LISA — a powerful program package for Listmode and spectral data analysis

Nuclear Instruments and Methods in Physics Research A 340 (1994) 379-383 North-Holland

NUCLEAR INSTRUMENTS & METHODS IN PHYSICS RESEARCH Section A

LISA - a powerful program package for LIstmode and Spectral data Analysis A. Oberstedt ', FA. Hambsch

CEC - JRC Institute for Reference Materials and Measurements (IRMM), Reteeseweg, B-2440 Geel, Belgium

Received 11 October 1993 LISA is a graphical program package which enables both off-line listmode and spectral data evaluation as well as on-line monitoring while multi-parameter experiments are running. It can be executed on every computer with a UNIX operating system and an X-WINDOW environment, running PV-WAVE from Visual Numerics Incorporation. This package is basically written in the language PV-WAVE CL, but integration of procedures written in the C-language and execution of UNIX shell commands lead to an additional increase of performance . 1. Introduction Experiments performed at the 7 MV Van de Graaff accelerator of the Institute for Reference Materials and Measurements (IRMM) in Geel, Belgium, are in the field of nuclear data . The types of data measurements range from simple acquisition of spectra up to the sequential accumulation of events of several parameters which might have been obtained from multiple coincidences between various detectors. In the second case data are acquired in listmode . Hence the objectives of the new LISA system are the following: 1) analysis of listmode and spectral data; 2) control of experimental parameters and data monitoring during data acquisition with count rates up to several 1000 s -1 ; 3) graphical representation like displaying, plotting and fitting, up to publication quality; 4) user-friendly handling during terminal sessions; 5) "automatic" data treatment without any user interference; 6) flexibility in adapting the data analysis to the needs of different experiments . The LIstmode and Spectral data Analysis system (LISA) is designed to meet these requirements . It is a completely revised and improved version of a program

* Corresponding author . * * Former : Central Bureau for Nuclear Measurements (CBNM). Present address: Chalmers University of Technology, Dept. of Physics, 412 96 Gdteborg, Sweden .

package, which has been started to be developed some years ago for the same purpose [1]. How these requirements were accomplished will be the subject of the following chapters.

2. Basic ideas and concepts The LIstmode and Spectral data Analysis system (LISA) is a system of programs and tools designed to facilitate the data processing associated with experiments at the Van de Graaff accelerator facilities (VG) at IRMM . It should allow an experimentalist to solve standard problems (e .g . spectrum evaluation, display, plotting, fitting) as well as the definition of non-standard data processing functions. So, LISA recognizes two distinct data types - listmode data (event by event) and spectral data (accumulated events), respectively . Here an event represents a set of numerical values associated with a "physical event" ; it is a collection of parameters measured in connection with the "physical event" . An event consists of several parameters correlated in time ; each parameter is a numerical value resulting from conversion of an analog signal produced by a detector . The parameters may represent physical coincidences (e .g . two detectors at different angles) or just different attributes of a physical event, like energy loss and residual energy of particles in a telescope detector . To keep track of the correlations between different parameters for later analysis, they are normally written event-by-event onto a storage device . Treatment of spectral data belongs to the group of standard applications like mentioned above, whereas

0168-9002/94/$07 .00 © 1994 - Elsevier Science B .V . All rights reserved SSDI 0168-9002(93)E1189-5

380

A . Oberstedt, EJ Hambseh /Nucl. Instr. and Meth. m Phys . Res. A 340 (1994) 379-383

listmode data treatment is a little bit more sophisticated, as will be described in more detail . LISA is designed to process listmode data, independently of whether they are coming either directly from a data acquisition system or from a storage device . In the second case the data have to be analyzed and the resulting configurations (one- or two-dimensional spectra) have to be stored . During data acquisition the (listmode) raw data have to be stored immediately ; nevertheless they may also be analyzed at the same time, in order to obtain at least preliminary results. The results, of course, have to be stored, too. A schematic view of the listmode data flow is shown in Fig. 1 . 3. The reaLISAtion The LIstmode and Spectral data Analysis system (LISA) has been developed and is currently running on an ETHERNET-based local area network (LAN) of SUN-sparc workstations and PC with X-terminal emulations including all necessary peripherals, available at VG of IRMM (see Fig. 2) . Thus, this program package is adapted to a UNIX operating system with an Xwindow environment . LISA is basically written in the interpreter language PV-WAVE CL from Visual Numerics Inc. [2]. Visual Numerics Workstation Analysis and Visualization Environment (PV-WAVE) is a powerful software system for the interactive analysis of scientific data, a programming language and a plotting and image display package. It is truly interactive enabling quickly to display, reduce, analyze, redisplay and interact with large multidimensional datasets . PVWAVE owes much of its semantics to the programming language APL. The power and conciseness of PV-WAVE is due to this APL influence . However, the main departure from APL is in the syntax, which is by far not so difficult to keep in mind as APL. The possibilities of PV-WAVE can be extended by combining the basic commands to new procedures and/or functions, offering superb possibilities for graphical presentation and easy treatment of large amounts of data . Thus, this major part of the software provides all options like displaying, plotting or fitting analytical functions to experimental data, necessary for preparation of diagrams worth for publication . Although the visualization is very fast, calculations (e .g. calibrations, iterations etc.) which have to be done event-by-event, are principally too time-consuming for any interpreter language . However, this problem, concerning only listmode data treatment, could be avoided by introducing special analysis modules (so-called "processors") written in the high level language C, which are strongly experiment dependent . The communication and data access to and from the "C-processor" is done within

PV-WAVE, so from the user's point of view the necessary C-programming could be kept to a minimum. So the only thing a user has to do, to analyze listmode data, is to write an analysis program according to his particular purposes . Many of those processors for standardized experiments at VG are already available and may be copied ; so actually some small modifications are sufficient to enable someone to analyze his experimental listmode data . Also a couple of service routines do exist, which makes a processor programmer's life much more comfortable . A detailed description is given in ref. [3]. All different applications may be selected by mouse/cursor from a pull down menu . For listmode data treatment however, it is necessary to give additional information about the kind of data to be treated (e.g . the number of parameters per event, the number of one- or two-dimensional configurations the data should be sorted in, their sizes etc.), and of course, the name of the processor which contains the routines of how to analyze the data . Finally, the names of the input data file as well as the output file for the resulting configurations to be stored in, have to be given. This can quite easily be done by editing a command-file, containing several keywords and their associated parameters, which performs the entire initialization. Having done this and after starting a LISA listmode data analysis session, the data are analyzed automatically, until either all data have been processed or the user terminates the session . Nevertheless, the user may in-

EXPERIMENT

Fig. l . Listmode data flow in LISA for both execution modes. Data may originate directly from the experiment (dashed lines). In this case the I/O-part of the system copies the measured raw data immediately to a file on an external device (disk or tape). Alternatively, the data may be fed in from a storage device . The dispatcher distributes them to several analysis sections . The output consists of one- or two-dimensional configurations which are stored, too.

A. Oberstedt, F-J. Hambsch /Nucl. Instr. and Meth . in Phys. Res. A 340 (1994) 379-383 terfere in order to change display modes or to zoom to particular areas of interest by single character keyboard input and/or movements of the mouse. The list

381

of all key-commands available and a brief description of their meanings is permanently displayed. Sometimes it might be useful, to be able to decide about the

Id(Illfl II'V VI!II ;Î~uül ~IÎ?

iIIIiIIIIIVIIElIII~IIIII'üll'I~Ilill9l

IUVIIfIUllllldNlIiII~UaIIID

SUN SCSI-BUS

Idefix

RS232 SCSI-BUS

i_ . . . -- . . ..._~P ~I POSTSCRIPT

LINE PRINTER

COLOUR POSTSCRIPT

Fig. 2. Schematic view of the computing environment at the Van de Graaff accelerator facilities . Besides the different SUN workstations PCs are connected which can be used with X-terminal emulations .

A . Oberstedt, F -J. Hambsch / Nucl Instr and Meth . i n Phys . Res . A 340 (1994) 379-383

382

h,ma n]mtQ :h~CC5_g .IjJt7 I . - ;~naleaaia~o~ 7~ . I

v x b T . 0 O O

y O

N O

O

O +J C C O .ca "Ô E â 0 .5 0 >

3

c

m _. ; N m L > O

." I a-I NN

rrss anses JJJJ

c

x o m a+

NO 01 V

mm ut ('1(D NN N P

m

ro C

C vro~ 7 O

¢ô

O m > V >â~iM1L v . .0 o ZO7 O\M ~ V N V m~ n i 0Wm G u m G ~m

--c . .» N mIL mr- 9 mZL7 m G Q NÔ C C m ~ Nm~~G Or 7> G OI F Cr uOß ~ ~m'r,V ..

p U

":: .Ç . . . . . . .C . . ..x .. . . .N . . . . .

m m

w

ôc

m ro m

" 7¢\n " Oro1+G0 " 2 I E OOU 22 . .v dP\dPvv

V N V7 u V

V

N m

S i a F ô

O

C

vC

V N N

â

a

V

O O

K H â

N

O S V ÔÔ a¢

FF NN Q ~F s GG OL7 O O Z ¢n

W O m d \u '-N r r\m~y7 y=~ L O o I ~OI~S -E -E m2 \ m ~CS a\I ""Y I N = .IUI OG 7 o 1~-r n m YGOo O N 1+ N V7IL IL

m

i. b0 O :-J C y U O U N

~ r- o n

rr mao NNZ 0 wwu. wrr Vm OG :r -r 0 0` V 7 GO . ô m 7m o-muv a o0 G O.G Y G N C C UVIOu Z G m c m . mr o o V 00 .~L "" O V Nm C "~ C O G~ (af 0 0 ollti ~~Mm mcN _cocococc _ __ mm v ro m7ro0(OION~NOO~vmvmemvv ss m m m m N 1* Gro ro mG ro ro mG mm mm m G7 G7 G V C vl~mus o : :(5 r ro roN 7 " " G r~mxVUr" = 1 Vvvvvvvv y N~ .11.v "r\

X C O ô 3 y C V .,J

U ä W 29

A. Oberstedt, F -J. Hambsch /Nucl. Instr. and Meth . in Phys. Res. A 340 (1994) 379-383

accumulation of events according to their physical significance . This can be done by defining conditions for two-dimensional configurations, so-called "regions of interest", either via mouse button clicks or by editing an ASCII-file . So, if selected, spectra are accumulated, depending on whether an event hits the indicated area or not. On the display, regions of interest are represented as polygons . A typical workspace for a listmode data analysis session is shown in Fig. 3. From the point of view of an experimentalist analyzing measured data, it is also desirable to use the same processor for evaluation of different data files originating from different runs of the same experiment. But considering e.g . experiments running over periods of days or weeks, it may most probably occur that calibration parameters like amplification factors, electronic offsets etc. may slightly change from one run to another. So the data treatment is the same, but the values of these parameters differ . This was taken into account with parameter-files containing the proper values associated to different runs, which may be created, added to the command-files, and used for the correct initialization of the needed parameters . This means that listmode data analysis of all runs of one experiment can be performed automatically within one LISA session . Hence, the treatment of large amounts of listmode data resulting from multi-parameter coincidence experiments is provided by LISA in a very convenient way. 4. Conclusions The objective to develop a program package, which combines the possibility of powerful data treatment for both, off-line analysis and on-line monitoring even at count rates of several thousand events per second, with user-friendly handling, has been achieved with LISA. It is installed and running on the cluster of SUN workstations at the Van de Graaff accelerator facilities at the

383

Institute for Reference Materials and Measurements (IRMM) in Geel, Belgium. Here the command language PV-WAVE CL provides fast visualization of data, where the user may choose between different plot modes by single character keyboard input and mouse operations . Always all possible commands at a moment are displayed to be selected. More time-consuming operations, like calculations in case of listmode data analysis, are handed over to processors written in the C-language in order to increase performance. Finally, the execution of UNIX shell-commands (invisible for the user) in the framework of LISA leads to an easy management of files created during an analysis session. All together, LISA represents a very useful tool for on-line and off-line data analysis of both, listmode and spectral data, which is used by an increasing number of experimentalists . Last but not least, another big advantage of LISA which should be mentioned is the portability to any type of experiment, to any other UNIXsystem and even - with some small modifications - to computer systems running with the VMS operating system . Acknowledgement One of the authors (A.O .) would like to acknowledge gratefully the warm hospitality at the Van de Graaff accelerator facilities at IRMM . The informative discussions with various users already in the state of development of LISA are much appreciated . References [1] FA . Hambsch and B.S . Jäckel, CBNM internal report GE/R/VG/68/90. [2] Visual Numerics Incorporation, Boulder CO, USA. [3] A. Oberstedt and FA . Hambsch, IRMM internal report GE/R/VG/75/93.