The fractional systematic error on FPi or sigma_pp is given by the sum in quadrature of the two vectors:systot_excinc560_r.dat and sys_wei60.dat.

This vectors are 60 bins from 0.35 to 0.95 GeV2.
The error given in systot_excinc560_r.dat is fully correlated.
The error given in sys_wei60.dat is 2x2 fully correlated starting from bin 2 to bin 58.
As an example to evaluate its contribution to amu, you can use the following FORTRAN code:

To open the cov matrix, just use the following FORTRAN code:

      open(UNIT=52,FILE=
     $'./systot_excinc560_r.dat', !bckg fully corr
     $  STATUS='OLD')

      open(UNIT=53,FILE=
     $'./sys_wei60.dat', !bckg 2x2 corr
     $  STATUS='OLD')

       DO I=1,60
          read(52,409) dbg(i)
          read(53,409) dbg2(i)
       ENDDO

 409   format(f10.5)

       close(52)
       close(53)

       dsy=0.
       dsy2=0.

       do i=1,60
          qqi = 0.355+(i-1)*0.01 
          amu(i) = facto*0.01*spp(i)*kern(qqi)
          dsys(i) = amu(i)*dbg(i) !full corr
          dsy = dsy+dsys(i) !corr
       enddo  
       dw2=(amu(1)*dbg2(1))**2
       do i=2,58,2
          j=i+1
          qqi = 0.355+(i-1)*0.01 
          amu(i) = facto*0.01*spp(i)*kern(qqi)
          dbamu(i)=amu(i)*dbg2(i) 
          dbamu(j)=amu(j)*dbg2(j) 
          dw2=dw2+(dbamu(i)+dbamu(j))**2
          print*,qqi,i,j,'dbamu(',i,')= ',dbamu(i),
     &  'dbamu(',j,')= ',dbamu(j),dbg2(i),dbg2(j)
       enddo
       dw2=dw2+(amu(60)*dbg2(60))**2
       dsyt =sqrt(dsy**2+dw2)  !corr+2x2corr

Finally if you don't care about correlation you can use systot2_excinc560_rt.dat
which is the sum in quadrature of the two above.