I fixed the loop issue in dbslmmfit.cpp that had been causing me to get results that differed from those of Sheng Yang's code. I now get dbslmm.txt results identical to those from Sheng Yang's code. I am also getting dbslmm.badsnps outputs that match exactly those from Sheng Yang's code. I verified this by the linux command line tool 'diff' on the pairs of output files.
Full Changelog: https://github.com/fboehm/DBSLMM/commits/v0.4.0