Building quantum competence in Bergen

We were thrilled to see the interest in quantum computing at the Department of Informatics in Bergen. No less than 24 students completed the course Introduction to Quantum Computing and Quantum Machine Learning! This is certainly a significant contribution when it comes to making our future workforce quantum literate.

The course responsible is Philip Turk, whom you can see in the picture – along with his Bloch sphere, is affiliated with SINTEF – in addition to the University of Bergen. He has every reason to be proud – both of the course he has developed and of his students’ performances at the oral examination. Many of them demonstrated convincing familiarity with advanced quantum algorithms such as quantum phase estimation and Shor’s algorithm – in addition to having simulated and trained their own quantum neural networks. Quantum machine learning coincides with Philip’s own research.

Being an informatics course, emphasis was put more on algorithms – and the formal distinctions between quantum and classical information – than on the physics behind it. We at the OsloMet Quantum Hub salutes the initiative and hope that it continues to build quantum competence in Bergen. Perhaps it would spread to the physics and the chemistry departments too?

On the origin of quantum advantage, by Maksym Teslyk

Quantum computing seems to be a popular trend all over the word nowadays. Many of us have heard that private companies, e.g., D-Wave Systems, Google quantum AI, IBM with its very latest quantum processor Condor, etc., are working hard on the construction and design of quantum computation facilities. Even more on that, the academic community is also involved in the race – and OsloMet is no exception here.

One may wonder why are we working so hard and investing so much time and resources to build a quantum computer? Maybe all this activity is nothing more but a hype, and the focus on common and well-known classical machines, which we know how to produce, would be a better strategy? What are the benefits of calculations based on quantum algorithms, instead of classical ones?

One may suggest pure math to answer these questions. However, algorithm complexity theory is not of great help. There are plenty of issues which it cannot solve. For example, one may try the NP problem, which is believed to be hard enough to be included in the Millenium Prize Problems listing. And introducing quantum complexity classes just makes the approach even more challenging.

The opposite solution is solely practice. One may design a quantum device able to solve some (abstract) tasks which is unsolvable by any existing (or forthcoming) classical computer. The approach attracts much attention today, but cannot provide any satisfiable answer – see, e.g., the recent efficient simulation of quantum processors.

Ok, but we know that the properties of elementary pieces of information exhibit significant differences. For example, any classical bit can be represented as a switch between two mutually excluding classical states. Quantum bits, or qubits, are encoded with the Bloch sphere – they are two-dimensional, unlike bits. Such a seemingly strange and counter-intuitive representation originates from the quantum superposition principle, allowing for a quantum system to contain both mutually excluding outcomes simultaneously. Combined with linearity of unitary operators performing quantum evolution, this allows to hack the computation. Namely, one may encode all the possible combinations of the onset data into a single input state and obtain all the possible outputs in a single run of a quantum processor. Obviously, such a recipe won’t work for classical switches.

So, have we found the answer? On the one hand, both quantum superposition and linearity of operators allow one to use quantum interference at its full power. This can be easily illustrated with the help of, e.g., quantum Bernstein-Vazirani algorithm which outperforms the classical one. On the other hand, the Gottesmann-Knill theorem challenges this and makes the whole approach a bit unclear.

If the properties of information cannot explain the quantum advantage, maybe we should look for something which cannot be reproduced within any classical device at all? How about entanglement? Indeed, this phenomenon, blamed by A. Einstein as a ‘spooky action at a distance’, exhibits the possibility of correlations at the level which is unattainable in the terms of non-quantum physics. It is also known as quantum non-locality and violates Bell inequalities which any classical system, regardless its properties and working principles, should obey.

Entangled states imply the existence of a quantum communication channel and can be used to transfer data far more efficiently than any existing communication networks. These are not just words: we do have such quantum protocols as quantum teleportation or superdense coding, and both are heavily based on entanglement.

So, we see that quantum non-locality may be treated as a quantum resource allowing more efficient information transfer. Moreover, it was shown that if the entanglement which is produced by some quantum circuit is upper-bounded, then it can be efficiently simulated classically. Therefore, one may conclude that entanglement speeds up quantum computation. However, another study could not detect any significant role of this resource in the Shor’s algorithm. Taking into account that the algorithm is one of the most efficient among the ones known to date, we must admit that the role of entanglement in the speed-up requires further clarification.

We may consider the problem at the very abstract level also. Any process obeys some set of rules, which can be formalized in the terms of underlying logic. The formalism was developed almost a century ago. It clearly demonstrates that any classical system obeys the rules of Boolean logic, which can be inferred from the relations among the subsets of a phase space, and that any quantum system is governed by the relations among the linear subspaces of the Hilbert space (quantum logic). The key difference between these two systems originates from the commutation properties of classical functions and quantum operators, respectively.

The Huygens-Fresnel principle allows to compare both logics with such different algebraic structures. Namely, let us consider a wave propagating from point A to B. In the terms of wave optics we know that the transition should take into account all possible paths through space. However, in case the wavelength is negligible, all these additional paths vanish but the one governed by far more simple ray optics. The same procedure may be applied to quantum circuits in terms of the path integral formalism. Taking the de Broglie wavelength to the zeroth limit determines their transition to classical calculations. Unfortunately, the estimation of information lost under the limit does not reproduce the corresponding increase of computational inefficiency. To sum up, we know that quantum advantage exists. This can be easily illustrated with the help of Grover’s search algorithm, which outperforms any classical analog. On the other hand, the origin of the advantage looks fuzzy. And everyone is highly invited to participate in solving the puzzle.

Congrats, Maksym

Yesterday, November 23rd, our visiting researcher Maksym Teslyk sucessfully defended his PhD dissertation at the University of Oslo.

Not only did he defend his thesis, he also gave two lectures on different topics this day, which was a long one for him, no doubt. I had the pleasure of being present at the first of these lectures, in which he told a very interesting story about the subtle quest for actual quantum advantage in computing. While we know that it does exists for several problems, it’s fundamental nature appears somewhat elusive.

In any case: We at the OsloMet Quantum Hub want to say CONGRATS, Maksym!

Qhub goes to Helsinki

Mid November a bunch of us had the pleasure of going to Helsinki, the captial of Finland and, arguably, the Nordic capital of quantum startups. In any case, what we learned did confirm that there is a lot of interesting Finish activities going on in the quantum area – both theoretically and experimentally.

Perhaps the most interesting part was a visit to the Low Temperature Lab at Aalto university. (By “low temperature” we are not referring not mean average Finish Winter temperature, we are talking milli Kelvins.) Several setups, involving extremely cold “fridges”, were constructed for various quantum sensing applications – including an ambitious quest to measure quantum effects in gravity.

Our trip, aiming both at learning more about Nordic quantum activities and at raising cohesion within our hub, also involved a visit to Helsinki University. In addition to an inspiring visit to a lab aimed at curious school children and youths, we learned about diverse topics such as quantum machine learning, post quantum cryptography and the benefits of using a combinatorial approach to quantum circuits – only to name a few.

For a few photos from the trip, please follow this link.

Quantum jumps at Holmenkollen

November 7th to 9th we had the pleasure of hosting a November School on Quantum Computing. At the risk of appearing cocky: Beforehand we were very proud of the program we had put together. And the lecturers did not let us down!

Several aspects of quantum computing were addressed. To name a few:

  • Quantum error correction
  • Quantum annealing
  • Quantum reservoir computing
  • Quantum hardware
  • Quantum computing for quantum chemistry
  • Quantum software engineering
  • Quantum states encoded in neural networks
  • Quantum noise

The lecturers included both academic researchers and representatives from the industry, specifically from D-Wave and IBM.

We believe it is fair to say that many large quantum leaps were made in the participants’ knowledge in quantum computing. In addition to our own PhD and Master students, people from Chalmers University of Technology, the University of Oslo, Lund University, Simula and the Norwegian School of Economics.

The venue provided a very nice atmosphere for getting to know more of the many flavours of quantum computing – and the growing community of people within the field.

Read more about it on the School’s website. Here you will also find the slides from most of the lectures – and a gallery. As small excerpt from this gallery is seen below.

Proud organizers: Andre Laestadius and Sergiy Denysov.
The venue: Holmenkollen Park Hotel.
Some of the participants at Roseslottet.

Our two-qubit computer in the Ruter  headquarters

AI and Machine Learning are vitally important for Ruter and  Quantum Advantage is something the company is paying a serious attention to. This year Ruter’s AI department runs the second project on gauging the potential of quantum computing as a tool to address the specific use-cases Ruter deals with on daily basis.

Last Friday we were invited to the Ruter’s  AI  Day and asked to present one of our quantum computers. We did so and gained a substantial interests from the event participants. It was nice to see that quantum computing provokes thoughts – and curiosity – even in the senior IT professionals.

Seminar: Black Hole Entropy

Our visiting researcher Maksym Teslyk is presenting work which is part of his Ph.D. dissertation. It relates to both classical and quantum physical information theory and to general relativity. The picture is Kip Thorne’s black hole visualization from the movie Interstellar.

Abstract:

A spherical system of mass M is represented as a set of Unruh horizons. The approach allows to estimate the total entropy of Unruh radiation from the set and calculate its ratio to the Bekenstein-Hawking entropy. The contribution of mass and spin s of the emitted particles is taken into account. For large values of M, the ratio exhibits susceptibility to the intrinsic degrees of freedom and varies from 0% (s = 0) to 19% (s = 5/2).

Time and place: Thursday Nov. 16th, room PS439 in Pilestredet 35.

Impressive Warsaw

In September this year, my good college Morten Førre and I had the pleasure of being invited by Katarzyna Krajewska to her group in Warsaw. Recently, they have developed a promising method for studying with photo ionization of atoms computationally. PhD student Mihai Suster is central in this development work.

Typically, atoms and molecules exposed to strong laser field, which, incidentally, was the topic of this year’s Nobel prize in physics, are described theoretically and computationally without considering magnetic interactions. However, with strong enough fields and low/high enough photon energies, this approach breaks down – rendering the problem much tougher.

This was one of several topics for discussion.

Morten and I also had the pleasure to learn about ongoing activities within quantum information technology at the University of Warsaw – both theoretically and experimentally. We were quite impressed by their various labs – and their efficient setups.

Impressive was also a relevant word for describing the city. Well worth a visit!

Something to stretch towards: Julia Derlikiewicz (left) and Marie Skłodowska Curie (right).

We thank our hosts, Katarzyna, Julia, Deeksha, Jurek, and Mihai, so much for their overwhelming hospitality and hope that we can return the favor at some point.

-Sølve Selstø

Seminar: Quantum Software Engineering

Tuesday Nov. 17th we have the pleasure of hearing Noah Oldfield, from Simula Research Lab, presenting results and research question related to his ongoing project. It involves software testing on actual quantum computers. See the abstract below for more details.

Quantum program outputs enable the development of unique quality assurance techniques. Our research focuses on efficiently distinguishing a specialized ideal state vector from the sampled state vector of a program using inference techniques.

To accomplish this, we utilized a hill climbing algorithm for stochastic searches between basis transformations, circumventing the exponential scaling of brute force searches with increased qubit numbers. We conducted tests on a suite of automatically generated faulty programs.

For those programs with state vectors representable in the Hadamard basis, we observed improved testing runtimes and enhanced phase gate fault detection.

Quantum-interest i Kode24

We are pleased to see that the Kode24, and online magazine for developers, has taken an interest for quantum computing – again. Under the heading “Hva er greia med …” [What’s the deal with …], this interview, conducted by journalist Kurt Lekanger, introduces a few of the basics.