Reproducing Ohlmann+16a results: 1st attempt

Recap of Last Post

  • In the last post, I performed several runs with varying separation and secondary mass.
  • However it was determined that point particles were not feeling the gravity of the gas, so the results were wrong.

New Work

  • I wrote a script in IDL which calculates the Kepler orbit given the initial velocities, and used it to show that the orbit obtained was just what would be expected if the point particles were not feeling the gas gravity, further confirming the above conclusion.
  • This problem with the code was corrected by Baowei and Jonathan.
  • The base case (run 084 from last blog) was rerun now with the code corrected, with the goal of reproducing the basic results from Ohlmann+16a.

Summary of New Results

  • The simulation now looks "reasonable" at the beginning (first ~dynamical time) but then deviates from the results of Ohlmann+16a.
  • I realized while writing this blog that the problem is caused by the mistaken re-introduction of the secondary (hence the introduction of a third particle with mass equal to that of the secondary) on restarts. In fact, there was a switch in the problem.data file called "secondary_already_present" to deal with restarts, but I had forgotten to change it to .true. before restarting.

Results

I) Circular orbit as in Ohlmann16a
Damp088) Extrapolated hydro BCs, Multipole expansion Poisson BCs, ambient dyne/cm
(comet compute 1728 cores, 2 cores per task to increase memory per task)
( cm, , 5 levels AMR)
Restarted from run Damp062, at s, just after Damping stopped, to s
​2d density
​2d density and velocity
​2d density and velocity (zoomed)
​2d temperature (zoomed)
​2d particle velocity (zoomed)
​2d particle position (zoomed)
​2d Particle mass (zoomed)

​Particle trajectories, (with circles representing softening radius)
​Particle separation with time
​Particle trajectories up to mistaken introduction of third particle
​Particle separation with time up to mistaken introduction of third particle

Discussion

  • What has happened is that another secondary got introduced on each restart. This happens at frame 107. It also happens at frame 92 but this won't matter much because I had to redo that frame only on bluehive because there had been an I/O problem at that frame on comet. So the simulation is correct up to frame 106, or elapsed time t = 2.12e6 - 1.5e6 = 6.2e5 s = 7.2 days, about 1 dynamical time, but is wrong after that.
  • Results up to the kink seen in the plot of separation vs time are very close to those of Ohlmann+16a (see below). The introduction of the third point particle occurs before the kink. So the results so far are consistent with those of Ohlmann+16a. (If one looks closely there is another point in the plot of separation vs time that is a little off, and this is due to the one frame that was done on bluehive where the 3rd particle also appeared.)

Next Steps
Once I've edited the problem.data file to make "secondary_already_present=.true.", I will redo the simulation from frame 106. I will then again compare the result with ​Ohlmann+16a ​Figure 1.

Here are the other figures from that paper for reference:
​Figure 2
​Figure 3
​Figure 4

Comments

No comments.