!--------------------------------------------------------------- 
!   This subroutine checks the number of external dipoles----
!---absorbing the correct number into the fragmenation functions 
!   it then returns the finite (msq_qcd*dip) ---------------------
!--------------------------------------------------------------- 
!--- Author C. Williams Feb 2011 
!-----------------------------------------------------------------

c--- Passed in
c---   p:      array of momenta to evaluate matrix elements
c---   p_phys: array of momenta to evaluate integrated dipoles

      subroutine qqb_gamgam_fragdips(p,p_phys,qcd_tree,msq_out) 
      implicit none
      include 'constants.f'
      include 'ewcouple.f'
      include 'ewcharge.f'
      include 'frag.f'
      include 'lastphot.f'
      double precision p(mxpart,4),p_phys(mxpart,4)
      double precision msq_qcd(-nf:nf,-nf:nf),msq_out(-nf:nf,-nf:nf)
      integer j,k
      double precision virt_dips,xl,dot,fsq 
      double precision aewo2pi,fi_gaq
      external qcd_tree


      aewo2pi=esq/(fourpi*twopi)            
      fsq=frag_scale**2

      xl=dlog(-two*dot(p_phys,2,lastphot)/fsq)
      virt_dips=+aewo2pi*(fi_gaq(z_frag,p_phys,xl,lastphot,2,2))

c--- fill underlying QCD matrix elements     
      call qcd_tree(p,msq_qcd) 

c--- fill output array
      do j=-nf,nf
         do k=-nf,nf
            
!   factor of two cancelled by statistical factor because two photons
            if((j.eq.0).and.(k.ne.0)) then
                  msq_out(j,k)=msq_qcd(j,k)*Q(k)**2*virt_dips
            elseif((j.ne.0).and.(k.eq.0)) then 
                 msq_out(j,k)=msq_qcd(j,k)*Q(j)**2*virt_dips
            endif
            
         enddo
      enddo
     

      return 
      end subroutine