One of the questions we hear occasionally asked at various quantum conferences is the following: If someone were to place online tomorrow a 1 Million qubit quantum computer, would anyone be able to start using it in the near term for a production application?
In the past, the answer was No. This was based upon the thinking that quantum software and algorithms are completely different from classical ones and you cannot just recompile a program and have it run faster on a quantum computer. In the classical world, for example, one might be able to change their computer from an AMD 2GHz microprocessor to an Intel 3 GHz microprocessor (or vice-versa) to get an immediate boost of speed. But with quantum, you can’t do anything like that because the basic algorithms are completely different.
But the situation is starting to change. Within the past year or two, we’ve seen an increasing number of efforts to make quantum computing capabilities accessible to people who may be called subject matter experts, but are not skilled in quantum technology or quantum programming. Recently, we’ve seen a number of companies developing offerings for people in this category. The concept is that a subject matter expert or data scientist will describe their problem at a high level and issue a command asking to the computer to provide a solution and then the answer pops out. No need to convert the problem to a gate level program or study quantum mechanics, quantum specific algorithms, error mitigation, hardware backends, or the finer details of quantum programming languages. The software will do it for you.
IBM has a term for this which they call Frictionless Quantum Computing. And this concept would extend not only to the quantum portion itself, but also to include the classical computing resources and workflow management for any hybrid algorithm that requires a quantum computer and a classical computer to work together to provide the solution.
The benefit of this approach to a user is that they can leverage quantum computing to get a solution to their problem much more quickly than they would otherwise. Their data scientists can spend more time on efforts to understand and describe their problem and what they want to achieve instead of figuring out how to program a quantum computer. This will also help alleviate a potential bottleneck and recruiting issues of a limited quantum trained talent pool which could prevent an enterprise from utilizing quantum technology. And the quantum providers benefit because they will get more customers, bring them on board faster and achieve a quicker revenue growth.
Here are a few examples of software companies that are pursuing this approach.
QC Ware is a software company based in Palo Alto, California that has invested in research to develop proprietary algorithms that can provide unique turnkey solutions in areas such as binary optimization, machine learning, linear algebra and chemistry simulation. These solutions are designed so that end users can be up and running quickly with a minimal amount of computer code needed to invoke the algorithm. They have included these algorithms in their Forge software which they are now offering as SaaS (Software as a Service) with Amazon’s Braket and the machines Braket supports including Rigetti, D-Wave, and IonQ. In addition, the software can flexibly support other backends including the IBM Q systems, classical computers, and various simulators.
Horizon Quantum Computing is taking a unique approach. They are developing software that can take a problem written in classical computer languages and convert it to run on a quantum machine.
Classiq is an Israeli company that is building a layer of the quantum software stack to bring automation and synthesis to the quantum algorithm design process.
Although this concept of software to enable utilization of quantum computing for people who are not steeped in the technology is a relatively new one in the quantum field, it is following a concept that has long been followed in the classical computing field. After all, how many of us are still programming spreadsheets in assembly code instead of using something like Excel.
June 2, 2021