Re: [sage-devel] Re: Apple on ARM

[Isuru Fernando]

conda-forge provides sage=9.2 binaries on linux-aarch64. Installers are at
https://github.com/conda-forge/miniforge#download
To install sage, follow the directions at
https://doc.sagemath.org/html/en/installation/conda.html

We have started initial support for osx-arm64 platform as well, but it's
only a small set of packages for now.

Isuru

On Wed, Jun 24, 2020 at 11:35 AM Samuel Lelievre 
wrote:

> And I missed this Sage Trac ticket:
>
> - Build failure of fplll-5.3.2 on Raspbian Buster (ARM)
>   https://trac.sagemath.org/ticket/29091
>
> where the fplll issue is discussed.
>
> --
> You received this message because you are subscribed to the Google Groups
> "sage-devel" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to sage-devel+unsubscr...@googlegroups.com.
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/sage-devel/ac994177-705c-4186-831b-de33a991751co%40googlegroups.com
> 
> .
>

-- 
You received this message because you are subscribed to the Google Groups 
"sage-devel" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to sage-devel+unsubscr...@googlegroups.com.
To view this discussion on the web visit 
https://groups.google.com/d/msgid/sage-devel/CA%2B01voOmVHWEEgyVKrnD4TCckv8TYciQbCJMh%3DD6FU73sXiT0Q%40mail.gmail.com.

[sage-devel] Re: Errors in determinant of 'large' symbolic matrices as opposed to working over better rings

[Linden Disney]

As a follow up on this, it seems that sage implements the determinant by 
evaluating the characteristic polynomials at 0, and the characteristic 
polynomial is calculated by maxima. Is it possible to edit the maxima 
source code in sage? 

On Monday, October 12, 2020 at 1:16:13 PM UTC+1 Linden Disney wrote:

> Attached is a jupyter notebook that runs Sage 9.1, a (slightly more) 
> minimal example of a problem that I discovered. When calculating the 
> determinant of a large (in the sense n>=9 I have currently found) symbolic 
> matrix the answer is not correct. To see this, run the notebook with 
> Qsimplify either True or False. When Qsimplify is false, the calculation is 
> done when variables lie in the symbolic ring, when true a specially 
> constructed ring is used instead. The output of the script shows a matrix 
> and the resulting characteristic polynomial after some simplification has 
> occured.  While the two matrices look the same regardless of Qsimplify, the 
> characteristic polynomial changes. This error goes away for smaller 
> matrices (it first turns up at rank=4, where the rank is of the Lie algebra 
> involved in the calculation, but this just gives the basis the matrix is 
> constructed from). General theory tells us that the answer when Qsimplify 
> is true is the correct one.  

-- 
You received this message because you are subscribed to the Google Groups 
"sage-devel" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to sage-devel+unsubscr...@googlegroups.com.
To view this discussion on the web visit 
https://groups.google.com/d/msgid/sage-devel/7aab0fb1-99af-4887-a35d-2161a3c79048n%40googlegroups.com.

Re: [sage-devel] Building Sage 9.2 on Mac fails due to cysignals

[Zachary Scherr]

Did you by any chance install fortran via:

brew cask install gfortran?

from your config.log file it finds gfortran version 8, but the most recent 
homebrew fortran is 10.2.

Maybe try uninstalling gfortran and then either brew install gcc or brew 
reinstall gcc to get a more up to date fortran.  For example on my Mac when 
I run gfortran --version I get:

GNU Fortran (Homebrew GCC 10.2.0) 10.2.0

On Thursday, October 29, 2020 at 12:34:30 PM UTC-4 modp...@gmail.com wrote:

> Okay rebuilding PARI without -mt=pthread option worked. 
> But now building the documentation fails. Please find the log file 
> attached.
> Anyway, I currently build it without documentation by running 'make build' 
> only.
> But was curious why documentation fails.
>
> On Wednesday, October 28, 2020 at 8:01:43 PM UTC+1 dim...@gmail.com wrote:
>
>>
>>
>> On Wed, 28 Oct 2020, 14:33 Zachary Scherr,  wrote:
>>
>>> Hopefully somebody with more experience can weigh in, but for the record 
>>> I have pari installed via homebrew and there the configure is only with 
>>> --with-gmp and --with-readline and that works with building sage.  It's 
>>> plausible that the -mt=pthread is what's causing the issues.
>>>
>>
>> Hombrew does not provide Pari packages. Unless you install them manually 
>> (these are just data files)
>> it will not be used by Sage.
>>
>> If you install Pari with  -mt=pthread
>> into /usr/local, with its packages, it will be accepted by Sage, causing 
>> this problem.
>>
>> Patches to fix this are welcome.
>> Ideally it can be a short C program to link against libpari, which would 
>> fail with the error message reported on this thread.
>>
>>
>>
>>
>>> On Wednesday, October 28, 2020 at 10:12:35 AM UTC-4 modp...@gmail.com 
>>> wrote:
>>>
 Yes, I have manually installed Pari 2.11.4 after configuring it with:
 Confgure --mt=pthread --tune

 On Wednesday, October 28, 2020 at 3:04:01 PM UTC+1 zsc...@gmail.com 
 wrote:

> Just wanted to make a couple of comments.  First off, brew installs 
> llvm as keg-only meaning that sage shouldn't even be able to find it 
> unless 
> you manually added something like:
>
> export PATH="/usr/local/opt/llvm/bin:$PATH"
> export LDFLAGS="-L/usr/local/opt/llvm/lib"
> export CPPFLAGS="-I/usr/local/opt/llvm/include"
>
> to your profile.  But even after I tried that myself, I failed to 
> reproduce your error on my own Mac machine.  This makes me think that you 
> might want to have a look at the version of Pari you installed on your 
> computer.  I noticed that ./configure finds a copy of Pari in /usr/local 
> on 
> your computer.  You can see that the specific error it references 
> mentions:
>
>  ld: illegal thread local variable reference to regular symbol 
> _PARI_SIGINT_block for architecture x86_64
>
> If removing llvm and make distclean doesn't solve your problem then 
> you may want to look into where you got your Pari.  It seems plausible 
> that 
> Pari might have been built in some funky way and that is what's causing 
> your issue. 
> On Wednesday, October 28, 2020 at 7:27:27 AM UTC-4 dim...@gmail.com 
> wrote:
>
>> try 
>> make distclean
>>
>> and the build 
>>
>> On Wed, 28 Oct 2020, 09:57 modp...@gmail.com,  
>> wrote:
>>
>>>
>>> I tried uninstalling LLVM and building it again, but it fails again 
>>> for cysignals. 
>>> But the problem now seems that it still sets -I and -L flags with 
>>> non-existent LLVM directories.
>>> On Tuesday, October 27, 2020 at 5:10:58 PM UTC+1 Matthias Koeppe 
>>> wrote:
>>>
 From config.log:

 configure:5906: gcc -v >&5
 Configured with: 
 --prefix=/Applications/Xcode.app/Contents/Developer/usr 
 --with-gxx-include-dir=/Applications/Xcode.app/Contents/Developer/Platforms/MacOSX.platform/Developer/SDKs/MacOSX.sdk/usr/include/c++/4.2.1
 Apple clang version 12.0.0 (clang-1200.0.32.21)
 Target: x86_64-apple-darwin19.6.0
 Thread model: posix
 InstalledDir: 
 /Applications/Xcode.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/bin
 configure:5917: $? = 0
 ...

 configure:6029: gcc -o conftest  -I/usr/local/opt/llvm/include 
 -L/usr/local/opt/llvm/lib conftest.c  >&5


 This looks like an unsupported configuration - injecting these -L 
 and -I flags via environment variables, but continuing to use Apple's 
 gcc.
 If you are trying to compile with homebrew's llvm, you should 
 probably configure by setting CC and CXX as well. But this is also 
 untested 
 - I have created https://trac.sagemath.org/ticket/30835#ticket for 
 testing such a configuration.


 On Tuesday, October 27, 2020 at 3:46:10 AM UTC-7, Dima Pasechnik