next up previous notation contents
Next: 3.2.26 Partial Binary Functions Up: 3.2 Constant Interval Arithmetic Previous: 3.2.24 Charts

3.2.25 Examples with a Binary Function

 

Consider the multiplication function, g(x,y) = xy,

math16326

math16336

An example evaluation follows:

math16340

For interval arguments that lie within one of the sections of tex2html_wrap_inline34489 , two applications of tex2html_wrap_inline33893 are used. For interval arguments which span two sections of tex2html_wrap_inline34489 , four applications of g are used, as was done in the above example evaluation. Since g is continuous, we may reduce the number of applications of tex2html_wrap_inline33893 .

Consider evaluating g(x,y), for the situation depicted below.

figure16376

The argument, (x,y), is covered by two sections of tex2html_wrap_inline34489 :

math16545

The sections share a common boundary, and points on the boundary are not strict local maxima or minima of g. This is true above, as both horizontal arrows point similarly, while the common boundary is vertical. With such a situation,

math16550

math16559

It follows that tex2html_wrap_inline35427 , tex2html_wrap_inline35429 , so:

math16563

The bounds of tex2html_wrap_inline35199 may therefore be computed with two applications of g, rather than four. With the common binary functions, this situation always occurs when an argument spans two sections, unless the function is discontinuous at the common boundary. If the function is discontinuous at the common boundary, then the function is tex2html_wrap_inline32653 -bumpy. Similar reasoning may be used when the argument spans four sections, to reduce the number of applications of g from eight to four. Concern for efficiency is focussed on evaluation of tex2html_wrap_inline35439 when tex2html_wrap_inline35441 is small.


next up previous notation contents
Next: 3.2.26 Partial Binary Functions Up: 3.2 Constant Interval Arithmetic Previous: 3.2.24 Charts
Jeff TupperMarch 1996