Abstract
Finding the energetic ground state of a magnet with disordered internal couplings is a complicated combinatorial problem. This so-called spin-glass problem has no analytic solution and even numerical techniques are found to be inefficient. It is known that many important optimization problems in machine learning, logistics, computer chip design, and DNA sequencing can be mathematically mapped to an equivalent spin-glass problem. A method for solving this problem can thus serve as a blueprint for approaching a large class of mathematical optimizations. This motivates research on analog spin-glass simulators as a new class of computational devices (coherent network computing). I will give an overview of the experimental progress in this area including the achievements of our group.