I have a question about using passive particles in my simulations of convection in the shells of icy moons. The documentation for active particles states that increasing the number of particles might influence the solution, but I was wondering if a similar situation might arise for passive particles as well. I ran simulations with different numbers of particles (1e3, 1e4, 1e5 and 1e6) and the convective pattern seems to change when the number of particles reaches 1e5 and upwards. I am using passive particles and was wondering if the number of particles in this case could also have some effect on the numerical solution. I have attached some images for reference if that helps. Thanks!
Welcome and thank you for posting the question to the forum.
In short, simply increasing the number of passive particles should not have any influence on the solution itself (it will affect run time, etc).
If the solution is changing as you change the number of passive particles, something else in the model setup must be changing as well.
I suggest doing a diff of the parameter files for each model run to determine exactly what has changed - e.g., diff file_1.prm file_2.prm.
If the diff only returns that the specifications for parameters associated with the number of particles has changed, can you please post your PRM files so we can take a closer look?
I ran that model and a second one with 100x more particles (1e5) to about 690,000 years and the results both looks visually identical and show no variation in the output statistics.
I also did not identify any part of the model setup that should be affected by the total number of particles (passive).
From your existing model run data, can you isolate where in the model evolution the solution begins to diverge, which presumably is after 690,000 years?
Thank you for taking a look at it! You’re right in saying they’re the same. The issue could have been with the time step so I re-ran the simulations with the same time steps and they ended up being identical.
