Re: Branch & bound for not everywhere defiend constraints
Dan Zuras Intervals wrote:
> The algorithm could be made more efficient by using
> many NaIs. Or by using none as I suggested.
I don't entirely follow what you've suggested. For example, any assertion
about the empty set is always vacuously true. So if f(X) is empty, then every
if-then clause in
If f(X) > 0 color me gray
else if 0 \in f(X) color me blue
else if width(f(X)) == +oo color me red
else if f(X) \subset [-oo,0] color me green
else if empty(f(X)) color me yellow
is true. For example, f(X) > 0 is true if f(X) is empty. So I don't see how any
yellow boxes will show up in your paving.
Nate Hayes