Add tutorial for setting up a simulation #514

efaulhaber · 2024-05-03T21:53:03Z

Find the rendered page here:
https://trixi-framework.github.io/TrixiParticles.jl/previews/PR514/tutorials/tut_setup_replaced/

svchb

Thanks. This also needs some suggestions for which modifications can be made to see some other results. For example moving the initial fluid up 0.1...

svchb · 2024-05-06T14:45:32Z

docs/src/tutorials/tut_setup.md

+At the beginning of most simulation files, we define the numerical resolution,
+so that it can easily be found and changed.
+First, we import TrixiParticles.jl and
+[OrdinaryDiffEq.jl](https://github.com/SciML/OrdinaryDiffEq.jl), which we will
+use at the very end for the time integration.
+```jldoctest tut_setup; output = false
+using TrixiParticles
+using OrdinaryDiffEq


This needs:

What does numerical resolution mean here?

What is the common name that we use?

What impact does that have?

Concrete values for the example that you are discussing.
...

svchb · 2024-05-06T14:47:39Z

docs/src/tutorials/tut_setup.md

+We also set the number of boundary layers, which need to be sufficiently
+large, depending on the smoothing kernel and smoothing length, so that
+the compact support of the smoothing kernel is fully sampled with particles
+for a fluid particle close to a boundary.


This would be much more concrete if you would just write compact_support < boundary layers and give here the common values for the compact support or how one can find them.

svchb · 2024-05-06T14:48:25Z

docs/src/tutorials/tut_setup.md

+
+# output
+1000.0
+```


Add a drawing.

svchb · 2024-05-06T14:53:52Z

docs/src/tutorials/tut_setup.md

+For stability, we need numerical dissipation in form of an artificial viscosity
+term.
+```jldoctest tut_setup; output = false
+viscosity = ArtificialViscosityMonaghan(alpha=0.02, beta=0.0)
+
+# output
+ArtificialViscosityMonaghan{Float64}(0.02, 0.0, 0.01)
+```
+We choose the parameters as small as possible to avoid visible viscosity,
+but as large as possible to stabilize the simulation.


This is formulated in a way that the only reason to use this type of viscosity is for stability reasons, which is not true it needs to be set according to the viscosity of the fluid to be simulated. It should also be mentioned here that beta is usually 0.0 in fluid simulations.

svchb · 2024-05-06T14:54:42Z

docs/src/tutorials/tut_setup.md

+The WCSPH method can either compute the particle density by a kernel summation
+over all neighboring particles (see [`SummationDensity`](@ref)) or by making
+the particle density a variable in the ODE system and integrating it over time.
+We choose the latter approach here by using the density calculator
+[`ContinuityDensity`](@ref).


This should also shortly mention the pro/cons of these methods.

svchb · 2024-05-06T14:55:40Z

docs/src/tutorials/tut_setup.md

+We will use the [`BoundaryModelDummyParticles`](@ref) with
+[`AdamiPressureExtrapolation`](@ref), which generally produces the best results
+of the implemented methods.


Since we don't know which methods will still be implemented this should be formulated more neutrally.

svchb · 2024-05-06T14:56:56Z

docs/src/tutorials/tut_setup.md

+
+The key component of every simulation is the [`Semidiscretization`](@ref),
+which couples all systems of the simulation.
+All methods in TrixiParticles.jl are semidiscretizations, which discretize


All methods? That is a bit too general.

svchb · 2024-05-06T14:58:44Z

docs/src/tutorials/tut_setup.md

+```
+
+See [Visualization](@ref) for how to visualize the solution.
+


Add image here

LasNikas

thanks a lot!

It would be nice to have a sketch of the tank at the very beginning. Also, a short description (or also sketch?) how we usually structure our example files.

LasNikas · 2024-05-06T16:16:52Z

docs/src/tutorials/tut_setup.md


 ```
+Now, we define the particle spacing, which is our numerical resolution.
+We also set the number of boundary layers, which need to be sufficiently
+large, depending on the smoothing kernel and smoothing length, so that


link to the corresponding doc section

LasNikas · 2024-05-06T16:17:44Z

docs/src/tutorials/tut_setup.md

+First, we import TrixiParticles.jl and
+[OrdinaryDiffEq.jl](https://github.com/SciML/OrdinaryDiffEq.jl), which we will
+use at the very end for the time integration.


put this somewhere above the section ## Resolution?

LasNikas · 2024-05-06T16:18:35Z

docs/src/tutorials/tut_setup.md


+For different setups and physics, have a look at [our other example files](@ref examples).


Maybe explain how we usually structure our example files. Resolution, experiment setup, ...

LasNikas · 2024-05-06T16:21:27Z

docs/src/tutorials/tut_setup.md

+First, we define the physical parameters gravitational acceleration,
+initial fluid size, tank size, fluid density, and simulation time.


Suggested change

First, we define the physical parameters gravitational acceleration,

initial fluid size, tank size, fluid density, and simulation time.

First, we define the physical parameters gravitational acceleration, simulation time, initial fluid size, tank size and fluid density.

LasNikas · 2024-05-06T16:22:00Z

docs/src/tutorials/tut_setup.md

+hydrostatic pressure gradient, we need to define a state equation, which
+relates the fluid density to pressure.


link to docs?

LasNikas · 2024-05-06T16:38:19Z

docs/src/tutorials/tut_setup.md

+In order to define a boundary system, we first have to choose a boundary model,
+which defines how the fluid interacts with boundary particles.
+We will use the [`BoundaryModelDummyParticles`](@ref) with


It might make sense to mention that the BC in SPH are modeled numerically with particles instead of having an analytical described surface

LasNikas · 2024-05-06T16:40:34Z

docs/src/tutorials/tut_setup.md

+information about the current simulation time and runtime during the simulation,
+and a performance summary at the end of the simulation.
+We also want to save the current solution in regular intervals in terms of
+simulation time as VTK, so that we can look at the solution in ParaView.


link to the visualization section

LasNikas · 2024-05-06T16:41:46Z

docs/src/tutorials/tut_setup.md

+simulation time as VTK, so that we can look at the solution in ParaView.
+The [`SolutionSavingCallback`](@ref) provides this functionality.
+To pass the callbacks to OrdinaryDiffEq.jl, we have to bundle them into a
+`CallbackSet`.


link CallbackSet to the corresponding doc page in SciML?

LasNikas · 2024-05-06T16:42:14Z

docs/src/tutorials/tut_setup.md

+Finally, we can start the simulation by solving the `ODEProblem`.
+We use the method `RDPK3SpFSAL35` of OrdinaryDiffEq.jl, which is a Runge-Kutta


LasNikas · 2024-05-06T16:46:10Z

docs/src/tutorials/tut_setup.md

+# output
+


maybe add something like plot(my_kernel) to visualize the Kernel in 1D (or more fancy in 2D)?

Add tutorial for setting up a simulation

d95a8f2

efaulhaber added the documentation Improvements or additions to documentation label May 3, 2024

efaulhaber requested review from svchb and LasNikas May 3, 2024 22:00

Add section about custom smoothing kernels

8cec32c

svchb requested changes May 6, 2024

View reviewed changes

LasNikas reviewed May 6, 2024

View reviewed changes

Merge branch 'main' into ef/tut-setup

83ae05f

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Add tutorial for setting up a simulation #514

Add tutorial for setting up a simulation #514

efaulhaber commented May 3, 2024 •

edited

Loading

svchb left a comment

svchb May 6, 2024

svchb May 6, 2024

svchb May 6, 2024

svchb May 6, 2024

svchb May 6, 2024

svchb May 6, 2024

svchb May 6, 2024

svchb May 6, 2024

LasNikas left a comment •

edited

Loading

LasNikas May 6, 2024

LasNikas May 6, 2024

LasNikas May 6, 2024

LasNikas May 6, 2024

LasNikas May 6, 2024

LasNikas May 6, 2024

LasNikas May 6, 2024

LasNikas May 6, 2024

LasNikas May 6, 2024

LasNikas May 6, 2024

		```

		See [Visualization](@ref) for how to visualize the solution.


		For different setups and physics, have a look at [our other example files](@ref examples).

		First, we define the physical parameters gravitational acceleration,
		initial fluid size, tank size, fluid density, and simulation time.

		hydrostatic pressure gradient, we need to define a state equation, which
		relates the fluid density to pressure.

		Finally, we can start the simulation by solving the `ODEProblem`.
		We use the method `RDPK3SpFSAL35` of OrdinaryDiffEq.jl, which is a Runge-Kutta

Add tutorial for setting up a simulation #514

Are you sure you want to change the base?

Add tutorial for setting up a simulation #514

Conversation

efaulhaber commented May 3, 2024 • edited Loading

svchb left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

LasNikas left a comment • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

efaulhaber commented May 3, 2024 •

edited

Loading

LasNikas left a comment •

edited

Loading