Hi Costas,
I'd like to participate in the discussion, but unfortunately Friday is
the busiest day of the week for me, and the only time I'm not already
scheduled is kind of early in Fermitime (8am-9:30am CST). :-(
If I'm the only one, though, I guess I can catch up later.
-Jeremy
On 03/02/2017 03:43 PM, Costas Andreopoulos wrote:
> Hi all.
>
> Sorry we were silent for a while regarding the tuning work. It has been a busy,
> period, and we also had to travel quite a bit since our last phone mtg.
>
> During the last two days Marco, Rhiannon and myself visited one of the main
> Professor authors (Holger) at Durham. Marco/Rhiannon, with the help of
> Holger, did some amazing work, taking us beyond the toy example of a single
> parameter / single dataset fit. We now have a more complex fit involving all
> MiniBooNE, T2K and MINERvA neutrino CC0pi data and 4 params
> (Ma and QE, MEC, RES normalisations). All
> * Best-fit parameters,
> * parameter covariance matrices
> * plots with best fit predictions
> * error envelopes in the fit kinematic variable space, and
> * the full fit covariance matrix in the space of all fit bins
> are outputs of this exercise.
>
> We also now know how to obtain the Professor parameterisation in each bin,
> and how to override their chisq calculation (in order to include penalty terms,
> add nuisance parameters that do not require MC in order to produce a
> parameterisation - e.g. relative normalisations for different datasets -).
> Marco and Holger are now working on the system to allow to specify input
> weights and, for example, switch on/off individual datasets from the fit.
> Hopefully this will be ready this coming week and Marco is also looking at adding
> the PGoF statistic for the fits. This is already beyond a toy and more firmly
> into the actual tuning territory.
>
> We suggest we have a mtg next week to discuss progress and plan the next
> (3rd) iteration which should produce results we would not hesitate to present
> at conferences. My aim is for a 4th iteration before NuINT17 which could be
> both presented there and published in a journal. After this is done, and with
> all the machinery already in place, we can turn our attention to the 1pi tuning.
>
> The next week is already mostly booked up. Is Friday a good day?
>
> cheers
> Costas
>
|