Miscellaneous things I did and learned at CERN plus some personal commentary.
During my stay at CERN we were able to set up a WIRED event display for the Linear Collider Detector (LCD). We had previously created our own simple LCD event display which had taken approximately 4 man months, but by using WIRED we were able to create a far more functional event display in 2 days. Subsequent work has enabled WIRED event displays to run inside JAS. Although initially done for LCD this work should be trivially applicable to WIRED event displays for any other detector.
A very fruitful collaboration with Mark Donszelmann was established and we have prepared a long list of future task to collaborate on (documented separately). The goal is to have a shared base library where classes of common interest to JAS and WIRED (and by extension other HEP Java projects) reside, as well as moving to a "HEP Java framework" into which both JAS and/or WIRED can be plugged.
I was very impressed with the reusability of the WIRED event display and think it highlights the tremendous savings in manpower achievable by eliminating costly duplication of effort in individual experiments. These savings are only achievable by supporting some software development projects that are not initially targeted to individual experiments.
I gave a talk to Atlas collaboration on JAS and received some useful feedback. They seem genuinely interested in JAS and have put effort into comparing in to other analysis tools such as LHC++ and Root (with some favorable conclusions from the JAS perspective). Feedback included the need for easier programmatic fitting and scripting. Atlas has some documentation on "requirements" for analysis systems.
I was also able to work with Julius Hrivinac to show him how to set up a JAS server. We discovered that the current JAS native libraries do not work on Redhat 6, and built new ones that do.
At the LHC computing review Atlas showed plans to use Java as their official graphics stategy, and showed slides of JAS and WIRED.
ToDo: I have been experimenting with adding scripting to JAS (work partly inspired by the contribution of a JAS beanshell plugin from Fred Gray at UIUC). This needs to be completed and demonstrations of how JAS meets the Atlas analysis requirements (with or without scripting) needs to be generated.
Jean Bourotte and his colleaques on CMS are using JAS and the JAS plot bean for monitoring status and data from the upcoming CMS test beam run. They were having problems due to the fact that only a very old release of Java is installed on the central Solaris machines at CERN. By copying the production release of JDK 1.1.8 for Solaris from SLAC I was able to show that using this release fixed the problems they were having.
ToDo: IT should provide better support for Java on machines it maintains. This may well be a good place for small scale collaboration between SLAC and CERN (and other labs?).
ToDo: They have several requests for enhancements, most of which are already planned or underway.
Gave a talk to the PS group about JAS. They are already using some Java for their control system (including a joint project with Jefferson Lab, "cdev"). Talked at length with Georges-Henry HEMELSOET about his interest in using JAS for real time displays. He had a list of requirements.
JAS for Retinal Scans??
Xavier Bourguignon and Dino Ferrero have created an JAS DIM for reading LHC++ HTM histograms and HepTuples from Objectivity databases. Xavier summarized the status at the last LHC++ meeting. The implementation needs some more work to bring it to a level where it could be used by end users.
The current interface works only in Local mode, but I was able to work with Xavier to set up a first version of a Remote DIM.
After trying to implement the proposed AIDA histogram interface, I instead worked with Mark Donszelmann and Wolfgang Hoschek and proposed a simplified histogramming interface, which is documented here. This will be discussed further at a meeting after I leave.
Unfortunately my stay at CERN did not overlap with time when Andreas Pfeiffer was both at CERN and not busy with other meetings, but we did get a brief opportunity to discuss the plans for AIDA. Hopefully we can continue to work together between now and my next visit to CERN. My plan would be to develop AIDA functionality in JAS which closely mirrors C++ AIDA and Lizard functionality. It would help if there were occasional AIDA meetings which were arranged as video conferences.
At a presentation on JAS at the API weekly meeting I mentioned LCD's/my interest in an Java "wrapper" around Geant4 to make it possible to use Geant4 from Java. There was some interest in this among the Geant 4 team, and as a result I was invited to attend their workshop in Japan in May.
Atlas has a related project to describe their detector geometry in XML (and possible XSL) and they plan to hold a workshop on this immediately preceding their software workshop in Berkeley at the end of May.
A meeting was organized by Steve Fished to discuss his plans for nudging HEP into the modern world. I presented a brief overview of the java.freehep.org site and our interest in a Java interface to Geant 4.
As a result of this workshop I also revived some old work on interfacing Java and C++, and discussed possibly trying to implement this is collaboration with Mark Donszelmann and Julius Hrivinac.
I discussed the use of JAS for analyzing retinal scan data with Alan Litke from Santa Cruz, and created a few demo plots from a sample data file he sent me.
I researched a little into archival LEP data formats, with a view to seeing if SLD could adopt the same format for long-term storage of its DST data. Earlier attempts to create a common format for all LEP experiments seem to have dissolved. Aleph has a project to convert their data to Objectivity format, which is being worked on by Guenther Dissertori. Unfortunately I did not get time to discuss this with him.