LINGO has two solution "crashing" methods that use different heuristics for generating starting points for nonlinear models: NCRASH and LCRASH. If at least one of the two crash parameters is set to 1, LINGO’s nonlinear solver will invoke a heuristic for generating a "good" starting point when you solve a model. If this initial point is relatively good, subsequent solver iterations should be reduced along with overall runtimes. It's possible, however, that runtimes may slow if the selected point turns out to be poor. It is difficult to know beforehand if a particular crashing technique, if any, will help performance. You will need to experiment with your model to determine the best settings for the two crash procedures.

LINGO defaults to not crashing an initial solution.