United, they compute -- and quickly, too

Few insurance company employees can say they help unravel the secrets of the universe in their spare time.

Jeff Renkar can. Day and night, Renkar’s six personal computers at his home outside Houston are helping run complex simulations of how the early universe evolved, as part of a new University of Illinois project called cosmologyhome.

In the eight years since the California-based SETI Institute thought of embedding software in people’s screen savers that would help sift through radio noise from space for possible signals from alien civilizations, some 40 research groups have launched projects based on the same principle. Hundreds of thousands of computers have been enlisted to study how proteins fold, to search for new prime numbers and to simulate climate change, among other efforts.

The Internet-based strategy is changing how scientists think about their largest computing projects. Tapping the unused processing capacity of a network of individual PCs offers the power of an expensive supercomputer at a fraction of the cost, and allows researchers to cut the lag time for some calculations from years to days.

Such “volunteer computing” studies have also spawned a quirky global subculture of home enthusiasts who are bringing their personal passion and the collaborative power of the Internet to the most refined domains of science. Like Renkar, they enjoy pushing their computers to the limit while donating something to basic science or the hunt for disease cures.

Many volunteers are computer programmers or scientists, but others have no link to the projects aside from their own enthusiasm. Their homes are scattered around the globe, from Sweden to subSaharan Africa, from East Timor to East L.A. Those helping the Illinois project include a train engineer in Britain, a retired correctional worker from Oklahoma with an interest in astronomy, and a Star Trek fan club “commander” in Iowa.

For Renkar, a Chicago native who built his own telescopes as a boy, contributing to the study of the universe’s origins fulfills a childhood dream. What started as a casual hobby has become something of a mission.

“I’ve always been kind of a wannabe scientist,” Renkar said. “Now I feel I’m really helping science produce new results and discoveries.”

To join the projects, computer owners download a program that lets their PCs work on small pieces of a larger research problem. When a PC finishes a chunk of the calculation, it sends it off to be double-checked and plugged into the big data set.

One initiative has wooed gamers who use the Sony Playstation 3 console, which contains a speedy graphics processor. Stanford University’s protein study called foldinghome gets the majority of its processing work from the consoles.

“I donate to other causes, and I just considered this a donation to society and science,” said Cory Parker, a product-assurance technician from Inverness, Ill., who runs foldinghome software on his four Sony game consoles. He said he uses only one Playstation for gaming; the other three are damaged units that he bought at a discount just to further the study.

The sheer number of people needed to make such projects worthwhile can be daunting, researchers said. Foldinghome, which began in 2000 and is one of the largest such undertakings, uses about 250,000 computers at any given moment.

Maintaining all those volunteers’ excitement about the often arcane questions that the projects are supposed to address is a major challenge, said University of Illinois physics professor Benjamin Wandelt, director of cosmologyhome.

“This is democratic computing, so it’s based on the goodwill of a bunch of people from all walks of life, all backgrounds,” Wandelt said. “If as a researcher you cannot communicate what’s interesting about your problem to the general public, this sort of thing probably isn’t for you.”

Wandelt wanted to test how minor changes to initial conditions after the Big Bang at the start of the universe could have affected how the cosmos would look billions of years later. Knowing how that process works could tell physicists more about what our own universe must have been like at the beginning.

To address that question, Wandelt’s team runs thousands of computer simulations of cosmological evolution. It also tinkers with the initial conditions, such as the speed of the universe’s expansion or the quantities of certain fundamental particles.

But he realized that the initial set of simulations would take about 300,000 hours of computer time -- about 30 years, using an ordinary home computer. By distributing the computing chore among 2,800 users from 78 countries, Wandelt’s team amassed more than 300,000 hours of work in just six months.

“I thought people would be enthusiastic, but the response swept us off our feet,” Wandelt said. He said 100 people registered on his project’s website before he had even announced it to the public.

Kathryn Marks, an American ESL teacher living in South Korea, said she decided to help the study because of her “intrinsic thirst for knowledge.

“I’ve always looked up at the stars and wondered where we came from and where we’re going,” Marks said.

For some volunteers, the online projects can become a slightly manic obsession.

“It’s a hobby. An addiction even,” wrote Guy Pauwels, a foldinghome devotee in Belgium, via e-mail. He said he belongs to a team of “folders,” as they call themselves; they organize competitions with teams in other countries to see whose computers can complete the most work.

Programs such as foldinghome provide among the most demanding tests of a PC’s performance, many users say. Amassing an impressive processing record is a mark of achievement, and it earns users “points” -- statistics that are worthless except as tokens of a user’s computing prowess.

To rack up more points, many volunteers buy computers solely to use on the projects. Dr. Michael D. McCord, a retired anesthesiologist from Beaumont, Texas, said he had seven PCs running the foldinghome software full time -- and many users devote even more computers to the cause.

“The competition and camaraderie got me interested,” McCord said. “I’m blessed that I can afford the electricity bill.”

That level of dedication came as a surprise to David P. Anderson, a research scientist at the University of California who wrote the original setihome software and developed the framework that many newer projects use. He said he’d like to make volunteer computing so easy for scientists to use that even an ordinary graduate student could rely on it to crack lengthy calculations.

“Our goal is to make volunteer computing into kind of a new paradigm,” Anderson said. “It could be a standard tool that all scientists who need a lot of computing power can turn to.”

Vijay S. Pande, director of foldinghome, said his team has published 54 studies based on its protein-folding simulations. Proteins must fold into precise shapes to function properly, and many diseases can be traced to errors in folding, including Alzheimer’s disease and ALS, also known as Lou Gehrig’s disease. Pande’s work is designed to offer insights into that process and to aid the design of drugs that could help crucial proteins fold correctly.

But not every scientific question lends itself to solution by thousands of computers, each working on one tiny piece of the problem, Pande said.

“A pregnancy takes nine months, but you can’t get nine women together and get it done in a month,” he said. “Some problems can’t be divided.”