Speaker: Johnny Gray
Affiliation: Caltech
Date: Wednesday, 8 November 2023 at 17:00
Location: Online seminar
There are multiple ways one can use tools from belief propagation for tensor networks, and these enable addressing a much larger scale of problem than usually possible – be it number of tensors or bond dimension. We introduce one algorithm that uses such tools, along with combining Schrodinger and Heisenberg pictures, to approximately simulate quantum dynamics on arbitrary graphs. We demonstrate the method on IBM’s recent 127 qubit ‘kicked Ising’ experiment, showing that it provides fast and converged results. The underlying compression and contraction primitives should be useful in a very wide range of settings.