Open ssiccha opened 3 years ago
But we already have "hints for the current node", namely the hints that we got from the parent node.
I think what you really mean is that you want to modify the list of hints currently in use, yes?
As to indicating that a method can never be successful: so far this is handled by one method (here: RecogniseClassical
) storing some data it computed into the recognition record; then other methods can check that data, and if that data implies they can never be called, they can immediately return NeverApplicable
. In how far do you consider this insufficient / worse than whatever you have in mind now?
As to what method to call next: so far we only had situations where a recognition method A determines that a specific method B should be called next; which we then handled by A simply calling B directly. Again: what situation do you have in mind where this is not sufficient (I do not doubt they exist! It just would help to talk about concrete instances of this, instead of staying in the abstract).
An idea @aniemeyer and I had: if call
RecogniseClassical
on a group, then also if the group is not classical, we get information about which Aschbacher classes it can not lie in and in which it probably lies. At the moment, all this information is not used by recog and must be "recomputed".I think we should have a mechanism in recog to tell
CallMethods
which method should be called next and that we know that some methods can never be successfull.