Present: Stepan, Nathan, Matt, Tim, John

Please send comments/corrections to the whole EC mail list.

1. We talked generally about how to expedite our first physics publications. Raphael (see copy below) had suggested that assignment of specific data sets to specific students had actually been an impediment to coordinating student help on the general analysis topics, since it could be interpreted as asking the student to concentrate solely on one data set.

2. Several disagreed with Raphael's premise, that assigning data sets to students was the fundamental problem. In the US, a student thesis will generally have to be the full analysis of a particular data set, though not all the work on the topic must be the student's. This fact does not prevent a student from contributing to various parts of other efforts or becoming the resident expert of some piece of the analysis, it just requires that the student fashion a coherent description of the entire analysis at the time of the thesis.

 Consequently, we did not see changing the assignment of data sets to students. But we did see requiring students to contribute substantially to the broader analyses, and certainly to getting HPS publications out. Matt noted that many of the students have already been playing this role: Kyle on trigger; Bradley on MC; Matt on tracking; Sebouh with some bump hunt details.

3. Matt noted that the analysis group is forming a bump hunt group and a vertexing group to push along those analysis/publication areas.The groups will enlist the current thesis students, who will be expected to contribute to the ongoing efforts. Matt hopes for more engagement from the advisors of those students and from other senior HPS members. The existing analysis review committee is a start.

4. In an effort to more directly engage help from the advisors of our graduate students, John will draft an email to them requesting more direct involvement. EC will comment and amend as necessary. We should also request more direct analysis involvement from our senior physicists, with the understanding that other responsibilities may put severe constraints on one's ability to help.

5. Stepan noted that much of our delay in publishing has not been due to lack of coordinating the students' efforts, but to fundamental problems we uncovered with the Monte Carlos and more minor problems with the reconstruction. We all agreed. Even so, several felt that more coordination will be very valuable for getting the 2015 bump hunt result out the door quickly, and for compiling the full 2015 results, 1.5 + 0.5 mm running, bump and vertex. It will also be a useful start for analyzing the 2016 data.

 

Comment from Raphael after last Wednesday's HPS meeting:.

Following your comment during today's meeting on the problem of having one point of contact only for the analysis. My experience with Ani is that we have been using a wrong method for assigning graduate students to projects. By assigning them a data set that they "own", we encourage them of doing everything on their data set and nothing on the others. We lose a lot of time because of that, because each student as to become expert in everything at some point and write codes for everything (exchanges is also complicated by the fact that some people are working at the java code level, others work on DSTs using different platforms).

A possibility to think about to reduce this issue would be to assign projects in a different manner, where each grad student would be contributing to each paper in their own ways. We can have a student working on cut optimization, a student on the statistical analysis methods etc. Note that we already have students working in this fashion like Kyle with trigger and Matt with simulation and tracking. I would suggest to have everyone "specialize" on a few aspects of the procedure in this fashion and not reserve data set to specific students in the future.

As an example, we spent the past year with Ani reproducing the data analysis of Omar with only minor modifications for 1.5 mm. This is formative for her certainly, but it is not very useful to the collaboration. If one had been taking charge of the statistical analysis, and the other the cut optimization and comparison to simulation, we would have saved a lot of time.

Such method would also help having communication as people will not be able to just do their analysis in their corner anymore. They will need to get data from people and provide their results to others.

Hope that will help for the reflection on the topic.

 

Respectfully submitted,

John

  • No labels