for the prediction of materials phase equilibria and molecular energies

We are seeking candidates to contribute to our research project with the following objectives: improving the prediction of quantum properties of molecules using quantum algorithms and developing and applying quantum machine learning algorithms for phase equilibria prediction. This position offers an exciting opportunity to contribute to the advancement of quantum algorithms for predicting molecular properties and the development of quantum machine learning approaches for phase equilibria prediction.

Tasks:

  1. Predicting materials phases and phase transitions through computations based on quantum machine learning.
  2. Improving the precision of predicting quantum properties (electronic structure, optical response, transport, etc.) of large molecules by complementing electronic structure calculations with machine learning approaches.

Scientific Training Goals:

  1. Gain knowledge and expertise in formulating quadratic unconstrained binary optimization problems (QUBO) to represent different metastable configurations of selected materials and using quantum annealers to solve these problems.
  2. Gain knowledge and expertise in translating classical machine learning algorithms for materials optimization to QUBO formulations and using quantum optimization algorithms for materials optimization for specific applications.
  3. Gain knowledge and expertise in a variety of electronic structure computational methods, both wave function-based and density-based, complemented by machine learning, to predict molecular properties.

Contact: Rosa Di Felice (difelice at usc.edu)

Project 10. Machine Learning for Quantum Information Processing in Hybrid Quantum Systems

The research project aims to optimize control of hybrid quantum systems using machine learning. Objectives include quantum state preparation, optimized measurements, and feedback protocols. Tasks involve developing reinforcement learning strategies for state control, finding optimized measurement strategies, and exploring feedback protocols. Training goals include understanding open quantum systems theory, mastering numerical simulations, and integrating technical skills with scientific knowledge. Expected results include ML-based strategies for state control, optimized measurements, and adaptive control techniques.

Tasks:

  1. Develop continuous reinforcement learning strategies for state preparation and stabilization in multipartite systems of linear and nonlinear oscillators (cavities, mechanical modes, qubits, etc.).
  2. Find optimized measurement strategies beyond the standard quantum limit for amplitude and phase estimation.
  3. Explore how feedback protocols can improve and/or speed up the protocols developed above.

Scientific Training Goals:

  1. Acquire knowledge about the theory of open quantum systems with a focus on noise and decoherence in hybrid quantum systems.
  2. Achieve competence in numerical simulations of quantum trajectories using classical stochastic algorithms to generate training data for neural networks and learn machine learning techniques.
  3. Combine the technical skills acquired under goal 2 with the scientific knowledge gained in goal 1 to develop protocols for quantum information processing.

Contact: Anja Metelmann (anja.metelmann at kit.edu)