In recent years, some big tech companies like IBM, Microsoft, Intel, or Google have been working in relative silence on something that sounds great: quantum computing. The main problem with this is that it is difficult to know what exactly it is and what it can be useful for.
There are some questions that can be easily solved. For example, quantum computing is not going to help you have more FPS on your graphics card at the moment. Nor will it be as easy as changing the CPU of your computer for a quantum to make it ‘hyperfast’. Quantum computing is fundamentally different from the computing we are used to, but how?
What is the origin of quantum computing?
At the beginning of the 20th century, Planck and Einstein proposed that light is not a continuous wave (like the waves in a pond) but that it is divided into small packages or quanta. This apparently simple idea served to solve a problem called the “ultraviolet catastrophe”. But over the years other physicists developed it and came to surprising conclusions about the matter, of which we will be interested in two: the superposition of states and entanglement.
To understand why we are interested, let’s take a short break and think about how a classic computer works. The basic unit of information is the bit, which can have two possible states (1 or 0) and with which we can perform various logical operations (AND, NOT, OR). Putting together n bits we can represent numbers and operate on those numbers, but with limitations: we can only represent up to 2 different states, and if we want to change x bits we have to perform at least x operations on them: there is no way to magically change them without touching them.
Well, superposition and entanglement allow us to reduce these limitations: with superposition, we can store many more than just 2 ^ n states with n quantum bits (qubits), and entanglement maintains certain relations between qubits in such a way that the operations in one qubit they forcefully affect the rest.
Overlapping, while looking like a blessing at first glance, is also a problem. As Alexander Holevo showed in 1973, even though we have many more states than we can save in n qubits, in practice we can only read 2 ^ n different ones. As we saw in an article in Genbeta about the foundations of quantum computing: a qubit is not only worth 1 or 0 as a normal bit, but it can be 1 in 80% and 0 in 20%. The problem is that when we read it we can only obtain either 1 or 0, and the probabilities that each value had of leaving are lost because when we measured it we modified it.
This discrepancy between the information kept by the qubits and what we can read led Benioff and Feynman to demonstrate that a classical computer would not be able to simulate a quantum system without a disproportionate amount of resources, and to propose models for a quantum computer that did. was able to do that simulation.
Those quantum computers would probably be nothing more than a scientific curiosity without the second concept, entanglement, which allows two quite relevant algorithms to be developed: quantum tempering in 1989 and Shor’s algorithm in 1994. The first allows finding minimum values of functions, which So said, it does not sound very interesting but it has applications in artificial intelligence and machine learning, as we discussed in another article. For example, if we manage to code the error rate of a neural network as a function to which we can apply quantum quenching, that minimum value will tell us how to configure the neural network to be as efficient as possible.
The second algorithm, the Shor algorithm, helps us to decompose a number into its prime factors much more efficiently than we can achieve on a normal computer. So said, again, it doesn’t sound at all interesting. But if I tell you that RSA, one of the most used algorithms to protect and encrypt data on the Internet, is based on the fact that factoring numbers are exponentially slow (adding a bit to the key implies doubling the time it takes to do an attack by force) then the thing changes. A quantum computer with enough qubits would render many encryption systems completely obsolete.
What has been achieved with quantum computing so far?
Until now, quantum computing is a field that hasn’t been applied much in the real world. To give us an idea, with the twenty qubits of the commercial quantum computer announced by IBM, we could apply Shor’s factorization algorithm only to numbers less than 1048576, which as you can imagine is not very impressive.
Still, the field has a promising evolution. In 1998 the first ord quantum drive (only two qubits, and needed a nuclear magnetic resonance machine to solve a “toy” problem (the so-called Deutsch-Jozsa problem). In 2001 Shor’s algorithm was run for the first time. Only 6 years later, in 2007, D-Wave presented its first computer capable of executing quantum quenching with 16 qubits. This year, the same company announced a 2000 qubit quantum quenching computer. On the other hand, the new IBM computers, although with fewer qubits, they are able to implement generic algorithms and not only that of quantum quenching. In short, it seems that the push is strong and that quantum computing will be increasingly applicable to real problems.
What can those applications be? As we mentioned before, the quantum tempering algorithm is very appropriate for machine learning problems, which makes the computers that implement it extremely useful, although the only thing they can do is run that single algorithm. If systems can be developed that, for example, are capable of transcribing conversations or identifying objects in images and can be “translated” to train them in quantum computers, the results could be orders of magnitude better than those that already exist. The same algorithm could also be used to find solutions to problems in medicine or chemistry, such as finding the optimal treatment methods for a patient or studying the possible structures of complex molecules.
Generic quantum computers, which have fewer qubits right now, could run more algorithms. For example, they could be used to break much of the crypto used right now as we discussed earlier (which explains why the NSA wanted to have a quantum computer). They would also serve as super-fast search engines if Grover’s search algorithm can be implemented, and for physics and chemistry, they can be very useful as efficient simulators of quantum systems.
The barriers that still have to be overcome
Unfortunately, algorithms and codes for classic computers couldn’t be used on quantum computers and magically get an improvement in speed: you need to develop a quantum algorithm (not a trivial thing) and implement it in order to get that improvement. That, at first, greatly restricts the applications of quantum computers and will be a problem to overcome when those systems are more developed.
However, the main problem facing quantum computing is building computers. Compared to a normal computer, a quantum computer is an extremely complex machine: they operate at a temperature close to absolute zero (-273 ºC), the qubits support are superconducting and the components to be able to read and manipulate the qubits are not simple either.
What can a non-quantum quantum computer be like? As we have explained before, the two relevant concepts of a quantum computer are superposition and entanglement, and without them, there cannot be the speed improvements that quantum algorithms promise. If computer disturbances modify overlapping qubits and bring them to classical states quickly, or if they break the interweaving between several qubits, what we have is not a quantum computer but only an extremely expensive computer that only serves to run a handful of algorithms. equivalent to a normal computer (and will probably give erroneous results).
Of the two properties, entanglement is the most difficult to maintain and prove to exist. The more qubits there are, the easier it is for one of them to deinterlace (which explains why increasing the number of qubits is not a trivial task). And it is not enough to build the computer and see that correct results come out to say that there are intertwined qubits: looking for evidence of entanglement is a task in itself and in fact, the lack of evidence was one of the main criticisms of D-systems. Wave in its beginnings.
What can we expect in the long run?
A priori and with the materials that quantum computers are being built with, it does not seem that miniaturization is too feasible. But there is already research on new materials that could be used to create more accessible quantum computers. Who knows if fifty years from now we will be able to buy “quantum CPUs” to improve the speed of our computers.
Join Our Telegram Group. Be part of an engaging online community. Join Here.
Subscribe to our NewsletterGet the latest updates and relevant offers by sharing your email.
Dean at Woxsen School of Business. He is a registered expert in Artificial intelligence, Intelligent Systems, Multi-agent Systems at the European Commission, and has been nominated for the Forbes 30 Under 30 Europe 2020 list.