Precision and Convergence tolerances, stopping conditions
Using the SDK for resolving a problem, I get a "Converged" Status. How do I set the Convergence setting using the SDK?
In C# for example you'd write myProb.Engine.Params("Convergence").Value = 0.00001; where myProb is a reference to your Problem object. The default value is 1.0E-4 or 0.0001; making it smaller will make it harder for this test to be satisfied, so the Solver will keep searching.
The other tolerance related to stopping, the local optimality or Kuhn-Tucker conditions, is the Precision tolerance, default value 1.0E-6 or 0.000001. You'd set this via myProb.Engine.Params("Precision").Value = desired value; Again, making it smaller makes it hard for this test to be satisfied. The Solver stops when either the Precision test or the Convergence test is satisfied, or the MaxTime or Iterations limit is exceeded.
Please sign in to leave a comment.
Comments
0 comments