shielding_par.f

c******************************************************************

c Do the SOR iteration in parallel

      subroutine fcalc_shielding_par(dt,rshield,sor_comm,myid2)

      include 'piccom.f'
      real dt
c     If rshield is not an integer, problem with the MPI routines
      integer rshield
c     sor_comm is the subset of MPI_COMM_WORLD communicator used for
c     the bloc sor
      integer sor_comm
c Correct the type of temporary variable (IHH). Better use initial letter.
      integer c

c Chebychev acceleration. Wild guess at the Jacoby convergence radius.

      rjac=1.-4./max(10,NRUSED)**2
      omega=1.
      maxits=2.5*NRUSED
      dconverge=1.e-5
      imin=1
      relax=0.5

c     fcalc_sor call : Do the iterations in parallel nrsize+1 is the 1st
c     dimension of the mesh (phi, rho, ...)  rshield+1 is 1+ the r
c     dimension really used. From 1 to rshield, but we add 1 to ensure
c     bloc communications of the BC at each iteration NTHUSED+2 is the
c     th dimension really used


      call fcalc_sor(nrsize+1,nthsize+1,rshield+1,NTHUSED+2,rjac,
     $     maxits,dconverge,k,sor_comm,myid2)

      
      
      if (myid2.eq.0) then
         write(*,'('':'',i3,$)')k
c     write(*,201)k,deltamax,relax
 201     format(' SOR iteration',I4,' delta:',f10.6,' relax=',f8.4)
c     Calculate electric force on probe. Moved to main.
      endif

c     Inner Boundary values
      do j=1,NTHUSED
         phi(0,j)=2.*phi(imin,j)-phi(imin+1,j)
      enddo
      do i=1,NRUSED
         phi(i,0)=phi(i,imin+1)
         phi(i,NTHUSED+1)=phi(i,NTHUSED-imin)
      enddo
c     write(*,*)'phi(rmax)=',phi(NRFULL,NTHUSED/2)

 123  continue

      end