Protein Structure Prediction Using Evolutionary Algorithms Hybridized with Backtracking

This work explores different evolutionary approaches to Protein Structure Prediction (PSP), a highly constrained problem. These are the utilization of a repair procedure, and the use of evolutionary operators whose functioning is closed in feasible space. Both approaches rely on hybridizing the evolutionary algorithm (EA) with a backtracking algorithm. The so-obtained hybrid EAs are described, and empirically compared to a penalty-based EA. The utilization of the repair procedure reveals itself as a very appropriate technique for tackling this problem.