> > But "the thing calling the restraint" is the model, which doesn't know > what the weights are. Whoever we tell the weights to knows :-)
> Are you suggesting that instead he should do: > model.evaluate(scale_factors=[1.0] * 999 + [0.1]) or model.set_weight(myrestraint_index, .5) or model.set_weight (my_restraint_pointer, .5); or restraintset.set_weight(r, .5);
> rather than passing the stdevs to the model. Of course, I am arguing > that a scale factor and a stdev should be treated similarly here. The stddev is an attribute of a particular type of Restraint. As I see it, weight is a transformation applied to the output of any restraint. Similar to adding up the logs of the restraint values instead of their their actual values.
>> Plus, this way we can reuse a restraint with different weights in >> different places if we should so desire. > > Surely the easiest way to do that is the put the restraint into two > different restraint sets, each with its own weight. Sure. But that requires copying the object and keeping them synchronized. Not that I see any particular application of this :-)