[dspsr-users] psrchive bug fix related to aux:rm and aux:dm

[Willem v. S. <van...@gm...>]

Hi everyone,

If you don't use aux:rm or aux:dm to store RM and DM variations, then this
update might not be relevant to you.

Before today, PSRCHIVE could incorrectly apply the corrections for aux:dm
(aux:rm) twice if the data were integrated in time or frequency before an
explicit call to dedisperse (correct Faraday rotation).  This would result
in a single phase shift (P.A. rotation) across all frequency channels.  As
a constant phase shift (P.A. rotation) is applied to all channels, this
would not lead to any smearing of the pulse (depolarization) in each
sub-integration; however, it would lead to smearing (depolarization) if the
data were later integrated in time with other sub-integrations that have
different values of aux:dm (aux:rm).

If the data are dedispersed (corrected for Faraday rotation) before
integration in time or frequency, then there should be no problem.

To address this issue,

   1. additional record keeping and state verification have been added in
   various parts of the PSRCHIVE library
   2. the code that implements the separate corrections for dm (rm) and
   aux:dm (aux:rm) has been simplified / clarified; and
   3. the "Expected Behaviour" documentation has been corrected / and
   clarified at
   https://psrchive.sourceforge.net/manuals/guide/design/cold_plasma.shtml

The additional state verification might trip over some edge cases that have
not been considered or tested.  Please let me know if you run into any
issues.

Cheers,
Willem