Springe direkt zu Inhalt

Jens Eisert's Research Group – Publications in Journals of the Nature and Science Group

Quantenphysiker Prof. Dr. Jens Eisert

Quantenphysiker Prof. Dr. Jens Eisert

The research group of Professor Jens Eisert has published six papers in the renowned scientific journals of the Nature and Science group in the last few weeks or is awaitung their publication. The studies, which were conducted with various international and German cooperation partners, deal with the topics of machine learning, measuring out quasi-local integrals of motion, the description of solid-state systems in non-equilibrium, machine learning, quantum error mitigation and quantum algorithms for combinatorial optimization.

News from Mar 13, 2024

In the past decade, Prof. Dr. Jens Eisert's research group has established a worldwide network of scientific collaborations with leading researchers in quantum information and technology. These collaborations consistently result in publications featured in prominent global scientific journals.

Publications in Nature Communications Physics

Measuring out quasi-local integrals of motion from entanglement (2024)

Bohan Lu, Christian Bertoni, Steven J. Thomson & Jens Eisert

DOI: 10.1038/s42005-023-01478-5

The publication presents the results of work at the Dahlem Center for Complex Quantum Systems. The quantum physicists have outlined an experimentally feasible procedure for measuring local integrals of motion based on their contribution to the slow growth of the negativity at long times following a quench from an arbitrary initial state. The proposed model paves the way for the application of spatially-resolved entanglement probes to phenomena in quantum simulation beyond many-body localisation. 

Read the publication

Publications in Nature Communications

Towards provably efficient quantum algorithms for large-scale machine-learning models (2024)

Junyu Liu, Jin-Peng Liu, Ziyu Ye, Yunfei Wang, Yuri Alexeev, Jens Eisert & Liang Jiang

DOI: 10.1038/s41467-023-43957-x

Jens Eisert collaborated with researchers from the University of Chicago, University of California, and Brandeis University to conduct a study on fault-tolerant quantum processing. The research introduces an efficient algorithm designed for complex machine learning models.

Building upon previously established quantum algorithms for dissipative differential equations, the authors demonstrated the applicability of similar algorithms to (stochastic) gradient descent – the primary algorithm in machine learning.

This innovative approach has the potential to mitigate substantial computational costs, power consumption, and time invested in both the pre-training and fine-tuning phases of machine learning processes.

Read the publication

News from Cluster Math+

Understanding quantum machine learning also requires rethinking generalization (2024)

Elies Gil-Fuster, Jens Eisert, Carlos Bravo-Prieto

DOI: 10.1038/s41467-024-45882-z

Generalization is an important concept in the context of machine learning and describes the transfer of learned knowledge to unknown data. In the light of advances in quantum-assisted machine learning, the question arises as to how generalization can be formulated here. The present work shows that we need to radically rethink this and that the known limits for relevant system variables do not apply.

Read the publication

Publication in Science Advances

An in-principle super-polynomial quantum advantage for approximating combinatorial optimization problems via computational learning theory (2024)

Niklas Pirnay, Vincent Ulitzsch, Frederik Wilde, Jens Eisert, Jean-Pierre Seifer

DOI: 10.1126/sciadv.adj5170

In a joint project with young scientists from Technical University of Berlin, Frederik Wilde, a doctoral student of Professor Eisert, sought a quantum algorithmic solution to the combinatorial optimization problem.

The research team resorted to the theory of computational learning theory and cryptographic notions and was able to demonstrate that quantum computers feature a super-polynomial advantage over classical computers in approximating combinatorial optimization problems. It was shown that quantum devices in principle have the ability to approximate combinatorial optimization solutions beyond the reach of classical efficient algorithms.

Read the publication

Press-release of Helmholtz Zentrum Berlin

Papers in Nature Physics in Print

Two further papers on quantum error mitigation and its limits and the classical simulation of solid-state systems in non-equilibrium – one of the central challenges of condensed matter physics – are in press in Nature Physics.

3 / 79

Keywords

  • computational learning theory
  • Dahlem Center for Complex Quantum Systems
  • entanglement
  • Jens Eisert
  • large-scale machine-learning models
  • Nature
  • Nature Communication
  • News
  • quantum algorithms
  • quantum computing
  • quantum error mitigation
  • quantum information
  • quantum physics
  • quantum science
  • quantum simulation
  • quasi-local integrals
  • Science
  • Science Advances
  • super-polynomial advantage