Type = Real(+)

Default = 0.0

Range = 0.0-0.0001

Convergence tolerance; this is the minimum α value that can be achieved before convergence, where α is the step length used in the conjugate gradient step of the estimation algorithm. In most cases, this value should be set rather low. If the user wishes to rely more on the MAXITER parameter for terminating the program, it should be left as the default value of 0.0. Note that using any tolerance value greater than 0.0 can cause the program to terminate before finding the minimum value of the cost function, and so is only recommended to be set for problems which run excessively large numbers of iterations due to slow convergence in which case terminating the program early is beneficial from a time cost standpoint.