The Acceleration And Application Of Markov Chain Monte Carlo Method

We attempt to propose a new architecture, a new idea to solve the problem. We willmap the calculation problem to a physical circuit; at the same time we map the evolution ofthe stochastic graph model in the circuit, a direct correspondence to the changing of circuitcurrent. Finally, the result we need is the structure composed of the direction of the currentin the circuit. The advantage of the new architecture is that it can get answers to yourproblems by observing changes in the circuit. If this architecture further research, as longas find the suitable methods of mapping and feedback. It is possible to solve complexproblems through the natural way in a short time.To realize a specific circuit corresponds to an abstract algorithm. We propose to usethe flow of currents in the circuits to exactly represent a directed graph and the order ofvoltages to represent the order of nodes. At any time, we randomly pick one node andperturb its voltage. As a result, some diodes are turned off and some are turned on. Thiscauses the change of current flows in the circuit. Correspondingly, the directions of someedges in the graph change. This gives us another graph of a different connectivity. After theflow of the currents in the circuit converges to a final state, we get the final graph. Notethat the convergence process is randomness-driven.To validate our design, we use digital circuit simulation. In our experiment, we use an example of16nodes to verifying our circuits. And we get the correct answer. Theimplementation of digital circuit has provided support to the rationality of our work.