Welcome to pyclaw Discussions!

[ketch]

👋 Welcome!

We’re using Discussions as a place to connect with other members of our community. We hope that you:

• Ask questions you’re wondering about.
• Share ideas.
• Engage with other community members.
• Welcome others and are open-minded. Remember that this is a community we
  build together 💪.

To get started, comment below with an introduction of yourself and tell us about what you do with this community.

[10726559]

Good afternoon, Mr. Ketch, I am a master student in fluid mechanics, now I am trying to use Pyclaw and specify boundary conditions by myself for shock tube type problems. However, the pressure value of the current time step on the boundary needs to be determined through the values of pressure and self-defined quantities in the previous time steps, I would like to ask that if there are some ways to extract and use them from the previous numerical results? Thank you very much!

[mandli]

How many time steps do you need?

[10726559]

• From t=0 (initial beginning time) to the current time step t=tn, or from the first time when the first pressure wave arriving at the boundary(if there exists time trigger in Pyclaw to help me identify it), I need to define the current pressure value on the boundary according to these waves in the boundary condition definition part, I don't know if there are some functions can achieve this. Thank you!

[mandli]

How are these boundary conditions dependent on the solution internal to the domain?

[10726559]

The quantities(density, velocity, pressure, etc.) of the last cell in the inner domain are known at the current time step, and the pressure on the ghost cell is just set to the ambient pressure; two Riemann invariants are used to connect the values of the last cell and the boundary, as well as those of the boundary and the adjacent ghost cell; under a certain assumption, the normalized Riemann invariant for the right-traveling wave can connect the incident pressure waves with ambient pressure, and it is known from the internal solution. Therefore, the strength of each incident pressure wave element at different time steps can be calculated; Then the pressure on the boundary at different time steps will be affected jointly by these previous pressure waves elements according to acoustic response theory, so I would like to extract and use the previous numerical results to define the current pressure on the boundary, afterwards Riemann invariants are used to determine the velocities on the boundary and ghost cell.

[mandli]

It sounds like all you need then is the previous time step's value of the pressure.  We do this often with Riemann based FV methods.  I am not sure why you would anything but the immediately available values in the cells though, especially given that this is a hyperbolic PDE.  These are easily available unless there is some sort of history you are looking for. Kyle

…

On Feb 24, 2023 at 10:39 AM -0500, Chen Liu ***@***.***>, wrote: The quantities(density, velocity, pressure, etc.) of the last cell in the inner domain are known at the current time step, and the pressure on the ghost cell is just set to the ambient pressure; two Riemann invariants are used to connect the values of the last cell and the boundary, as well as those of the boundary and the adjacent ghost cell; under a certain assumption, the normalized Riemann invariant for the right-traveling wave can connect the incident pressure waves with ambient pressure, and it is known from the internal solution. Therefore, the strength of each incident pressure wave element at different time steps can be calculated; Then the pressure on the boundary at different time steps will be affected jointly by these previous pressure waves elements according to acoustic response theory, so I would like to extract and use the previous numerical results to define the current pressure on the boundary, afterwards Riemann invariants are used to determine the velocities on the boundary and ghost cell. — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you commented.Message ID: ***@***.***>

[10726559]

Thank you very much, Mr. Mandli. I am considering a partially reflective boundary on which incident pressure waves at different times will act on it. I will look up the state.problem_data dictionary and guage in Pyclaw document mentioned by Mr. Ketch.

[ketch]

In general, if you need values from times earlier than the previous step, you could use the state.problem_data dictionary to keep track of them, or put a guage at the boundary.

[10726559]

Thank you very much, Mr. Ketch. It can be helpful for me. I will look at it.

[pavelkomarov]

I'm Pavel Komarov, PhD student at UW in the Electrical and Computer Engineering department, advised by Nathan Kutz. (There are corners of ECE that are basically applied math: systems, controls, signal processing, ML). I'll be using Clawpack to do PDE simulations. So far I've found it pretty confusing. I'm making a taxonomy of important objects for myself, which I may turn into some documentation for you all to host on the website, so hopefully new users can get up to speed faster than I've been unable to. I also hope to submit some PRs if I think of ways to make the interface of this project more intuitive. Who is in charge? I gather @ketch and @mandli and @rjleveque are pretty important players. (Thanks for moving the build system over to meson last year. I wasn't even aware at the time that setuptools had serious limitations. I was just pretty frustrated at the time that installation was broken, but I've had less problem lately.)

[mandli]

Welcome Pavel! Have you taken a look at the documentation at clawpack.org? There's some break down of the structures of PyClaw there. Otherwise the examples are a good place to start, as you have already found. Please feel free to submit PRs as you see things that you think would make it easier for people to get using the code. I also would suggest taking a look at Randy's books on finite differences and finite volumes as they provide the theoretical basis for the methods and use the language and notation in those texts. I would also suggest taking a look at the Riemann Problems in Jupyter book that @ketch, @rjleveque and @maojrs wrote.

[ketch]

@pavelkomarov Welcome! Improvements to the documentation would be very welcome; @mandli has pointed out some resources, but I agree that the "new user" experience could be greatly improved and you are at the best stage to make contributions in that regard. I'm glad to hear that installation went better this time.

For proposals that would change the user interface, it will be best to have a discussion first as those can affect people who don't use PyClaw directly but only through GeoClaw or AMRClaw.