Quantum software stack is ready for business, says IBM
Higher level functions and managed services are starting to lower the bar to entry
As perhaps the classic example of a "holy grail" technology, quantum computing is advancing on multiple very different fronts at the same time, with researchers pursuing alternative approaches to the task of stabilising qubits and reducing noise and errors.
The lack of standardised approaches, interfaces and protocols complicates the task of developing a software stack for nascent quantum computers, and error correction measures required to obtain meaningful results have proved hard to implement. Researchers commonly write quantum programs at the circuit level - the quantum equivalent of writing assembly language - which requires a deep understanding of the hardware at a very low level.
This means that so far meaningful applications for quantum computers have been few and far between and extremely specialised. Shor's algorithm, the feared approach which could in theory crack cryptosystems like RSA that rely on factorising very large numbers, has only been successful in factorising numbers up to 21, and that was a few years ago.
Using quantum and quantum-adjacent approaches, researchers have managed to factor much larger integers since then, but nothing like the 2048-bit numbers used by current RSA systems, and without any quantum advantage - meaning there's nothing gained over using classical methods.
For several reasons, the software stack has been a limiting factor in creating practical applications for quantum computers, but, insisted Blake Johnson, distinguished engineer and quantum engine lead at IBM, progress is being made. IBM, one of the names most associated with the field, has been pumping considerable resources into this area, he said.
"At the same time as our team has had great success advancing the quality of our hardware, we have also been heavily investing in improving our software stack," Johnson told Computing.
"As our partners and users expand the complexity of the programs and algorithms they are studying in the quest to find quantum advantage, a highly performant and stable software stack is essential."
IBM launched Qiskit back in 2017, an open source software development kit (SDK) for IBM quantum computers and simulators that's based on Python, which (based on IBM benchmarks at least) is a top performer.
According to Johnson, Qiskit is also "enabling a broader and global base of industry experts to explore new algorithms and applications.
"We have recently expanded the Qiskit stack to include AI-powered services and tools, including the use generative AI to assist users in mapping their problems to quantum circuits and executing on hardware."
In addition, last month IBM launched the Qiskit Functions Catalog which, together with IBM’s cloud-based platform allows premium subscribers to access quantum hardware and software as a managed service, and make use of higher-level functions written by IBM others for use cases including chemistry and data science.
The combination of managed services and software provides "the performance of Qiskit with partner services that enhance the execution experience through advanced error mitigation and suppression, and also functions, providing easy to use domain-specific interfaces for chemistry and optimisation problems," said Johnson.
"We can’t wait to see what our users can do with these increasingly powerful tools as they push forward in their discovery of quantum algorithms and applications."