Tolerances for Numeric Comparisons

USAGE:

std.tolerance(what="double.eps", A=1, B=0.5) 

ARGUMENTS:

what
the particular machine characteristic to be used.
A, B
the parameters in the formula for the computed tolerance. See below.

VALUE:

This function is used by the all.equal methods to provide tolerances for comparing standard objects with their recomputed forms. It returns the result of A * eps ^ B where eps is the larger of the machine characteristics given by what on the current machine and on the originating machine. This choice of tolerance is suitable for applying the regression tests supplied with S. To use a different regimen for testing, either supply the tolerance argument to all.equal or re-define std.tolerance

SEE ALSO:

, , .