Quantum Computing: enthusiasm and scepticism still coexist
Logical, Physical, Topological Qubits, and error-correcting codes towards 2024
2023 was the year of Large Language Models (LLMs) without a doubt. On the other hand, several interesting things have happened in the field of Quantum Technology that may lead to an exciting 2024. As usual when speaking about quantum, enthusiasm and scepticism coexist as in a superposition of states. What has happened recently? What do the experts say?
The purpose of this post is to guide my AI community through the quantum realm and to present some recent exciting findings, tempered by other cold statements from leading experts.
The outline of the post, will be:
Some definitions to better grasp the quantum world;
Two relevant Quantum breakthrough in 2023: Logical and Topological qubits;
Enthusiasm and scepticism about the future of Quantum Computing;
Job & Research opportunities, talks, and events in AI.
Let’s start with some basic definition, that allows me to give you some context.
Quantum Definitions
Quantum computing is a new paradigm of information processing that exploits the fundamental laws of Quantum Mechanics, where new features are allowed that have no classical analogue, such as superposition of states and entanglement.
Superposition is a quantum phenomenon that describes the fact that a quantum entity (think of an atom or a particle such as a photon) can co-exist in two or more states. Think of a ball that can only be red or yellow, but as it flies it is orange. Only after touching the ground it can become permanently red or yellow.
Entanglement is a quantum phenomenon involving two or more quantum entities. When two particles are entangled, the state of each particle cannot be separated from that of the other. It cannot be described independently, even when the particles are separated by large distances. This is why entanglement is said to lead to non-locality, where entangled particles exhibit related properties regardless of the distance between them.
Qubits play the role of information units, like classical bits, but are endowed with the special characteristics mentioned above. Superposition and entanglement are extremely powerful, but at the same time fragile. A noisy environment can break them. I always like to compare qubits to quiet children. It is easy to ask a child to be quiet and silent, although it can sometimes be a challenge if there are distractions around him. On the other hand, it can be more challenging to ask two children to be silent, at least for a few minutes. From the moment the children become five or ten, it starts to become difficult to keep them silent even for a few seconds. When there are a hundred children, it is almost impossible. In this case, a quantum computer can only function if its qubits, like the children, are not too disturbed and remain silent (they are said to maintain a coherent state). Only then do superposition and entanglement not expire and show their superpowers.
However, maintaining a quiet environment to protect fragile quantum properties is difficult. Noisy interactions and disturbances are difficult to eliminate and make quantum hardware error-prone, corrupting calculations before they can achieve the desired results. This is why fault-tolerant (i.e. error-stable) quantum computers are still a dream. It is said that we are now in the Noisy Intermediate-scale Quantum era (NISQ), where our current quantum processor are noisy and not advanced enough for fault tolerance or large enough to achieve quantum advantages over classical computers. There are several approaches to try or hope for fault-tolerance. Among them, the protagonists of 2023 were error-correcting code methods for physical/logical qubits and topological qubits.
Logical, Physical qubits and Error-correcting codes
Such as an idea shared by two or more people can survive even if one of them forgets it, in quantum computing a logical qubits might be supported by several physical qubits.
Physical qubits are the actual quantum bits implemented with physical systems (an atom, a photon, etc). Physical qubits, like all quantum systems, are subject to errors due to decoherence, environmental interactions, and imperfections in quantum gates. These errors can cause a qubit to lose its quantum information, thus presenting a major challenge for scaling up quantum computers.
Logical qubits are used to perform calculations. They must therefore be as stable as possible. For this reason they are formed by grouping several physical qubits together. The idea is to encode the state of one logical qubit into a highly entangled state of several physical qubits (as if you were sharing your idea with friends. If you forget it, the idea is still in someone else's head). With enough physical noisy qubits, it is possible to generate a smaller number of useful, noise-free logical qubits. This encoding allows errors in individual physical qubits to be detected and corrected without disturbing the information stored in the logical qubit. The methods that allow detection and correction of error in physical qubits that guaranteed the stability of a logical qubit are named Error-correcting codes (here a nice blog post).
In other words, quantum error-correction overcomes the noisy limitations by creating logical qubits, groups of physical qubits that are entangled to store information redundantly. This redundancy allows for identifying and correcting errors that may occur during quantum computations. By using logical qubits instead of individual physical qubits, quantum systems can achieve a level of fault tolerance, making them more robust and reliable for complex computations.
Topological qubits
To better grasp what a topological property is, let’s start with this example. Think of a pair of jeans and a scarf. The former has two holes (where the legs fit), the latter is flat and has no holes at all. The number of holes is a topological property, and as such is not localised but distributed throughout the object. If we now put the jeans and the scarf in the washing machine, despite the fact that the washing machine produces a lot of disturbance and shuffling, the number of holes remains the same for both garments.
Similarly, topological qubits are based on quantum states that are not localized to a specific physical location but rather are defined by the global properties or topology of the system. Because the information is encoded in the global topological properties of the system, it is inherently protected from local errors or disturbances, such as minor changes in material properties or external perturbations. Examples of topological qubits are Majorana fermions or Anyons, which exist in certain types of two-dimensional materials or structures.
Topological qubits are distinct in that their error-resilience is not achieved through error correction codes but rather is an intrinsic property of the topological state itself. This could potentially allow for simpler quantum circuits and more stable quantum computation, as fewer physical qubits might be needed to form a robust logical qubit.
2023 Quantum Breakthrough
Hoping that the above context and definitions will help you better understand the quantum realm, we are now ready to explore two important discoveries of recent months. The first is related to Topological qubits, the second one is related to Logical qubits.
Non-abelian Topological Qubits
Scientists, at Quantinuum, in collaboration with researchers from Harvard University and Caltech, created and manipulated non-abelian Anyons (which are topological qubits) for the first time, which may pave the way for quantum computers with very low error rates.
Their main results are shown in this paper, and well explained in these blog post: first blog post and second one.
The error-tolerant nature of topological qubits results from the intrinsic topological properties of non-abelian Anyons. When they are wrapped around each other, they trace error-resistant paths, and although disturbances such as magnetic fields slightly alter the paths, their qualitative linkage or topology remains unchanged.
This increased stability that they were able to show, enables the development of large-scale quantum computers capable of performing complex computations with a high level of accuracy, potentially outperforming classical computers in a variety of applications. The error-tolerant nature of topological qubits also reduces the need for error-correcting codes, simplifying the design and implementation of quantum computing systems.
Logical Quantum Processor
“I think this is one of the moments in which it is clear that something very special is coming!”
This is what Mikhail Lukin said after his Quantum Optics Lab in Harward, in collaboration with colleagues from MIT and QuEra Computing, succeeded in creating the first programmable Quantum Logic Processor. How was this possible? Lukin and his collaborators successfully executed large-scale algorithms (such as surface codes) on a quantum architecture known as the Neutral Atom Array (which they had previously built). This is a block of 280 ultra-cold suspended rubidium atoms that act as physical qubits, simulating up to 48 logical qubits. How? By controlling the 280 physical qubits with error-correcting code methods, after grouping them into groups of 5/6 entangled qubits to store information redundantly. This is an incredible research progress, being that the best computing systems so far have demonstrated to work with two logical qubits. More about their results can be found in this Nature paper and this blog post.
Will this open the door to what Simone Severini calls in his newsletter the era of Logical Intermediate-Scale Quantum (LISQ)?
Enthusiasm and Scepticism
Despite the enormous achievements of this year and previous years, quantum technologies, and quantum computers in particular, are still at an embryonic stage if we assess them according to their market maturity.
As in a quantum superposition, enthusiasm and scepticism coexist at this stage. It is worth mentioning Scott Aaronson, a world-renowned expert in quantum computing, who has recently devoted most of his time to Artificial Intelligence. In a recent blog post, commenting on Lukin's group achievements, Scott said:
"I’m now completely convinced that AI will transform civilization and daily life in a much deeper way and on a shorter timescale than QC will — and that’s assuming full fault-tolerant QCs eventually get built. I’d like to contribute if I can to helping the transition to an AI-centric world go well for humanity."
Along with Scott Aaronson, other leading experts say that caution is wiser than hype. In this recent interview, you can find more sceptical opinions, including Yan LeCun, head of Meta's AI (although he is not a quantum expert, but certainly someone to listen to when it comes to technology).
Personally, as a physicist, I am extremely enthusiastic about this technology and am optimistic about its (long-term) future, perhaps also because I am motivated by dreams of being able to process information in the same fundamental way that our Universe does. That sounds to me amazing and worth exploring, regardless of everything else!
Opportunities, talks, and events
I share some opportunities from my network that you might find interesting:
🚀 Job opportunity: An Italian well-founded startup, applying Deep Learning in the fashion sector is looking for a Machine Learning Scientist or AI developer (more info here).
🚀 Job opportunity: Ecosmic, a startup empowering sustainable space operations, has several open positions you might be interested. Checkout their website!
🔬Research opportunity: Two Postdoctoral positions in AI for rare diseases @ Alessaandro Laio's Group at SISSA (Trieste). More info here.
🎙 Tech Talk by Norman Di Palo at Pi School: Robotics Meets Foundation Models.
If you would like to get in touch with me or view my lectures and courses (technical and non-technical), you can find everything here.
You can find me on LinkedIn or Twitter, where I share Science & Tech news. If you wish to book a call or view my lectures and courses (technical and non-technical), you can find me here.
