SET_NUMERICAL_LIMITS

This command is available for the POLY, Scheil, and ED-EXP modules.

Change the criteria for convergence. This is to speed up a calculation in a complex system.

If there is any convergence problem in finding a stable solution at any stage during a calculation procedure enforced by COMPUTE_EQUILIBRIUM, STEP_WITH_OPTIONS, MAP or ADVANCED_OPTIONS), this message displays:

Convergence problems, increasing smallest site-fraction from 1.00E-30 to hardware precision 2.00E-14. You can restore using SET-NUMERICAL-LIMITS

This implies that smallest site fraction in the current POLY workspace is automatically increased from the default value 1.00E-30 to the hardware-dependent precision (under Linux, as 2.00E-14). For other subsequent POLY module calculations, you can use this command to restore or reset the smallest site fraction to the previous or another preferred value, as well as to reset other numerical limits.

Syntax	set_numerical_limits
Prompt	Maximum number of iterations /500/: By default, the program tries 500 iterations before it gives up. As some models give computation times of more than 1 CPU second/iteration, this number is also used to check the CPU time and the calculation stops if 500 CPU seconds/iterations are used.
	Required accuracy /1E-6/: This is a relative accuracy, and the program requires that the relative difference in each variable must be lower than this value before it has converged. A larger value normally means fewer iterations but less accurate solutions. The value should be at least one order of magnitude larger than the machine precision.
	Smallest fraction /1E-12/: This is the value assigned to constituents that are unstable. It is normally only in the gas phase you can find such low fractions. The default value for the smallest site-fractions is 1E-12 for all phases, except for IDEAL phase with one sublattice site (such as the GAS mixture phase in many databases) for which the default value is always as 1E-30 (unless you have used the SET_NUMERICAL_LIMITS command to reset an even-lower value (e.g. 1E-45, that is naturally enforced to all the phases in the system).
	Approximate driving force for metastable phases /Y/: Y is the default. Enter N to change the default as required and based on the options described below. This setting involves the convergence of metastable phases and affects their driving forces. It can also have an effect on when an equilibrium is considered successful. The default is to allow an equilibrium with metastable phases that have not converged, as long as the stable phases have converged. This is efficient but often causes approximate values of the driving forces for the metastable phases. If you change the default, it enforces metastable phases to converge. This gives accurate driving forces for metastable phases as well as stable phases. It can however take a slightly longer time, and if metastable phases do not converge it causes the equilibrium calculation to fail.