I am running a parallel paraview server on a large cluster. I am trying to use the ParticleTracer to visualize a flow field with a plane as a source. This works completely fine for serial runs of paraview, whether that be locally or through pvserver. However, anytime I run in parallel, I immediately get a seg fault error when applying the particle trace filter. I have attached the error to this post. It seems the problem is in the ParticleTracerBase class when trying to call getCacheDataTime. I have tried to turn on “Ignore Time” in the pipeline browser, but that also does not work.
I have tried this both with ParaView 5.10.1 and 5.11 on two completely different clusters. I am wondering if this is a known issue and if there might be some way to fix it. Unfortunately, I am short of time as I have a deadline soon and have very little time to debug over the next few days. This is also the reason I am not able to provide much in the way of helping you all reproduce my error.
Thanks for getting back. I attached a few time-steps of my data. I get the error right when applying the ParticleTracer filter on every dataset that I have tried. Again, this only happens for parallel runs.
I appreciate the help with this. I found that this helps in preventing immediate crashing; however, I have found the particle tracer to be generally unstable no matter what if a significant number of particles are used. Unfortunately, this makes it pretty much unusable for me.
I am trying to create a smoke tracer effect to visualize my flow field, so I obviously need a lot of particles for this. I am assuming that the particle tracer wasn’t designed for this use though, so I can’t really complain.
I haven’t been able to find another of way doing this in Paraview, unfortunately. Is there perhaps a way to compute an Eulerian advection (as opposed to the Lagrangian approach of the ParticleTracer filter) of some species inside of Paraview given the time-dependent velocity field that I compute from my CFD code? I obviously could do this myself with my own advection-diffusion solver, but it would be extremely convenient to be able to do in Paraview.
I am trying to create a smoke tracer effect to visualize my flow field, so I obviously need a lot of particles for this. I am assuming that the particle tracer wasn’t designed for this use though, so I can’t really complain.
The StreamTracer was not designed to produce an effect but to simply simulate particles behavior in a vector field.
The Lagrangian particle tracker let you implement your own model if that helps you.