​EAS 2020 Interactive Schedule S8 is the common envelope section.

Luke's talk on Thursday, July 2, at 11 am Eastern Time ​abstract

Amy's talk on Friday, July 3, at 4:15 am Eastern Time ​abstract

Created maps of the magnetic field on the front and back side.

(Couldn't quite get them in the same image yet.)

Realized this can be used to visually confirm convergence of simulations to steady state.

Still re-running all simulations and extending some of them to reach convergence. Results all match with what we expect.

Also working on a paper for the results. Might need to run a couple of high resolution cases for proper figures.

Charge Exchange

Collisional ionization appears to be working. Definitely a strong upper bound without it…

Frame 300, old run

Frame 299, new run

I'll queue the GJ 436b run (as soon as I figure out why the slurm script isn't working…) and the new HD 209458b run.

MHD

Officially stuck, so I'll turn the diffusion up some more. See ​last week's post for latest frame.

I've also set up a run to try restarting from the final frame of the original run with an extremely high field. Need to figure out how to reload a hydro run with MHD, and add to total energy when I add the field.

Goal: Figure out how to make the constant parameters that are currently in the Fermi FORTRAN program instead normally distributed.
Summary of Progress: ​Click here to see the PDF which summarizes my progress so far and my next steps.
Video Of Current Model Output (Colored by Technology): ​http://www.pas.rochester.edu/~esavitch/fermiProject/results.ogv

MHDCollidingFlows runs with the new analytic cooling and the ambient tracer. Need better way to handle parameters Temp0 and tracer in the cooling code though. Also need to double-check the scales..

​Mach15 result

      A_alpha = alpha
      !cs=sqrt(gamma*Boltzmann*TempScale/Xmu/muH/amu)
      !power=.5d0*(1d0-2d0*beta)
      !A_alpha=alpha*4.76e-20*&! (ergs*cm^3/s/K^.5)
      !   (3d0/16d0*Xmu*muH*amu/Boltzmann*(cs*vx)**2)**(power)
      A_beta = beta

Current version of code -- need better ways of handling Temp0 and trailer!!!

  FUNCTION AnalyticCoolingStrength2(q, Temp)
     REAL(KIND=qPREC) :: AnalyticCoolingStrength2
     REAL(KIND=qPrec) :: q(:)
     ! Local declarations
     REAL(KIND=qPrec) :: Temp, Temp0, T0

     Temp0 = 100000  !10^5 Kelvin
     T0 = Temp/Temp0
     AnalyticCoolingStrength2=q(1)**2 * A_alpha*T0**A_beta*ScaleCool

     !! only for ambient traer > 1/3 which is the scaless value for initial ambient density
     if( q(9) <= 1d0/3d0 ) AnalyticCoolingStrength2=0d0


  END FUNCTION AnalyticCoolingStrength2

MHD

Only got one frame last run, still getting tons of pressure protections (despite not being close to MinTemp anywhere, as far as I can tell). Perhaps diffusion needs to be turned up a bit? But things are starting to look very interesting. Also, realized the stellar field wasn't being applied correctly at the left boundary, so I've fixed that.

GJ 436b

Slurm script seems messed up. Need to fix. Otherwise working well.

​Charge Exchange Paper

Made most of the changes we discussed at the meeting. Waiting on results of collisional ionization test to start the 1.2x solar mass loss rate run, with more solar-wind-like properties.

Objectives

• Debugging
• Refine dust routines

Progress:

• HPC Proposal: Still really busy with this, need to get through 1m core hours by the end of June…
• Dust Grain Size Distribution: Previously the code was designed to read in a dust.data file which contained the dust grain size distribution information. I realised that might not be so great cause it would be hard to reproduce a simulation if that file were to be lost + it should scale with the volume of the cell. After some edits, the dust grain size distribution is now set via input parameters in physics.data and then calculated during setup initialisation.

(I'll try and add the equations for the different distribution types when I get a chance; a lot of unexpected meetings today so I might not manage before the meeting tonight.)

Up Next:

• Continue Debugging

Relevant papers

​Krumholz+2007

​Leenaarts 2020 review paper

Radiation transfer and flux limited diffusion in AstroBEAR

​AstroBEAR wiki page reproduced

I've checked format up to section 2.2 only. Later sections may contain typo and mistakes.

My note for today's meeting

​My note on Leenaarts section 2

​My note on Marshak wave (following Castor chapter 12.1)

These runs are currently using the parameters from Run 2 of the ​parameter space paper, to wit:

The initial B field excludes the center of the planet.

Here are some relatively current movies:

Fixed left boundary

B field

rho, temp, P

beta, no ram

beta, with ram

CEJet

​Notes

Run time on Stampede2

Frame rate for the jet modal on 120 cores:

Time for one modal with 120 framesis ~170 hours in wall time = 20 400 SUs.

I have 3 runs planned for this week, which will use about 81 200 SUs.

PN talk for EAS2020

Presentation time is 15 min, including a 4-5 min Q&A and 1 min for transition. I will prepare 10 slides max for the talk, and maybe <2 extra slides for questions.

Outline:

1. Introduction to the problem. Show ~2 examples of bipolar PPN with collimated jets.

2. Simulation setup (1): initial condition from the SPH model, choice of physical parameters.

3. Simulation setup (2): technical details in the AMR simulations: relax the ejecta on AMR grid, fallback during the quiescent period, regions of refinement, and application of the cooling curve.

4. Results (1) - model A, the central pillar of jets and inner shock

5. GISW model and the elliptical shock

6. Results (2) - model B and model C

7. Results (3) - model D, compare to OH 231.8+04.2

8. Asymmetries in line plots

9. Momentum budget

10. Conclusions

Meeting on rad hydro

Thursday (6/18) at 11 am. Zoom coordinates to be emailed.

A line of code which was supposed to zero out EMF flux in the y direction inside the boundary of any object was set to ey instead of e2y causing magnetic field to diffuse inside the asteroid.

It disproportional affects high magnetic Reynolds number cases and lowers the max amplification by a little bit (< 0.5).

The results can be contrasted in these movies (Different R_M):

Before:

After:

Currently restarting all simulations. Could take upto 2 weeks.

Objectives

• Debugging
• Refine dust routines

Progress:

• HPC Proposal Had to do some work there for the current & next round. We want to do different setups next year to look at different regions of dust destruction
• Simulations We're heading towards the end of Q1 so I'm currently really busy setting up some final simulations (post-processing)
• Debugging Made some progress regarding gas drag. Now the gas in the ambient medium is moving (see ​screenshot) but not in the clump. Not sure whether that is still density related or not, will need to check. I'm also adding a dust distribution routine so that those can now be calculated inside the code instead of with an external Python script beforehand (makes it more compatible with the post-processing).

Up Next:

• More debugging

Goal: Figure out how to make the constant parameters that are currently in the Fermi FORTRAN program instead normally distributed.
Summary of Progress: ​Click here to see the PDF which summarizes my progress so far and my next steps.

​Charge Exchange Paper

Haven't gotten any new comments yet. Still working on the collisional ionization run on Stampede.

MHD

Still lots of pressure protections, but running reasonably ok (> 1 frame per day). Looks like the high-temp vacuum is starting to smooth out, so hopefully we're over the hump.

GJ 436b

Still no issues at equivalent of frame 4 (where it crashed with pressure protections before frame 1 last time). If it makes it to t=1 (frame 10), I'll just rename the files and have it restart from there.

Jobs, etc.

Started updating my (long long form) CV based on suggestions from AAS, also finishing up the requirements for the university's teaching recognition. This includes a first draft of a teaching philosophy, which I think is sufficient for the requirements but needs some revision.

Also probably going to apply to that job Laura sent around - may want people to start too early, but at least some gov't contractors have very long hiring processes, and worst case it's good practice and I might (maybe) get some useful feedback.

Spring code

Lots of tests fail with OpenMP. Need to determine which fail because of bugs and which fail because of pseudorandom ordering. On the backburner for the moment.

​Charge Exchange Paper

Made the changes discussed, responded to Luca's comments. Have a job queued on Stampede to test changes with collisional ionization.

MHD

Pressure protections started immediately, so there must have been some in the missing portion last time, too. But it still seems to be running reasonably ok. Here's the current frame:

GJ 436b

Haven't hit a problem yet, it seems, so I'm continuing the current run to see where it does. If we do run into an issue, I'll run some tests with the outer boundary out farther, see if I can find a better balance between keeping the atmosphere hydrostatic and still having a decently-low ambient temperature.

​Spring code

Pretty much completely done with current to-do list. Merged into master branch on GitHub. Would still like to figure out how particles are distributed using MPI (how is an essential tree implemented?) and possibly adapt the spring portion to be a hybrid code, but OpenMP works well enough for now (though I'm segfaulting in the tests while using OpenMP, so need to fix that).

Summary of Asteroid Magnetization Cases

Compiled the results of all the cases for the Journal Club. Probably should shift to using L8 or L16 norms for the magnetic field instead of using the global maxima.

Summary of Moon Magnetization Simulations (probably not Steady State)

Focused on cases Resistive limit of the moon resistivity profile (from the Shimizu2013 paper, which in turn references a 1982 model) due to long standard queue.

Link to results:
​http://www.pas.rochester.edu/~aanand/movies/moon2.0/

The proposal we submitted just had the plot of the resistive case at 3.8 Ga with atmosphere.

Prof. Blackman believes adding crests and valleys to the lunar surface to simulate pockets of magentic field is more important than the resistivity profiles we select. Need to add a custom shape function for that.

Objectives

• Debugging
• Refine dust routines

Progress:

(A lot of research clusters in the UK have been down for almost three weeks now due to some security issue. I've moved my stuff to a working cluster but queue times have been long so progress has been slow.)

Dust Drag Progress & Issues

• Comparison with Dust Post-Processing: I've got the necessary simulations to compare my routines with the results from the post-processing
• Debugging: I fixed some stuff this morning but it must have broken something because suddenly I'm somehow getting an inflow from the lower x boundary about halfway through the simulation with passive advection: ​Start of Simulation → ​Middle of Simulation. For active advection, the velocity is suddenly completely gone ​Gas and Dust Momentum (FIXED! Forgot to remove some debugging stuff from last week and that messed stuff up there)

Update

• Debugging: The inflow was a bug in the dust wind initialisation. Some time ago I modified the wind so that the shock could come with or without dust. Turns out that causes an issue if the conditions inside the domain allow the bow shock to hit the lower boundary. I went back and changed that case to allow dust inflow and I found that for dust density 1e-16 the dust in the wind has no noticeable effect on the dust in the domain so effectively that makes the wind dustfree. ​Dust Clump evolution with negligible dust in the wind and passive advection
• In other news: My velocity values are now only 1e4 off down from 1e8… getting there

Up Next:

• More debugging

Ionization of the envelope

• ​https://www.pas.rochester.edu/~yishengtu/research_files/CEE_gp_meeting_06012020/mesa_profile.png

Equation of State

• MESA-OPAL-SCVH eos covers region -11.0 ⇐ logRho ⇐ 3.64; 2.4 ⇐ logT ⇐ 7.8
• I carved out the region -11.0 ⇐ logRho ⇐ 0; 2.4 ⇐ logT ⇐ 7.8
• Then extended the EOS to lower temperature using ideal gas, with gamma = 5/3 and mu = 2.21 (0.69 + 0.29 * 4 + 0.02 * 18)

Result

• Completed obtaining and inverting eos:​https://www.pas.rochester.edu/~yishengtu/research_files/CEE_gp_meeting_06012020/combined_mesa_eos.png
• Inverted EOS: T(D, E) and T(D, P):
  • T(D, E)​https://www.pas.rochester.edu/~yishengtu/research_files/CEE_gp_meeting_06012020/TDE.png
  • T(D, P) ​https://www.pas.rochester.edu/~yishengtu/research_files/CEE_gp_meeting_06012020/TDP.png
• Inversion shows self-consistency
  • E(D, T) & T(D, E):​https://www.pas.rochester.edu/~yishengtu/research_files/CEE_gp_meeting_06012020/EDT-TDE.png
  • P(D, T) & T(D, P):​https://www.pas.rochester.edu/~yishengtu/research_files/CEE_gp_meeting_06012020/PDT-TDP.png
  • E(D, P) & P(D, E):​https://www.pas.rochester.edu/~yishengtu/research_files/CEE_gp_meeting_06012020/PDE-EDP.png

• All EOS tables are ready

Next Step

• Running a star sitting on the grid with new tabulated EOS on Stampede for higher resolution?
  • Low resolution simulation with older table EOS, there are some small disagreements with current version but very similar (4 level AMR, concentrate refinement at the center)
  • Ideal gas rho: ​https://www.pas.rochester.edu/~yishengtu/research_files/CEE_gp_meeting_06012020/ideal_rho.gif
  • Ideal gas vr (radial velocity):​https://www.pas.rochester.edu/~yishengtu/research_files/CEE_gp_meeting_06012020/ideal_vr.gif
  • Table eos rho:​https://www.pas.rochester.edu/~yishengtu/research_files/CEE_gp_meeting_06012020/table_rho.gif
  • Table eos vr:​https://www.pas.rochester.edu/~yishengtu/research_files/CEE_gp_meeting_06012020/table_vr.gif
• Prepare for CE simulation

Perhaps future project?

• If we want to run simulation with a lower ambient, we might need an EOS at lower density (-18 ⇐ logRho ⇐ -11)
• To extend to lower density region (-18 ⇐ logRho ⇐ -11), we need to use PTEH eos, which is also part of MESA-eos.
• However, the code of PTEH is different from OPAL-SCVH, so some work are needed to get the additional outputs (non-trivial)
• Current python code I'm using to get the current EOS can be extended to implement the additional region (not too difficult)