povanberg / dgfem-acoustic Goto Github PK

Dear author, by downloading the content you uploaded, I understand that you have studied the content of the intermittent Galerkin finite element in depth, and the information you uploaded has been of great help to my study. As a beginner, I express my gratitude to you again.
But when I run the example you gave, the results I get are a bit unintelligible.
./dgalerkin ../../doc/2d/square.msh ../../doc/config/config.conf

The parameters in the config.conf file are as follows:
timeStart=0
timeEnd=0.1
timeStep=0.0001
timeRate=0.001
elementType=Lagrange
timeIntMethod=Runge-Kutta
Reflecting = Reflecting
Absorbing = Absorbing
MyPhysicalName = Absorbing
numThreads=4
v0_x = 0
v0_y = 0
v0_z = 0
rho0 = 1.225
c0 = 343.3
initialCondtition1 = gaussian, 0,0,0,1,1
saveFile=data.msh
（I see that in the config.conf file file, the Source command is commented out, and I have not changed this.）
Use the command gmsh data.msh to view the generated .msh file.
The results I got are shown in the figure below. I haven't seen the phenomenon of sound wave propagation in the picture. As a beginner, I don't understand it here. I hope you are busy to answer my doubts.

Thank you again for the information you have shared and your contribution in this research direction.

hello,when I install the Prerequisites by script,there occurs:
[1] Get depedencies and external libraries.
./DG_setup: 10: ./DG_setup: module: not found(module load cmake /3.11.1)
./DG_setup: 11: ./DG_setup: module: not found(module load gcc/4.9.2)
mCmake found.
G++ found.
Gfortran found.
Lapack/Blas found.
Gmsh found.
Eigen found.
[2] Build sources.
CMake Error: The source directory "/home/chenzecai" does not appear to contain CMakeLists.txt.
Specify --help for usage, or press the help button on the CMake GUI.
make: *** No targets specified and no makefile found. Stop.
[end] Error!
what should I do?thanks a lot~

Can this code run in two layers? I want to use it to Simulate seismic waves

Hi I am also trying to use the code and I am having a little bit of trouble replicating your convergence plots in your report.

I would love to get some extra details or just to have a chat.
My email is [email protected].

Thank you guys so much

Hi authors,

Thank you for sharing your great work! I mainly work on geometric acoustics and find your work very promising for follow-up research. I can compile and run the example locally. I have one question while reading your code and report where it says "The walls, the stairs and the furniture are modelled as hard, reflective, nonporous surfaces with an absorbing coefficient of 5%". Does that mean all your reflecting condition assumes 5% absorption and all your absorbing condition assumes 5% reflection? If so, how are they enforced in the code? I do see the m_fBC buffer is relevant but am not sure how these coefficients affect the calculation numerically.

If the current code does not reflect that, can you kindly point me to the right place to modify the code? Basically, I'm wondering how I can use different coefficients by slight modifications.

Thanks,
Zhenyu

Dear author，
By running the program you uploaded and learning the documents you shared, it has greatly helped me in my study. When I was studying the document, I saw the sentence:

I have several questions about this place.

1. After studying your documents and other articles, I think the role of Jacobian matrix is to convert the grid from actual coordinates (x, y, z) to reference coordinates（ξ,η,φ)∈[-1,1] ³，and it satisfies the following formula:
   
   When I use the examples you uploaded, square.msh and config.conf. the Jacobian matrix of the first element and the node coordinates of the element are printed out, and the calculation is performed:
   
   Wherein, (-4.278517,1.721583,0) is the coordinate of the first node of the first element (the node tag is 165). (-0.6971,1.45557,0) is the parameter coordinate of the node. I found that the parameter coordinates are outside the range of [-1,1].
   I found that the Jacobian matrix is calculated by using the function of gmsh. I also used the relationship that 2 times the area of the triangle element is equal to detJ to verify, and found that it is consistent. In this way, the determinant of Jacobian matrix seems to have no problem. I wonder why the transformed parameter coordinates are not within the range of [- 1,1], so that the Gaussian integration of formula (5) cannot be performed.
   2 I have some different understandings of Formula 5. I think that the integral of function with respect to K is converted to [- 1,1] ³ Times detJ. For an element, convert to [- 1,1] ³ When, its Jacobian matrix is fixed, and detJ is a fixed constant. When using the Gaussian integral formula, it seems that it is OK to multiply detJ at the outermost. The formula I understand is as follows. I would like to ask the author if I have misunderstood it.
   

3 Formula (9) is obtained after discretization. In this formula, uk is the node solution of each element. I think your program saves the node to the msh file for output. If I want to solve for μ at any point in the element, I need to use the following formula:

If I want to solve the μ。 My idea is: first, locate the element according to the coordinates of m, find the coordinates of the element nodes, and then calculate the area of the triangle according to the coordinates.

According to the definition of Lagrangian basis function, the area coordinates A1, A2, A3 at point m are needed (A1 is the area of small triangle m23, A1 is the area of small triangle m23, and A1 is the area of small triangle m23). S is the area of the element triangle. The basis function can be expressed as N1=A1/S; N2=A2/S; N3=A3/S。At point m, the solution formula of μ is:
μm=N1u1+N2u2+N3*u3
I want to use this method to solve μ at point m in Descartes coordinates. I want to ask you if there is any problem with my thinking.
I thought that the coordinates at point m could also be converted to parameter coordinates for solution, but there might be a problem with the Jacobian matrix in question 1, which made me uneasy to convert to parameter coordinates for solution.

My question may be a little childish. I hope you will give me some advice. Thank you for checking my question in your busy schedule, and look forward to your reply.
Thank you again for sharing the program selflessly. I wish you good health and good luck!

Dear author
Through your explanation and the explanation file you uploaded, I have a deeper understanding of the program.
When I ran the program, I found the parameter m_elOrder can be used to calculate the Gauss2 to determine the accuracy of the integration.
I want to ask you, what should I do if I want to improve the accuracy of integration? Do you need to change the triangular element from three nodes to six nodes when Gmsh divides the mesh? But I didn't find the relevant operation in Gmsh software.
I hope you can reply to my email when you have time, and look forward to your reply.
I wish you a happy life and smooth work.

Dear author
By running the program you shared, I got a good simulation effect. The information you uploaded has been of great help to my study. But I still have some questions to ask you about the numerical flux between elements.
（1）In the figure, the direction of ns points to the left. When we study the K+ element, its normal direction points to the outside and the ns direction is left. When our analysis element is the k- element, does the normal direction of the face point to the right?
（2）I can't understand the normal vector ng in the formula below very well. Where does ng point and what is its relationship with ns?

These are some of my doubts. Thank you very much for checking my email in your busy schedule. Hope to hear from you.
Thank you again for your selfless sharing. I wish you success in your work and a happy life.