Thanks a lot Sir.....for such a detailed answer..... this has really encouraged
be to delve deeper in these areas.... I will try my best to meet your
expectations, while writing my proposal....One thing is for sure, whether I get
selected or not... your concept clarification and your advices, regarding how
to approach these projects, are really going to be helpful throughout my
life.... Thank u once again....
Regards,Aarjavee S. Kamdar
On Thursday, 26 March 2015 7:03 PM, John Pye <[email protected]> wrote:
Hi Aarjavee
For the proposal format, please see http://ascend4.org/GSOC2015.
Some comments below:
On 25/03/15 19:50, Aarjavee S. Kamdar wrote:
Respected John Pye Sir,
I am Aarjavee Kamdar, a third year Computer Engineering student from
Government Engineering College, Rajkot. Firstly, I would like to thank you and
your team for maintaining such good open source software for the public use. I
just read the ideas page of ASCEND and I am keenly interested in contributing
to your software.
I am familiar with the languages: c, c++, python and am willing to learn
other languages required in the following projects that interested me a lot:
1) Improvements to conditional Modelling syntax - willing to learn Flex /
Bison
This work would start with scanner/parser work, but it would go deeper than
that, into the data structures that support our CONDITIONAL, WHEN, EVENT
syntax, and into how we pass back debugging messages to the user. I will
require an elegant approach. I think that the hard part is probably not the
scanner/parser part. You might like to compare/contrast ASCEND syntax with that
of Modelica. We can do a lot better here on usability!
2)Dynamic Modelling Improvements
Our state of the art here is the 'ksenija2' branch in our repository. I am
very keep to improve our support for hybrid discrete/continuous modelling.
Ksenija and I recently identified what looks like some bugs in the handling of
ksenija2/models/johnpye/solardynamics_simple.a4c model. We would like to
improve the debugging tools associated with the new boundary-handling IDA
solver, because debugging these event-driven models is difficult; we need to
give more support to the modeller to help them fix their bugs.
3)GUI Improvements - ready to learn to read Tcl/Tk code
The main GUI development is in Python these days. A grab-bag of ideas right
now:
* better debugging of EVENTs in the ksenija2 branch (mentioned also above)
* better handling of solver interruption
* possible live-plotting of observed variables during integration progress
* improved line-routing in the canvas GUI, giving better visual feedback from
canvas model solving (eg green box means that part of the model is solved)
* various fixed for the 'tree view' of the main GUI window, eg hiding/showing
variables/relations/WHENs, etc, improved usability for entering values
* some solution for allowing celsius temperatures to be entered/reported in
the GUI. Currently we can only do absolute temperatures. We need a nice elegant
solution. without too much hacking.
4)Parallel optimization solver for ASCEND
We need someone with strong parallel programming experience for this one. We
would like to assess options for solvers like particle swarm optimisation,
genetic algorithnms, etc. This is quite a different class of solver to what we
have currently. I think it would be very interesting to see how productively
ASCEND can be used for that type of problem, and whether we can build a
sensible framework that's useful for the modeller. This kind of problem become
especially interesting when there are non-continuous functions being optimised,
eg models with WHEN statements, or models with many local minima.
Please kindly guide me how to approach these projects. Also, Please do let me
know what format of proposal would you expect.
Thanking you, Regards, Aarjavee S. Kamdar
Hope this is helpful (also to others who might be working on submitting ideas).
As I always say to people interested in our project, *please* don't just
repeat what I'm describing above. Break it down and show that you can go to the
next step, eg what exactly would be implemented, where is the code that does
this stuff and what needs to be changed, The more you can show you have done
already, and the more you can show that you *really* understand what you would
be doing, the better your proposal. And of course, *use* the program. Don't
write a proposal if you haven't at least built and run ASCEND, and played
around with some models.
Look forward to receiving your proposal!
Cheers
JP
------------------------------------------------------------------------------
Dive into the World of Parallel Programming The Go Parallel Website, sponsored
by Intel and developed in partnership with Slashdot Media, is your hub for all
things parallel software development, from weekly thought leadership blogs to
news, videos, case studies, tutorials and more. Take a look and join the
conversation now. http://goparallel.sourceforge.net/
_______________________________________________
Ascend-sim-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/ascend-sim-users
