local coordinates in output can be wrong when volume is offset, rotated or asymmetric

[Jochem Snuverink]

discussed with Laurie. Can be fixed with using the (local) coordinates from marker volume.

[Jochem Snuverink]

to be checked with offsets and tilts of element/markervolume

[Jochem Snuverink]

current idea for a solution is to create a parallel world with boxes without material without offsets to get the coordinates. this will work for any layer of volumes and any geometry / offset.

[Laurie Nevay]

Can get transform from higher up in the grometry but need to kmow how many levels to go up. This is hard and may involve a string compare which will be very inefficient for the large number of secondaries that will exist.

As mentioned by Jochem, a better solution, and the one implemented, is a parallel read out geometry. Strictly, this is a read out geometry rather than a parallel world in geant terminology althoug conceptually it is a parallel world. In tis read out geometry, simple cubes or cut tubs are used in place od the elements and these are always aligned to the reference trajectory. The output gets the transform from this volume. Note boundaries in the read out geometry limit step size in the 'mass world' too.