Question on Jacobi implementation in NR3

[Devinzhang]

From page 573 to 575, some lines of code in implementation part are not so clear to me.
1. tracking the accumulated rotation of eigen vectors matrix v[0..n-1][0..n-1] should be as in equation 11.1.24, but the actual implementation is rot(v,s,tau,j,ip,j,iq). It is not clear to me how 11.1.24 is translated as rot(...)
2. I cannot see why we need vector b(n),d(n),z(n), since all accumulated rotation will be in the elements of matrix a[0..n-1][0..n-1] and the diagonal will be the eigenvalues after convergence(the matrix a is modified through rot anyway).

Thanks!