Universal Quantum Intermediate Representation

Quantum computing offers exciting promises regarding how it can benefit society. However, it is also still in its infancy. In many ways today’s environment is similar to the state of classical computing in the early 1950s: each system is different, and capabilities and resources are highly limited. We face several major technical challenges throughout the stack that yet need to be overcome, and it is not obvious which hardware technologies are likely to succeed and which applications will benefit society. To advance the field of quantum computing we need to make it possible to easily connect emerging technologies, and nurture the growth of an ecosystem that facilitates to develop and experiment with different approaches and analyze potential applications.
To that end, we present an LLVM-based Quantum Intermediate Representation (QIR) intended to serve as a universal interface between programming languages and targeted backends. Our goal is to facilitate the development of a common infrastructure including advanced optimizations and code transformations that take full advantage of the classical compute power and tools at our disposal in addition to quantum resources. In the coming years, we expect there to be exciting progress in how classical and quantum computations can interact at the hardware level. With QIR we provide a foundation for software frameworks that can target today’s devices as well as more powerful systems of the future.