Quantum Computing Faces Software Gap as Hardware Surges Ahead

Commercial quantum computing is just over the horizon, but most of the news in the sector focuses on breakthroughs in hardware processing power or new facilities opening to support the supply chain.

Quantum computing software risks lagging behind hardware, exacerbated by a skills gap. In this Q&A, Erik Garcell, technical marketing manager at quantum software company Classiq, explains how bridging the quantum software gap will make quantum computing accessible.

Enter Quantum: Why is quantum computer software lagging so far behind hardware development?

Erik Garcell: Quantum programming for the most part is at the assembly-level programming language level now. It's Boolean logic, almost like programming with punch cards.

I’m often asked what’s the largest quantum circuit you've ever built, 100 qubit systems are becoming common and IBM just made its 1,125 qubit Condor processor. But you need to be able to use it.

You cannot get there with the current programming that's available. That’s the problem Classiq is trying to solve.

How does Classiq’s tool tackle this problem?

Classiq has a higher-level modeling language that makes it easier to program. It's trying to lower the barrier to entry to program so you don't have to have as much quantum computing knowledge.

Related:Citi, Classiq, AWS Test Quantum Finance Portfolio Optimization

Computer scientists can get into this using regular code, they don’t need to know rotation angles to get certain probabilities or how to mark gates and CNOT gates. You tell the system what you want the quantum algorithm to do – give it an equation and variables, for instance – send it to the server and it gets compiled and run.

What quantum hardware can it run on?

Any of the hardware vendors that are out there. I could tell the system to optimize for IBM, Microsoft or AWS hardware. Managers don't want to teach their developers three different programming languages; they want to explore and test. Classiq aims to bring the software to where the hardware is.

If you were doing this by hand in assembly-level language, you'd have to understand what is optimal for that particular hardware. How does that play with the gate set or the conductivity of the qubits? Maybe on one particular set of hardware sometimes qubit two doesn't talk to qubit three. If you build your quantum circuit with that in mind, it's going to have to get un-optimized at the end.

Hardware is fancy and it gets a lot of media attention, but software is what lets you use it.

Classiq’s tool includes a quantum resource estimator. What does that do?

You take your quantum circuit and tell it the error budget that you want this algorithm to run within and the error rate of the system you’re putting it on. Then it tells you what real resources would be needed to run this quantum circuit you've built within that amount of error.

A lot of people in the industry like this because it gives them a quantum finish line. They can say when this particular hardware exists, our company will see quantum utility.

The utility aspect is important because utility means different things for different companies. It could mean the answer is better than with classical or it runs with less energy costs of a faster speed to answer even if it's the same solution.

It lets companies know when quantum utility will matter for them. People can say I'm going to put this on a sticky note so that when I see a media article saying that this hardware exists, that's what I need to tell my IT developers to switch over all our classical to quantum.

What would your advice be for companies on the software side wanting to embrace quantum?

If you're an engineer building bridges and doing stress load analysis it is important to learn and understand things at a basic level. You take a class to learn how to do this by hand then use AutoCAD, an industry-relevant tool that's going to do all that for you. You don't do that by hand.

I see the same thing in quantum. It’s important to get a basic understanding of what's happening on the back end, the same way that computer scientists take one course on Boolean logic. You're never going to need to use that unless you're building fundamental components for computers.

Most programmers have learned C++ from then on. I say use Qiskit [IBM’s open-source software development kit] or PennyLane AI [a Python library for quantum machine learning], two excellent learning tools.

Let's get away from just the fundamental learning of quantum and let's become users of this great technology, for people in industry. They want to use it so let's give them a tool and start developing the skills necessary to take advantage of that.