[yt-users] Trouble with SciPy Install

Matthew Turk matthewturk at gmail.com
Mon Jan 27 05:23:58 PST 2014


Hi Nathan,

On Sun, Jan 26, 2014 at 10:24 PM, Nathan Butcher <butchernate at gmail.com> wrote:
> Nathan,
>
> When I copied over the end of the error message to a new file only 50 of the
> 100 lines I tried to copy made it. I have attached file that actually
> contains the last 100 lines which includes the error I mentioned.
>
> Devin,
>
> I am trying to install yt with pip and getting an error stating "Blas
> libraries not found" and telling me that I can set the BLAS environment
> variable to resolve it. The HPCC has a BLAS module but the module help
> command gave me no information so I have contacted the HPCC people about
> that module. I will let you know what I hear back and if I get it working.
>
> Nathan Butcher

This is one of the harder things to get around with the SciPy install,
which is getting a working shared/static setup that can be linked
against by SciPy; in fact, it's one of the reasons that we stopped
doing scipy by default.

-fPIC is a compiler optimization, only available to statically linked
libraries (which Python by default does not use) that improves
portability -- essentially, when running a function from a library,
you end up in a situation where the position of the code in memory
does not govern how it executes.  More info can be found here:

http://en.wikipedia.org/wiki/Position-independent_code

The reason this is done is that if the code is *not* position
independent, you can optimize it slightly further than if it needs to
retain information about its relative position.

Regardless, the solution tends to be to build your own
BLAS/ATLAS/LAPACK stack, or disabling them during the build of SciPy.
If you are doing a large amount of linear algebra, you probably want
to be careful about how you choose to do this.  But, if you're not,
you can probably even get away with installing SciPy from source and
disabling them, like so:

ATLAS=None BLAS=None LAPACK=None python2.7 setup.py install

which will build workalike libraries inside scipy rather than using
the system ones.  But, again, this may not be right for your use case.

Best of luck,

Matt

>
>
> On Sun, Jan 26, 2014 at 6:53 PM, Nathan Goldbaum <nathan12343 at gmail.com>
> wrote:
>>
>> Hi Nathan,
>>
>> I don't see the error you're referring to in the log file you attached.
>>
>> Can you paste the full install log into an unlimited pastebin
>> (gist.github.com or pastebin.com) and send us the link?
>>
>> Cheers,
>>
>> Nathan Goldbaum
>>
>> On Sunday, January 26, 2014, Devin Silvia <devin.silvia at gmail.com> wrote:
>>>
>>> Hi Nathan,
>>>
>>> I've recently encountered the same issue.  I've tried a number of
>>> things (unsuccessfully).  I'm not entirely sure what it will take to
>>> get things working.  I'll let you know if I manage to solve the issue.
>>>  Please let me know if you manage to do the same.  At some point, it
>>> might be worth talking to the HPCC folks directly.
>>>
>>> Cheers,
>>> Devin
>>>
>>> On Sat, Jan 25, 2014 at 8:18 PM, Desika Narayanan
>>> <desika.narayanan at gmail.com> wrote:
>>> > Does installing yt without scipy and then afterward pip installing
>>> > scipy work?
>>> >
>>> > (This worked for me on a similar problem on an OS X install some time
>>> > ago)
>>> >
>>> >> On Jan 25, 2014, at 5:43 PM, Nathan Butcher <butchernate at gmail.com>
>>> >> wrote:
>>> >>
>>> >> Hello,
>>> >>
>>> >> I am trying to install yt with SciPy on Michigan State University's
>>> >> HPCC system and am running into errors that I cannot resolve by following
>>> >> the instructions in the install script. I have tried uncommenting each of
>>> >> the NUMPY_ARGS statements and had no success. I have attached the log file
>>> >> from my most recent installation attempt as the error message is long. My
>>> >> guess is that it has to do with the "recompile with -fPIC" message but I
>>> >> cannot figure out what that means.
>>> >>
>>> >> Can anybody explain what this error message means and how to resolve
>>> >> it.
>>> >>
>>> >> Thank you,
>>> >> Nathan Butcher
>>> >> <short_yt_install_471.log>
>>> >> _______________________________________________
>>> >> yt-users mailing list
>>> >> yt-users at lists.spacepope.org
>>> >> http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org
>>> > _______________________________________________
>>> > yt-users mailing list
>>> > yt-users at lists.spacepope.org
>>> > http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org
>>>
>>>
>>>
>>> --
>>> Devin W. Silvia
>>> Research Associate
>>> Department of Physics and Astronomy
>>> Michigan State University
>>> www.devinsilvia.com
>>> _______________________________________________
>>> yt-users mailing list
>>> yt-users at lists.spacepope.org
>>> http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org
>>
>>
>> _______________________________________________
>> yt-users mailing list
>> yt-users at lists.spacepope.org
>> http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org
>>
>
>
> _______________________________________________
> yt-users mailing list
> yt-users at lists.spacepope.org
> http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org
>



More information about the yt-users mailing list