Bug#969785: ITP: fenics-performance-tests -- solvers for testing the parallel performance of DOLFIN

2020-09-07 Thread Drew Parsons
Package: wnpp
Severity: wishlist
Owner: Drew Parsons 
X-Debbugs-Cc: debian-de...@lists.debian.org, debian-science@lists.debian.org, 
lu...@debian.org

* Package name: fenics-performance-tests
  Version : git20190115.e866927
  Upstream Author : Chris N. Richardson 
* URL : https://bitbucket.org/fenics-project/performance-tests/
* License : MIT
  Programming Lang: C++
  Description : solvers for testing the parallel performance of DOLFIN

This package contains solvers for testing the parallel performance of
DOLFIN and the underlying linear solvers. It tests elliptic equations
- Poisson equation and elasticity - in three dimensions.

The intention of this package is help demonstrate and monitor the HPC
performance of Debian's FEniCS/DOLFIN packages. HPC compute time has
been graciously offered for HPC package testing by France's Grid5000
consortium, grid5000.fr.

To be maintained alongside fenics under the Debian Science Team.



Re: RFS: fflas-fpack

2020-09-07 Thread Andreas Tille
Uploaded.  Thanks for your work on this, Andreas.

On Fri, Sep 04, 2020 at 06:42:24PM +, Torrance, Douglas wrote:
> On 9/4/20 3:42 AM, Andreas Tille wrote:
> > sorry for the long delay.
> 
> No worries!
> 
> > I've checked this but the autopkgtest fails in my pbuilder invironment:
> > 
> > 
> > autopkgtest [07:03:26]: test upstream-tests: [---
> > g++   tests/test-multifile1.o tests/test-multifile2.o  -fopenmp -lblas 
> > -llapack -lgivaro -lgmp -lgmpxx -o test-multifile1
> > /bin/sh: 6: ./test-fdot: not found
> > FAIL: test-fdot
> > /bin/sh: 6: ./test-finit: not found
> > FAIL: test-finit
>  > [ etc. ]
> 
> The problem was that with pbuilder, we still have the leftover build 
> artifacts, and so the Makefile that autopkgtest was calling wasn't 
> rebuilding the tests.  And the ones that it did find were in the wrong 
> directory.
> 
> I've updated the Makefile so that it cleans out the tests in the build 
> directory first.
> 
> It should be good to go now if you're available to look at it again.
> https://salsa.debian.org/science-team/fflas-ffpack
> 
> Thanks!
> Doug

-- 
http://fam-tille.de



Re: Introduction to causal dynamics

2020-09-07 Thread Ralph Alexander Bariz
Thats ok, then development will stay on my own gitlab instance and official 
mirrors nonetheless are hosted by tuxfamily.
I just released v0.2 and would like to package it and push it to debian repos. 
However I have no idea how to do it right. Is there some guide?
The output will be multiple packages:

- libcausal-cpp.deb
- libcausal-cpp-dev.deb
- libcausal-cpp-examples.deb (examples shall stand alone, since they are nice 
to just play with... gol implementation n stuff. probably also splitting them 
up once and putting them into packages like causal-gol.deb)
- causalviz-cpp.deb (analyzes causal structure and renders it into graphviz dot 
graphs... in v0.2 defunct)

And in future also:

- libcausal-d.deb
- libcausal-dev.deb
- libcausal-examples.deb

BR
Ralph

‐‐‐ Original Message ‐‐‐
On Thursday, September 3, 2020 10:25 PM, Anton Gladky  wrote:

> I have just added you into the Debian Science Group on salsa.
>
> Usually salsa is used to maintain the packaging stuff, but the upstream is 
> hosted mostly
> on github/gitlab and similar.
>
> Best regards
>
> Anton
>
> Am Mi., 2. Sept. 2020 um 10:20 Uhr schrieb Ralph Alexander Bariz 
> :
>
>> Hi Anton,
>>
>> Since I'm using Debian(in truth Parrot OS, a Debian Testing based 
>> derivative) its just natural, that I want to take care to get it into Debian 
>> repositories. Also I want to get it away from my quite insecure own gitlab 
>> instance, having it on Debian Salsa Git would be perfect. Also I'd like to 
>> pass ownership, or at least get some push from somewhere else making it 
>> impossible to (get forced to) re-license it. Who knows where home-office 
>> rules of theese times lead to, just want to be sure it stays AGPL. Also for 
>> sure I'd like to join the team.
>>
>> BR Ralph
>>
>> ‐‐‐ Original Message ‐‐‐
>> On Tuesday, September 1, 2020 9:35 PM, Anton Gladky  wrote:
>>
>>> Hi Ralph,
>>>
>>> thanks for the introduction. Could you please shortly formulate how the
>>> Debian Science Team can be useful for you?
>>>
>>> Best regards
>>>
>>> Anton
>>>
>>> Am So., 30. Aug. 2020 um 14:13 Uhr schrieb Ralph Alexander Bariz 
>>> :
>>>
 Hi all,

 My name is Ralph Alexander Bariz. I've written a, I think quite usable, 
 proof of concept for a runtime which should introduce a new kind of 
 algorithmic dedicated to the graph oriented modeling and execution of 
 complex non-linear systems.
 Please see 
 https://gitlab.ralph.or.at/causal-rt/wiki/-/blob/ralph/debconf/debconf.odp
 Please see the C++ POC Implementation 
 https://gitlab.ralph.or.at/causal-rt/causal-cpp
 I request to move over the whole project group to salsa 
 https://gitlab.ralph.or.at/causal-rt
 My salsa username is "udet".

 Below I've written, for people interested in the why and probably a way to 
 some kind of new discrete and, error-resistant discretely, executable 
 physics, the thesis. I would also like this post to be seen as an official 
 pre-publication of this thesis.

 Thanks.

 Preface:
 I'm system analytics and architect, no mathematician. So this wont contain 
 a lot of numerical math what probably also is not necessary but instead 
 the results of a structural analysis of what Germans call "Wirklichkeit".

 While this journey begun with working out a methodology to model and 
 execute symmetric interaction simulations on GPU's utilizing definite 
 integrals I was not convinced it could allow to model and execute the 
 aimed complex systems observed to be real.
 It continued passing by actor model systems which were more what I seek 
 for but still very data oriented while lacking for a definition of "the 
 how".

 At that time I came into contact with Werner Heisenberg's and Hans-Peter 
 Dürr's "last assumption" defining a virtual entity they called "Wirks". 
 This, for me, was the key to understand what we seem to have missed all 
 the time. Here a discrepancy between the German and the English language 
 got very obvious. While a certain understanding of "the how" seems to be 
 deeply integrated into German language, the English language seems to 
 completely lack it. This discrepancy gets most obvious when thinking about 
 the classic definition of causality in both languages. While the English 
 language defines causality as the implication cause -> effect, while cause 
 and effect are both about the "what", the German definition is 
 "Ursache"(cause) -> "Wirkung" while "Wirkung" is not about the "what" but 
 about the "how". Also one might note, the English "reality" covers the 
 German "Realität" but not the German "Wirklichkeit" while the reality is 
 about the set of all being and the "Wirklichkeit" is the set of all 
 happening.
 When trying to model this thought of a "Wirks" there came up a few 
 implications which made such a model very attractive not only