Disconnectivity graphs in planted spin-glass problems

A disconnectivity graph is a simplified representation of a high-dimensional energy landscape consisting of low-energy pathways between different minima in the landscape. As such, one can use disconnectivity graphs to obtain insights into the accessibility of different minima in the energy landscape, and therefore make predictions for the computational effectiveness of different optimization methods. In this talk different disconnectivity graphs for different planted spin-glass problems are presented, and their similarities and differences discussed.