Tutorial 4 - Lattice-Boltzmann #3893
Conversation
Store the lag time and MSD series separately. Remove the superfluous LOOPS variable. Indicate which section in Frenkel 2002 discusses the diffusion formula.
Add solutions and literature references. Test the solutions in CI.
|
Check out this pull request on Review Jupyter notebook visual diffs & provide feedback on notebooks. Powered by ReviewNB |
doc/tutorials/04-lattice_boltzmann/04-lattice_boltzmann_part1.ipynb
Outdated
Show resolved
Hide resolved
Rewrite the solution to run faster with shorter polymers. Calculate several polymer statistics (R_F, R_g, R_h) and explain they depend on the number of monomers.
Integrate the autocorrelation function to remove the variance bias in time series.
|
@KaiSzuttor Part 3 is now ready for review. The idea would be to hide the solution of the plots for Rf, Rg, Rh. Once the participants have a working solution for Rf, they can copy-paste it for Rg and Rh. Regarding the diffusion coefficient, they should re-use the solution they wrote in Part 2 and write a fitting function based on the Kirkwood-Zimm model. I'm still working on the electrophoresis part, but if I can't get it to work out in time, plan B is to only do the monomer+polymer diffusion in the advanced group, and the monomer + Poiseuille flow in the beginner group. I tried Part 3 with Langevin Dynamics instead of LB, and I got similar results at a fraction of the computational time. Maybe in the future we could create a polymer tutorial with Langevin. |
jngrad
changed the title
jngrad
force-pushed
the
tutorial_04_summer_school_2020
branch
from
8841251 to
02e102f
Compare
|
@schlaicha @christophlohrmann @reinaual the latest changes are in, I'll hide solutions later |
|
I only had a quick look so far. One thing I find very important is to make it very clear in the text, what part of the observations actually depend on hydrodynamics and which don't. Stuff like r_g shouldn't really be calculated with active LB, since the results can only get worse due to longer simulation times. Long term, it might even be reasonable to split this tutorial. The equilibrium polymer properties could be handled in a separate "Polymers"tutorial which wouldn't use LB. |
|
@RudolfWeeber I get the same results with Langevin Dynamics in place of LB for all observables in the polymer part. The same is probably true for the monomer part, although I didn't try it. |
| @@ -9,11 +9,12 @@ | |||
| "#### Before you start:\n", | |||
There was a problem hiding this comment.
I guess the equation should read
\begin{align}
\rho &= \sum_i f_i \
\vec{j} = \rho \\
\vec{u} &= \sum_i f_i \vec{c_i} \\
\Pi^{\alpha \beta} &= \sum_i f_i \vec{c_i}^{\alpha}\vec{c_i}^{\beta}
\label{eq:fields}
\end{align}
for better readability?
Reply via ReviewNB
There was a problem hiding this comment.
ReviewNB's support for LaTeX seems incomplete, the original equation looks good in Jupyter, with one equation per line
|
The numbering is a bit weird as part 1 is from 1-3, part 2 starts with 5 and part 3 with 5.2; splitting into several files seems like a good idea but should be introduced in the notebook of part 1. |
Very good point! |
|
In Part 4 I think we should switch the exercise, such that the students have to set up the LB stuff and we give them the comparison to theory |
The participants should focus on setting up an LB fluid instead of writing the matplotlib code.
Also give more details on tasks, add links to the user guide and renumber sections.
|
@schlaicha the numbering now restarts from 1 in every part |
The choice of solvent implementation only affects the Rouse vs. Zimm regime. Chain properties such as the hydrodynamics radius, end-to-end distance and radius of gyration do not change.
jngrad
force-pushed
the
tutorial_04_summer_school_2020
branch
from
14cf61d to
c7d42c8
Compare
|
I suppose, we merge this PR this morning, to avoid confusion? |
|
Yes, that would be more convenient for tutors. |
Description of changes: