SOLVCON seminar invitation on December 8, 2015, Tuesday

12 views
Skip to first unread message

Taihsiang Ho

unread,
Nov 27, 2015, 2:13:42 AM11/27/15
to sol...@googlegroups.com
Hi,
there will be a seminar event and
we are glad to invite anyone who is interested in the CFD FOSS[1][2], SOLVCON[3].

The target audience of this seminar mainly focuses on CFD software developers,
but the seminar also includes the introduction for beginners who have no background of programming and the CFD knowledge for SOLVCON development.
The speaker, You-Hao Chang will introduce the necessary CFD knowledge for those beginners.

The detailed seminar information could be found at:

Looking forward seeing you!

[1] computational fluid dynamics
[2] Free and open-source


Cheers,
Taihsiang

Yung-Yu Chen

unread,
Dec 8, 2015, 7:47:43 AM12/8/15
to sol...@googlegroups.com
Hi,

Slides are updated to the seminar page https://solvcon.github.io/seminar/2015/first/index.html .

yyc

--
You received this message because you are subscribed to the Google Groups "solvcon" group.
To unsubscribe from this group and stop receiving emails from it, send an email to solvcon+u...@googlegroups.com.
To post to this group, send email to sol...@googlegroups.com.
Visit this group at http://groups.google.com/group/solvcon.
To view this discussion on the web visit https://groups.google.com/d/msgid/solvcon/CAGpVV1jfNLqQogLCK_Kv8fL6yB1ghXCLZSqs%2BCD0wc770u%2BM1A%40mail.gmail.com.
For more options, visit https://groups.google.com/d/optout.

--
Yung-Yu Chen
http://yyc.solvcon.net/
+886 (99) 129 4763

david.m...@gmail.com

unread,
Dec 9, 2015, 7:43:39 PM12/9/15
to solvcon
I enjoyed the meeting, thanks everyone. 

I hope to participate more in the future!

David

david.m...@gmail.com

unread,
Dec 10, 2015, 8:19:17 PM12/10/15
to solvcon
I'd like to wrap my head around the structure of all data tables (slide 8 of YY's).

Would it be possible for someone to just pickle a mesh (grid?) object for a small 2D rectangular mesh? A few hundred points max, like the left side of slide 5. 

Or just np.save() all 13 tables if pickling won't work. Pickle preferred (it's something I can just use without having to read a bunch of code first.)

Either post it or just e-mail to me at this addres.

Thanks! 

david.m...@gmail.com

unread,
Dec 10, 2015, 8:23:20 PM12/10/15
to solvcon
A small 3D example would be great also if it is easy and convenient - something like the right side of slide 6.   Thanks!

Yung-Yu Chen

unread,
Dec 10, 2015, 8:53:38 PM12/10/15
to sol...@googlegroups.com
The RHS at slide 6 was from an old mesh file that I can’t find right now.  But https://github.com/solvcon/solvcon/blob/master/test/data/cubic_t200mm.g is a 3D mesh that you can play with.

If you get SOLVCON built (it takes some time figuring out the dependency), you can use the following commands to view the mesh in browser:

$ scg mesh $SCSRC/solvcon/test/data/cubic_t200mm.g cubic_t200mm.html
$ open cubit_t200mm.html

I assume $SCSRC is the source directory you cloned to, and the runtime is OSX.

yyc


--
You received this message because you are subscribed to the Google Groups "solvcon" group.
To unsubscribe from this group and stop receiving emails from it, send an email to solvcon+u...@googlegroups.com.
To post to this group, send email to sol...@googlegroups.com.
Visit this group at http://groups.google.com/group/solvcon.

For more options, visit https://groups.google.com/d/optout.

Yung-Yu Chen

unread,
Dec 10, 2015, 8:56:18 PM12/10/15
to sol...@googlegroups.com
I think a solvcon.block.Block object is picklable, but it’s not an efficient data format so that I didn’t really use pickle for mesh data in SOLVCON.  Instead, I developed a home-grown data format for Block, and the code is at https://github.com/solvcon/solvcon/blob/master/solvcon/io/block.py .

https://github.com/solvcon/solvcon/blob/master/solvcon/command.py#L431 demonstrates how to load the Block file with the BlockIO class.

yyc


For more options, visit https://groups.google.com/d/optout.

David Mikolas

unread,
Dec 11, 2015, 3:03:01 AM12/11/15
to sol...@googlegroups.com
Thanks! But I think in this particular case pickle is THE efficient method because I actuall know how to use it. Or just np.save 13 times. 


You received this message because you are subscribed to a topic in the Google Groups "solvcon" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/solvcon/Dme4xtDPCMM/unsubscribe.
To unsubscribe from this group and all its topics, send an email to solvcon+u...@googlegroups.com.

To post to this group, send email to sol...@googlegroups.com.
Visit this group at http://groups.google.com/group/solvcon.

Yung-Yu Chen

unread,
Dec 11, 2015, 3:24:58 AM12/11/15
to sol...@googlegroups.com
Mere np.load() or np.save() won’t work for Block objects.  See https://github.com/solvcon/solvcon/blob/master/solvcon/io/block.py#L621 .  The boundary condition information needs to be taken care of.  I didn’t mention the boundary conditions in the talk because it complicates things too much.

Pickling probably works, but I don’t have a unit test for it because pickle isn’t a supported format.

There are already some sample mesh data in https://github.com/solvcon/solvcon/tree/master/test/data that you can play with.  But without loading them into memory there’s not much you can do.  The easiest way is to get SOLVCON working.  Then in solvcon.io.tests the unit tests will serve as examples.

yyc


For more options, visit https://groups.google.com/d/optout.

David Mikolas

unread,
Dec 11, 2015, 3:52:02 AM12/11/15
to sol...@googlegroups.com
YungYu, you are answering questions that I did not ask.

I'd like to see those 13 numpy arrays. That is precisely what I want to do. 

Yung-Yu Chen

unread,
Dec 11, 2015, 4:09:27 AM12/11/15
to sol...@googlegroups.com
Once you load the sample data into memory, you can get all arrays including those 13 ones.

yyc


For more options, visit https://groups.google.com/d/optout.

David Mikolas

unread,
Dec 11, 2015, 4:25:34 AM12/11/15
to sol...@googlegroups.com
OK, you've got the "what" part.  Now the rest...

"Would it be possible for someone to just pickle...
...Or just np.save() ..."

I'm hoping that someone who already has solvcon installed and running could just type a few lines - either

pickle.dump(object, filename)

or do something like:

names = ['ndcrd', 'fccnd', 'fcnml', 'fcara', 'clcnd', 'clvol',
          'fctpn', 'cltpn', 'clgrp', 'fcnds', 'fccls', 'clnds',
          'clfcs']
for name in names:
    array = np.array(getattr(the_object, name))
    np.save(name, array)

So the expected format of an answer to "Would it be possible for someone to..."

would be something the lines of "yes" or "no".


Yung-Yu Chen

unread,
Dec 11, 2015, 4:54:37 AM12/11/15
to sol...@googlegroups.com

david.m...@gmail.com

unread,
Dec 11, 2015, 10:27:55 PM12/11/15
to solvcon
Got them, thanks!!!


On Friday, November 27, 2015 at 3:13:42 PM UTC+8, Taihsiang Ho wrote:
Reply all
Reply to author
Forward
0 new messages