I think the lensing results are OK. The ISW-redshift distortion correlation only comes into the counts-CMB cross-correlation. (and of course I think CAMB sources is correct anyway!)
I think at some blissful point in the past, with intel compilers, deallocate(A,stat=err) would set the error flag if A was not allocated rather than crashing. Unfortunately this seems to have changed around version 10.0, and presumably also problems in other compilers, which neccessitates lots of tedious nullify statements, or consistently tracking what is allocated. However in this particular case I don't get the bus error with the intel compiler. The fix is probably to add
Code: Select all
call MatterPowerdata_Nullify(CAMB_Pk)
at the top of MakeNonLinearSources in cmbmain.f90 - does that help?