NuTools

In short: the problem is that atmospheric neutrino fluxes are parametrised by angles respect to the zenith, and GENIE in its wisdom decides that the zenith is the direction of the z axis.
On the other end, LArSoft definitely sets z direction as the beam direction.
My request is for GENIEHelper to rotate the flux for us.

• first, NOvA should weigh in, in that if their z is toward the zenith, this request can be rejected immediately
• GENIE allows a rotation to be specified, and GENIEHelper passes one through from the configuration; the rotation that seems to do the trick is

  [
    +1.0,  0.0,  0.0,  # new x axis in old coordinates
     0.0,  0.0,  1.0,  # new y axis in old coordinates
    -1.0,  0.0,  0.0   # new z axis in old coordinates
  ]

  
  this rotation swaps y and z, resulting in z pointing west and x pointing south (because x is not changed, and GENIE x points south).
• this request is for such a matrix being used by default by GENIEHelper when dealing with atmospheric fluxes, to be combined with the user-specified one.

Of course this is a breaking change that needs to be carefully pondered and well advertised.
Also of course, this does not fix the azimuth angle: in LArSoft z should point away from the beam source, usually not westward.
A new parameter (zFromSouth) might even be added to simplify this; if I am not mistaken, the transformation matrix would become

[
  +sin(zFromSouth),  -cos(zFromSouth),  0.0,  # new x axis in old coordinates
         0.0,               0.0,        1.0,  # new y axis in old coordinates
  -cos(zFromSouth),  -sin(zFromSouth),  0.0   # new z axis in old coordinates
]

with zFromSouth being defined as the angle respect to south (west is π/2) the beam comes from (in the [0,+π/2] range for DUNE, MINOS and NOvA).