SCIP integration into sage

[moritz]

SCIP is an optimization software, that can solve mixed interger (non)linear programs: http://scip.zib.de/
It has a python interface: PySCIPOpt https://github.com/SCIP-Interfaces/PySCIPOpt
#24662
We have now added this as a sage package and written a backend, see #21003, which is based on #24662.

It would be great if some people could test it  and possibly review it!

Cheers
 Moritz

[Francesco Biscani]

http://scip.zib.de/academic.txt

"""
1. This license applies to you only if you are a member of a noncommercial
   and academic institution, e.g., a university. The license expires as
   soon as you are no longer a member of this institution.
"""

So basically these acedemics, I imagine funded with taxpayers' money, are preventing the general public to use software they paid for?

Talk about ivory tower...

--
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+unsubscribe@googlegroups.com.
To post to this group, send email to sage-...@googlegroups.com.
Visit this group at https://groups.google.com/group/sage-devel.
For more options, visit https://groups.google.com/d/optout.

[Vincent Delecroix]

On 13/04/2018 11:20, Francesco Biscani wrote:
> http://scip.zib.de/academic.txt
>
> """
> 1. This license applies to you only if you are a member of a noncommercial
> and academic institution, e.g., a university. The license expires as
> soon as you are no longer a member of this institution.
> """
>
> So basically these acedemics, I imagine funded with taxpayers' money, are
> preventing the general public to use software they paid for?
>
> Talk about ivory tower...

This is indeed a problem with PySCIPOpt. I don't think this is a bad
intention. My point of view is that they wanted to have a form of dual
licensing for making happy (paying) customers that in many situations do
not like open source. Not to prevent users to use it. Moritz might
infirm or confirm.

If the package appears to be useful (I think it is) we could ask
kindly to package PySCIPOpt under GPL or LGPL.

I know of at least the CGAL project where the distinction is made
cleaner [1]: they offer a GPL and a commercial version... and they *do*
have customers (enough for 8 salaries).

Best
Vincent

[1] https://www.cgal.org/license.html

[Francesco Biscani]

I can sort-of sympathise with the intention. But I cannot understand why they decided to roll their own license instead of going with the usual dual licensing everyone is doing. Perhaps because they wanted to enforce the "thou shalt cite me" clause?

As an academic myself, I also cannot understand how they could possibly think that putting artificial barriers for the use of the software by non-academic "non-commercial" individuals could possibly be a good idea ethically.

[Dima Pasechnik]

We had this discussion about SCIP licensing before...

Academy is f*ed up by the citation chase and the grant system...

[jplab]

Hi,

Maybe I just give my basic knowledge about SCIP and ZIB.

Le vendredi 13 avril 2018 11:20:25 UTC+2, bluescarni a écrit :

http://scip.zib.de/academic.txt

"""
1. This license applies to you only if you are a member of a noncommercial
   and academic institution, e.g., a university. The license expires as
   soon as you are no longer a member of this institution.
"""

Perhaps including the very first sentence of the license would put a bit more context:

''This license for ZIB software is designed to guarantee freedom to share and 
change software for academic use, but restricting commercial firms
from exploiting your knowhow for their benefit.''

... this somehow reminds me of some long discussions I've seen pass by on sage-devel. Anyhow, the
discussion here is about announcing that it would now be possible to use SCIP inside Sage. I think this is
a good news. As far as I know, it has been possible to use other Ma's softwares inside Sage, and I don't
see why it couldn't be the same with SCIP.

According to Gurobi, they are the best non-proprietary LP optimization package that you can get. [see p.17 of http://www.gurobi.com/pdfs/benchmarks.pdf]

So basically these acedemics, I imagine funded with taxpayers' money, are preventing the general public to use software they paid for?

As for ZIB, in 2016, it had almost 8M€ in third party funding, see http://www.zib.de/institute/zib-in-numbers
If you go in the Research section, and you know that it is a German based institute, you may guest some companies that fund them.

That said, of course, the publish "Publish or Perish" has a role to play in the choice of license. But since I
don't want to spend time rewriting an optimization package, I will gladly cite them and say  that I used their
software.

My 2 cents...

JP

[Francesco Biscani]

On 14 April 2018 at 17:59, jplab <jeanphil...@gmail.com> wrote:

Hi,

Maybe I just give my basic knowledge about SCIP and ZIB.

Le vendredi 13 avril 2018 11:20:25 UTC+2, bluescarni a écrit :

http://scip.zib.de/academic.txt

"""
1. This license applies to you only if you are a member of a noncommercial
   and academic institution, e.g., a university. The license expires as
   soon as you are no longer a member of this institution.
"""

Perhaps including the very first sentence of the license would put a bit more context:

''This license for ZIB software is designed to guarantee freedom to share and 
change software for academic use, but restricting commercial firms
from exploiting your knowhow for their benefit.''

That's a declaration of intent, which is fine and all, but what matters is the rest of the license, where the exact terms are laid out.

 

... this somehow reminds me of some long discussions I've seen pass by on sage-devel. Anyhow, the
discussion here is about announcing that it would now be possible to use SCIP inside Sage. I think this is
a good news. As far as I know, it has been possible to use other Ma's softwares inside Sage, and I don't
see why it couldn't be the same with SCIP.

According to Gurobi, they are the best non-proprietary LP optimization package that you can get. [see p.17 of http://www.gurobi.com/pdfs/benchmarks.pdf]

So basically these acedemics, I imagine funded with taxpayers' money, are preventing the general public to use software they paid for?

As for ZIB, in 2016, it had almost 8M€ in third party funding, see http://www.zib.de/institute/zib-in-numbers
If you go in the Research section, and you know that it is a German based institute, you may guest some companies that fund them.

That said, of course, the publish "Publish or Perish" has a role to play in the choice of license. But since I
don't want to spend time rewriting an optimization package, I will gladly cite them and say  that I used their
software.

Actually what pissed me off the most was the fact that apparently you have to be a "member of a noncommercial and academic institution" to use SCIP for free (and "the license expires as soon as you are no longer a member of this institution"). So, again, if you are, say, an interested citizen, an independent researcher or a developer who does open source in his/her free time while being employed in the industry as your day job, apparently you cannot get access to a piece of software you paid for with your taxes (or, you have to pay for it twice - perhaps this sounds familiar to a certain other situation going on in academia :)). This seems to be a completely gratuitous restriction to me.

Cheers,

  Francesco.

[moritz]

Just a clarification regarding the licences:

PySCIPOpt actually is under a MIT licences.
https://github.com/SCIP-Interfaces/PySCIPOpt/blob/master/LICENSE

"Only" SCIP (and scipoptsuite) is under this academic license that is criticized in the replies below.

[Ralf Stephan]

On Friday, April 13, 2018 at 11:28:25 AM UTC+2, vdelecroix wrote:

On 13/04/2018 11:20, Francesco Biscani wrote:
> So basically these acedemics, I imagine funded with taxpayers' money, are
> preventing the general public to use software they paid for?

This is indeed a problem with PySCIPOpt. I don't think this is a bad
intention.

Ignoring casual users without bad intention? I begin to see a pattern.

Interestingly, Wolfram added a buy option for the hobbyist so realism is possible after all.

[moritz]

The licensing issue also makes installing more cumbersome, since you are supposed to open a browser and click around.

This could possibly be improved by
displaying the licensing information given and linked and the page scip.zib.de/download.php?fname=scipoptsuite-5.0.1.tgz directly in the terminal during installation process and letting the user input name, country and email right there (perhaps with meaningful defaults).
Then a curl request like the following could be used
`curl   -d tname='Your Name' -d email=yo...@email.address -d country='Mars' -d license=academic -d fname=scipoptsuite-5.0.1.tgz -L scip.zib.de/download.php  -o path/to/your/sage/upstream/scipoptsuite-5.0.1.tgz`
where the entries for tname, email and country have to be replaced with the user's input.

We definitely don't simply want to use a command like

`wget http://scip.zib.de/download/release/scipoptsuite-5.0.1.tgz`

to put the file in the upstream directory, because the user needs to acknowledge and agree to the license.

[Jeroen Demeyer]

I totally agree that it's a pity that this is not free software, but
that fact should not prevent it being an *experimental* package with
Sage bindings.