next up previous notation contents
Next: 3.2.13 Examples with Monotonic Functions Up: 3.2 Constant Interval Arithmetic Previous: 3.2.11 Monotonically Decreasing Functions

3.2.12 Lower Bounds

 

We have concentrated on upper bounds since lower bounds may be easily constructed using the rules given for upper bounds. This is achieved with the following identity:

math13481

which follows directly from the definition of extremal bounds. Given that tex2html_wrap_inline34803 is an upper bound of tex2html_wrap_inline34805 , it follows that tex2html_wrap_inline34807 is a lower bound of tex2html_wrap_inline34809 :

eqnarray13488

The proof is valid for all tex2html_wrap_inline33809 since it only relies on properties of tex2html_wrap_inline32451 .

So both lower and upper bounds for tex2html_wrap_inline34127 may be constructed from the upper bounds of tex2html_wrap_inline34817 and tex2html_wrap_inline34127 :

math13498

A similar process allows construction with lower bounds:

math13507

We do assume that the number system underlying the interval number system has an exact negation operator which is total. Although the above construction could be taken literally, it is mainly a device to simplify exposition. In practice, upper and lower bounds are usually computed simultaneously by a single procedure.


next up previous notation contents
Next: 3.2.13 Examples with Monotonic Functions Up: 3.2 Constant Interval Arithmetic Previous: 3.2.11 Monotonically Decreasing Functions
Jeff TupperMarch 1996