Flint 2.7.0 release candidate 1

[Bill Hart]

Hi all,

We are planning to do a Flint release this week or early next week and
so I have tagged flint-2.7.0-rc1.

The main big change in this release is multivariate factorisation by
Dan Schultz. However a full changelog and list of authors will be made
available in the next rc (if there is one) or in the final release.

All bugs should be reported on our GitHub issue tracker:

https://github.com/wbhart/flint2/issues

Note that the function nmod_mat_scalar_mul_add has been deprecated and
replaced with nmod_mat_scalar_addmul.

Bill.

[Bill Hart]

I forgot to mention that there is a big change to the fmpz_mod_poly
module. We have finally made the long promised change to using a
context object.

Hopefully nobody was using this module other than projects I have been
directly involved in, as the module was in quite a bit of disarray and
pretty unreliable. It's somewhat improved now after focusing on it for
two release cycles, though still has some cleaning up to be done,
which we hope won't result in major changes to the interface. I'll say
more on this when we do the final release.

But for now, if you have code written against this interface, you now
need to add context objects to every call.

Bill.

[Bill Hart]

Hi all,

I have now tagged flint-2.7-rc2. There are some minor fixes and in
particular some CMake issues fixed in this RC. It is not expected that
there will be a further RC and the final release will be made some
time on Friday if no more serious issues are reported.

Bill.

[Dima Pasechnik]

On Thursday, December 10, 2020 at 8:25:08 AM UTC Bill Hart wrote:

I forgot to mention that there is a big change to the fmpz_mod_poly
module. We have finally made the long promised change to using a
context object.

Hopefully nobody was using this module other than projects I have been
directly involved in, as the module was in quite a bit of disarray and
pretty unreliable. It's somewhat improved now after focusing on it for
two release cycles, though still has some cleaning up to be done,
which we hope won't result in major changes to the interface. I'll say
more on this when we do the final release.

But for now, if you have code written against this interface, you now
need to add context objects to every call.

Well, there is semi-abandoned "deformation" project by Seb Pancratz, 

https://github.com/dimpase/deformation

which also features as an experimental Sage package, which we would like to keep alive.

https://trac.sagemath.org/ticket/30325

It turns out it works with Flint 2.5.2 and recent GMP (it was coded to use MPIR explictly),

with Flint 2.6.3 some tests crash, don't know why yet, (and it has a name clash with flint - it has its own nmod_mat_charpoly()

which looks directly replaceable by Flint's 2.6+ offering here), and with Flint 2.7.1 there is fmpz_mod_poly()

problem, as you got this context thing in now.

I guess it should be easy to add this context thing there, and try if 2.6.3 crashes disappear.

Any pointers to what this context thing should be there?

https://github.com/dimpase/deformation/blob/master/deformation/frob.c

Dima 

[Bill Hart]

You just initialise the context object with the modulus p (or p^n or
whatever he is using as a modulus), using fmpz_mod_ctx_init(ctx, p).

Bill.

> --
>
> ---
> You received this message because you are subscribed to the Google Groups "flint-devel" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to flint-devel...@googlegroups.com.
> To view this discussion on the web, visit https://groups.google.com/d/msgid/flint-devel/12aa810c-e38b-44fd-b6a4-0b7e898b9c80n%40googlegroups.com.

[Dima Pasechnik]

On Wednesday, April 28, 2021 at 8:50:02 PM UTC+1 Bill Hart wrote:

You just initialise the context object with the modulus p (or p^n or
whatever he is using as a modulus), using fmpz_mod_ctx_init(ctx, p).

Thanks, we now have a more or less working update of deformation

at https://github.com/sagemath/deformation

Probably it can be further simplified by using Flint's recently added multivariate

polynomials functionality rather than  deformation's own.

Dima

[Bill Hart]

Great news. Yes, Sebastian did a lot of work to implement his own. I
imagine it'd be faster now too, assuming it's any kind of bottleneck.

> To view this discussion on the web, visit https://groups.google.com/d/msgid/flint-devel/9b13996d-3f8b-4ae0-be9b-804a41633886n%40googlegroups.com.