5 Spectropolarimetry Reduction Algorithms

Spectropolarimetry reduction involves determining the polarization of the data from observations in two polarization states, the E and O rays produced by a polarizing prism, and two pockels cell states or waveplate positions (for each Stokes parameter) which are referred to as A and B. Thus the polarization is derived from four measurements AE, AO, BE and BO. The reduction should be such that it is insensitive to polarization within the spectrograph, and flat field effects which will give rise to systematic difference between the E and O states, and to time variations in transmission and seeing which will give rise to changes between A and B.

The IPCS2STOKES program uses the simple difference algorithm where the polarization is given by:

P = AEBE (AOBO) AE + BE + AO + BO (1)

This algorithm does not fully correct for transmission changes between the polarization states, but since the Pockels cell polarimeter modulates rapidly such effects normally average out and do not cause problems.

The CCD2STOKES and CCD2POL programs give a choice of algorithms. The first algorithm, referred to as the OLD algorithm, since it is the original one used by these programs is a modification of the difference method where the O data is scaled by a factor F as follows:

F = AE + BE AO + BO (2)
P = AEBEF(AOBO) AE + BE + F(AO + BO) (3)

This scaling makes the algorithm much less sensitive to transmission variations

The alternative algorithm is the RATIO algorithm which is as follows:

R2 = AE/AO BE/BO (4)
P = R 1 R + 1 (5)

The RATIO algorithm works very well on bright stars, but can fail on faint objects (or on 100% polarized calibration sources) through attempting to take the sqaure root of a negative number. Under these circumstances the OLD algorithm should be used.