In early November, a team led from McGill University in Montreal announced a computer whose processors are living bacteria, the product of two decades of work spanning several countries and disciplines. Rather than electrons racing through silicon, microbes drift through a small etched structure of branching tunnels, each fork a choice and each route a possible answer. It is a quiet reminder that computation has always been a physical act, and that nature may still offer ways of thinking that our machines have not yet learned. Whether such living engines can ever rival the speed of conventiona
Researchers Use Living Bacteria as a Computer, Solving Subset Sum Problem Through Tunnel Networks
Each fork in the tunnels is a decision point; each route, a possible solution.
Mark
So what actually happened here? Bacteria solved a math problem?
Mimi
Roughly, yes. The team turned the Subset Sum Problem into a branching tunnel network, released bacteria into it, and tracked which routes they took with image analysis.
Mark
Why bacteria instead of just a faster chip?
Mimi
Parallelism. They explore many routes at once, and as they reproduce they add more processing units. A regular computer works through options sequentially.
Luke
Before we get carried away: the bacteria are slow. The reporting says so plainly, and nothing here shows this beating a conventional machine.
Mark
Fair. Why two species?
Mimi
E. coli moves through fluid predictably and measurably. V. natriegens reproduces very quickly. Each covers a different need.
Mark
Nicolau says it needs no RAM. Does that mean it could ease the RAM shortage?
Luke
That's his suggestion, and it's a forward-looking claim. The article doesn't give evidence that it would, or at what scale.
Mimi
And he admits the Doom caveat: it would have to be huge.
Mark
What about his definition of a computer?
Luke
That's his framing. It's a provocative one, but it's a view, not a finding.
Der Puls
- Hard problems like the Subset Sum Problem can grow so tangled that conventional computers would need centuries to untangle them.
- Researchers from McGill and partners across Canada, Europe, the UK and Australia have turned living bacteria into a branching problem-solver.
- E. coli and Vibrio natriegens race through etched tunnels, exploring many possible answers at once rather than one after another.
- Individual bacteria are sluggish, so the system would have to grow very large to match digital machines, and its real speed is still unproven.
In early November, a team led from McGill University in Montreal announced a computer whose processors are living bacteria, the product of two decades of work spanning several countries and disciplines. Rather than electrons racing through silicon, microbes drift through a small etched structure of branching tunnels, each fork a choice and each route a possible answer. It is a quiet reminder that computation has always been a physical act, and that nature may still offer ways of thinking that our machines have not yet learned. Whether such living engines can ever rival the speed of conventional computing remains an open question, one the researchers themselves have not yet answered.
On November 3, researchers at McGill University in Montreal unveiled a computer whose processors are living bacteria. It is not a simulation or a metaphor, but a small physical structure etched with tunnels, where microbes travel toward the solution to a math problem. The idea has been developing for about twenty years, and the results appeared earlier this year in Nature.
The test case was the Subset Sum Problem, which grows exponentially harder as its inputs multiply. The team translated it into a branching graph and built a tunnel network that mirrors it, with each fork a decision point and each complete route a possible solution. Into this network they released two species: E. coli, which moves through fluid in a predictable way, and Vibrio natriegens, which reproduces quickly. Instead of testing routes one by one, the bacteria explore many at once, and automated imaging traces which tunnels satisfy the constraints. Temperature was adjusted to prevent overcrowding.
The trade-off is speed. Each bacterium is slow compared with digital processing, but reproduction creates more processing units when a problem demands more power. Dan Nicolau, the paper's senior author, argued that conventional sequential machines need vast amounts of RAM for such problems, while the bacterial system needs none, and he suggested this could ease the global RAM shortage.
He also addressed the inevitable question of whether it could run Doom, and said it probably could, though it would have to be enormous to compensate for slow processors. The approach remains experimental, and how large and how fast such machines could truly become is still to be shown.
On Nov. 3, researchers at McGill University in Montreal announced a computer whose processors are living bacteria. The machine is not a metaphor or a simulation. It is a small physical structure, etched with tunnels, in which microbes travel toward the answer to a math problem.
The work comes from an international team that has spent 20 years developing the idea. Scientists in Canada, Europe, the United Kingdom and Australia contributed, drawing on biology, computer science and engineering. The results appeared earlier this year in a paper in Nature. The researchers say the approach could handle problems that conventional computers find punishing.
The test case was the Subset Sum Problem, which becomes exponentially harder as the number of inputs grows. Once the inputs get large enough, traditional computers can take centuries, or even millions of years, to finish it. The team converted the problem into a graph that branches at every possible combination of inputs, then built a tiny structure that mirrors that graph. Its tunnels follow the branches. Each fork is a decision point, and each complete route through the network stands for one possible solution.
Into those tunnels the researchers released bacteria. They used two species for different reasons: Escherichia coli, because it moves through fluid in a predictable, measurable way, and Vibrio natriegens, because it reproduces very quickly. As the bacteria spread, they explored many routes at the same time instead of working through them one after another, as an electronic computer does. The team used automated image analysis to follow the bacteria down the correct tunnels, which showed which paths satisfied the problem's constraints. They also adjusted the temperature to keep the bacteria from overcrowding as they moved through the network.
The trade-off is speed. Compared with digital processing, individual bacteria are slow. But reproduction supplies a kind of built-in scaling: if the problem needs more computing power, growth inside the tunnels creates more processing units.
Dan Nicolau, a McGill bioengineering professor and the paper's senior author, explained the logic in an email to CNET. Electronic sequential computers, he said, are not meant or designed to solve decision trees. He noted that if they are to solve such problems, they need a great deal of RAM. The bacterial system, by his account, needs none, and he suggested biocomputing could help address the global RAM shortage.
Then came the inevitable question about Doom. Fans have run the 1993 shooter on everything from E. coli to a thermostat, and Nicolau said it could probably run on this system too, with a caveat about size. "(The biocomputer) must be huge!" he said, since slow processors mean you would need enough bacteria to compensate. He also offered a blunt definition of the field's ambitions: a computer, in his words, accepts numbers and outputs numbers, and every other use of the word is, from his point of view, poetical.
The approach remains experimental. How large such machines would have to be, and how fast they could really run, has yet to be shown.
Bemerkenswerte Zitate
Electronic (sequential) computers are not meant, or designed, to solve decision trees.— Dan Nicolau, McGill bioengineering professor and senior author
(The biocomputer) must be huge!— Dan Nicolau, on running Doom
Möchten Sie die ganze Geschichte?
Das Original lesen bei CNET ↗
Education