Traditional computers struggle with solving complex optimization problems involving large numbers of interacting variables due to inefficiencies like the von Neumann bottleneck.
A new approach called collective state computing maps these problems onto the Ising model from magnetism called the Ising problem.
The Ising problem represents a problem as a graph where nodes have two states (+1 or -1) and the goal is to minimize the total energy.
Researchers are exploring physical systems like light-based techniques that could solve the Ising Hamiltonian more efficiently than traditional computers.
The article describes a study using an array of VCSELs (vertical-cavity surface-emitting lasers) where information is encoded in laser polarization states to represent solutions.
Interactions between lasers encode the problem structure and the system aims to quickly find the correct solution using effects like interference and feedback.
Testing on small problems showed promise but challenges remain like reducing VCSEL anisotropy – overcoming this could enable an optical Ising computer surpassing traditional architectures.
Source: SciTechDaily
