1 of 28

AP Bootcamp @ CERN :

The Post Mortem

9 March 2020

Sam Meehan, Clemens Lange

Savannah Thais, Lukas Heinrich

1

2 of 28

2

Lunch

ATLAS

[Lukas & Sam]

Lunch

Dinner

@ Meyrinoise

Cont. Integration

[Giordon]

Docker

[Danicka]

Monday

Tuesday

Wednesday

CERN Re-Ana

[Tibor]

Lunch

Discussion with Theorists + Reception

CMS

[Savannah & Clemens]

ATLAS

[Lukas & Sam]

CMS

[Savannah & Clemens]

ATLAS

[Lukas & Sam]

CMS

[Savannah & Clemens]

Catch-up Time

Catch-up Time

Catch-up Time

Awesome H(tautau) Analysis

Pre-workshop

3 of 28

Action Photos

3

4 of 28

Who Came

4

About as expected

(overflow room caused bias to ATLAS)

20% of people were sent to CERN *specifically* for this

5 of 28

Who Came

5

This was generally the target audience → “teach the teachers”

We did not accept ~5 masters students during pre-registration

6 of 28

Who Came

6

Large span of ages [23,39]

Better than the status quo in ATLAS, but this shouldn’t necessarily be our “benchmark”

7 of 28

Did they skip out?

7

They came for the mornings

They came for the afternoons

Missing Participants : (out of 30 registered)

  • 3 cancellations & 1 no show
  • “funky” numbers from overflow room - 5 people

8 of 28

Base Knowledge

8

They were “alright-ish” programmers

They were “alright-ish” with basic version control

9 of 28

Base Knowledge

9

Most didn’t know what we wanted to teach them

The language they were meant to learn was completely foreign

They didn’t really know if their code was “well-written” for workflows

10 of 28

CI/CD Mentalities

10

Novice

Expert

Novice

Expert

Before

After

11 of 28

CI/CD Technicalities

11

Before

After

Novice

Expert

Novice

Expert

12 of 28

CI/CD Technicalities for my Analysis

12

Novice

Expert

Novice

Expert

Before

After

13 of 28

Docker

13

Novice

Expert

Novice

Expert

Before

After

14 of 28

Docker for my Analysis

14

Novice

Expert

Novice

Expert

Before

After

15 of 28

Workflows

15

Before

After

16 of 28

Containerization for my Analysis

16

Novice

Expert

Novice

Expert

Before

After

17 of 28

Winning hearts and Winning minds

17

Novice

Expert

18 of 28

Constructive Feedback

18

The material became available too late - it would've been better if we could have the setup procedure for REANA ready beforehand. If that was the case I could know it wouldn't work for me outside CERN. Also, in a tutorial where the attendees are not used to the actual commands, it is very confusing to have "autofill" turned on in the command line. Personally, this feature made me lose my focus, and I could not use the commands on my own terminal due to technical host problems.

If possible, please provide some online instructions in case someone gets left behind (Videotape the sessions)

Was a few issues with the prepared material that lead to some confusion (for me) - otherwise excellent!

As previously mentioned, the afternoon sessions were somewhat difficult to follow as they were seemingly off the cuff. Also, the fact the CI/CD afternoon session resulted in a broken pipeline was less than helpful as it wasn't obvious what working code would have looked like

For the ATLAS specific part on monday afternoon, I think it would have been very useful to have a similar structure as the CI/CD and Docker tutorials, with hidden solutions available if needed. The ATLAS specific part, especially the first afternoon, was very difficult for many to follow, as most time was spent trying type in what we thought Giordon was typing, which just resulted in many errors. So it would have been nice with a tutorial with solutions, or a shared document with the right commands/.yml file content to fall back on if you ended up behind for some reason.

Things to keep in mind for the future

(ATLAS, material, logistics, organization)

My CI sometimes failed because I didn't know the h--> tautau analysis well enough. Some amount of the time that I would have liked to learn about gitlab/docker, I had to debug my mistakes in using the scripts. Having scripts that are easier to execute would help. For example: not having to copy and paste arguments like xsec into command line, not having compilation and execution in the same skimming script, maybe having numbers in the beginning of the scripts like 1_skimming.sh, 2_histograms.py... .

19 of 28

Anecdotes - Constructive

19

It was a really long day. It would have been nice to be in a room with windows, or to have a full lunch break. I was really tired and running out of steam by Wednesday. I don't know what the solution is here, this is just how I felt, sorry!

Perhaps having a planned afternoon break to get some fresh air and keeping afternoon structured more like the morning in my opinion.

More time for working on own analysis would've been good to finish

Since it was a workshop that required me to switch between looking at the not-so-big projector screen and my own laptop, I thought it would have been nicer if we could be in a room where we all faced the screen, instead of a round table. I also found the TV where the speakers share their terminals too small.

There were times when the speaker started talking, but people were still talking in small groups to try to fix some problems, and I could not hear the speaker at all, but felt too rude to ask them to stop talking.

It was sometimes difficult to see the command line / terminal when projected

I can not see the screen clearly

Things to keep in mind for the future

(ATLAS, material, logistics, organization)

20 of 28

Anecdotes - Good

20

I really appreciated all the commands and outputs shown on the workshop git. They really helped me to follow the speakers, and reassured that I was doing the right things/seeing the right outputs.

Well done. Explained an extremely new concept in a way that I still remember and can use.

Very useful to have people around to help pack my analysis into a dockerfile.

I enjoyed trying to apply what I learnt in the morning session to my own analysis!

I liked the motivation for reproducing analyses in the intro.

Interesting workshop, well organised, really helpful material/tutorials on the indico page & well communicated by the instructors, significant contribution by everyone who was there to help the participants be able to complete the exercises, friendly atmosphere, extremely long survey :-)

Great work. Love the software carpentry setup.

But in general I thought it was super useful, I feel way less overwhelmed by discussions about these tools and do feel empowered to make changes/improvements to the setup my analysis already uses. Thank you so much for all of your hard work! I would definitely recommend this to other people.

I did, however, appreciate that it didn't last the full week so that it didn't feel like taking so much time out of my work.

Things to keep in mind for the future

(ATLAS, material, logistics, organization)

“When I attended the ATLAS software tutorial at the beginning of my PhD, it only lasted 3 days. No analysis preservation was mentioned (or maybe it was and I just zoned out). Later on, the tutorial for expanded to a week. I think it is really to have the workshop because I did not get this during my software tutorial time. And I got to do hands on stuff!”

21 of 28

Anecdotes - Good

21

I really appreciated all the commands and outputs shown on the workshop git. They really helped me to follow the speakers, and reassured that I was doing the right things/seeing the right outputs.

Well done. Explained an extremely new concept in a way that I still remember and can use.

Very useful to have people around to help pack my analysis into a dockerfile.

I enjoyed trying to apply what I learnt in the morning session to my own analysis!

I liked the motivation for reproducing analyses in the intro.

Interesting workshop, well organised, really helpful material/tutorials on the indico page & well communicated by the instructors, significant contribution by everyone who was there to help the participants be able to complete the exercises, friendly atmosphere, extremely long survey :-)

Great work. Love the software carpentry setup.

But in general I thought it was super useful, I feel way less overwhelmed by discussions about these tools and do feel empowered to make changes/improvements to the setup my analysis already uses. Thank you so much for all of your hard work! I would definitely recommend this to other people.

I did, however, appreciate that it didn't last the full week so that it didn't feel like taking so much time out of my work.

Things to keep in mind for the future

(ATLAS, material, logistics, organization)

Seemed like it was a nice alpha but needs work to become user friendly

22 of 28

Do it again?

22

“Loved it! More workshops like this please!”

“Make this mainstream”

CMS

ATLAS

Disagree

Agree

Disagree

Agree

23 of 28

Documentation : Always Important

23

Let’s start by making sure everyone knows where the docs are

24 of 28

We have future instructors!

24

This is because everyone was *great* when teaching

25 of 28

Thank You!

25

“I would like to thank you for the organization. I really like the topics even if i was a beginner. I was able to follow the lecturers and I really appreciated the heavy hands on approach. Congrats to all of you”

“Great time and thanks”

“I'd like to thank all the organizers and mentors. This workshop was definitely the most awesome one I've ever been to.”

“Excellent workshop - well done!”

26 of 28

Future Plans

  • Success breeds success : Future iterations already in the works
    • @CERN : Leonora Vesterbacka (NYU postdoc) as potential lead
      • Many local mentors to help with instruction
    • @Valencia : Emma Toro (UW postdoc → Valencia faculty) as lead
  • Orthogonal bootcamp idea : “VIP Workshop”
    • Workshops seem targeted at younger audiences thus far
    • Make “space” for more senior colleagues
  • Caveat : Everything is contingent on current COVID-19 stuff
    • Guidance at CERN

26

27 of 28

CMS afternoons and view of services (1)

27

28 of 28

CMS afternoons (2)

28