Meeting Notes for 20 August 2008 CMS e-Lab Swift – Analysis Advice on GUI for analysis (Dan) ROOT Versions Student Testing CMS/LIGO pretest/posttest Link fixes System Updates – Argonne (Mihael) SC08: Particiants Changes from last year ELabs Meeting New hire update Pretest answers/students/ databases LIGO report: Multi-channel plotting Project: Quake Catcher Network at Stanford Machine Changes for LIGO

Meeting Notes for 20 August 2008

Present: Mihael, Eric, Liz, Tom L., Tom J., Dan K, Bob
Absent: Mike W., Randy, Kris, Joao

CMS e-Lab

Swift – Analysis

Dan K. joined the telecon. Tom L. reported that he did a full soup to nuts test of the Swift/OGRE analysis (data, plot, poster and rerunning of the analysis from the plot). Mihael reported that he did not actually have all the data, but that OGRE just uses what it can find.

The following was decided:

• Dan will maintain the OGRE code in the SVN repository.
• Tom L. will document a small problem in the analysis interface involving a couple of columns.
• Mihael will investigate the bug.
• Mihael will get a leptoquark account and copy the full collection of data to Argonne. For now we will be keeping it locally and shipping data to the ELabs analysis nodes when a user runs an analysis using the grid.
• Tom L. will get the results of Doug Berry's study of data sizes that CMS might encounter when CMS moves from testbeam data. This will influence how we decide to handle data in the future.
• Mihael will put the changes he made in the common code currently in the trunk to the branch so we can test it with Cosmic. He wants to be sure that these changes do not break the Cosmic analysis.

Advice on GUI for analysis (Dan)

Dan K. has a new version of OGRE that uses graphical cuts. It is implemented with a Java Applet and keeps a history of the cuts the user tries. It all works with a web browser which loads the applet that creates the plot. The user must have the Java Virtual Machine installed. Eric pointed out that it has to have a failover to the version we currently have which does not rely on a java applet. Graphical cuts are temporary and you commit at some stage. Dan wants advice on how to present the history in the interface. In a future meeting we can have a talk-aloud so everyone has a better idea of how to solve it. Maybe we could use EVO.

We discussed what are the requirements for an e-Lab. Many of us think it is o.k. to require Javascript being turned on. Eric has tried to allow users without Javascript to use his.

ROOT Versions

Dan reported that ROOT 4 is sufficient for OGRE; In fact, it breaks a little with ROOT 5, according to Mihael. Eric will tell Dan what version he is using, to see if they are compatible. If so, we only have to build one version of ROOT for the grid nodes (ie. Teraport, for starters), and Eric is going to have to do that anyway.

Eric: Bluestone uses ROOT 4.04/02g of 20050902

Student Testing

Tom L. is working with a student, Ike, from a local suburb who is using OGRE to do a test beam study. Mihael will let Tom L. know when the data is available on www13 so he can move over to it and test. Eventually we will move him to the Monte Carlo version.

CMS/LIGO pretest/posttest

We need these from Tom L. working with Jean.

Link fixes

Tom is still working through his piles to find the link fixes Liz will put into the CMS e-Lab.

System Updates – Argonne (Mihael)

Mihael has not heard from Rick yet. Today or tomorrow should be the start. Bob will need to update the news item to extend the dates.

SC08:

Particiants

Dan is not planning to go. Bob and Tom J. (100%); Tom L (Wednesday only), Mihael (availability depends on his commitments to SWIFT at conference), Eric if there is money for him to go). Ask Marge – Kris/Beth/Ken have not attended for a couple years. New hire may want to go. Eric thinks going to Chicago would be more useful than SC2008.

Changes from last year

Last year we had 4 monitors from one pc; Dispense with single PC and use independent laptops. Big monitor is useful. One for each laptop. Tie-wrap the laptop on the desktop. We will reproduce the café sign.

ELabs Meeting

We need to get together sometime. Would a day during Oct 21st- 24th be a good date? Talk to Marge.

New hire update

Mihael and Mike interviewed six new candidates on the phone on Monday; they concluded we should make an offer to one of the candidates that came to Fermilab on the previous Friday. Liz has set that in motion.

Pretest answers/students/ databases

We decided that we are going to have an annual cycle to testing. We will start all over each year. Liz asked for a final decision on whether we totally clean out these databases each year or rename old tests and add new test. Eric pointed out the rule that you never delete data. Liz has exported that data so we do have a copy. Eric suggested we add a year column to the answer database. Liz has already added a timestamp. We could add a year column to the question database, but if we name the database with the year (e.g., pretest2007) and insert new questions with the generic name pretest and posttest, we should not have to change the code.

LIGO report:

Multi-channel plotting

A "transformation" to plot multiple channels in the same graph is almost done; He has it working in a ROOT script. Examples are posted to Gladstone room. Very neat.

Project: Quake Catcher Network at Stanford

This project collects data from the accelerometers on newer laptops. They detected the recent Los Angeles earthquake. You get can add accelerometers to use with desktop computers if you don't have the newer laptops, but it may require some minor soldering to do so.

Tom L. noted that the staff and teachers need to understand the funding cycle we are in where assessment is critical. We have to keep our focus on this and not get to diverted by neat additions..

Eric responded that he was only pointing out another neat project which might have some overlap with ours, given that the LIGO e-Lab involves seismic data. We would certainly not support the other project, but would classroom participation be a help or a hindrance to use of the LIGO e-Lab?

Machine Changes for LIGO

A machine at Caltech will replace tekoa. Tekoa currently collects data and runs Bluestone; Bluestone will run at Argonne. Eric and Fred hope to move the data acquisition function to a machine at Caltech and ship data to Argonne. Fred is looking for specs for the machine and talking to the Caltech people. Advantages – we can get the Livingstone data as well. A UMinn scientist named Vuk Mandic is setting up sensors in the Homestake Mine, to study conditions for a possible underground gravitational wave detector. Those data will go to Caltech and they could be accessible to us. Eric outlilned the tier structure of the LHC data centers, as first described in the Monarch report, and also as used by CMS and ATLAS. In our case Argonne could be thought of as a "Tier2" site for LIGO (Tier 0 is Hanford and Livingston while Tier 1 site is Caltech), but separate from existing LIGO Tier 2 sites (UW Milwaulkee and Penn State).

-- Main.LizQuigg - 28 Aug 2008
-- Main.EricMyers - 03 Sep 2008