quantum optimization

The Path To Quantum Optimization

Rebel Brown

First –  let’s all recognize that quantum optimization is not just around the corner. Yes, you can run some problems today. But quantum computers aren’t yet ready to process real-world business problems, such as logistics or supply chain optimization. The hardware needs to scale and deliver more reliable results (e.g., we need error correction for quantum)  before we can approach a production-ready quantum processing world.

We at QCI doubt that supply chains will ever run their computations on pure quantum – the data volumes would require massive quantum hardware that would be cost prohibitive.

Even when we do scale quantum systems, we doubt that pure quantum systems will be the best answer for Supply Chain and Logistics optimization. Why? Here’s one example.

Classical computers are better at running the search-based computations that find the best answers to these problems. Especially for what are known as ”sparse” problems where the captured data doesn’t fill a majority of the variable fields. For example, if you have a large set of options with each product only checking the box for a few, your data is sparse. If you have a large set of options with each product checking the box for many of them, your data is more dense.

Quantum computers are able to process dense data sets and constrained optimization problems much more effectively and accurately than classical systems. The challenge today is that quantum computers aren’t scalable enough to process large data sets. Since quantum computers ingest only a single variable for each qubit, the data volumes a quantum system can process are limited by qubit count. That’s why we have to wait for hardware scale (as well as error correction) before production-ready quantum computing becomes a reality.

What is My Best Approach to Quantum Today?

Despite the above limitations, you can start reaping the benefits of quantum power before we’re able to scale to full production quantum. And you can do so without spending enormous time and money to write new, complex quantum computing programs. 

Right now, there’s an opportunity to apply quantum techniques to classical processing to accelerate performance, increase accuracy and provide a diversity of answers that solve the problems that drive optimal business decisions. This means you can begin to leverage software that offers computational solutions leveraging quantum on your classical systems, without having to re-program your problems and requests.

There’s also an opportunity now to evaluate quantum hardware vendors and their software development kits to decide if you want to invest in the expensive resources to program custom quantum algorithms and low level coding for the distant future. As with prior computing platforms and IT infrastructure, you need to make a “build vs buy” choice when it comes to quantum software.

  • Do you want to work with a software vendor who has developed ready-to-run software that delivers the results you need, without the complexity and expense of quantum programming?
  • Or do you want to hire the quantum expertise and consultants you’ll need to write your own software from the hardware up?

It’s a big decision with long term ramifications. The costs and time required to develop and maintain quantum programs are  significant so be sure to gather a full understanding of the impacts.

What is the Most Likely Path to Quantum Optimization?

We’ve already discussed applying quantum techniques to classical systems to get better results for optimization problems or other computations.

The next and most likely use of quantum computers to process true production optimization problems comes with hybrid computing. That’s the blending of classical and quantum computers, working together to solve complex computations, provide powerful simulations and solve new problems we can’t even solve today.

Imagine being able to send a request to solve a problem, with an intelligence that sends the appropriate pieces of the problem to quantum systems (aka the really complex subproblems) while sending the rest of the subproblems to classical systems?  Your problem is seamlessly solved in parallel across multiple instances of quantum and classical systems, and you get more accurate and more diverse results, faster and with more certainty that you can actually get a valid result. Because you get multiple results that meet your problem constraints, you get to choose what best fits the needs of your specific situation at any point in time.

Hybrids promise to solve complex problems and perform sophisticated simulations in the nearer term than pure quantum.

Pure quantum computing will become a reality when hardware scales and error-correction is in place for these larger systems. But you won’t need to rely on pure quantum to solve your optimization problems. You’ll be able to solve them today using quantum-inspired classical. Tomorrow, you’ll find ever better solutions using hybrid architectures.

Quantum optimization is not a question of “if.” It’s a question of “when and how.”