Information science and theoretical physics have a long history of successful cross-fertilization: from the development of modern computers until the introduction of new protocols and standards that formed the root of the modern world wide web. Nowadays, one of the new frontier of physics is the development of quantum technologies: from already well developed quantum cryptography and sensors, to the more challenging and foreseeable quantum simulators and eventually computers.
In the last two decades, quantum science has been mostly driven by theory and proof of principle experiments. However, with the increasing complexity of the experiments, it is now clear that detailed numerical simulations of many-body quantum systems (the quantum hardware) are required. However, this is one of the
most challenging problem in computational physics, since a quantum system of N elements is described by a N-body wave function (a N-rank tensor) and the dynamical equations form a system of coupled partial differential equations with a number of variables that grows exponentially with N.
A promising approach to attack this challenge exploits tensor network methods: they provide a compressed but faithful description of many-body quantum systems in a wide range of scenarios by means of a decomposition of the N-rank tensor into a network of low-rank tensors. Algorithms are then applied to optimize the
tensors to reach a description of the physical process using minimal resources at a given precision.
Although tensor network methods already proved to be highly effective, they
raise a number of theoretical physics, computing science, optimization and
high-performance-computing challenges. In the scope of my Heisenberg Fellowship
I plan to develop novel algorithms and implement HPC solutions to attack applied
and fundamental open problems in quantum science, exploiting the possible
synergies with experts in the different fields needed to solve such challenges.
Kurt Mehlhorn, 11/02/2016 08:00 AM
Kurt Mehlhorn, 10/06/2016 11:58 AM -- Created document.