Difference in algorithms of full-way bounce back and half-way bounce back boundary conditions in LBM?