Quantum Computing through a physicist’s eye: An Interview with Dr Shangjie Guo

This is a physicist’s interview. A physicist who is trying not to be a physicist for the good of the world. I challenged Dr Shangjie Guo to explain the complicated concepts burried within the field of quantum computing and he took on that challenge. The result is this interview. He is a PhD in quantum computing from the University of Maryland, College Park and a practitioner of qunatum computing. His knowledge about the software, hardware, theory, application, challenges, opportunities of quantum computing is profound. Find the excerpts from his interview before it becomes book in few months.

Mihir: The concept of qubit. A bit has only two states: 0and 1. A qubit has more than two states. By nature, a qubit is more powerfulthan bit. Is that true?

Shangjie: Remember, it also takes more to measure a qubitthan a bit. If you are talking about just one qubit, it is not that powerfulcompared to a probabilistic bit. If you flip a coin and you don’t look at it.Than it is a probabilistic bit and it does everything that one qubit does. Ifyou want to measure another probabilistic bit using flipping the coin, you willhave to measure the first and then flip the coin again. In other words, youflip the coin twice and measure twice. In quantum computing, because ofentanglement between qubits, you only have to measure it once – you only needone gate. Quantum computing advantage extends when you have multiple qubits.

Mihir: You talked about gates there. In classical computingwe have gates like AND, OR, NOT. In quantum computing, we have CNOTGate orHGate. Fundamentally, is the concept of gate similar in classical and quantumcomputing?

Shangjie: More or less, yes. For example, to add numbers in classical computing you have a routinethat consists of several gates, like AND, OR. To add numbers in quantumcomputing there is a routine with multiple quantum computing gates, like CNOT.Think of the complexity. Certain algorithms are faster because they have fewergate operations. It may not matter for many common algorithms right now becauseof increased computational power. Think about complex algorithms. If you can dothe same routines and operations in quantum computing using considerably lessgates, it may improve the speed and performance.

Mihir: Let’s pivot to the hardware. A limitation that we areapproaching in the classical computing is hardware. We can’t squeeze the sizeof chips anymore and we have to keep stacking them to achieve more power. Howdoes the quantum computing hardware differ in breaching that limitation?

Shangjie: Some of the hardware for quantum computing is sameas classical and some different. We have both superconducting qubit andsemiconducting qubit. Semiconducting qubit is very similar in that it issilicon based. Supercomputing qubit have supercomputing electrode. You have toput them in refrigerator for operation. There is another option of vacuum chamber.You try to trap some natural ions with magnets. You capture an ion.

Mihir: An each ion becomes one qubit?

Shangjie: I would say some of the electrons on those ionsbecome qubits. To be more specific, some of the energy levels of some of theelectrons are qubits.

Mihir: Can you really control how many qubits are going tobe in such a system?

Shangjie: Yes. So basically, you don’t care about the noisefrom the vibration of these atoms. The information is coded on the electrontransitions. You need to be more careful about the electron field. Going backto the question you can control the number of qubits depending on how manyatoms you are having in the system.

The trouble is if you have a many qubits than it creates anengineering problem. For example, there will be a crosstalk noise. If you lettwo qubits talk to each other they will affect the third one.

Mihir: My introduction to classical computing was throughpascal language and had a little exposure to assembly programming. Theclassical computing evolved big time to low level languages and high leveladvanced languages. Many of quantum programming is python based these days.Does that mean we skipped the equivalent of assembly programming and startedwith high level programming in quantum computing?

Shangjie: No, we have not. It is very basic programming.Even if we are using python base, we are coding gate by gate. That python isactually working for the control of the quantum state. The control is reallycomplicated. How do you control energy of an atom? How do you controltransition of electronic states? That’s the problem we are solving. You use electronicpulse. Those electronic pulses are complicated. If you are measuring a qubitfrom 0 to 1, than there are many physical states to measure. Python in quantumcomputing is actually working for that not for computing.

Mihir: You did your PhD in quantum computing. How many yearsdid you study quantum computing?

Shangjie: Six.

Mihir: Was there enough information then to study quantumcomputing for six years? Being a program, there were multiple people doing aPhD in quantum computing. Was there enough basis information available and opportunityfor multiple PhDs in the field?

Shangjie: Most of the quantum computing researchis about how we can actually have a quantum computer. So moving from theory toreality. Fundamentally, a PhD student work on one of the two problems: 1) Doyou want to build a quantum computer? Or 2) Do you assume that a quantumcomputer exists and want to build applications of it?

Mihir: When am I getting a quantum computer on my desktop?

Shangjie: Probably, gonna be a long time because there is noreason to get it to the desktop. One reason is a quantum computer is big andneeds a lot of power, so you might just want to rent it for certain amount oftime. Also, it is not meant for day to day things like streaming, surfing, orwriting. It doesn’t serve that purpose. It is meant to solve some hardproblems. I don’t think you will get a quantum personal computer in the foreseeablefuture. You will get it on cloud quite sooner. In fact, there is alreadysomething on the cloud. It is just not as good or useful.

Mihir: Can I really trust a quantum computer on cloud thatit is indeed a quantum computer and not a quantum simulator running onclassical computing?

Shangjie: This is an academic problem of verification. Youjust use another quantum computer on another cloud. You would want them to deliversame result or be similar enough. This problem is called quantum verification.It is again a big problem. It is a mathematical problem.

Mihir: Quantum mechanics and quantum chemistry have beenusing approximations to do quantum calculations for decades now. How do thecurrent quantum algorithms differ from them? Why can’t we use those samealgorithms in quantum computing?

Shangjie: Good question. There are approximations in quantumcomputing but there are fewer approximations. That’s why we are hoping that inquantum computing we can do large calculations more accurately.

Mihir: Is the key to success in quantum computing in futureis better hardware or better software?

Shangjie: Both. Because we don’t have efficient algorithmsand software, we don’t have motivation to build better hardware. For betterhardware, we need to find better algorithms that can use that. We need to makeprogress in both. However, my opinion. Finding a really useful quantumalgorithm is more important right now than building a better hardware.

Mihir: And my last question, what talent is going to definethe success of quantum computing in near future?

Shangjie: I actually think the bottleneck isproject managers. We need project managers that can identify, define and manageproblems for quantum computing. Everybody is working on their own stuff. Thequestion is how can we bring everyone together and solve quantum problems. Yes,we need project managers. IBM is building biggest quantum capabilities isbecause they are doing a good job of project management.