Non-integral genera for ZLat
#891
Conversation
Codecov ReportBase: 73.70% // Head: 73.73% // Increases project coverage by
Additional details and impacted files@@ Coverage Diff @@
## master #891 +/- ##
==========================================
+ Coverage 73.70% 73.73% +0.03%
==========================================
Files 345 345
Lines 100578 100684 +106
==========================================
+ Hits 74128 74244 +116
+ Misses 26450 26440 -10
Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here. ☔ View full report at Codecov. |
src/QuadForm/Quad/ZGenus.jl
Outdated
| rank = sig_pair[1] + sig_pair[2] | ||
| out = ZGenus[] | ||
| local_symbols = Vector{ZpGenus}[] | ||
| # every global genus has a 2-adic symbol | ||
| if mod(determinant, 2) == 1 | ||
| push!(local_symbols, _local_genera(ZZ(2), rank, 0, 0, even)) | ||
| if mod(numerator(determinant)*denominator(determinant), 2) == 1 |
There was a problem hiding this comment.
| if mod(numerator(determinant)*denominator(determinant), 2) == 1 | |
| if mod(numerator(determinant),2)==1 && mod(denominator(determinant), 2) == 1 |
| end | ||
| # collect the p-adic symbols | ||
| for p in prime_divisors(determinant) |
There was a problem hiding this comment.
I think you do not yet catch the example L = diag(1/3, 3) because det(L)=1. Instead you need the bad primes where the lattice is not unimodular. This is now given by the min and max scale.
Thus I think that your default arguments for the scale bounds are also questionable.
There was a problem hiding this comment.
So iterate over the bad primes instead.
There was a problem hiding this comment.
I just modified it so that we take in account also the primes for which the valuations of min_scale and max_scale are non trivial... As for the bounds on the scales, if the determinant is integral and min_scale is not specified, we get the same as before, if the determinant is strictly rational, then we adapt by imposing the maximal scale to be 1 if not specified (it is a choice to be consistent with what is happening on the other side). Of course we can change this "convention" but I guess it makes sense in general.
|
For automorphous number stuff you mentioned, you mean that all functions having a |
Probably it works even if it is not, but I did not check the details and so let us better stay on the safe side. |
This is a first PR for a bigger import, namely to be able to have similar functionalities for
GenusQuadandGenusHerm. According to the tests, the new changes do not break the previous results on integral lattices.Note: I have added
RationalUnionwhich behaves asIntegerUnionbut we also getRational{<: Integer}andfmpqwith it. It allows one to have a common type for objects that can be coerced tofmpq.There are some cleanups all the way to keep consistency and "file-style". In particular I have tried to replace all accessors by the appropriate (already defined) functions.