Here's a list of all the priors available to use. The default priors
(same as "prior_default.pro") are:

phase:  flat between -3 and +3 days, a Gaussian rolloff with sigma=0.5 d
        beyond that. always use this prior, because if you're more than
        3 days off in the time of maximum, the C matrix you are using 
        is wrong. In the next iteration, the fit should move to the flat
        part of the prior (with a much better estimate of the C matrix).

delta:  flat between -0.3 and 1.6, a Gaussian rolloff with sigma=0.1 on
        either side. this keeps objects within the training set deltas.

A_V:    negative A_V disallowed, exponential falloff with tau=0.457 mag
        for positive A_V. this is the correct prior for the low-z sample.

R_V:    Gaussian in 1/R_V, with <R_V> = 3.1 and stddev(R_V) = 0.4

Plots of all the priors are available, see prior_<name>.ps


File                    Priors 
------------------------------------------------------------------------------
prior_adam.pro          default phase, R_V; A_V prior matches Adam's homebrew
                          version (an exponential with tau = 1 mag, convolved
                          with a Gaussian sigma = 0.2 mag; negative A_V is 
                          allowed because of the convolution), delta prior 
                          also matches Adam's hard limits -0.4 < delta < 1.5

prior_avnear0.pro       default phase, delta, R_V; A_V prior is approximately
                          a delta function (like fixing A_V = 0)

prior_avpeak.pro        default phase, delta, R_V; A_V prior is peaked near 0,
                          a one-sided Gaussian with sigma = 1 mag, plus a 
                          "delta" function getting 20% weight (A_V >= 0)

prior_default.pro       default phase, delta, A_V, R_V

prior_d19.pro           default phase, delta, A_V (exponential 1.9*0.138 mag),
                          R_V constrained near 1.9

prior_flatav.pro        default phase, delta, R_V; A_V prior is flat (A_V >= 0)

prior_flatnegav.pro     default phase, delta, R_V; flat A_V, negative A_V ok

prior_flatrv.pro        default phase, delta, A_V; flat R_V (1 <= R_V <= 7)

prior_flatrvav.pro      default phase, delta; flat A_V (A_V >=0) and 
                          flat R_V (1 <= R_V <= 7)

prior_gauss1.pro        default phase, delta, R_V; A_V prior is a one-sided 
                          Gaussian (sigma=1.0 mag), A_V >= 0 only

prior_glos.pro          default phase, delta, R_V; A_V prior mimics calculations
                          based on galaxy lines-of-sight ("delta" function plus
                          exponential with scale length of 0.4 mag)

prior_glosz.pro         default phase, R_V; A_V and delta prior same as glos,
                          except they get truncated by selection effects present
                          in the ESSENCE survey from Peter Garnavich's simulations

prior_nodel.pro         default phase, A_V, R_V; there is no delta prior so
                          the fits may go to regions extrapolating out of
                          the training set. watch out for this!

prior_none.pro          default phase; no delta prior, flat A_V (including
                          negative A_V), flat R_V (1 <= R_V <= 7)

prior_nonegav.pro       default phase; no delta prior, flat A_V (A_V >= 0),
                          flat R_V (1 <= R_V <= 7)

prior_rv15.pro          default phase; no delta prior, flat A_V (including
                          negative A_V), R_V constrained very near 1.5

prior_rv19.pro          default phase; no delta prior, flat A_V (including
                          negative A_V), R_V constrained very near 1.9

prior_rv31.pro          default phase; no delta prior, flat A_V (including
                          negative A_V), R_V constrained very near 3.1 

prior_snls.pro          same as glosz, except cuts take effect at a deeper  
                          magnitude, to simulate SNLS selection effects (MWV)

prior_tonry.pro         default phase, delta, R_V; extinction prior is based
                          on Tonry et al. 2003: Gaussian with sigma = 1.0 mag 
                          for positive A_V, Gaussian with sigma = 0.2 mag for
                          negative A_V, plus 50% of those in a "delta" function

