Resolution Study and Updated Figure

I noticed the field amplification changes noticeably with grid size, so conducted a resolution study for the max field. It seems to indeed change. All field is on the surface of the asteroid for uniformity.

(Maybe the scale height at the surface is not being resolved yet?)

Another Memory Allocation Bug in Regrid

Updated the code and figure.

Charge Exchange

Here's the split of absorption between hot and cold neutrals:

Shows pretty much what the theoretical calculation does - the hot neutral density just isn't high enough to get significant absorption. You can see a lot more broadening in the hot case, the total absorption is just very low compared to the cold.

Papers

Summary: After reviewing all of the papers, it's starting to look like everyone can fit GJ 436 but not HD 209458.

(Not sure why only the first attachment link is working…. Open post to download attachments.)

Spring code

Basically done. A few more problems to update, some minor refactoring and upgrades for the spring part (particularly, need to figure out the correct places to parallelize with OpenMP).

Objectives

• Grain-grain collisions routine (draft)
• Source term integration (debugging)

Progress:

• Sod Shock Tube: As suggested, I ran the dusty Sod Shock Tube problem setup with different interpolations methods: ​PCM, ​PLM, and ​PPM.
• Dust Processing: Made progress regarding grain-grain collisions & source term integration but not enough to run simulations right now

Up Next:

• Continue source term integration & draft grain-grain collisions

Charge Exchange

With a theoretical optical depth as a function of frequency of

\tau = \sigma_{\nu_0} n_{HIhot} d \frac{c}{\nu_0} \sqrt{\frac{m_H}{2 \pi k_B T}} e^{{-\frac{(\nu - \nu_0)}2 c^{2 m_H}{2 \nu_0}2 k_B T}}

0.125704 e^{{-8.92598\times10}{-25} (\nu - 2.47\times10¹⁵⁾2)

Which gives an optical depth at \nu_0 of .126, an optical depth at \pm 100 km/s of 0.061.

I've attached a Mathematica notebook with my work.

I've also made it possible to output both the unsplit and split (as separate files) Lyman-alpha absorption from postprocessing. Should be able to get some preliminary results in a day or two.

Spring Code

Split into 10 modules now, approximately 6ish left to refactor. May be able to finish by the end of the week. Almost certainly by the end of the month.

Desity and Field Plot of Final Frame

Tried to reproduce this figure from the 2018 paper:

Now, however, the Cartesian grid is too grainy and leads to ring like structures (Nodal plot is slightly better).

Trimmed down the boundary and increased resolution (2x) to get this:

But max field has reduced by ~22%, possibly due to new boudary conditions.

Regrid

Regrid was successful to Lv. 4 for the simulation we were using.

The int data type was going into overflow by a bit.

Here are the resulting field lines:

However, when I try to do the same with the higher resolution simulation, it runs out of time. So it probably needs >5x the time from the lower resolution case or there could be another bug.

Free Electron

• Added free electron fraction to Fields.f90
• Interesting problem with tabulated free electron

​https://www.pas.rochester.edu/~yishengtu/research_files/CEE_gp_meeting_04202020/Figure_1.png ​https://www.pas.rochester.edu/~yishengtu/research_files/CEE_gp_meeting_04202020/MESA_EOS_where_free_e_constant.png

Clump

Upper: ideal gas
Lower: Tabulated
Sound speed ~ 1.2e7

• 1e7 ​https://www.pas.rochester.edu/~yishengtu/research_files/CEE_gp_meeting_04202020/1e7_rho.gif
• 2e7 ​https://www.pas.rochester.edu/~yishengtu/research_files/CEE_gp_meeting_04202020/2e7_rho.gif
• 4e7 ​https://www.pas.rochester.edu/~yishengtu/research_files/CEE_gp_meeting_04202020/4e7_rho.gif
• 6e7 ​https://www.pas.rochester.edu/~yishengtu/research_files/CEE_gp_meeting_04202020/6e7_rho.gif
• 8e7 ​https://www.pas.rochester.edu/~yishengtu/research_files/CEE_gp_meeting_04202020/8e7_rho.gif
• 1e8 ​https://www.pas.rochester.edu/~yishengtu/research_files/CEE_gp_meeting_04202020/1e8_rho.gif

StarOnGrid

• The star blew up at ~frame 39, perhaps resolution too low
• Refinement doesn't seem to work with tabulated EOS but works with ideal gas EOS

​https://www.pas.rochester.edu/~yishengtu/research_files/CEE_gp_meeting_04202020/StarOnGrid_figs/

Objectives

• Properly integrate dust source routines
• Continue with grain-grain collisions
• MHD production runs for post-processing dust treatment

Progress:

• Sod Shock Tube: Finished (for now). The issue before was that the gas movement is triggered via initial values of the energy and as the dust has now energy component it remained at rest. Thus active advection + Sod Shock Tube is currently not possible (it would require writing a different initialisation routine and I've noted it down as something to be done in the future but I don't feel it's necessary right now), passive advection + Sod Shock Tube, on the other hand, works: ​Setup | ​Gas Density (​Video)| ​Gas Velocity in x (​Video) | ​Dust Density (​Video)
• Dust Post-Processing: We were awarded HPC time so I'm currently setting up MHD (instead of just HD as in the previous papers) runs for the dust processing done by a colleague. (Not sure about the best way to visualise field lines…)

Up Next:

• Source terms!! plus further development of dust routines but source terms now take priority

Code

• compilation magically works when I asked for an interactive session last week …
• Not sure why but the code is now compiled and executable.

Clump

• Ran table EOS simulations. Planning on running corresponding ideal gas EOS this week.
• cs ~ 1.2e7
• v_wind (table EOS) = 1e7 ​https://www.pas.rochester.edu/~yishengtu/research_files/CEE_gp_meeting_04132020/out_1e7/1e7_rho_fig.gif
• v_wind (table EOS) = 1e8 ​https://www.pas.rochester.edu/~yishengtu/research_files/CEE_gp_meeting_04132020/out_1e8/1e8_rho_fig.gif

Star On Grid

• Putting a single RGB star on the grid using Luke's profile
• WinSCP stopped working this morning.. So everything on bluehive for now

​Charge Exchange Paper

Figures and captions (except for postprocessing) done. Also basically done with methods. Need to talk about other papers in intro still (which? of Khodachenko19, Khodachenko17, Shaikhislamov16, Christie16, Tremblin13, Ekenback10, Bourrier16).

Postprocessing: Completed up to frame 276 for the split-species high wind. Completed up to 226 for the split-species low wind. High wind is taking 7-8 hours per frame, so should be done late next week (~10-12 days). Low wind takes closer to 10 hours per frame, but hopefully can be done by the end of the month.

​Updated PDF

Updates:

• Colorbars on contour plots.
• Temperature sensitivity studies.

Regrid to Highest Resolution

Works till upto Lv. 3.

Get the follwoing error for Lv 4. Callback to exact issue with gdb debug has not been successful so far.

 level          -2
 level          -1
 HDF5 error:  function h5dwrite_f failed with error code           -1 .
HDF5-DIAG: Error detected in HDF5 (1.8.17) MPI-process 0:
  #000: H5Dio.c line 254 in H5Dwrite(): not a dataspace
    major: Invalid arguments to routine
    minor: Inappropriate type
forrtl: severe (174): SIGSEGV, segmentation fault occurred

High Resistivity Bug

We added some checks for the new diffusion, but I do not see any difference in .out or the output.

Equivalent Resistivity

Computed the table for the cases I had so far. The comet case seems to cause more amplification for a given equivalent (or even just) resistivity, followed by the shell case, and then the uniform case.

Also, can cases with different winds parameters be directly compared?

Here is the table for cases with 300 #/cc, 500km/s, 1e6 K.

(Image and table in Excel)

High resistivity

The present day moon case at least ran completely with some noise. The case with Magnetic Reynolds' number of 0.5 (needed for the above table) runs into major issues:

3D Plot of Field Lines

This is to be added to the Supplementary information of a paper:

Need suggestions on how to add more field lines in the front. Will also need to patch the data into fixed grid to stop the field lines from breaking.

​Update of Current Progress

Objectives

• Set up Sod Shock Tube Problem for dust
• Properly integrate dust source routines
• Continue with grain-grain collisions

Progress:

• Grain-Grain Collisions: Made some progress here while I didn't have internet access would like to pause it now though to focus on the source routines from next week onwards.
• Sod Shock Tube: The default Sod Shock Tube problem is incompatible with the dust modules so I'm trying to integrate the dust by setting up my own Sod Shock Tube using split region. It did run but due to lack of internet, I couldn't check whether it was doing the right thing and unfortunately, the starting setup looked like ​this. I will try and rectify that next and maybe reduce the problem to 1D as I don't really need 2D here.

Update:

• Sod Shock Tube: Turns out the issue above was a) due to using 2D in global.data when problem.f90 was for 1D and b) colourmap choice. Using 1D and a different colour map I get ​this for the gas which looks alright (density cut ​here). However: While the dust is initialised it's not given any momentum at the beginning so it just remains stationary. Need to find out why the dust velocity initialisation is not triggered.

Up Next:

• Figure out dust velocity issue for Shock Tube problem
• I really want to get the source routine integration working next so that will be my priority over the next weeks.