Avoiding traps in nonconvex problems

Iterative projection methods may become trapped at non-solutions when the constraint sets are nonconvex. Two kinds of parameters available to help avoid this behavior and study gives examples both. The first kind parameter, called a hyperparameter, includes any parameter that appears in definition iteration rule itself. second comprises metric sets, feature arises problem be solved has two or more variables. Through we show importance properly tuning both offer heuristic interpretations observed behavior.