Meeting Notes for 16 January 2008
Present: Dan K, Tom L., Tom J. Bob, Liz, Mihael, Eric;
Absent: Mike W., Randy., K. Whelan, Joao
Problem with Uploading in Cosmic e-Lab
Bob has teachers who are encountering problems uploading. Mihael asked them to provide more detail.
CMS e-Lab Status
CMS e-Lab Status amd databases for e-Labs (Report from Dan K.)
Dan reconfigured his code with an eye towards jsp rather than php.
The tarball on leptoquark with code is ready for Mihael. It can be dropped into any Tomcat server.
Jsp version of the analysis setu works on leptoquark It is using tags. It builds a run list either by using xml or a database; it is hardwired to leptoquark serving the database. ROOT has to have access to the files that it needs for the analysis. Currently there is a problem that the server does not want to make cookies. Mihael says he should not worry about it because he thinks he can solve it. Currently user selects whether they want to use database or xml. This can be stored in the session. Assume that it will work. He mentioned a problem with Cgi servet. It's not a problem exactly. The CGI servlet is disabled by default in Tomcat5 (at least on the Fedora distribution). So it has to be manually enabled by editing the main web.xml file and the servlet has to be renamed (that's noted in the web.xml file in the servlet section). So if they haven't used the CGI on Tomcat before then they may have to enable it..
He had made a new version of the OGRE page so the the user can do graphical cuts. It is a Java applet. User will have to have java-enabled or it won’t work.
Databases for CMS and other e-Labs
Currently it is using MySQL. Dan did get it to work with PostgreS.
Eric is going to add MySQL to data1 for LIGO so Dan can use this for his e-Lab.
Databases are going to be necessary; we will have to go from flat file (XML ) to database.
Mihael advocated an abstract unified way of doing this as an ideal.
VDC is not maintained anymore. The motivation for making a simpler version of the VDC is no longer there with caching.
Eric: let’s use them wisely and coordinate our efforts.
Data catalog interface exists. It may not be rich enough.
Both MySQL and PostgreS can coexist in an e-Lab.
Eric is concerned about how can we tie the two databases together so that we can coordinate between the wiki and e-Lab and provide a common login. Mihael warned people to not use mediwiki with PostgreS.
Work on new CMS e-Lab
A teacher is currently working with Dan on a CMS e-Lab more like what will on the experiment.
CMS EO collaborators at CERN is excited, but have no plan yet; Trigger list still not set. We will have to wait for it. QuarkNet will have its own trigger.
Dan talks with Michael Barnett and Kris Whelan. It looks like the experiments are going to do their own things. The ATLAS e-Lab is very different. The two groups are aiming at very different things; The two approaches may complement each other.
- AMELIA – aimed at looking at event pictures with tracks where user can see where the particles go. Similar to what is done in ethe Master Classes. Visually trying to figure out what it is. Pictorially representing them. It uses a small dataset. Event viewer – any ATLAS event can be displayed. And it shows what happens in the detector and where the pieces went. You could identify a Higgs and look at it. Students will try to find the patterns visually.
- CMS – analyzing large datasets by looking at things statistically. Students will sift through data and find things or measure things the way a scientist would. Z-mass is the first approach. How do we know that? How do we identify the Z from the large dataset? Take the students through the process of sifting through the data. Uses a Gaussian peak to find the z-mass. Introduce the students to the more foreign notion of mass vs. weighing something on a scale. What is mass? How do we find it and measure it. The leptonic decays are very clean. Walk them through the whole process and then give them more data. It forces kids to think. They will start with Monte Carolo data. Isolate the students from particle identification. Give them electrons; do not make them find them.
Eric: What kind of conversation can students have with physicists? Giving them a trigger channel without guidance can be a problem. Dan K: Perhaps have graduate students available. They have five years of funding for the grad students. The kind on the trigger level are very rudimentary.The trigger will be shared by QuarkNet. We will not be able to change them very often. There will have to be a big discussion including students as to what the trigger should be. Tom L: provide a rubric for posters that is used by grad students to judge the work of the students. Students get invited into a forum based on their fine poster. Perhaps every semester we can change the trigger based on the input of this forum.
Status of Testing of Refactored Code (Nick, Bob, Tom)
There is a bug in SWIFT. . There is a problem in swift in passing lists..
Nick Dettman, Bob and Tom made a great deal of progress; formal list – save plots of those tests.so that they can be rerun and we can compare the results.
Tom J. advocated using the last raw datafile (table for the histogram). Golden standard – easier to do a diff between two plot files rather than a visual comparison. By saving plots and rerunning them good to compare
Bob – very close to making it production.
Mihael will try to fix the bug and we can do final testing on www12 and then put it on www18.
Article on i2u2 in International Science/Grid
Reverse Proxy for Production
Mihael implemented the reverse proxy on the production machines and the Seatlle teachers were able to use the e-Lab. Port 8080 was being blocked.
-- Main.LizQuigg - 16 Jan 2008