Brian Hayden

I have had this conversation with David Rubin before, and we agreed that putting something out that we know is biased is actually worse than just not putting out anything...

Essentially the bias comes from the fact that you are assuming that the measured x1 and c value for every supernova is the "true" x1 and c value. Since the...

Those are not the priors I'm talking about. I'm talking about hyperpriors describing the true distribution that x1 and c are drawn from for all supernovae.

From what I've seen, the true distribution of x1 and c are pretty close to gaussian. I have not tested beyond this. This discussion is becoming a bit technical, and...

On first glance, I do think it could work. I still haven't convinced myself that we can modify a chi^2-like likelihood by just adding the likelihood of the observed x1...

I just want to clarify; I'm not trying to intentionally stifle development of SNCosmo. I'm just not ready to fully share my thoughts on this since the work is on-going...

Do you mean like a fitter that works on distance moduli? So if you have the distance modulus of each SN (i.e. from MLCS), then it's already 'standardized' and all...

The cosmology bias in the SALT case comes from the covariance between beta, sigma_int, and omega_m. If you build a bias into the distance moduli, you will still get a...