| Thread Links | Date Links | ||||
|---|---|---|---|---|---|
| Thread Prev | Thread Next | Thread Index | Date Prev | Date Next | Date Index |
On 05/06/2012 07:01 PM, Michel Hack wrote:
The proposed new Level 1 is clear: plain Moore intervals. That's not what earlier motions decided to use, but then new motions can override old ones.
If unbounded intervals are excluded from the standard, or replaced by intervals with a vague overflow bound, the standard will be useless for
rigorous global optimization.This would be a severe step backwards. It would mean that people in optimization working with true unbounded intervals (a frequent situation) have no support through the standard.
Better no standard than such a standard! Arnold Neumaier