Pseudopotential MRT LBM

I’m try to transform my debbuging BGk LBM code to MRT. I found some problems the results is not in agreement with the BGK model. I use this subroutine

Have any ideas ?