Profile - People behind the LHC grid: Wesley Smith
iSGTW: What do you do?
Wesley Smith: I work at the Compact Muon Solenoid (CMS), where my job is to throw away 99.999 percent of the data we record.
What we’re hoping to do is like looking for a needle in a haystack, only with multiple haystacks. I want to remove as much hay as possible, without losing any needles.
iSGTW: How did you become interested in physics?
Wesley Smith: I grew up in San Francisco, where we lived one block away from Mel Schwartz of Stanford University, who won the Nobel Prize in Particle Physics. I knew his son, we went to elementary school together, and our parents spent a lot of time together. It was a connection that stayed almost 40 years, and a powerful influence.
Then when I was in college at Harvard, I studied under Carlo Rubbia (former Director General of CERN,1989-1994, and Nobel prize-winner in physics) and Larry Sulak.
My first summer job, as a freshman in college, was working at Fermilab’s Neutrino Experiment E1A, where I sat in a trailer in a cornfield on hot summer days and looked for neutral-current events in spark chamber film. Basically, we were looking for situations in which a neutrino comes in, a shower of particles is produced, and no muon comes out. And contrary to prediction, we were finding lots and lots of these events—research which hinted at the existence of the Z particle. And then, of course, the bubble chamber neutrino experiment at CERN published this discovery first. It was an extraordinary summer. That was it; I was hooked on particle physics.
iSGTW: When did you first start on the Large Hadron Collider?
Wesley Smith: I worked on the Superconducting Supercollider in Texas, and then shortly after Congress cancelled that in late 1993, I came to CERN at the invitation of my future CMS colleagues to work on the trigger. At the same time that I was working on the trigger design for CMS, I was running a working trigger system at the ZEUS experiment for the HERA collider in Germany—a trigger system that had to operate at the highest speed of any particle physics collider detector until the LHC. So, I could immediately take what I knew from the field in working with ZEUS, and apply it to the design of the CMS trigger at CERN.
iSGTW: What is a trigger?
Wesley Smith: It inspects detector information and decides whether to keep the event or discard it. There is such a large volume, you just cannot keep it all. But you don’t want to throw away something that proves interesting later.
Essentially, if we’re looking for things like the Higgs Boson, we’re talking about very rare events that occur maybe 100 or 1,000 times a year. In order to be sure to catch that, we have to capture a lot of collisions with our detector, which is essentially a very high-speed camera. But there’s so many collisions happening—about 1 billion per second at full luminosity—that there is no way we can save all that data. The flood never stops.
So, we have to decide what to keep, what to discard, what we want to look at right away, what we want to look at later, and what is a false alarm.
After processing in dedicated hardware for 3 microseconds, we forward about 100,000 beam crossings of data to a farm of CPUs, where each CPU has 40 milliseconds to detect and analyze 1 megabyte of data.
On the grid
iSGTW: If you had to describe your project in two or three sentences, you would say . . .
Wesley Smith: We probe deeply into the nature of matter, to find why particles have the masses they do, and to understand more about our universe.
For example, astrophysicists have the embarrassing situation of not being able to account for 90 percent of the matter of the universe, for example. This research will help us solve that.
iSGTW: What is the role of grid computing in all this?
Wesley Smith: At first, it was taking us years to do the simulations and accumulate the statistics. Then, in June 2001, Miron Livny—principal investigator of Open Science Grid—introduced us to Condor; in one month with grid-computing, we could do what had taken us a year.
I was sold.
I’m not interested in something new in computing unless it helps me do my science, and this clearly did that.
Computing is not a vending-machine model, where you pay money and get results. It’s a collaboration. It depends on software, middleware, hardware, resources, people and organization.
iSGTW: Where do you see the grid going?
Wesley Smith: At the University of Wisconsin—where I work, when I am not on sabbatical as I am now —we have an organization called the Grid Lab of Wisconsin, or GLOW. We brought together scientists from all kinds of disciplines, including chemical engineering, forest science, genetics, astrophysics, and other fields, and looked at problems we had in common. Much of it turned out to be computing power. So our needs may have looked disparate at first, but in many cases, our problems were the same—with the same solutions. And with the same software, middleware and hardware, we found that we could all do more efficient science . . . the grid changed the way we do our science.
iSGTW: After 15 years of anticipation, what’s going through your mind right now?
Wesley Smith: I’m so eager to see this really, truly beautiful machine operate.
—Dan Drollette, iSGTW