next up previous notation contents
Next: 3.3.26 Examples with a Binary Up: 3.3 Linear Interval Arithmetic Previous: 3.3.24 Examples with Binary Functions

3.3.25 Binary Functions: One-Step Method

The framework presented results in the two-step method, which does not directly apply to the general exponentiation function tex2html_wrap_inline35139 . With appropriate extensions, the two-step method may generate valid bounds for tex2html_wrap_inline35139 but it will still generate bounds which are not optimal.

With the two-step method an upper bound h, of g, was found; h was then treated as a unary function so that our previous methods may be applied. We will remove the intermediate step, and consider g as a unary function of tex2html_wrap_inline32761 :

math23332

The previous methods may now be applied.

The relevant derivatives appear in the table following. Positive multiplicative factors were removed from some table entries. Throughout the table tex2html_wrap_inline37179 while tex2html_wrap_inline37181 .

math23336

The one-step and two-step methods produce identical algorithms for addition, subtraction, multiplication, minimization and maximization. The sign of tex2html_wrap_inline37183 is independent of the sign of bd, as expected. The sign of tex2html_wrap_inline37187 reverses as the sign of bd reverses, as expected. With either the one-step or two-step method, efficiency may be improved by considering the tex2html_wrap_inline34489 regions tex2html_wrap_inline37193 overlap, as was done in section gif.


next up previous notation contents
Next: 3.3.26 Examples with a Binary Up: 3.3 Linear Interval Arithmetic Previous: 3.3.24 Examples with Binary Functions
Jeff TupperMarch 1996