In 2012, sitting in a hot pool, physicist Seth Lloyd proposed a quantum internet application to Google's creators, Sergey Brin and Larry Page. He called it Quoogle: a search engine that uses mathematics based on the physics of subatomic particles and shows results without knowing the queries themselves. Such a leap would require a completely new type of memory - the so-called QAMM, or quantum random access memory.
Although the idea intrigued Brin and Page, they abandoned it, Lloyd told "Gizmodo". According to him, they reminded him that their business model is based on knowing everything about everyone.
But KOSU, as an idea, did not die. Modern computers well remember information in billions of bits, binary digits equal to either zero or one. RAM, or random access memory, stores information for a short time on silicon chips, assigning each piece of information to a specific address, which can be accessed randomly and in any order to refer to this information later. This makes the computer much faster, allowing your laptop or mobile phone to immediately get to the data stored in RAM, often used by applications, instead of searching for it in storage, which is much slower. But sometime in the future, computer processors may be supplanted or augmented by quantum computer processors, machines capable of embedding giant databases.machine learning and artificial intelligence. Quantum computers are still a nascent technology, but if they are ever able to execute these potentially lucrative algorithms, they will need a whole new way of accessing RAM. They will need a BODY.
“KRAM can be a terrific application that makes quantum devices from Google and IBM instantly useful,” Lloyd told Gizmodo.
Classic computers like the ThinkPad, Iphone, and the most powerful supercomputers do all of their operations by translating data into one or many combinations of bits, zeros, and ones. Bits interact with each other, ultimately producing another combination of zeros and ones. Quantum computers also produce the end result in the form of ones and zeros. But as the counting proceeds, their quantum bits, or qubits, communicate with each other in a new way, through the same laws of physics that govern electrons. Rather than just being zero or one, each qubit can be both when counting, using a mathematical equation that encrypts the probability of getting zero or one only when you test its value. Several qubits use more complex equations,which refer to qubit values as single mathematical objects. The result is one or more possible binary strings, the final value of which is determined by the probabilities in the equations.
This weird mathematical approach - qubits are equations until you calculate them, and then they look like bits again, but their values can include an element of randomness - allows you to solve problems traditionally difficult for computers. One such challenge is the decomposition of large numbers into prime numbers, which breaks the algorithms used to store large amounts of encrypted data - a development that can be “catastrophic” for cybersecurity. It can also serve as a new way to process large datasets, such as those used in machine learning (such as advanced face recognition systems).
Quantum computers are still no better than conventional computers. IBM gives scientists and entrepreneurs access to a working 20-qubit processor, and Rigetti to a 19-qubit processor, while traditional supercomputers can simulate quantum powers up to 50 qubits. Despite this, physicist John Preskil recently announced that technology is entering a new era in which quantum computers will soon be useful for more than entertaining physics experiments. The US government takes quantum technologies seriously because of their importance to cybersecurity, and many physicists and programmers are looking for new niches for them.
Many researchers also hope to find applications for quantum computers in the development of artificial intelligence and machine learning using quantum algorithms. Such algorithms are complex and involve a significant amount of information, thus requiring a quantum alternative to RAM: qRAM.
Promotional video:
Quantum RAM is not billions of bits stored in multiple qubits. Instead, it's a way for quantum computers to apply their quantum operations to large lists of data found in machine learning problems. Ultimately, regular random access memory is made up of data needed to run programs, and programs access it by specifying the address of the bits - in the same way you can get the sum of cells by typing (A2 + B2) instead of typing numbers each time. manually. Quantum algorithms will have to access ordinary random access memory at the quantum level - in the most primitive sense, they create a superposition in which the cell is both A2 and B2 at the same time, and only then, after the calculation is completed, shows the value of either A2 or B2. There is nothing quantum about memory as such - quantum is the way you access and use it.
Basically, if you have a lot of stored data - as, for example, in databases for training chatbots - then there may be a quantum algorithm that can do more than a regular computer when it comes to searching through data or a message of something important. … This can be very lucrative for both the financial industry and companies like Google, and of course it will require quantum RAM.
An article on QRAM, written by Lloyd and his team ten years ago, described one way of accessing only those addresses in memory needed for superposition, using something they called a "quantum fire chain." Basically, since each address in RAM is just a sequence of bits, it can be thought of as a branching tree, in which each qubit is a pointer telling the computer to turn left or right. This works in conventional computers as well, but a quantum computer with only two choices will inevitably entangle extra paths at every turn, ultimately leading to an incredibly large and fragile quantum state that can easily disintegrate in a non-quantum environment. Lloyd and his colleagues proposed a tree structure,in which each branch is automatically held in standby mode, allowing the computer to move only on the right or left branch (side) to access the desired memory without entangling unnecessary information. The difference is quite technical in nature, but it is designed to significantly reduce the power required to solve this kind of problem in machine learning.
"Most of the algorithms used in research require some kind of quantum memory," commented Michelle Mosca, a scientist at the University of Waterloo in Canada who also researched quantum memory, for Gizmodo. "Anything that reduces the cost of applied quantum RAM can also dramatically reduce the time before the advent of everyday quantum computers."
But we're still at a very, very early stage in the development of quantum programming. Today, the way old computers remember information seems almost ludicrous. RAM consisted of magnetic loops connected by wires, where each loop corresponded to one bit, and the orientation of the magnetic field in the coil represented its meaning. The first commercially available American computer, UNIVAC-I, was known for storing data by converting electrical impulses into sound waves using liquid mercury. That memory had no random access - you couldn't get any data you wanted at any time, but only in the order in which it was stored. And it was considered cutting edge technology.
“It was a work of art,” explained Chris Garcia, curator of the Computer History Museum. "At that time they tried everything they could and hoped that some of it would work." At that time, such solutions were superior to all previous ones. Today computers store memory on microchips made of a special material called "semiconductors", which became possible not only because of the advancement of science, but also thanks to the processes that made silicon storage much cheaper than storage from tiny magnetic coils.
What will quantum memory look like? Most likely not in the way Lloyd and colleagues imagined it. At last year's conference, physicists joked that the field of quantum computing may well turn to another analogue of vats of liquid mercury. Surely we will have new technological and mathematical advances that will optimize computers and their methods of storing information.
Lloyd agreed with this. “I would love to see someone spread our idea,” he said. "If we could translate ordinary information into a quantum state, this would be an amazing application of quantum computers in the short term." After all, computers are about more than just their ability to execute fancy algorithms. They enable these algorithms to be used to process and organize data to create something useful.
And maybe someday we will really use quantum Google.
Ryan F. Mandelbaum