Closed davidfarmer closed 8 years ago
I agree the current labelling is sub-optimal and under-documented. The discussion for Dirichlet characters should come afterwards, for a shorthand once the general scheme for all dimensions is decided. The current labelling works like this: http://beta.lmfdb.org/ArtinRepresentation/dimension/conductor/tim_index/ tim_index is chosen by Tim, in a way that I never really asked him to describe. I have assumed throughout this might change between his uploads (the tim_indices are used to refer to another internal database of galois groups of number fields, with some additional data), but he assured me last week he could fix that ordering once and for all. I will try to talk to Tim before leaving, and anyway forward him this issue.
Yes, as Paul says, I am trying to fix this issue once and for all. I spoke to John Jones and Dave Roberts, and we agreed to try to standartize all Galois fields in the database, using Pari's polredabs. If this works (I am in the process of checking that), then we'll get more canonical labels.
One reason to do this is that currently, if you want to rebuild the same database, you have to add fields in a specific order - that's not very good. With new canonical labels, the order would not matter.
(Sorry, I am a bit tied up with admin at the moment, and not sure I'll make it to the meetings. But if you really want to see me, I can.)
I think I'm in a good person to advise against an ordering which is at all arbitrary, even if it is fixed once and for all, as that caused so much grief with elliptic curve labelling -- which was much more about the ordering of the isogeny classes for fixed conductor and of the curves in each class than what the labels actually looked like.
So please try to find an ordering which can be computed and documented!
John
On 18 September 2013 12:23, pdehaye notifications@github.com wrote:
I agree the current labelling is sub-optimal and under-documented. The discussion for Dirichlet characters should come afterwards, for a shorthand once the general scheme for all dimensions is decided. The current labelling works like this: http://beta.lmfdb.org/ArtinRepresentation/dimension/conductor/tim_index/ tim_index is chosen by Tim, in a way that I never really asked him to describe. I have assumed throughout this might change between his uploads (the tim_indices are used to refer to another internal database of galois groups of number fields, with some additional data), but he assured me last week he could fix that ordering once and for all. I will try to talk to Tim before leaving, and anyway forward him this issue.
— Reply to this email directly or view it on GitHubhttps://github.com/LMFDB/lmfdb/issues/127#issuecomment-24657225 .
working on this
This is done
The Artin representation (definition) knowl should describe the naming scheme for Artin representations, or better yet, make a new knowl with that information.
That naming scheme should probably have some discussion, since right now it appears to just be some numerical order. When the Artin representation is 1 dimensional, the label of the Artin representation is (currently) not the same as the label of the corresponding Dirichlet character.