Electric Sail Simulations

[sandroos]

Posting here the first "reasonable" (=seems to work) electric sail simulation:

Some details:

• external charge density at origin that creates ~radial tether electric field
• 2 populations, protons and electrons, streaming from left to right at 400 km/s and 7 particles/m3
• I put extra protons circularly around the origin to shield the tether electric field

Movie shows:

• electron density (color)
• electric field (vector arrows)
• you can see electrons shielding the tether electric field and waves emanating from the tether

[sandroos]

Issues:

• should remove electrons near the tether instead of adding protons, now you can see an electron cloud flowing out with the extra protons
• electrons still not hot enough and tether voltage needs to be higher
• some waves get stuck at inflow boundary

Vlasiator issues:

• BUG: there is a bug somewhere triggered by load balance, this simulation was run without balancing load at all
  • species-specific MAXVDT values were not copied during load balance
• PERF: spatial translation does not scale with OpenMP at all, I suspect there are two reasons:
  • there is something (prob. dccrg calls) preventing good scaling
  • number of blocks per cell is quite low here, so a #pragma over blocks in cell does not scale well
• PERF: Poisson solver takes a lot of time
  • it is called many many times
  • one of the functions doesn't seem to scale well with OpenMP
• IMPR: the current dt evaluation routine doesn't work, spatial dt_max needs to be evaluated during acceleration to prevent CFL violation
  • dynamic dt can now be used
• PERF: Load balance needs to take Poisson solver into account, the balance is quite bad if only the number of velocity blocks is used