Be so good they can’t ignore you

The first time I read this sentence was on the cover of a notebook, I love motivational stationary and this one caught my attention. It differed from other inspirational sayings in that it didn’t encourage believing in yourself but rather working on yourself. Later, when reading Eric Barker’s blog, of which I am a long term fan, I learnt that this was advice given by Steve Martin to aspiring performers. I am, no longer, a performer, but have since made it a motto of my own work. I find that whenever I’m demotivated, it gives a tangible goal, as it forces me to ask myself what can I do to become a better physicist and whatever that is, I do. It also shows a defiant attitude toward preconceptions and restrictions imposed from outside, but one that is intended to prove them wrong. It will make you become the dark horse of your work endeavours.

Cal Newport, a Georgetown professor in Computer Science and blogger of Study Hacks, wrote a book of the same title. The book’s goal is to identify the steps one can take to build a successful career. It analyses the careers of a number of people, both successful and failed, and finds what were the steps taken in each case that determined the career’s fate. One of the first things that got me interested in the book, is the rejection of the “courage culture”- people who promote the idea that the only thing standing between you and your goal (in this case your dream career) is yourself, and all it takes is for you to believe in yourself and build up the courage to step off the expected path. I have always been suspicious of the belief that all it takes is courage, for any situation, and much more with respect to a career path. First of all, because many endeavours won’t succeed unless we have a background set of skills that can help our bid, but mostly because I don’t believe most things come easy, it’s resolve and hard work what makes them possible. I believe in research, preparation, planning and effort. It is true that in many cases, talent plays a big role, but I’ve never regarded natural ability as the ultimate deciding factor. I wouldn’t be running if that was the case 🙂

I must say, however, that some amount of believing in yourself and your abilities is necessary, as otherwise, one might fall sick with “imposter syndrome”, so common in academia. In my opinion, the motto “be so good they can’t ignore you” can be used to soothe such feelings, as a survey of one’s worked-for abilities should put at ease any feelings of inadequacy. But, we won’t speak any more of imposter syndrome here, as that deserves a whole other post by itself. Let’s turn our attention instead to Newport’s 4 rules that can help us build a fulfilling working life.

Don’t follow passion: We are told time and time again to follow our passion: “do what you love and you won’t have to work a day in your life”. It’s one of those inspirational sayings we see everywhere and what makes Steve Job’s  commencement address so popular that it has more than 24M views on youtube. However, it is not useful advice for someone unsure about which career is best for them. I’m sure you have experienced before surprise at enjoying something you didn’t think you would. By solely advocating to pursue the things you *know* you like, there are many other enjoyable endeavours that are left out. Moreover, there is little evidence that people have pre-existing passions and this kind of approach to finding a fulfilling work life can lead to a lot of unhappiness. “Choosing your career should not be treated as finding your true calling.

Have the mentality of a craftsman: Instead of focusing on whether the jobs fulfills any dream-job fantasies we may have, we should focus on the value that we offer, enjoy the process of the work and be proud of the output generated. This is, of course, easier said than done. When our focus shifts to what we produce, the goal becomes much clearer: improve the outcome of our work. To do that, we must engage in deliberate practice, which is the style of difficult practice that is required to improve in a task. It is the kind of practice that will involve learning new techniques, practice for hours and continually face our ignorance. It is the difference between a master and a middle of the pack. It requires that we stretch beyond what is comfortable and we are willing to accept ruthless criticism. Most of us have experienced this kind of practice when going through the educational system and trying to come to grips with the material we didn’t understand. However, once we are out of education, it becomes increasingly difficult to do so, because we are not forced to, and because it’s easier to tell ourselves we have too much email. But forcing ourselves to engage in deliberate practice will increase our abilities and help us become so good that we will get noticed.

Leverage skills to obtain more control: Some people enjoy tremendous freedom and control over their working life, while others don’t. It begs the question: why do they get those perks? In most cases, those people have rare and valuable skills, that are so priced by their employers, that they can leverage them to their advantage. That is what Newport calls “career capital”. Through deliberate practice, those people have built up career capital – a valuable set of skills that allows them to trade them for more control of their own working life. However, attaining control is tricky. If it is attained before we have enough career capital to make it sustainable, we will fail, but also, once we have built enough career capital, we might face a pushback from the very people we have become invaluable to, they might try to prevent change that benefits us and not them. So how can we decide if a bid for more control is the right step to take? Newport’s law of financial viability: “When deciding whether to follow an appealing pursuit that will introduce more control into your working life, ask yourself whether people are willing to pay you for it. If so, continue. If not, move on.”

Find a mission: “A career mission is an organising purpose to your working life”. Finding a mission can be condensed in two actions: doing a series of little bets to scout out different ideas that might succeed, and having the mindset of a marketeer, i.e. being able to identify why some ideas catch while others fall flat. True missions require a specific lifestyle: patience to build career capital as well as being constantly searching for the next big idea. These ideas tend to lie in the space just beyond the cutting edge of a field, which has been referred to as the “adjacent possible”. Identifying these big ideas requires dedication to brainstorming and exposure to new ideas. But how can we figure out if a chosen mission is likely to succeed? The answer is given by Newport’s law of remarkability: “For a mission-driven project to succeed, it should be remarkable in two different ways. First, it must compel people who encounter it to remark about it to others. Second, it must be launched in a venue that supports such remarking.

Reading this book made me have a different perspective on the career decisions that lead me to where I am now. I can see how hard work and little bets allowed me to move forward and I can also see the mistakes I made. I found the book very compelling because it gives me a framework that will help me achieve a fulfilling career, it gives me tools to achieve that desire rather than just the inspiration. Also, Newport relates his personal journey as a researcher and what deliberate practices he engages on, as well as how he develops his career mission. These are practices I can directly apply myself to grow as a physicist. While reading I felt I couldn’t wait to put this framework in practice, this blog is a result of that.

What is a quantum computer?

This is a question you might get asked fairly often if you ever mention that you do research in quantum physics. You might even be asked if you are a Prime Minister visiting a research facility on the topic. So among all the questions I would expect to be asked, this is not a particularly surprising one. Except on my PhD viva exam, when I really did not expect it. Particularly in the form it was phrased: “How can you tell if a machine is a quantum computer? If aliens came to Earth and tried to sell you a quantum computer, what could you do to be certain they were not fooling you?”. I didn’t expect alien sales to be part of my viva, that’s for sure! Fortunately for me, some conversations I had previously had with experimentalists at the Centre for Quantum Photonics helped me give an answer that satisfied my examiners. However, I came out from the exam with the nagging feeling that that should have been a much easier question to answer.

In any standard course on quantum computing and its implementations, usually one learns about the DiVincenzo criteria, which states that a quantum computer should have:

  1. A scalable physical system with well-characterized qubits.
  2. The ability to initialize the state of the qubits to a simple fiducial state.
  3. Long relevant decoherence times, much longer than gate operation times.
  4. A “universal” set of quantum gates.
  5. A qubit-specific measurement capability.

These criteria were formulated in the year 2000 with a specific computational model in mind, the circuit model. Since then, other ways of implementing quantum computing have sprung out, from computational models (such as cluster state and adiabatic models) to physical implementations that don’t deal with qubits but rather with continuous variable systems. Many of these new developments don’t quite fit the DiVincenzo criteria, so are there any better definitions of what quantum computers are?

Searching through the scientific literature, and through statements of experts to the media, we can find five distinct ways in which to specify what a quantum computer is:

  1. Abstract theoretical definitions: Such is Deutsch’s definition in his 1984 paper where he formulates a physical version of the Church-Turing thesis that is compatible with quantum theory: “a quantum computer is a […] quantum generalization of the class of Turing machines”. Or Feynman’s: “It’s not a Turing machine, but a machine of a different kind”. These definitions are very abstract and lack details and specifications on the structural components, making them not very useful in practical scenarios.
  2. Implicit definitions: A quantum computer is not defined necessarily by its components but rather by stating that it uses the laws of quantum mechanics to perform computation. As true as this definition is, it is no help when trying to decide whether a computer is quantum or not. How can we assert from outside which laws govern the logical operations inside?
  3. Comparative definitions: A quantum computer is a device that can outperform classical computers. While we certainly expect them to be able to solve problems that would otherwise be outside our reach if we only had access to classical computers, relying on the classification of problems in complexity theory is uncertain business, as this classification is not written in stone and evolves as the field develops.
  4. Constructive definitions: These definitions specify what a quantum computer is by stating their components or the way information is processed. For example, defining the quantum computer as a machine that fulfils the DiVincenzo criteria falls in this category. This kind of definitions share the characteristic of being narrow and tied to a specific implementation, and therefore not general enough to apply to all architectures, physical implementations and computational models.
  5. Operational definitions: The quantum computer is fully defined by what it does, if a machine acts like a quantum computer, it is a quantum computer. The definition makes no assumptions about the theory of computation or the nature of physical reality, and therefore different interpretations of quantum mechanics should agree that the machine is a quantum computer.

There is an excellent paper which intends to figure out a useful operational definition for quantum computers that will stand the test of time, and by making no reference to how the computer exactly works, it should still hold in years to come when new techniques have been developed. As end users we care mainly about performance, and not necessarily about the nitty-gritty details of how that performance is achieved. As an example, how many of you understand all the complexity of the device you are using to read this blogpost? I certainly don’t.

The proposed operational definition of a quantum computer in this paper is as follows:

“We define a quantum computer as a device that can run a quantum algorithm efficiently, where a quantum algorithm is a classical bit string that encodes a series of quantum operations. The quantum computer should be able to take this string as input and produce another one as output. The probability distribution of the output should be consistent with the predictions of quantum theory. Finally, the time it takes the computer to produce the output should be in agreement with the difficulty of the algorithm.”

Note that this definition makes no mention of the way in which the quantum operations are performed, and therefore is open to different models of computation. The time constraint in this definition is crucial, as it excludes classical computers, because, for example, if we ran Shor’s factoring algorithm in a quantum computer we would expect an answer in polynomial time and a classical computer would require exponential time. But also this definition doesn’t depend on the current classification of complexity problems, as “the difficulty of the algorithm” mentioned in the definition would refer to the difficulty of a particular problem at the time of the test.

It is also worth noting that both the input and output are classical bit strings. After all, the input and output are the interactions of the machine with the controller, which lives in a classical world. The quantum program will be the encoding of the operations of a particular quantum algorithm, which will be part of the input string along with the initial state of the quantum computer. The instructions for generating the initial state on the quantum computer must have an efficient classical representation, this is the case for all computational models. It is worth noting that for all the current proposals of quantum computers, some classical pre and post-processing are assumed, which agrees with the above definition as long as this classical processing takes only polynomial time in the problem size.

In this paper they also have a set of criteria for building a quantum computer, that is general enough to fit all computational models (currently known) and all physical implementations proposed. It is based on 4 statements:

  1. Any quantum computer should have a quantum memory. Quantum memory is the broad term used to state that the quantum computer must have the capability of efficiently representing any computable pure quantum state (of a size accordant with the size of the computer) in its internal degrees of freedom. This quantum state will not have, in general, an efficient classical representation.
  2. Any quantum computer must facilitate a controlled quantum evolution of the quantum memory, which allows for universal quantum computation. By controlled quantum evolution, the authors mean that the evolution of the internal state of the quantum computer must follow the laws of quantum mechanics, and will ultimately be controlled by the end user.
  3. Any quantum computer must have a method for cooling the quantum memory. Entropy is accumulated in the quantum memory as a product of a previous computation or because of the presence of errors due to noise from the environment. Cooling refers to information-theoretic cooling, where the entropy is extracted from the quantum memory. It encompasses the process of initialisation of the quantum memory as well as the error correction procedure.
  4. Any quantum computer must provide a readout mechanism for subsets of the quantum memory. The computer must have a mechanism to translate the outcome of the computation to classical bits for the controller to obtain the result of the computation. The authors refer to subsets of measurements as during most error correcting procedures, there are intermediate measurements used to assess the presence of errors, and this kind of measurements are not of much use to the end user.

There are two other essential characteristics for a quantum computer the authors require:

  • Fault-tolerance: Fluctuations from the environment can cause stochastic errors to appear in the quantum computer. If a quantum computer still works according to its definition in the presence of such errors, it will be deemed fault-tolerant. This will be the case if the computer uses some form of error correction that has an error threshold (maximum size of individual errors) higher than the errors caused by the environment.
  • Scalability: Currently any claims of scalability in any particular physical implementation of a quantum computer are predictions, as no reasonably large quantum computer exists. Theoretically, a quantum computer is scalable if the resources required to run the same polynomial time algorithm on a larger input scaled polynomially on the size of the input. What makes a particular architecture scalable depends heavily on the architecture and technological prediction and it is difficult to make general statements without getting into the detail any given implementation. An excellent read on this subject can be found in the Quantum Pontiff blog post “Climbing Mount Scalable”.

It is perhaps because we don’t have a functioning large-scale quantum computer that it is difficult to give accurate definitions without hiding behind the “spooky” laws of quantum mechanics. At the end of the day definitions are not the most important thing, and proof of that is that Nielsen & Chuang, the must-read book for any quantum information scientist, does not define what a quantum computer is and rather lets the reader build up an intuition. But it is important to know what we talk about when we talk about quantum computers, for us to be able to make informed decisions about whether a machine is a quantum computer or not, in case aliens or someone else came to us with sales pitch.

Welcome to my blog

Hello Internet and welcome to my blog!

The idea for starting this blog comes from the group meetings we had at the Controlled Quantum Dynamics Theory Group at Imperial College, where I did my PhD. Every Wednesday morning, we’d gather in a meeting room with beautiful views over London, had breakfast and talked about physics… most of the time. Every week a member of the group would prepare a talk on a subject of their choice, it could be about their own work, an interesting paper they had read, something they had recently learnt about or anything else they thought was interesting. We had fascinating talks on art, bike physics, coffee, songbirds and many other topics.

I took the approach of using my allotted talks as an opportunity to learn about some new physics, what better incentive to get your facts straight than a room full of physicists ready to argue? Jokes aside, preparing an hour of spoken material on a new subject allowed me to grasp the basic concepts of the new topic, and I don’t think I would have tried so hard to learn about something which had (most of the time) not much to do with my PhD work, had it not been for these talks. Also, the questions raised by some members of the audience, made me look at a topic in a different or find out connexions I didn’t know about. Overall, I think those talks were one of the most formative experiences I’ve ever had.

This blog will be my online replacement for those group meetings. Topic-wise, I think it’s reasonable to say that most posts will be quantum-related, but I also hope to include some posts on computer science, outreach and other miscellaneous topics. I will sometimes post about things I know about, but my intention is to learn new things for each post. So if I ever get something wrong in a post, please point it out in the comments!  Hopefully, with time I will get the chance to learn from the comments on my posts as well.