Related with my PhD study I had to generate my own C++ code to solve 2D wave equation with NRBC (Non-Reflecting Boundary Condition). Now I am able to simulate a wide range of geometry inside my simulation area (limited with physical memory, no HDD swap 
), thanks to Carnk-Nicolson method. Sources, boundaries and measurement points can be selected arbitrarily, giving a big opportunity to model almost every "symmetric" enclosure. The data can be collected from the simulation field in time domain, and then can be processed with a FFT routine. The code enables the user to observe the radiation pattern of a wave-guide also, but the setback is, the wave-guide must be symmetric, like a cone or cylinder. For non-symmetric wave-guides the simulation must be done in various axis separately.
The code will be presented on my web page, (open source) when I am done with my PhD. Before that, I will be performing some measurements in order to show that the software is generating similar results with the measurements, and can not spend much time on the gui. I will surely appreciate for those who can code me a simple gui for to place the sources, boundaries and measurement points on the 2D simulation field, or a semi-finished open source gui can be useful also. Since the current software is coded using OpenGL routines, I would be looking forward keeping it that way.
-Onur
), thanks to Carnk-Nicolson method. Sources, boundaries and measurement points can be selected arbitrarily, giving a big opportunity to model almost every "symmetric" enclosure. The data can be collected from the simulation field in time domain, and then can be processed with a FFT routine. The code enables the user to observe the radiation pattern of a wave-guide also, but the setback is, the wave-guide must be symmetric, like a cone or cylinder. For non-symmetric wave-guides the simulation must be done in various axis separately.The code will be presented on my web page, (open source) when I am done with my PhD. Before that, I will be performing some measurements in order to show that the software is generating similar results with the measurements, and can not spend much time on the gui. I will surely appreciate for those who can code me a simple gui for to place the sources, boundaries and measurement points on the 2D simulation field, or a semi-finished open source gui can be useful also. Since the current software is coded using OpenGL routines, I would be looking forward keeping it that way.
-Onur
Attachments
Wow, very cool! This is basically what I use Matlab and the PDE toolbox to do. However, I regularly run out of memory for high frequencies even with 3GB of swap file and 1GB of RAM. I actually find the GUI to be not very useful - I like programming my shapes in the code and just using the GUI to verify that I got things in the right place and the right size. Then I do boolean operations to produce the final geometry. Doing things in the code also allows me to directly program things like tractrix horn flairs, etc., so I can just input a flair frequency and throat diameter, and everything else is automatically calculated in a loop.
MatLab, Brrrrr!
I know MatLab since many of my friends suggested using matlab for to solve pdes. But, learning how to code matlab is far more difficult then finding a way to solve pdes! It is not something useful unless your aim is to learn how to code matlab as your phd thesis I don't even want to talk about the memory leakages, memory management faults, and writing data to FAT as if writing to a swap file.
What I've coded does also need high amounts of physical memory, since I can not code a swap file routine. On the other hand, you can choose any frequency you like, you can add boundaries anywhere you like and yet the boundaries are non-reflecting. You can define sources with any geometry, anywhere you like. All this, open source 3.2 MB Unlike the case with matlab 7.1, which needs 3 CDs.
Numerical Recepies and Maciej Matyka is all you need But the open source he provides is depending on a ready interface, which is hard to get along with. Therefore I am desperate for a gui!
http://panoramix.ift.uni.wroc.pl/~maq/eng/
www.nr.com
I know MatLab since many of my friends suggested using matlab for to solve pdes. But, learning how to code matlab is far more difficult then finding a way to solve pdes! It is not something useful unless your aim is to learn how to code matlab as your phd thesis
I don't even want to talk about the memory leakages, memory management faults, and writing data to FAT as if writing to a swap file. What I've coded does also need high amounts of physical memory, since I can not code a swap file routine. On the other hand, you can choose any frequency you like, you can add boundaries anywhere you like and yet the boundaries are non-reflecting. You can define sources with any geometry, anywhere you like. All this, open source 3.2 MB
Unlike the case with matlab 7.1, which needs 3 CDs.Numerical Recepies and Maciej Matyka is all you need
But the open source he provides is depending on a ready interface, which is hard to get along with. Therefore I am desperate for a gui!http://panoramix.ift.uni.wroc.pl/~maq/eng/
www.nr.com
Excuse me...... matlab must be the most simple programming language availeble.. for a Phd student.. such a thing souldn't be the least obstacle....... there's just one drawback : slow execution... but mostly you win that time back with the abality to write your code so much faster in matlab...
BTW
1 cd for the software, 1 cd for the docmentation, 1 cd for the documentation PDFs....... compare that to 3.5Gb of Visual Studio.
Thijs....matlab fan!
BTW
1 cd for the software, 1 cd for the docmentation, 1 cd for the documentation PDFs....... compare that to 3.5Gb of Visual Studio.
Thijs....matlab fan!
Lets not bring this to a holly war between matlab and C There can be no winners, and my gui goes up in flames in between !
I guess this software may be used by many of us with ease with a proper gui. I have to prepare a txt file for sources, boundaries (enclosure geometry) and measurement points. It is hard when you consider dealing with co-ordinates for each enclosure.
There has been many postings about reflected waves in enclosures and maybe one of us may optimize an enclosure and if that someone doesn't have any commercial concern, he may be willing to share it with us for DIY purposes. That is what I am after. That must be what the one, who will prepare the gui, must be after too
The code will be freeware and a sample exacuteable will be presented on my web page right after I am done with my Phd.
-Onur
There can be no winners, and my gui goes up in flames in between !I guess this software may be used by many of us with ease with a proper gui. I have to prepare a txt file for sources, boundaries (enclosure geometry) and measurement points. It is hard when you consider dealing with co-ordinates for each enclosure.
There has been many postings about reflected waves in enclosures and maybe one of us may optimize an enclosure and if that someone doesn't have any commercial concern, he may be willing to share it with us for DIY purposes. That is what I am after. That must be what the one, who will prepare the gui, must be after too
The code will be freeware and a sample exacuteable will be presented on my web page right after I am done with my Phd.
-Onur
So as presented, it will be possible to load geometry from the text file? That would still be my preferred scenario, or at least to have that option. Of course I will probably keep going with Matlab since I have so much stuff developed for it already, but it sounds like basically the same thing.
I will risk exposing my ignorance of these matters, but are you asking about a pleasant method to define geometry for use in your simulation code? One can construct geometry in autocad or any other software which is capable of exporting such geometry as a DXF file. Some of these software packages may be free, and possibly open source.
DXF is very well documented, geometry contained within can easily be accessed.
DXF is very well documented, geometry contained within can easily be accessed.
importing geometry from such software can be a good idea. I am familiar with DXF file format, but making a stand-alone software can be more attractive, no one has to deal with AutoCad, or a file converter software, just 3 or 5 Megs of download and you can simulate the radiation pattern of your horn.
As I have said before, the purpose of the hobbiest should not be learning the software, but using it for his purpose. As the programmer makes the software easy to use, he helps the user to finish his design in a shorter time.
But, anyway, thank you, you have given me a very good idea I will use DXF for all my data input purposes. On the other hand, I don't think I can publish the code in that fashion, I will surely try to fix a simple editor.
-Onur
As I have said before, the purpose of the hobbiest should not be learning the software, but using it for his purpose. As the programmer makes the software easy to use, he helps the user to finish his design in a shorter time.
But, anyway, thank you, you have given me a very good idea
I will use DXF for all my data input purposes. On the other hand, I don't think I can publish the code in that fashion, I will surely try to fix a simple editor.-Onur
Two links to check out, may be some useful information.
It is GPL .
Perhaps the user interaction required to define the geometry is not too difficult, and also you might supply a template cad file with standard elements which can be modified by the user.
http://www.ribbonsoft.com/qcad.html
http://www.ribbonsoft.com/dxflib.html
It is GPL .
Perhaps the user interaction required to define the geometry is not too difficult, and also you might supply a template cad file with standard elements which can be modified by the user.
http://www.ribbonsoft.com/qcad.html
http://www.ribbonsoft.com/dxflib.html
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.