The Washington giant's Redmond Corporation is competing with Google, IBM, and a holding company of small companies specializing in developing quantum computers - machines that are theoretically much more powerful than current computers by bending the laws of physics.
Microsoft says it has a different approach that will make the technology less error prone and more suitable for commercial use. If it works. On Monday, the company unveiled a new programming language called Q # - pronounced Q Sharp - and tools that help programmers craft software for quantum computers.
Microsoft is also releasing emulators that will allow programmers to test this program on a traditional desktop computer or through the Azure cloud computing service.
Machinery is one of the advanced technologies, along with artificial intelligence and augmented reality, that Microsoft CEO Satya Nadella considers crucial to his company's future. Microsoft, like IBM and Google, will most likely rent computing time on these quantum machines through the Internet as a service.
D-Wave Systems Inc. In 2011 the first company to sell a quantum computer, although the technology was controversial and could only perform a certain subset of mathematical problems. Google and Heb have produced machines that are believed to be close to achieving "quantum supremacy" - the ability to tackle a very complex problem to solve on any standard giant. IBM and startup regity computing also software for its machines.
Microsoft, in contrast, is still trying to build a working machine. It seeks to design a novel based on the control of an elusive particle called fermion majorana that no one was sure until it existed a few years ago. Todd Holmdahl, head of quantum computing efforts at Microsoft, said engineers are close to being able to control fermion majorana in a way that enables them to perform calculations. Holmdahl, who led the development of the company's Xbox and HoloLens glasses, said Microsoft has a quantum computer on the market within five years.
"We are talking to multiple clients today and we are proposing quantum-inspired services for some problems," he added.
These systems push the boundaries of how atoms and other small particles work. Whereas traditional computers process bits of information as 1s or zeros, quantum machines rely on "qubits" that can be 1 and zero at the same time. Even two bits can represent four numbers simultaneously, three bits can represent eight numbers, and so on. This means that quantum computers can perform mathematical operations much faster than standard machines and address problems that are way more complex.
Applications can include things like creating new drugs and new materials or solving complex chemistry problems. Krista Sfor, who oversees the software aspects of Microsoft’s business, says the "killer application" of quantum computing is discovering a more efficient way to collect ammonia for fertilizers, a process that currently consumes 3 percent of the world's natural gas.
The technology is still emerging from a long research stage, and its potential is widely discussed. Only researchers have been able to keep the bits in a quantum state for fractions of a second. And when kbites fall from the quantum state, they produce errors in their calculations, which could negate any benefit from using a quantum computer.
Microsoft says it uses a different design - called a topological quantum computer - that in theory will create more stable copets. This could produce a machine with an error rate of 1,000 to 10,000 times better than the computers other companies are building, Holmdahl said.
Jonathan Breeze, a research fellow working on advanced materials at Imperial College London, said minimizing or correcting errors in quantum computations is essential for the technology to realize its commercial potential. A lower error rate in Microsoft's design could mean that it could be more useful for with only a few qubits being processed - perhaps less than 100.
Savory said her team has already demonstrated mathematically that algorithms that use a quantitative approach can speed up machine learning applications considerably. Enabling it to run up to 4000 times faster. (Machine learning is the type of artificial intelligence behind recent developments in the ability of computers to identify objects in images and translate languages and driving cars.)
Comments
Post a Comment