BLonD code development meeting

Friday 4 Aug 2017, 10:25 → 12:00 Europe/Zurich

864/2-B14 - SALLE J.B.ADAMS (CERN)

* Danilo cleaned up slices
* Now slices use some functions from cedar
* should rf params be an option for slices?
    * Alex is using it but it doesnt have to be mandatory
    * Danilo had it removed but will put it back
    * Alex: convert coordinates should be in cut options as it used only there
* Action point Danilo: move set_cuts(), track_cuts() to the cuts object
* Impedances will have to be cleaned up next
* bunch length scaled to be 4 sigma or not?
    * it has to be explicit and documented
* Action point Danilo add an option to all of the functions computing bunch
  length to scale by 4 or not
* Remove try except in fwhm(), replace with an if then else on t1, t2
* we need functions to identify where the bunches are
* What alex did in the past:
    * option 1: trying to find the bunch anywhere
    * option 2: you give a bunch pattern

* Danilo for slices cleanup: in toolbox there is a folder:
    * cern/tomoscpope
    * filters_and_fitting/cedar files
* TC21 is the testcase for the new slices
* Danilo used subtree from gitlab
    * You don't need to use your credentials cause cedar files will be by
      default copied under blond
* The cedar files have to be cleaned up as they will be distributed with blond
* the cedar files has to be benchmarked for correctness as well
* Danilo on new slices:
    * added cut, filter, slices, fit options object to pass to the constructor
    * eval is used to call a fit method by its name as a string
* Alex: rescaling the profile is better for the fit, converges faster
* Helga: for the moment we maintain the fits as they are but we change only the
  options

* Helga's presentation:
    * after the cleanup of separatrix the basic tracking will be ready
    * suggests to change some names in the code to be more descriptive
    * logger:
        * the runtime errors remain the same
        * for the rest you use logger
        * new logger class in toolbox
        * A sub-logger in each class reports to the root logger
        * We need to check whether the performance is getting worse while more
          log messages are added
* Alex: we should move the counter from rf_params to the tracker

* Konstantinos presentation on gpu profiling and pre-push script
    * Slides on indico page

* Helga: next week meeting for one-turn feedback(?)
* Next blond meeting in two weeks

 

10:30 → 11:00 Code clean-up

Speakers: Alexandre Lasheen (CERN), Danilo Quartullo (Sapienza Universita e INFN, Roma I (IT)), Helga Timko (CERN), Ivan Karpov (CERN), Mr Konstantinos Iliakis (CERN), Markus Schwarz (CERN), Simon Albright (CERN), Theodoros Argyropoulos (Instituto de Fisica Corpuscular IFIC (ES))

BLonD_clean-up_ideas.pdf

11:00 → 11:30 Profiling & GPU tests

Speaker: Mr Konstantinos Iliakis (CERN)

blond-meeting-040817.pdf

11:30 → 12:00 A.O.B.

Speakers: Alexandre Lasheen (CERN), Danilo Quartullo (Sapienza Universita e INFN, Roma I (IT)), Helga Timko (CERN), Ivan Karpov (CERN), Mr Konstantinos Iliakis (CERN), Markus Schwarz (CERN), Simon Albright (CERN), Theodoros Argyropoulos (Instituto de Fisica Corpuscular IFIC (ES))