Now I'm trying to simulate colloidal particles with hydrodynamics using fix srd command. If I set all the boundary conditions to periodic, the simulation runs well. But if I just add two walls (use fix wall/srd) and change the corresponding boundaries to fixed, the simulation will always run a little while and end with errors of lost atoms (and a big number of atoms, so must be srd particles). I'm really surprised and I thought it should be easy to bound the srd particles between walls by simply using fix wall/srd command. What did I miss? BTW, I use fix wall lj/126 to bound colloidal particles, but as they are far from the wall, I don't think they are interacting with it when the error occurs.