Update 9/17

Radiation Pressure

No real progress to report, other than that stampede is beginning to test my patience. But I believe I've solved all of the problems (except the MPI issue - more below), so we should at least start getting some frames.

MPI issues

Located all of our calls to IRECV and ISEND, and all matching WAITs are there. No MPI errors are going ignored. Is there anything else we can do to be convinced that the problem's not on our end? (Here's ​one example of a case where irecv is called by RECV, which could potentially cause the issues we're seeing but not be anything we can affect).

I've found the environment variable that controls the request descriptors (not sure why I couldn't find it before, but I distinctly remember not being able to set it), and upped it to 1014, so hopefully that will be sufficient to allow us to run for the full two days we're allowed.

HD209458b ionizing flux only

Got some analysis done here. Mass loss rates are about an order of magnitude lower, which is to be expected given the tighter binding of the atmosphere.

Mass loss rate: ~5.3x109 g/s

It is (almost?) just Roche lobe overflow. Here's the Hill sphere (in red) overlaid on the wind, with contours of the dimensionless Coriolis length.

http://www.pas.rochester.edu/~adebrech/PlanetIonization/HD209458b_steady_state.png

MHD planets

Here's the global steady state.

http://www.pas.rochester.edu/~adebrech/HD209458b/MHD/mhd_planet_hydro_steady_state_global0150.png

Global simulations are actually much worse than local. The run stalls out after 6 frames. Running for 2 more days, we get three more frames; here's the last one output:

http://www.pas.rochester.edu/~adebrech/HD209458b/MHD/mhd_planet_global_159.png

Stellar wind implementation

I have a Matlab script that's solving for the (radial) velocity, pressure, and density of a pseudo-stellar wind with adiabatic gamma. I have noticed a couple of odd jumps that seem to be artifacts of the numerical solver, however. Also, I can't solve for any points inside the reference radius. This seems like a Matlab (or, specifically, vpasolve) limitation, but I haven't looked into other numerical methods in Matlab or otherwise. I've attached a sample solution file​.

Current list

I'll just keep this as a running tally, to keep us all organized a bit:

To Do

  1. Continue radiation pressure runs
  2. MHD planet runs (use the larger local runs as good-enough?)
  3. Invert subcycling for line transfer
  4. Add MHD to line transfer run
  5. Test John’s stellar wind for charge exchange
  6. Test charge exchange
  7. Run charge exchange
  8. Development for AMR line transfer/point source line transfer
  9. Development for metastable He

Complete

Debug MPI issue on stampede (ish)
Set up global MHD run (w/ stellar wind)

Attachments (1)

Download all attachments as: .zip

Comments

No comments.