c c file resc3.f c c . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . c . . c . copyright (c) 2008 by UCAR . c . . c . UNIVERSITY CORPORATION for ATMOSPHERIC RESEARCH . c . . c . all rights reserved . c . . c . . c . MUDPACK version 5.0.1 . c . . c . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . c c c subroutine resc3(nx,ny,nz,work,res) c c c ... purpose c c c subroutine resc3 computes the fine grid residual in the nx by ny by nz c array res after calling cuh3 or cud3 or cud3sa. if c c L * p = f c c is the n by n (n = nx*ny*nz) block tri-diagonal linear system resulting c from the pde discretization (done internally in cud3 or cuh3) and phi c is the approximation to p obtained by calling cud3, then resc3 computes c the nx by ny by nz residual array c c res = f - L * phi. c c one of the vector norms of res, c c || res || c c can be computed as a "measure" of how well phi satisfies the c discretization equations. for example, the following statements c will compute the location and size of the maximum residual in res c on cray computers: c c ijk = isamax(nx*ny*nz,res,1) c c kmax = (ijk-1)/(nx*ny) + 1 c c jmax = (ijk-(kmax-1)*nx*ny-1)/nx + 1 c c imax = ij - nx*((kmax-1)*ny-jmax+1) c c resmax = abs(res(imax,jmax,kmax)) c c c *** please note: c c let pe be the exact continuous solution to the elliptic pde c evaluated on the nx by ny by nz discretization grid c c let p be the exact solution to the linear discretization c c let phi be the approximation to p generated by the mudpack solver c c then discretization level error is defined by the condition c c || phi - p || < || p - pe ||. c = c c a common measure of multigrid efficieny is that discretization level c error is reached in one full multigrid cycle (see references [2,9] in c the mudpack file "readme"). this can happen before the residual is c reduced to the level of roundoff error. consequently, || res || is c a conservative measure of accuracy which can be wasteful if multi- c grid cycles are executed until it reaches the level of roundoff error. c c || res || can be used to estimate the convergence rate of multigrid c iteration. let r(n) be the residual and e(n) be the error after c executing n cycles. they are related by the residual equation c c L * e(n) = r(n). c c it follows that the ratio c c || r(n+1) || / || r(n) || c c estimates c c || e(n+1) || / || e(n) || c c which in turn estimates the convergence rate c c c = max || e(k+1) || / || e(k) ||. c k c c notice c n c || e(n) || < c || e(0) ||. c c c ... assumptions (see cud3.d or cud3sa.d or cuh3.d) c c (1) nx,ny,nz have the same values as iparm(14),iparm(15),iparm(16) c (used to set the fine grid resolution when calling cud3 or cuh3 or cud3sa) c c (2) work,phi are the same parameters used in calling cud3 or cuh3 or cud3sa c c (3) work,phi have not changed since the last call to cud3 or cud3sa or cuh3 c c c (3) assures a copy of the last approximation phi is contained in work. c if these assumptions are not true then resc3 cannot compute the c residual in res. c subroutine resc3(nx,ny,nz,work,res) c c compute fine grid residual in res after calling cud3 or cud3sa or cuh3 c implicit none integer nx,ny,nz,ic complex work(*),res(nx,ny,nz) c c set coefficient pointer c ic = 1+(nx+2)*(ny+2)*(nz+2) ic = 1+(nx+2)*(ny+2)*(nz+2) call rec3(nx,ny,nz,work,work(ic),res) return end subroutine rec3(nx,ny,nz,phi,cof,res) implicit none integer nx,ny,nz,i,j,k complex cof(nx,ny,nz,8),phi(0:nx+1,0:ny+1,0:nz+1),res(nx,ny,nz) do k=1,nz do j=1,ny do i=1,nx res(i,j,k) = cof(i,j,k,8) - ( + cof(i,j,k,1)*phi(i-1,j,k)+ + cof(i,j,k,2)*phi(i+1,j,k)+ + cof(i,j,k,3)*phi(i,j-1,k)+ + cof(i,j,k,4)*phi(i,j+1,k)+ + cof(i,j,k,5)*phi(i,j,k-1)+ + cof(i,j,k,6)*phi(i,j,k+1)+ + cof(i,j,k,7)*phi(i,j,k) ) end do end do end do return end