|
| |
Project: "Development of a GOOSY-PAW Successor System"
Minutes of the 4th Meeting on August 10, 1998
Participants: R. Barth, H. Brand, H. Essel, S. Hofmann, M. Kaspar, C.
Kozhuharov, N. Kurz, M. Richter, K. Schmidt, K. Sümmerer, H. Winter
Distribution: P. Braun-Munzinger, S. Glückert, H. Göringer, N.
Herrmann, R. Holzmann, H. J. Kluge, P. Koczon, W. Koenig, U. Krause, U. Lynen, V. Metag,
W.F.J. Müller, G. Münzenberg, V. Schaa, C. Scheidenberger, C. Schlegel, K.H. Schmidt, H.
J. Specht, R. Steiner, E. Stiel, R. Fox, K. Suzuki
1. Various Topics
- Several external individuals should be invited for giving talks about their systems,
ideas, and requests. Since in September 98 there will be therapy and also holiday for
several people of DV&EE most invitations should start not before the last week of
September. After some time of learning details of different systems some of us should go
to other institutes for detailed discussions about possibilities, trends, and
cooperations.
- A person to be invited should be K. Rehlich of DESY, co-author of DOOCS, an oo-based
control system for the TESLA test facility TTF, and some experiments at DESY. This
invitation has to be coordinated with the GSI accelerator control groups.
- Other guest speakers may come from Eurogam (Liverpool), Euroball, DESY (M. Gasthuber,
ZDV), MPI Hd (TSR), TUM und LMU München.
2. Experiments with Euroball by M. Kaspar
Euroball is a large European collaboration (http://www-gsi-vms.gsi.de/eb/html/eb_welcome.html)
running for over 8 years. It is not just a single experimental setup but more a collection
of different Gamma- and particle detectors running together in different combinations.
Items are single crystals, Clusters (7 crystals of HPGE + BGO), Clovers (4 crystals of
HPGE + BGO), particle Si telescope detectors, PPAC counters, neutron detectors, and
others. Currently, the first prototypes of segmented Clovers and segmented Clusters are
being tested. Once included into the set-ups, they will increase the number of channels
considerably. The event parameters are E, t, dE, and patterns in coincidence leading to
large multi-dimensional histograms (matrices, cubes, hyper-cubes).
Euroball III operates at 250 * (2 E + 1 t) from Ge crystals, several
hundred * (E + t + pattern) from BGOs, 40 * (dE + E + t) from Si, 10 to 50 t from PPAC, 3
* 15 E from a neutron wall, depending on configuration. At about 10 Gammas/event, 50 to
100 parameters/event with 10+ kHz event rate, 2-3 MB/s are written to tape (e.g. DLT
7000).
Euroball at GSI run with different combinations of Clusters and the
Crystal Ball at X7 or MSEP. First tests with flash ADCs (Struck) for Gamma-pulse analysis
had been done. About 20 to 200 channels, 5 parameters/event at 5 kHz event rate. Data
Acquisition had been done with the MBS and GOOSY.
Plans for future experiments are:
- VEGA (4 segmented Clovers = (4 + 9)*4 channels Ge + BGO or
- the Miniball (Munich): 6 * (7 + 7*6) = 294 channels or
- various combinations of subsets of these arrays.
Finally, 500+ channels with 50 to 100 parameters/event at 10 to 50 kHz
event rate are planned.
This data rate should not only be written on tape but also be analyzed
off-line.
Specific needs for Ge on-line data analysis:
- 8 k spectra (histograms) for Gamma
- 200 – 300 raw spectra
- 200 – 300 calibrated spectra
- patterns for quick overview
- pictures as a collection of many small, stamp like spectra for quick overview
- fast and reliable peak search for
- quasi automatic energy calibration of n channels
- data base for calibration sources
- general fit procedures (1/x, sqrt(x), exp(x),...) in addition to polynomials
- administration of calibration factors within a data base
- fit requests:
- gaussian
- optional Gamma peak tails, left/right
- multiple peaks with conditions (pre-given width, equal width,...)
- "real" re-binning of spectra: transformation from n channels to m channel (a
la GOOSY overlay transformation)
Specific needs for Ge off-line data analysis (in addition to the
on-line requests):
- matrices -> cubes -> hyper-cubes -> N-tuples
- efficient accumulation of large 4k * 4k long words = 64 MB matrices. Current problems:
- only 2-3 matrices in memory but 50 – 200 matrices to be accumulated
- therefore 10 – 100 times reading and scanning of all data from DLT
- therefore the need of local DLTs rather than remote mass storage in automatic tape
libraries
- large local disk space (> 200 GB) could help to get rid of DLT scanning
- optional symmetrical creation of multiple-dimensional spectra
- projections, gates, background subtraction
- export/import interfaces handling simple formats
- all object names must be indexed in any dimension. Possibly in ASCII as done by Tcl/Tk
- the analysis speed must be equivalent to the data acquisition speed
Specific needs for other parameters:
- scatter plots (point by point, i.e. event by event 2-dimensional matrix plot)
- mixed pictures with scatter plots and spectra
- simple set-up of scatter plots
- color used in scatter plots for the visualization of true/false conditions
- scatter plot with several points per event
- several scatter plots should run in the background
- about 200 – 300 conditions
- about 200 – 300 polygons
Controls are needed in addition to the data acquisition and analysis.
Setting, controlling, and logging of some hundred HV channels, Constant Fraction delays
and thresholds, PT100 temperature channels, valve control of liquid N2 system with
history, trending, graphical interface, data base. It is not clear whether the displays
should be merged with the acquisition/analysis windows (possible confusion).
Question concerning root:
- where are and how to handle conditions (windows, multi-windows, polygons)?
- where are polygons?
- where is a peak search facility?
- how to ensure data integrity (object persistency) like Global Section under VMS?
- are there scripts possible to control root?
- is asynchronous analysis with parallel event taking, histograming, display, control,
etc. possible?
- where is the user friendly command interface (e.g. instead of pseudo hbook
nomenclature)?
3. Next meeting
The project members agreed in first regular meetings twice a week on
Monday 14:00 and Thursday 14:00, always in the meeting room of the
DV&EE department
The next meetings and their agenda:
Thursday, August 13, 1998, 14:00
S. Hofmann: Experiments at SHIP
Monday, August 17, 1998, 14:00
K.H. Schmidt: Data Analysis of specific FRS Experiments
M. Richter, August 11, 1998
|
New GSI
Analysis System GO4