particle parallel run crashes - values out of range
i experienced strange behavior of my "Pope-particles".
I have massless particles following the flowfield. They carry scalars, which are used to compute stuff...
The solver works fine on one CPU, but switching to parallel it crashes immediately.
Investigating the problem I found that, once running in parallel, I have values totaly out of range!
Example: The constructor assigns for the temperature either 300K or 400k. Applying some mixing I have (at least on one processor) 300<T<400. Once I switch to parallel I have values totally out of range...eg T=1e-40 , or T=0...thus dividing something by T gives the solver a hard time then...
I thought this might be related to not initilized member variables (e.g. T) on particles cloned to occupy ghost cells...but the number of values out of range is not correlated to the number of processor boundaries.
The "clone()" function, explicitly used in the code, works just fine.
So I am kind of lost right now..
Any hints would be appreciated!!
Thanks a lot!
P.S.: I am using OF-1.7.1
P.P.S.: And where are the ghost cells and ghost particles created anyways ? ( is it Pstream or the Decompose or ...?)
|All times are GMT -4. The time now is 21:25.|