Relative convergence tolerance

Type : Floating point number

Range : [1e-010,1e+019]

Default : 1e-008

This option determines the convergence tolerance for the algorithm. The algorithm terminates successfully, if the (scaled) NLP error becomes smaller than this value, and if the (absolute) criteria according to the options Dual Infeasibility Tolerance and Complementarity Tolerance are met. See also the option Acceptable Relative Convergence Tolerance as a second termination criterion. Note, some other algorithmic features also use this quantity to determine thresholds etc.

Learn more about

• Acceptable relative convergence tolerance

• Complementarity tolerance

• Dual infeasibility tolerance