Re: Page and line penalties

[Joe Neeman]

On Sat, 8 Apr 2006 00:52, David Feuer wrote:
> On 4/7/06, Joe Neeman <address@hidden> wrote:
> > In constrained-breaking, I use the square of the force rather than its
> > absolute value -- I made the change for precisely this reason.
>
> The sum of absolute values will generally be more stable than the sum
> of squares, or so the stat books say.
When I said "stable" before, I meant that, starting from the minimum, the 
weighting function is monotonic in any parameter. Probably "stable" wasn't a 
good word to use.

In any case, I think convexity is important. Particularly since the 
constrained-breaker has to deal with worse spacing than the gourlay-breaker, 
I think it makes sense to put large penalties on things that are farther from 
optimal.