09:17:14	 From Nicolai Ree : Will you provide us with example solutions for last weeks exercises?
09:17:31	 From Nicolai Ree : Cool :)
09:22:28	 From Jaime Caballer-Revenga : Don't do these considerations assume that you "know" kind of the shape of this curve (where the local minima are, kind of)?
09:22:55	 From Jaime Caballer-Revenga : or you are supposed to try until you grow an "intuition" of where the local minima are
09:22:56	 From Jaime Caballer-Revenga : ?
09:24:04	 From Laurits Høgel : Are you not supposed to maximise the likelihood?
09:36:02	 From Peter Andresen : Then don't you get a good p-value due to the calibration data? without necessarily fitting the signal well?
09:37:24	 From Nicolai Ree : Shouldn’t it be theta_i in the equation
09:37:34	 From Norman Pedersen : How could one implement the penalty term in Minuit ?
09:38:11	 From Peter Andresen : If you simoultaneously fit the calibration and the data you are interested in, Then don't you get a good p-value due to the calibration data? without necessarily fitting the signal well?
09:38:44	 From Jaime Caballer-Revenga : could you repeat the implementation on minuit?
Just, adding the penalty term to the result ?
09:38:54	 From Nicolai Ree : The extra term is inside the sum, right?
09:39:11	 From Nicolai Ree : ok
09:39:27	 From Nicolai Ree : thanks
09:39:33	 From Victor Valera : What if instead of a partially known value, I have a forbidden range? e.g. I don't want negative values
09:40:42	 From Anna S : You usually can set the range for which you want to minimize 2LLR
09:41:32	 From Anna S : (that was to Victor :P)
09:44:03	 From Jakob Riber Rasmussen : Can you repeat how one would fit/test the correlation of the parameters?
09:44:55	 From Jakob Riber Rasmussen : Ahh great
09:48:50	 From Jaime Caballer-Revenga : How do you obtain the template in first place?
09:48:56	 From Jaime Caballer-Revenga : heuristics?, trying?
09:49:04	 From Jaime Caballer-Revenga : ok thanks
09:52:56	 From Laurits Høgel : why is the error bar increasing with higher value in general? Shouldn’t it be opposite
09:53:50	 From Laurits Høgel : Not gonna argue with thtat
09:59:57	 From Peter Andresen : Could you say a bit about the parabola shape of the Chi-2 value, and what it could be used to say?
10:00:20	 From Peter Andresen : yea
10:02:05	 From Peter Andresen : And where did that come from=
10:02:06	 From Peter Andresen : ?
10:02:35	 From Peter Andresen : Alright, thanks :) 
10:02:59	 From Nicolai Ree : How is it that the errors on a parameter is calculated by Iminuit, I remember that is was when chi^2 increased by one w.r.t. the minimum chi^2 value, but how is it really happening in Minuit?
10:04:10	 From Peter Andresen : In the problem set, there was a question about how much data you would need to get a certaint error on your parameter, how do you figure this out ? :-) We can also take that when you go through it of course
10:04:24	 From Nicolai Ree : Yes :)
10:05:37	 From Peter Andresen : So the error decreases with sqrt(N)?
10:05:59	 From Peter Andresen : Perfect, thanks