Amorphica: A Multi-Policy-Based Annealer for Solving Real-World Combinatorial Optimization Problems

Multi-policy based annealer to solve real combinatorial optimization problems

Researchers at Tokyo Tech have developed a fully-connected annealer that can be extended to a multiple-chip system, and features a multipolicy mechanism to solve CO problems in real-world scenarios. The Amorphica annealer can fine-tune the parameters to match a target CO problem. It has applications in finance, logistics, machine learning and other areas.

Modern society has become accustomed to receiving goods efficiently at their doorsteps. Did you know that achieving such efficiency involves solving a mathematical puzzle, namely the best route between the various destinations? The \”traveling-salesman problem\” is a mathematical problem that belongs to the class of \”combinatorial Optimization\” (CO).

The number of routes increases exponentially as the number destinations grows. A brute force approach based on exhaustive searches for the best route is no longer practical. An approach called \”annealing computing\” is used to quickly find the best route without having to do an exhaustive search.


Leave a Reply

Your email address will not be published. Required fields are marked *