15:01:14 From Matthew Feickert to Everyone:
	yeah, I'd try JAX if you haven't yet
15:05:35 From Matthew Feickert to Everyone:
	Boyang, have you gone through the user guide? https://pyhf.github.io/pyhf-tutorial/
15:05:59 From Lukas Alexander Heinrich to Everyone:
	possibly interesting research direction
15:06:00 From Lukas Alexander Heinrich to Everyone:
	https://github.com/gradhep/neos/blob/main/examples/binning.ipynb
15:06:25 From Matthew Feickert to Everyone:
	add 😀
15:08:31 From Boyang Zhang to Everyone:
	Boyang, have you gone through the user guide? https://pyhf.github.io/pyhf-tutorial/
	yeah, but I still don’t quite understand… I will try to ask more questions later
15:09:37 From Matthew Feickert to Everyone:
	As I'll mention later, please open GitHub Discussions to ask questions. If things aren't clear we want to discuss it to make it better
15:10:28 From Boyang Zhang to Everyone:
	👍
15:21:58 From Ramdas Makhmanazarov to Everyone:
	Some additional question about modifiers. Am I right, that the default constraint term for staterror is on constrained by normal? And only this constraint term can be used for this modifier
15:49:27 From Alexander Held to Everyone:
	if you have zero rate but non-zero observed data I imagine you’ll never get around the problem
15:49:40 From Alexander Held to Everyone:
	you just have an ill-defined model to begin with for such a case
15:54:40 From Matthew Feickert to Everyone:
	yeah
16:00:18 From Matthew Feickert to Everyone:
	HistFactory v2 is something we've been discussing for some time, but we need to decide what is going to go in there. So getting input from belle II users will be quite important
16:01:57 From Lukas Alexander Heinrich to Everyone:
	a key question is also how to encode in JSON .. but all solvable issues
16:33:33 From Géraldine Räuber to Everyone:
	I would have two questions that are related to plotting style (might be relevant for publications):
	1. is it possible to customize the colours in the plots produced by cabinetry?
	2. is it possible to choose the ordering of the templates in which they stack? (my observation so far was that they are ordered alphabetically)
16:33:53 From Matthew Feickert to Everyone:
	It is all matplotlib
16:33:55 From Matthew Feickert to Everyone:
	so yeah
16:36:43 From Alexander Held to Everyone:
	here’s the issue about colors: https://github.com/scikit-hep/cabinetry/issues/381
16:36:51 From Géraldine Räuber to Everyone:
	thank you for your answer!
16:37:27 From Alexander Held to Everyone:
	a more general idea we had was to allow the user to provide custom rcParams and instead of overriding them (which we do now) just take the information from them to apply to plots
16:38:11 From Alexander Held to Everyone:
	feel free to ping these issues to help prioritize features!
17:05:37 From Ramdas Makhmanazarov to Everyone:
	Has anyone compared pyhf and trexfitter?
17:05:55 From Alexander Held to Everyone:
	yes, extensively (I’m a TRExFitter developer)
17:08:26 From Alexander Held to Everyone:
	see https://github.com/scikit-hep/iminuit/blob/develop/doc/notebooks/template_fits.ipynb and https://github.com/scikit-hep/iminuit/blob/develop/doc/notebooks/template_model_mix.ipynb for the iminuit template stuff
17:08:33 From Ramdas Makhmanazarov to Everyone:
	Reacted to "yes, extensively (I’..." with 👍
17:09:12 From Alexander Held to Everyone:
	@Ramdas: out of curiosity, are you on ATLAS, or where do you know TRExFitter from?
17:22:05 From Nikolai Hartmann to Everyone:
	get_labels_and_handles maybe
17:23:36 From Ramdas Makhmanazarov to Everyone:
	Replying to "@Ramdas: out of curi..." 
	
	 Yes, I am from ATLAS. I also had some presentation about my experience with pyhf. Bud I'm afraid that I can't present today. My main questions were about gamma parameters pruning(setting thresholds), constraint types changing, reading roostats produced by trexfitter, some questions about asimov data fitting (differences between default pyhf api and cabinetry api - why there inconsistencies).
17:26:01 From Alexander Held to Everyone:
	Replying to "@Ramdas: out of curi..."
	
	Feel free to follow up on that via issues or email, I’m particularly curious in the API changes you mention.
17:27:15 From Ramdas Makhmanazarov to Everyone:
	Replying to "@Ramdas: out of curi..." 
	
	 Your email alexander.held.@cern.ch
	?
17:27:29 From Alexander Held to Everyone:
	Replying to "@Ramdas: out of curi..."
	
	yes
17:27:35 From Nikolai Hartmann to Everyone:
	seems the function for the legend handles and lables i had in mind is on the axes ;)
https://matplotlib.org/stable/api/_as_gen/matplotlib.axes.Axes.get_legend_handles_labels.html#matplotlib.axes.Axes.get_legend_handles_labels
17:29:13 From Alexander Held to Everyone:
	Replying to "seems the function f..."
	
	that gives me two empty lists `([], [])
	` but I am probably using it wrong
17:30:35 From souravdey to Everyone:
	Replying to "seems the function f..."
	
	could you try something like this:
	proxy1 = plt.Rectangle((0, 0), 1, 1, fc="#b27cc9", linewidth=3)
	handles = [proxyh1]
	labels=['label']
	ax.legend(handles,labels,loc='upper left')
17:31:08 From Alexander Held to Everyone:
	I have to switch to another meeting, thank you very much for organizing this workshop! It was really interesting to see the use in pyhf and the interest from Belle II.
17:31:39 From souravdey to Everyone:
	Reacted to "I have to switch to ..." with ❤️
17:31:52 From Ramdas Makhmanazarov to Everyone:
	Replying to "@Ramdas: out of curi..." 
	
	 Some of my results I presented in github. We had conversation there.
	I'm newbie in hep soft and some kin of analysis. And my supervisors told me to start with pyhf, as alternative for trexfitter, and to provide faster analysis (now we do reanalysis).
	Sorry for my English.
	I'll send you my slides
17:36:02 From Ramdas Makhmanazarov to Everyone:
	Replying to "@Ramdas: out of curi..." 
	
	 Unfortunately I leaved from of my working PC, because it too late on my location. Tomorrow I'll send
17:36:50 From Ramdas Makhmanazarov to Everyone:
	Reacted to "I have to switch to ..." with ❤️
17:42:47 From Ramdas Makhmanazarov to Everyone:
	Thank you all for great workshop and presentations!
17:55:07 From Arul to Everyone:
	Thank you to all the speakers and organizers for a  great workshop :)