Quantum computing has the capacity to solve problems that today’s computers probably never can. However, since it works very differently from a regular computer, software will have to be rewritten and new ways of writing algorithms will have to be created. In a detailed conversation with BusinessLine , Arvind Krishna, Senior Vice-President, Hybrid Cloud, and Director of IBM Research talks about how quantum computing will deliver a quantum leap in problem solving. Excerpts:
When regular computers and supercomputers are getting faster each passing day, what’s the need for a quantum computer?
Quantum computer is not about making problems that you can solve on today’s computers go faster. There are a set of problems you can never solve on a regular computer but you can on a quantum computer. If you want to simulate a molecule of something like caffeine which has 160 electrons, it would require a computer that’s 1/10th the volume of our planet. But that can be done with a quantum computer the size of a small room.
Similarly not everything can go on a quantum computer because it is inherently probablistic. If its probablistic, you cannot do banking on that. But it will work for molecules because properties of elements are statistical in nature. As an aggregate they have a certain property but not everything inside is the same.
There are a set of problems that we believe are amenable to quantum computing. We can use them for modelling risk computation for a financial institution. Computing risk is something quantum computers are very good at.
What would be the impact of quantum computers on IT security as it will be able to break encryption codes within seconds?
Encryption as widely used today are based on certain cryptographic algorithms that depend on factoring large numbers into two primes. We depend on that because it is a hard problem to solve. Could quantum computer break that code? The answer is yes. But you don’t have to base encryption on factoring prime numbers. There are other known techniques. which appear to be quantum proof. So, if you would like that what you encrypt today to still remain encrypted 5-10 years from now, I suggest you look at those techniques.
How far are we from adopting quantum computing?
I look at it this way. We were at five qubits (computing unit in a quantum computer) in 2017, later in 2017 we put out 16 qubits. In December of 2017, we announced we had 50 qubits but not robust enough to put on cloud. Looking at that progression, at somewhere between 50 and 100 qubits, you’ll find quantum computers solve problems that normal computers are unable to solve. Within five years, we’ll find significant commercial uses of quantum computers.
What is the kind of interest you are seeing from India for quantum computing?
When we put the quantum computer on cloud, we thought a few thousand trials will happen from the subset and we’ll be done. But 5 million experiments have been done from nearly 100,000 users so far.
There’s a pretty good overlap from the Indian subcontinent. India was among the top five countries in terms of usage.
Is access cost the biggest hurdle to taking to quantum computing?
No, knowledge is a bigger issue than cost. For these computers to work, they have to work at quantum mechanics level. For that, you should be able to manipulate the state of an electron or the state of a photon. It is more of a question of who has the kind of knowledge to be able to do that. People who work on super conductors, solid state physics, people who are experts on quantum information theory and people who work on nano technology and people who work on super computing — you require the intersection of at least five of those skill sets to be able to work on quantum computers.
How would the algorithms for a quantum computer differ from today’s algorithms?
The algorithms for quantum computing is about finding the minimum energy function of a Hamiltonian. This is not something that you can visualise. You can visualise X+Y or solving a quadratic equation. But with quantum computing it isn’t the same algorithms at all.
Moreover, there’s no perfect quantum computer. There are errors in physical infrastructure of a quantum computer. Nobody has yet found a way to build a perfect one. I would assert that at least for the next decade, most quantum computers will have these errors underneath. Therefore, we’d need to design an algorithm that can work with the presence of errors. So those who can design algorithms that can function with the presence of errors will be the real winners.
An error free quantum computer will take a decade. How will the usage work?
Within the next five years, we’ll see significant use cases. You also need to be careful what problems you select to solve on quantum computer because those problems have to live within this presence of errors.
How will you make quantum computing affordable?
If your cost is compared with hiring six researchers for six months versus solving it with quantum computer within five minutes, you would find quantum computer to be very cost effective.