Dear all,
Thank you for the suggestions, Jeroen! I've implemented them in two comments, one about the dataset and a second one on toy studies. To be clear: my issue with their systematics was not that they do not vary their signal model, because they actually do. The comments can be found at http://cds.cern.ch/record/2120681.
Cheers,
Mick
Op 05/01/16 om 19:37 schreef Jeroen Van Tilburg:
Hi Mick,
My 2 cents. I agree that their answer on the fit systematics is quite brief. In such cases, it is best submit a reply where you repeat our concerns. More specifically:
- Gaussian constraints for the signal parameters: they tried this, so you can leave out this point, I believe.
- Using a larger data set: They ignore this point. So it is good to iterate it. They will not redo their analysis on a larger data set, but it is good to make the remark, such that it is properly answered.
- If I understand you correctly, they only change the background model. I think that they should at least try a few different models, where they vary the signal model as well. Just answer them that we think this should be done.
- If the systematic uncertainty from this variation is large, then they better assess this systematic uncertainty with toys. I.e. generate O(100) experiments with at least 10x more data in each toy than they have. Then try out different plausible fit models.
Cheers Jeroen
On 5 Jan, 2016, at 16:17 pm, Mick Mulder mick.mulder@cern.ch wrote:
Dear all,
The proponents have responded to our comments on PAPER-2015-053. They can be found at https://cds.cern.ch/record/2116308. Please take a look and see whether your comments were adequately adressed.
Personally, I think most of their response is good, but I am not satisfied with their answers to our questions on fixing most of their fit shapes and determining a systematic uncertainty for their fit model. They do not answer the question on fixing their shapes, and on the topic of determining the systematic uncertainties they only discuss why they choose now to vary their shapes to estimate fit model uncertainties, and ignore the topic of possible toy studies. What do you think we should do about this?
Cheers,
Mick
Op 17/12/15 om 13:42 schreef CERN Document Server Submission Engine:
Dear LHCb Colleague, The comment (LHCB-PAPER-2015-053-001-COMMENT-004) that you made on LHCB-PAPER-2015-053-001 (entitled: 'Measurement of the inclusive $B_{s}^{0} \rightarrow D_{s}^{(\ast)+}D_{s}^{(\ast)-}$ branching fraction') has itself been commented on by Adrian Andrew Pritchard [CERN - PH/ULB] (adrian.pritchard@cern.ch).
This new comment (LHCB-PAPER-2015-053-001-COMMENT-012) may be seen at http://cds.cern.ch/record/2116308
Best regards, The CERN Document Server Server support Team
Bfys-physics mailing list Bfys-physics@nikhef.nl https://mailman.nikhef.nl/mailman/listinfo/bfys-physics
Bfys-physics mailing list Bfys-physics@nikhef.nl https://mailman.nikhef.nl/mailman/listinfo/bfys-physics