Dear author
it's my honor to see the discontinuous Galerkin finite element program you shared. However, I encountered the following problems during installation：

After my query, it seems that the latest tbb does not contain tasks. Since I have not used tbb before, I want to ask you which version of tbb you use. I have searched the Internet for a long time but haven't solved this problem, so I take the liberty to ask you.
Then, I have two questions for you:

1. I want to use the discontinuous Galerkin finite element method to solve the acoustic equation. I also found the original version of the program at github and want to modify it on this basis. However, I see that its flux matrix seems to have less negative signs than the equation used at ordinary times (as shown in the figure below). So I want to ask you, is there a problem with my understanding or is there an omission in the procedure?
   
   

2. When I tested the program, I found that the effect of wave propagation was good. However, the solution between adjacent elements is not smooth, and there is obvious triangle trace. I checked the program. I guess this is caused by the numerical flux term, but I don't know how to modify it. So I want to ask you whether this problem has been alleviated in the program you modified?
   
   
   Thanks again for sharing the code! I hope you will give me your advice and look forward to your reply! I wish you a happy life and all the best!

Dear author
I am very glad that you can reply to my question. The "build. sh" you wrote is really helpful for us to compile programs. I just compiled with the "sh build. sh" method. I found that some packages were still missing during the compilation process. I solved the other problems. The "tbb:: task" error always exists.
When running build.sh, the following statement appears:

When this statement appears, I think libvtk9 dev has been successfully installed.
So I don't know exactly where the error caused the program to fail to compile?

Is there a problem with my installation? I did not find libtbb2-dev after searching.

But after searching, I found no way to install libtbb2-dev by ubuntu.

Thank you. I am using ubuntu18.04 now. I found that there is no libtbb2-dev in this version after searching the website you recommended.

I'll try to upgrade the system.

Dear author
I conducted the following tests, but there are still related errors:

I found that ubuntu18.04 does not contain the libvtk9-dev file, so I made relevant changes to build.sh:

I don't know if the mistake is due to this reason.

How long ago did you update the file. I checked. The latest file was updated two hours ago. Is this the last version?

Dear author
When I updated the system to 22.04 for compilation, the original problem was solved. It seems that the problem is really related to the lack of necessary packages in the system. However, a new problem has arisen:

To solve this problem, I modified CMakeLists.txt in the src file:

In this way, the program can be compiled successfully.

But when I use "run_caa" to run the program, I will report an error:

Dear author
After your guidance, the program has run successfully. Since I have not used the vtk file before, I checked it online and displayed it with 'paraview', as shown in the figure:

I found that the sound wave propagation pattern can be displayed, but then there are very serious triangular traces. I guess this is because the third item in the text has a problem in the calculation, that is, there is a problem in the calculation of the numerical flux term between elements. This leads to the discontinuous solution of adjacent elements, which shows that the trace of triangle element is too serious.

Dear author
I have carried out relevant operations according to your guidance. Through comparison, I found that the effect was indeed improved. However, there are still traces of triangles, but for such a simple model, such problems should not occur.

And the value of u obtained in the original program should be the value of each node. If I understand correctly, the value at the center of your output is calculated by Lagrangian interpolation using the u value at each node.
In Paraview, the value at the center is used to solve the value of each node. I can't figure it out. If the calculation is carried out in each element, how to interpolate one value to get three values? If we use the values at the centers of several adjacent elements to interpolate the values at the nodes, we can get smoother values, but it seems unreasonable.