Physics


Stanford Linear Accelerator Center
High-energy physics researchers and administrators meet over NeXT technology

Reproduced with permission from NeXT Computer, Inc.
A Reference Guide to NeXT in Higher Education, Fall 1992
ยช 1992 NeXT Computer, Inc


At Stanford Linear Accelerator Center (SLAC), physicists and administrators alike use NeXT technology in all aspects of their work, from developing computing tools for data analysis to organizing an annual two-week summer institute.

The primary interest of Staff Physicist Paul Kunz, a NeXT computer user since 1989, is developing computer applications so scientists can visualize their manipulation of data in a highly interactive manner. For example, with Reason and HippoDraw-applications developed by Kunz and his colleagues-scientists can create analysis chains with a mouse rather than by writing computer code. Scientists select an input (for example, a file containing data about physics events) and set up the necessary analysis, which could take minutes or hours depending on the size of the data sample. The output of the analysis is a large table of numbers, displayed as histograms, scatter plots, or graphs. Because these plots are produced almost instantly, users see the results of their work immediately.

"We chose the NeXT platform because of the NeXTSTEP development environment, particularly Interface Builder, which enables us to build applications quickly," says Kunz. "Scientific workstations today have a minimum of 10 MIPS, 8 MB memory, and a 1-megapixel display. This kind of power is relatively new in our environment. However, the so-called interactive interfaces we usually see are really accelerated batch interfaces; we type something, then wait for results. NeXTSTEP's graphical interface and development environment let us create truly interactive tools. As we move a slider to change a variable, we see our plots change immediately."

Using Reason and HippoDraw, Kunz and his colleagues have already analyzed many data sets containing thousands of physics events. "It used to be the case that working with a computer involved a trade-off," Kunz says. "We'd have an idea of what we wanted to do, but completing the program for it would often take weeks or months. The time it takes to develop a program with NeXTSTEP is faster by a factor of at least five. When a program is this much faster to develop, researchers can drive to work in the morning, thinking, 'What new thing should I create today?' With the previous generation of platforms, they drove to work thinking, 'I wish I could finish today what I started last week.' That's a big difference in approach and productivity."

In designing Reason, Kunz was careful to separate general data analysis functions from functions specific to physics. The application provides a palette of objects that perform operations such as input, output, and looping. HippoDraw, meanwhile, handles the plotting of histograms. Physics-specific objects are organized on a user palette. Users can also write their own objects to perform any number of functions and include these on the user palette. As a result, the application can be customized to perform tasks applicable not only to physics but to any field requiring statistical analysis.

Kunz says SLAC physicists have also developed a NeXT computer application that is a front-end interface to the Minuit kernel, the standard minimization program. Given a set of points that should match a known function-for example, a Gaussian distribution-the program plots the curve that best describes the function and calculates the function's parameters.

There are three advantages to the NeXT version of the Minuit program, according to Kunz. "First, it's really fast to do a fit. Second, it uses an existing program that many scientists regard as a trusted friend. Third, the program shows the curve that would result when taking into account the error on any parameter. For example, if the error is + 1 percent, the user can see the resulting curve simply moving the slider on a scale from -1 to +1."

Staff Physicist Bill Atwood is using the NeXT machine to model particle physics experiments. "Particle physics research used to be at the cutting edge of computer technology," he says. "Today, however, we're years behind other sciences: most of our code for particle research is written in FORTRAN 77. While on-line data acquisition techniques and end-user data analysis tools have improved, the core computing environment is outdated. One of my projects is to take the core FORTRAN procedural language codes used to model particle physics experiments and make them current in today's technology. For us, that means using NeXT."

During a particle physics experiment, particles are shot through accelerators to collide with one other or with particles in a target. The collision occurs in a detector that is constructed by layering tubes of different size, material, and physical properties. The particles that result from the collision can be recognized by their trajectories and by the layer of the detector in which the ionization trails are recorded.

Gismo, a program which was co-author by Atwood and other scientists from the United States and Europe, lets physicists model a detector that specifies the radii of its layers, materials, and magnetization, and then simulates particle collisions. The application registers as many as 20,000 points from a typical collision, then uses pattern recognition algorithms to plot straight and curved lines showing the trajectory of each particle. The resulting graph is a dramatic display with different colors representing the various particles and their energies.

"Until recently," says Atwood, "detectors were designed by people who guessed on the optimal detector design based on previous experience. Gismo lets us tell ahead of time what benefits we'll derive from a particular design."

At SLAC, the administrative staff also uses NeXT computers. According to Steve Williams, assistant director of the research division, the display on his NeXT computer on a typical day might contain windows for a terminal session with the mainframe, a Lotus Improv spreadsheet containing lists of people in the SLAC user organization, a FrameMaker document for creating form letters for job offers, an organizational chart created with Diagram!, a to-do list created in WriteNow, and a calendar program.

SLAC Administrator Jane Hawthorne uses a NeXT machine to coordinate annual summer institute. She uses FrameMaker to manage a database of summer institute participants, correspond with students in the postdoctoral research program; and maintain a list of caterers, entertainers, and other vendors she uses in coordinating the institute. Says Hawthorne, "The NeXT gives me the ability to do three or four things or more at once much more effectively than other computers I've used."

For more information, please contact:

Paul Kunz
Staff Physicist
Stanford Linear Accelerator Center
Stanford University
Stanford, CA 94305
(415) 926-2884
pfkeb@kaon.slac.stanford.edu