January 2021: Quantum Computing
Quantum physics is creating the potential to create quantum computers and networks that may one day greatly outperform current computers.
This discussion featured a panel of four researchers who talked about their work in quantum computing and how it could change the work we do in IT.
During this event, we talked with researchers Diana Franklin, Liang Jiang, Robert Rand, and David Shuster.
About the Speakers
Diana Franklin
Diana Franklin leads five projects involving computer science education involving students ranging from 3rd grade through university. She is the lead Principal Investigator (PI) for quantum computing education for EPIQC, a National Science Foundation (NSF) expedition in computing. Her research agenda explores ways to create curriculum and computing environments in ways that reach a broad audience. Her research interests include computing education research, architecture involving novel technologies, and ethnic and gender diversity in computing. She is the author of “A Practical Guide to Gender Diversity for CS Faculty,” from Morgan Claypool.
Liang Jang
Liang Jang investigates quantum systems and explores various quantum applications, such as quantum sensing, quantum transduction, quantum communication, and quantum computation. His efforts are helping to make quantum computing and communication technology scalable and more accessible. Professor Jiang is Principal Investigator (PI) for the Jiang Group, which investigates quantum control and quantum error correction to protect quantum information from decoherence for various physical platforms. He has worked on modular quantum computation, global-scale quantum networks, room-temperature nano-magnetometer, sub-wavelength imaging, micro-optical quantum transduction, and error-correction-assisted quantum sensing and simulation.
Robert Rand
Robert Rand is part of the small but growing community of researchers creating quantum programming languages. Today’s programmers of classical computers have a vast library of languages they can use, from high-level languages such as Python and Java to more targeted languages for specific tasks such as working with databases or spreadsheets. But quantum computing languages sit closer to the early, pre-Fortran days of computer science, where many options exist but no consensus has emerged. Professor Rand authored an online, interactive textbook, “Verified Quantum Computing,” which takes a mathematical approach to its topic, teaching concepts by asking students to prove theorems, which can then be verified by an automated proof assistant.
David Schuster
David Schuster is the Principal Investigator (PI) for the Schuster Lab, which specializes in quantum information, with research efforts in quantum computing, hybrid quantum systems, and quantum simulation. His research focuses on understanding and controlling the unique properties such as superposition and entanglement of quantum systems in a variety of platforms.
Q and A with Panelists
Diana Franklin: Decoherence is a bit separate from measurement. The belief is that measurement will *always* destroy the state. Decoherence is when the state changes even in the absence of measurement . Decoherence is somewhat an engineering problem - they are trying to better fabricate and protect devices so they have less decoherence. On the other hand, some are working on algorithms that know about the types of errors that occur in the hardware.
Liang Jiang: Sometimes we may also engineer decoherence to assist quantum information processing or error correction, which is an active research direction.
Diana Franklin: When you have a superposition, that is a number of simultaneous states at once (in a single qubit, just two states, but in n qubits, it’s 2^n states). However, you can only ever read out one of those 2^n states, and the one you measure is a probabilistic outcome. So there is a certain probability of reading out each state, and quantum operations modify those probabilities.
David Schuster: It is very unlikely that it is possible (though we don't even know for sure that NP is not P). But it is thought there is a class of problems (BQP) that are soluble on a quantum computer but are not in P.
Diana Franklin: Theoretically, there is not a distance limitation. However, because entanglement and superposition are at the quantum level, a very small disturbance can modify the state. So it’s not distance that’s the problem - it’s the time and environment at which the distance is traveled.
Robert Rand: It’s probably worth noting that quantum entanglement cannot be used to send information faster than light. This is surprising, since what happens to one qubit happens to the other simultaneous, but the destructive properties of measurement prevent actual communication without using classical communication as well.
David Schuster: Google and IBM are pretty similar (though details matter), grids of superconducting circuits but there are also ones that have entirely different platforms such as atoms, or electrons, and also are architected fairly different in how they interact.
Diana Franklin: Yes, definitely. It is not anticipated to ever replace general-purpose computers.
Diana Franklin: There is something called a qutrit that can have three states, as opposed to the traditional qubit, which has two. This depends on the architecture, but in superconducting machines, it’s for the same reason that bits have two states: binary is convenient for computing.
Robert Rand: This depends on the architecture, but in superconducting machines, it’s for the same reason that bits have two states: binary is convenient for computing.
Diana Franklin: Entanglement is actually required to do even the simplest quantum calculations. If you think about getting qubits to cooperate together to form a variable, you need entanglement. For example, let’s imagine that you have 3 bits that can store a single number from 0-7 in a classical computer but, when in superposition, multiple numbers simultaneously with some probability of reading out each number. I’d like to store 000 with 25% probability and 111 with 75% probability. Without entanglement, that means that each individual qubit would have 75% probability of reading a 1. Which means that the number 101 could be read with 0.75 * 0.25 * 0.75 probability. But that’s not what we wanted. If the first measurement reads a 1, we want the other two to also measure a 1. That requires entanglement.
Diana Franklin: Yes, people are thinking of creative ways to do things. Ant Man had a remarkably accurate (to a certain point) description of the Quantum Realm. I just put in a proposal to NSF to create an app that has several games that have analogies to quantum concepts. There are some videos from Europe about Quantum Kate that relates them to everyday concepts. I think if some major media company were to decide to do it, and got together the right team, we could make something really cool. We were trying to write comic books with comic book heroes with quantum-inspired skills, but it was too heavy of a lift, so we did zines instead.
Robert Rand: David Schuster and I are different parts of the same large grant (EPiQC). I lead the research on QC learning, and he is building the physical computer, and a bunch of other people are working on bridging the gap between algorithms that assume perfect hardware and incredibly imperfect hardware. It brings a computer science systems approach to bridging that gap through technology-aware compiler optimizations, detecting machine errors and mapping qubits with that knowledge, communication between qubits, specializing the program to particular nuances of hardware, etc.
David Schuster: I think they have to some extent, but the main reason that its not more ubiquitous is power consumption, not the resistor itself but measuring the very small signal it produces.
David Schuster: Not including the R+D its probably ~$1-3M. This is largely due to "low volume." Right now there are several types of quantum computers which all have a few tens of quantum bits. Most likely the challenges will be similar in terms of making sure there aren't mistakes in the libraries, user behavior, etc.