The calculation of the initial amplitudes (normalisation) fails for this case. The amplitudes (eps1_0,eps2_0) are offset in the horizontal plane (see fma_emit_lhc2016_scan_qx22_qy10_p27.png). One reason could be that this case tracks for a fixed dp and the fma analysis uses the 6D matrix.
Plotting the normalisation for one case, doesn't reveal any problems ( -> tnorm_all.png) -> to check one should plot for small angles.
The input files for this case are attached in input_files.tar.gz
input_files.tar.gz