Puma-EM Activity
Brought to you by:
ides
Activity for Puma-EM
-
Idesbald Van den Bosch committed [68e373] on Code
Changed MPI C++ bindings to C bindings, as the C++ bindings have been deprecated in recent MPI
-
Changed import from scipy to numpy.
-
Puma-EM released /puma-em/0.6.8/Puma-EM-0.6.8.tar.gz
-
Changed python to python3 in makepackage.py.
-
Changed copyright year to 2024.
-
Bernd committed [e5e99f] on Code
Code was broken due to various software changes (gmsh, python3, numpy/scipy). Various fixes just to enable the code to run again.
-
Nicolas Ruel posted a comment on discussion Help
Hi Ioanna, I'm trying to do this simulation but I'm faces some troubles. I don't find this graph for the monostatic_plate. Could you send your simulation_parameters file or explain me how you simulate those graphs ? Thanks !
-
Hi all, I'm actually in the same case as you were both and don't understand how to fix it. Can you explain with some more details your procedure Thomas ? Emily if you found the solution I would be glad to hear it :) Looking forward to your reply. Kind regards, Nicolas
-
Ioanna Did posted a comment on discussion Help
Dear Ides, I was testing a target I created myself (.msh file created in GMSH) at 1 GHz.
-
Hi ioanna, what target/freq were you testing? For me, the simulation did not succeed at all. On Thu, Sep 26, 2019 at 4:52 PM Ioanna Did ioanna@users.sourceforge.net wrote: Hello, I am trying to find the monostatic RCS of a UAV using Puma-EM. I tried to run a few simulations with a different angle step size to decide how big the step size can be without greatly deteriorating the accuracy of the plotted result. What I noticed was that I got different RCS values for the same angle and, most importantly,...
-
Hi Ioanna, thanks for the feedback. Indeed, the mesh is often the issue in this type of simulations, and spotting it can be difficult, especially with millions of edges! On Thu, Sep 26, 2019 at 5:11 PM Ioanna Did ioanna@users.sourceforge.net wrote: Hello, In case someone faces a similar issue and goes through this thread, the issue was that the CAD model was too detailed and had self-intersections. Commercial software also failed to complete a simulation with it. In the end, based on the original...
-
Hello, I am trying to find the monostatic RCS of a UAV using Puma-EM. I tried to run a few simulations with a different angle step size to decide how big the step size can be without greatly deteriorating the accuracy of the plotted result. What I noticed was that I got different RCS values for the same angle and, most importantly, some outliers, which seem to be 100 or 1000 times bigger than the values produced by other simulations at that value. I have highlighted the values in question in Image...
-
Hello, In case someone faces a similar issue and goes through this thread, the issue was that the CAD model was too detailed and had self-intersections. Commercial software also failed to complete a simulation with it. In the end, based on the original design, I built a simpler model using basic shapes such as ellipsoids and spheres and the simulations ran smoothly. Thank you, Ides, for all your help and suggestions! Kind regards, Ioanna
-
Hello, I am trying to find the monostatic RCS of a UAV using Puma-EM. I tried to run a few simulations with a different angle step size to decide how big the step size can be without greatly deteriorating the accuracy of the plotted result. What I noticed was that I got different RCS values for the same angle and, most importantly, some outliers, which seem to be 100 times bigger than the values produced by other simulations at that value. I have highlighted the values in question in Image 1 and...
-
Hi Ioanna, No I don't have an idea. Normally this stage takes a few seconds, even less here since there is no exchange at all since you only use one process... Short of trying the simulation myself, I cannot help you, sorry. Maybe try with two processes? On Tue, Jun 25, 2019 at 9:47 AM Ioanna Did ioanna@users.sourceforge.net wrote: Dear Ides, In order to scale the UAV that I would like to run a simulation on, I used -meshscale in GMSH, as -scale did not seem to work. The simulation stops progressing...
-
Dear Ides, In order to scale the UAV that I would like to run a simulation on, I used -meshscale in GMSH, as -scale did not seem to work. The simulation stops progressing at 'Exchanging Z_near blocks for preconditioner construction", although it doesn't crash altogether. I've left it on for days, which is why I believe it is not a matter of waiting for the results to be calculated. I have attached the output log so far. Do you have any suggestions or any thoughts about what this means? Thank you...
-
Dear Ides, In order to scale the UAV that I would like to run a simulation on, I used -meshscale in GMSH, as -scale did not seem to work. The simulation stops progressing at 'Exchanging Z_near blocks for preconditioner construction", although it doesn't crash altogether. I've left it on for days, which is why I believe it is not a matter of waiting for the results to be calculated. I have attached the output log so far. Do you have any suggestions or any thoughts about what this means? Thank you...
-
stem posted a comment on discussion Help
and this is my code
-
Dear professor now ，I use puma for calculating the monostatic RCS，but I find whatever the side length of the cube ，the calculation results are the same ，I also find different diameter of sphere also have the same monostatic RCS.I can not find a solution. I am here to ask for your help. Looking forward to hearing from you,thank you very much
-
Dear Ioanna, this is always a problem when the design and/or mesh is from a third party. The best is to design it yourself, that way you control the process. If this is not possible, then the designer should do his work according to your requirements. If this is not possible, then a back-and-forth is as inevitable as it is inefficient :( As to your question: sometimes it is easy to see what the mesh problems are, sometimes it is not. Some CAD software duplicate points, and so the surfaces are artificially...
-
Dear Ides, The CAD file in question is in STEP format and I have tried both - meshing it with GMSH and then using the .msh file as input to Puma and - merging the .stp with a .geo file and then meshing the .geo file through Puma. The result was the same. The mesh size is indeed undoubtedly wrong, however I didn't make the mesh finer because my computer didn't have enough memory to complete fine meshing. I previously assumed that the mesh size would only affect the correctness of the result, not crash...
-
Dear Ioanna the number of unknowns is given at the start of the run: Number of RWG = 7656. A very small number. So something is wrong with your input mesh file. A first hint is that the simulation frequency is 3GHz, and that the wingspan of the drone is 6.5m. So the number of unknowns should be around 0.5-1 million (with an average mesh size of 1cm per element). A second hint is that your target seems to be made of 12 unconnected surfaces (look at lines 98-99 of your log file). a third hint is that...
-
Dear Ides, Thank you for your quick reply. My computer is a laptop with 8 GB of RAM. How can I compute the number of unknowns? I have attached the output file, in case it can give you further insight. I understand that the average RWG length is unacceptable, but I suppose that would only affect the accuracy of the result and I was wondering if there is another problem (eg with the model geometry) before looking for a computer able to mesh a shape of such dimensions. Mine runs out of memory and turns...
-
Dear Ionna, can you tell me how many unknowns you have, and what are the characteristics (memory) of your computer? Kind regards Ides On Thu, Mar 14, 2019 at 3:26 AM Ioanna Did ioanna@users.sourceforge.net wrote: Hello, I am trying to run Puma (monostatic RCS) for a UAV, whose largest dimension is 2.1m. My efforts so far have been unsuccessful, as the simulation stops either at the integration part or during the cube distribution. In detail: - The model is in STEP format, so I cannot use the lc parameter...
-
Hello, I am trying to run Puma (monostatic RCS) for a UAV, whose largest dimension is 2.1m. My efforts so far have been unsuccessful, as the simulation stops either at the integration part or during the cube distribution. In detail: The model is in STEP format, so I cannot use the lc parameter to setup GMSH. I merged it to an empty .geo file, as suggested online by GMSH users, to handle it more easily. I thought of adding "-clscale" to your GMSH command and using it or using Mesh.CharacteristicLengthFactor....
-
Thomas Pairon posted a comment on discussion Help
Hello Emily, I had the same problem as you. The problem was I did not have right to create the folder tmp0 in simuDir. I solved the problem by extending permissions on the code files. Hope it will help, Thomas
-
Dear Emily, no I don't know: it is the first time that I see it. As I said, it is near impossible for me to debug at a distance. I use Ubuntu 18.10, and I have no issues at all.... Sorry. On Wed, Oct 31, 2018 at 8:13 AM EmilySun emily-sun@users.sourceforge.net wrote: Dear Ides, Thank you for your advices. I run the "run.sh" program step by step, and check files in Simdir folder. And in file "CPU_time_read_MLFMA_mesh_part1.txt", I found the following error: File "/home/jiajingsun/Puma-EM/code/read_mesh.py",...
-
EmilySun modified a comment on discussion Help
Dear Ides, Thank you for your advices. I run the "run.sh" program step by step, and check files in simuDir folder. And in file "CPU_time_read_MLFMA_mesh_part1.txt", I found the following error: File "/home/jiajingsun/Puma-EM/code/read_mesh.py", line 264, in read_mesh_GMSH_2 triangles_nodes[index, :] = tmp[-3:] ValueError: cannot copy sequence with size 2 to array axis with dimension 3 Also, in the tmp0 folder of simuDir, almost all the folders are empty except for the "octtree_data" with one CUBES_DISTRIBUTION.txt...
-
Dear Ides, Thank you for your advices. I run the "run.sh" program step by step, and check files in Simdir folder. And in file "CPU_time_read_MLFMA_mesh_part1.txt", I found the following error: File "/home/jiajingsun/Puma-EM/code/read_mesh.py", line 264, in read_mesh_GMSH_2 triangles_nodes[index, :] = tmp[-3:] ValueError: cannot copy sequence with size 2 to array axis with dimension 3 Do you have any idea why is this happening? Looking forward to your reply. Kind regards, Emily Details: Traceback...
-
Dear Emily, it is hard to tell you what's wrong without having access to the computer. My advice is: run the "run.sh" program step by step, and check where it fails. It seems here that the mesh is not read properly, although I don't see any error message... As I said, I cannot investigate at a distance, sorry. Kind regards Ides On Tue, Oct 30, 2018 at 2:44 PM EmilySun emily-sun@users.sourceforge.net wrote: Dear Ides and PUMA-EM Community, I recently downloaded Puma from Github and am now trying to...
-
Dear Ides and PUMA-EM Community, I recently downloaded Puma from Github and am now trying to install it on my Ubuntu 16.04. I followed the guide and installed all the 3rd part libraries. And I runed "make clean" and "make libs" successfully with no error. However, when I run "./run.sh", I got the following errors : IOError: [Errno 2] No such file or directory: '/home/jiajingsun/Puma-EM-master/simuDir/tmp0/mesh/N_RWG.txt' The details of runing "./run.sh" are as follows. number of processes = 1 input...
-
Hello Ioanna, the plates have an analytical solution that is only an approximation which is good near perpendicular angles of incidence. So no surprise to see big differences the further one goes away from the nadir. The sphere is a different case, as there is an analytical solution. The differences may come from the discretization of the sphere for Puma, or from the precision chosen, or from the step in frequency; or from so many different sources. To me the results look pretty good, but one can...
-
Attaching the images because the links don't seem to work, at least for me
-
Hello, I've been trying to simulate a square strip of 5λ x 5λ and a sphere and compare the result to the analytical (textbook) solutions using Matlab. 1) In the first case, I get So there is a difference at high observation angles and also the lower values are not as low as the ones from the analytical solution. Trying to investigate what might be wrong, I found that for the analytical solution, the values change greatly depending on the angle step size. The graph below is the same equation plotted...
-
Hello, I've been trying to simulate a square strip of 5λ x 5λ and a sphere and compare the result to the analytical (textbook) solutions using Matlab. 1) In the first case, I get So there is a difference at high observation angles and also the lower values are not as low as the ones from the analytical solution. Trying to investigate what might be wrong, I found that for the analytical solution, the values change greatly depending on the angle step size. The graph below is the same equation plotted...
-
Dear, can you provide me with more data? the RCS graph with frequency, the number of unknowns with frequency, the number of iterations per frequency. On Tue, Jul 10, 2018 at 8:03 AM, Ticket 2 2@bugs.puma-em.p.re.sourceforge.net wrote: [bugs:#2] https://sourceforge.net/p/puma-em/bugs/2/ Abrupt change of monostatic RCS as a function of frequency* Status: open Group: v1.0_(example) Created: Tue Jul 10, 2018 06:03 AM UTC by Anonymous Last Updated: Tue Jul 10, 2018 06:03 AM UTC Owner: nobody When calculating...
-
Anonymous created ticket #2
Abrupt change of monostatic RCS as a function of frequency
-
Hi Jeson, Did you compile the Puma-EM code by running "make libs"? If yes, please send me the complete output message of the code, so I can try to help. Kind regards On Wed, May 23, 2018 at 11:16 PM, Jeson jchen@users.sourceforge.net wrote: Hello, Since 2017, Microsoft implement Winodws 10 Linux subsystem. With Ubuntsu app, It seems possible to install Puma-EM in Linux subsystem of Windows. (At lest installation of blitz is fine. No error in testing with "puma-em$ make libs") After installing all...
-
Jeson modified a comment on discussion Help
Hello, Since 2017, Microsoft implement Winodws 10 Linux subsystem. With Ubuntsu app, It seems possible to install Puma-EM in Linux subsystem of Windows. (At least installation of blitz is fine. No error when testing with "puma-em$ make libs") After installing all the libraries listed in manual, I tried to run default setting with " puma-em$ ./run.sh " in above Winodws 10 Linux subsystem and there is an error poping out: IOError: [error 2] No such file or directory: '/mnt/c/Users/jchen/Downloads/Puma-EM/simuDir/tmp0/mesh/N_RWG.txt'...
-
Hello, Since 2017, Microsoft implement Winodws 10 Linux subsystem. With Ubuntsu app, It seems possible to install Puma-EM in Linux subsystem of Windows. (At lest installation of blitz is fine. No error when testing with "puma-em$ make libs") After installing all the libraies listed in manual, I tried to run default setting with " puma-em$ ./run.sh " in above Winodws 10 Linux subsystem and there is an error poping out: IOError: [error 2] No such file or directory: '/mnt/c/Users/jchen/Downloads/Puma-EM/simuDir/tmp0/mesh/N_RWG.txt'...
-
Hello, Since 2017, Microsoft implement Winodws 10 Linux subsystem. With Ubuntsu app, It seems possible to install Puma-EM in Linux subsystem of Windows. (At lest installation of blitz is fine. No error in testing with "puma-em$ make libs") After installing all the libraies listed in manual, I tried to run default setting with " puma-em$ ./run.sh " in above Winodws 10 Linux subsystem and there is a error pops out: IOError: [error 2] No such file or directory: '/mnt/c/Users/jchen/Downloads/Puma-EM/simuDir/tmp0/mesh/N_RWG.txt'...
-
Dear Mike, no it only supports flat triangles. On Tue, Nov 7, 2017 at 11:32 PM, mike mike1@users.sf.net wrote: Does Puma-EM support a geomatry that is mshed with curvilinear triangles where each element has six nodes? Thx High-order Curvilinear Triangles https://sourceforge.net/p/puma-em/discussion/744307/thread/14a49db1/?limit=25#44f3 Sent from sourceforge.net because you indicated interest in https://sourceforge.net/p/puma-em/discussion/744307/ To unsubscribe from further messages, please visit...
-
mike posted a comment on discussion Help
Does Puma-EM support a geomatry that is mshed with curvilinear triangles where each element has six nodes? Thx
-
tony_tin modified a comment on discussion Help
Dear ides: I am very thank to you for your Puma-em. I have some questions. 1. I dont...
-
Dear ides: I am very thank to you for your Puma-em. I have some questions. 1. I dont...
-
Dear ides: I am very thank to you for your Puma-em. I have some questions. 1. I dont...
-
Dear ideas: I am very thank to you for your Puma-em. I have some questions. 1. I...
-
Dear ideas: I am very thank to you for your Puma-em. I have some questions. 1. I...
-
Hello Christophe, it is possible, as demonstrated by Guillaume Sylvand in his thesis...
-
Christophe Picard posted a comment on discussion Help
Dear Ides and PUMA-EM Community Do you know if it's possible to adapt PUMA-EM to...
-
Anonymous created ticket #10
Simulate EM field given radiation pattern
-
Puma-EM 0.6.7
-
-
Changed the year for the copyright.
-
Phase correction of initial currents guess from...
-
Changed the way the sampling points are calcula...
-
Hello Bruno Normally the code is run like this: ${PYTHON_CMD} code/RCS_MLFMA.py --inputdir...
-
Bruno Carpentieri posted a comment on discussion Help
Hello Ides. I have successfully compiled and run Puma-EM on my Ubuntu 15.10 local...
-
xiangdaopu modified a comment on discussion Help
Here is the testing result: The testing model is cubi.geo, the frequecy is 3GHz,...
-
Here is the testing result: The testing model is cubi.geo, the frequecy is 3GHz,...
-
Here is the testing result: The testing model is cubi.geo, the frequecy is 3GHz,...
-
Here is the testing result: The testing model is cubi.geo, the frequecy is 3GHz,...
-
Hi all Puma-EM users may concern, Intuitively, the accuracy of the octree in Puma-EM...
-
Hi Anders, good that you found out. However, sincos is very handy, as it computes...
-
Anders Petersson posted a comment on discussion Help
Hi Ides, It turns out that the functions sincos() and sincosf() are not standard...
-
Hello Anders, I am a bit baffled: sincos is a standard function in gcc. It works...
-
I have not been able to find a solution. The routines sincos() and sincosf() must...
-
Hello Anders, did you solve your problem? On Thu, Mar 17, 2016 at 10:20 PM, Anders...
-
Hello Anders, did you solve your problem?
-
I recently downloaded Puma from sourceforge and am now trying to install it on my...
-
Small code changes, no consequences.
-
Replaced a 2D Blitz Array with a std vector of ...
-
Changed array variable to reference to this array
-
Small fixes and changes, nothing important, jus...
-
Small changes in far-field multiplication
-
Puma-EM 0.6.6
-
-
Added a new target EMCC_pipe
-
Corrected a bug that prevented high numbers of ...
-
Puma-EM 0.6.5
-
-
Changed copyright year
-
Corrected an int overflow bug in mesh_cubes.cpp
-
Small changes to RWGs_renumbering.cpp
-
Hello all Puma-EM users, The mesh_functions_seb[mesh_functions_seb.cpp] needs to...
-
Sorry for all, I'm wrong when I do the simulation in FEKO (my simulation setting...
-
You are absolutely right! From that single comparison, I can't say which one is correct!...
-
You are absolutely right! From that single comparison, I can't say which one is correct!...
-
jpadam posted a comment on discussion Open Discussion
technically, you didn't prove that Puma-EM failed... only the comparison failed :)...
-
This is the model(trihedral illuminated by plane wave, incident angle is theta=45...
-
Compare with FEKO, the result is significant different. Who can find out the problem...
-
Compare with FEKO, the result is significant different. Who can find out the problem...
-
Compare with FEKO, the result is significant different. Who can find out the problem...
-
The simulation result is:
-
This is the model(trihedral illuminated by plane wave, incident angle is theta=45...
-
When I test the RCS calculation in Puma-EM, I find out this problem.
-
Added broadcast of params_simu, for computers w...
-
For RCS scan plane, if the scan angle respect to \theta, the easiest way is to rotate...
-
For RCS scan plane, if the scan angle respect to \theta, the easyest way is to rotate...
-
You are right. I have been thinking about it since you mentioned it. I will correct...
1 >
