Problem Descriptor Reference > The Sections of a Descriptor > Variables > The THRESHOLD Clause

The THRESHOLD Clause

<< Click to Display Table of Contents >>

Navigation:  Problem Descriptor Reference > The Sections of a Descriptor > Variables >

The THRESHOLD Clause

Previous pageReturn to chapter overviewNext page

An optional THRESHOLD clause may be associated with a variable name.

 

The THRESHOLD value determines the minimum range of values of the variable for which FlexPDE must try to maintain the requested ERRLIM accuracy. In other words, THRESHOLD defines the level of variation at which the user begins to lose interest in the details of the solution.

 

Error estimates are scaled to the greater of the THRESHOLD value or the observed range of the variable, so the THRESHOLD value becomes meaningless once the observed variation of a variable in the problem domain exceeds the stated THRESHOLD.  If you make the THRESHOLD too large, the accuracy of the solution will be degraded.  If you make it too small, you will waste a lot of time computing precision you don't need.  So if you provide a THRESHOLD, make it a modest fraction of the expected range (max minus min) of the variable.

 

The THRESHOLD clause has two alternative forms:

 

variable_name ( THRESHOLD = number )

variable_name ( number )

 

Note: In most cases, the use of THRESHOLD is meaningful only in time-dependent or nonlinear steady-state problems with uniform initial values, or that ultimately reach a solution of uniform value.