To solve the problems of small molecules could take lifetimes to Crack
#Cosmoread: Molecules that help muscles contract a day may help drive a new kind of molecular supercomputer, the researchers said.
The biological computer quickly or more complex problems that traditional supercomputer births will have to be solved to crack, the scientists added.
Modern supercomputers are staggeringly powerful. The world’s fastest supercomputer, Tianhe-2 in China, about 55 quadrillion calculations per second, which is a desktop computer or video game console is up to many thousands of times more capable of carrying out.
However, traditional supercomputers typically operating sequence, one at a time to perform. In contrast, the brain simultaneously, or can perform multiple operations in parallel. The human brain chemical also powers other molecular forms, an energy-efficient process that generates far less heat than the silicon chip in the molecule adenosine triphosphate, or ATP, turning these cellular processes.
These factors partly why the brains of some of the problems of traditional supercomputers much faster, while consuming less power can remedy is understandable. For example, the human brain is only about 20 watts of power, which is barely enough to run a dim light bulb, the power of Tianhe-2 is about 17.8 MW, which is about 900,000 such light consumption is enough to run the bulb consumes.
Now researchers have suggested that a new computer in parallel computing power of ATP, to some extent, that is what the human brain can help out.
“The problem that can solve electronic computers are very well. We have just not good at solving problems that are electronic computers are aiming to solve,” said study senior author Dan Nicolau Sr., a chemical engineer at McGill University in Montreal.
A decade ago, his son Nicolau, lead author of the study with Dan Nicolau Junior more than working on the idea for this device, of California, Berkeley, the University began. “It looked like mazes with pictures of minor bugs discovered, a back-of-the-envelope idea started as much rum I think that after the” big Nicolau said in a statement.
A class rum fueled those scribblings eventually, turn into glass coated silicon chip about 0.6 inches (1.5 centimeters) wide, two researchers at the microscopic channels etched, each less than 250 nanometers wide. (It is thinner than the wavelength of visible light.) Chip, miniscule with its network of channels, a city street grid looks a bit like a mini version of.
Researchers channels swimming around inside the fibers of protein referred to is much like moving cars to drive on city streets. These “agents,” as scientists call them, actin filaments and microtubules, proteins that make up the internal structure of cells included. Such agents myosin, which helps muscles contract, and kinesin, which transport cargo around inside cells helps as were driven by molecular motors. The researchers used these molecular motors to power the ATP, and the fluorescent labeled agents to track them visually.
Agents in a corner of the device to enter and leave the many different exits. They randomly several junctions inside the chip can be redirected under a variety of channels. Channels scientists solved a problem of the device layout corresponds to want, and choose the exit agents representing potential answers.
Scientists known as NP-complete problems on a class of problems to test your new device. In this type of puzzle, one quickly to confirm that any solution or may not be able to work, but could not find the best solution to the problem quickly.
Puzzle is a classic example of an NP-complete “traveling salesman problem,” which is a list of some of the cities and a city that is visited every other city exactly once and returns to the starting position at must find the shortest possible route. An early exit is a way for all of the cities and does not go into any city more than once, confirming that this is the shortest route, trying every combination that included you may be able to find. This brute-force strategy significantly increases the number of cities grows as complex.
Thus solving the problem of shipping of the goods and can improve the routing of packets of data, the researchers said.
Researchers traveling salesman problem, wants to use their tools to attack, moving within these networks so they would send countless molecules, “running amok in the city, the city is like traveling salesmen to send millions, and which paths look the most promising, “Nicolau said.
The researchers’ latest experiments, the subset sum problem is NP-complete version of their new device tested. Such as one half of the whole number, such as 1 and negative 1, but no degree – – the problem, one is given a set of integers and if there is a subset of those integers whose sum is zero must find.
2, 5 and 9 – – In experiments with a set of three integers researchers showed his device had almost the right answer all the time. Device than the electronic computer, the researchers in the journal Proceedings of the National Academy of Sciences reported in a study published online February 22 will count is approximately 10,000 times less energy consumption.
Find an answer to the simple problem may seem trivial, but the new device is a proof-of-concept chip that can solve the trickier problems as for more complex versions of works, the researchers said. For example, the subset sum problem faster and are more difficult to analyze there are more integers. “The best possible out laptops now include a subset of the first 30 prime numbers have failed to address is,” Nicolau said.
Previous research has suggested that “by an NP-complete problem, one can solve them all,” Nicolau said. “Certainly, if our work can solve the traveling salesman problem, it can be very practical application.”
However, other methods such as quantum computing, it also simultaneously carry out many calculations, the components used in quantum computers more easily disrupted by molecular machines are used in the new study, the researchers said.
A potential limitation of this approach is how the agents currently all devices are fed into a corner of each chip, the researchers said.
“The more agents, more time to feed them and takes the count out,” Nicolau said. “Ways we can solve that problem, a number of devices such as the sharing of each device in each part of the problem is that there are a number of solutions.”