Hi, I would like to write something similar to: for(t in 1:100) v[x[t]] <- v[y[t]] + v[z[t]] in a vectorized form. The x, y, and z vectors may contain duplicates (so v[x] <- v[y] + v[z] has different semantics). The for loop is not efficient enough for my purposes and I would like to avoid using C/Fortran. This problem occurred to me on several occasions and I feel it is quite general. Does anyone have an idea how to solve it nicely? Thanks, bk