Closed urp1 closed 7 years ago
You've incorrectly identified QP
as coinciding with vertex V2
in your call to my code, whereas in fact it seems QP
coincides with V3
in the example against which you are comparing.
If I modify your call to my code to read
double QP[3] = {0.05, 0.1, 0.0};
then I get -1.86352e-6
which is the answer you want.
Yes you are right. The result with Analytic formula for that case was -1.03e-6, I jumbled the vertex.
Nevertheless, if set QP to the value you indicated then I get: (1.96546e-06,0).
Is it possible that the code on the website (http://homerreid.dyndns.org/scuff-EM/SingularIntegrals/) is different then the code that you have? Perhaps some issues were resolved afterwards?
I appreciate your help.
I don't think so, but you can easily test this by downloading and installing SCUFF-EM and then compiling the test code against libscuff.
To do this, add the following two lines to the top of the test code:
#include <TaylorDuffy.h>
using namespace scuff;
Then compile and link with flags such as
CPPFLAGS=-I/path/to/scuff-em-installation/include/scuff-em
LDFLAGS=-L/path/to/scuff-em-installation/lib
with -lscuff
added to the linker command line.
The version in scuff-em package gives correct results! So there must be some bug in the Singularity extraction package on the website.
Thanks for your help!
Thanks for identifying this issue.
I updated the website documentation to eliminate the old buggy code and to clean up the notational confusion that was the subject of your original issue.
Let me know if you have any further questions.
Also, if you are interested, I have an upcoming publication presenting a Taylor-Duffy method for tetrahedron-product integrals (for volume-integral-equation solvers).
Thanks!
Yes, I am interested in reading your upcoming publication, where can I download it? Singularity handling is always tricky, but your package is very useful.
The publication will be up on ArXiV next week. The code is already available for download in my BUFF-EM package but is not as thoroughly documented as in SCUFF-EM.
I noticed another issue with the singularity extraction routine when calculating the vector potential entries of the EFIE. Here are the parameters I have set:
Executing the code: 2.4263e-7
Analytic result: -1.8635e-6 obtained with the formulas in the following paper: http://ieeexplore.ieee.org/document/563344/?arnumber=563344&tag=1
Here is the c++ implementation of the code for the formulas in this paper
Which is executed by calling the following:
The singularity routine with scuff-em does provide correct result compared to analytic formulas when alpha = beta, i.e. when the Q = QP.