Standardised Quanten Software Stack
The development of quantum algorithms with a potentially exponential speed advantage over classical algorithms has sparked widespread interest in industry and science. Advances in hardware development in recent years have shown that this potential can be exploited. However, there is still a large gap between accessible and user-friendly programming, as known from classical computer science, and quantum programming, which is often still on an experimental stage. To close this gap, a higher-level quantum programming language that goes beyond the currently established assembler-like languages and that automates many of the small-step elements is being developed as part of the Qompiler project. Matching the quantum programming language, a compiler will be developed to allow optimization of the generated quantum circuits and ideal integration of hybrid algorithms. Via a firmware for an ion-based quantum computer and an interface to the higher software levels, both of which are being developed in the current Qompiler project, the generated and compiled circuits can then also be executed on a German quantum computer. Towards the end of the Qompiler project, its results, in particular the developed interface between firmware and compiler, will be transferred into standardization activities.
Challenge and innovation
The state of development of quantum computing is still at an early stage, even beyond the well-known open questions about the hardware. Everyday programming with the established interfaces turns out to be repetitive and small-step, which makes for a high barrier to entry and susceptibility to errors. In order for quantum computing to offer added value to as many companies as possible in the long term, it is of the highest priority (in addition to hardware develop-ment) to also enable non-quantum physicists to work with quantum computers. In this respect, an easily accessible higher-level quantum programming language based on established paradigms is an important step in this direction. Many of the intended functions do not consist of particularly complex approaches, but are based on the principle of automating as much programming work as possible. In cases where complexity is nevertheless necessary, it is to be hidden from the user/programmer.
Furthermore, quantum compilers and quantum firmware currently only exist as proprietary approaches of individual vendors, and uniform interfaces are not yet available. This bears the risk of vendor lock-in. By developing a quantum firmware for an ion-based computer with a standardized interface (targeted DIN SPEC) and the quantum compiler, it will be possible for users to access a quantum computer developed in Germany.
In the Qompiler project, several levels of the quantum software stack are developed in parallel. The lowest develop-ment level in the project is the firmware for an ion-based quantum computer. A hardware-specific compiler takes care of the translation of general quantum circuits into the natively available gate set and ensures that the advantages of the ion-based quantum computer are fully exploited. Based on this, the Qompiler project defines and develops an interface to address the higher software levels. These consist of the categorical level and a higher quantum programming language. Features of the programming language envisioned in this project include: Automated Memory Management, Automated Gate Error Correction, and Automated Uncomputation (i.e., Garbage Collection). For everyday programming, the language is also integrated into existing development environments. Directly connected to the functional aspect of the language is the categorical level CAP of the University of Siegen, which enables a check of the type consistency - i.e. a verification of the programs before the actual execution within a certain framework - and thus an optimization of the created quantum circuits. Promising hybrid algorithms are connected across the stack by using benchmarking protocols to determine an optimal allocation between classical and quantum resources. Thus, the full range of application potential of hybrid algorithms, for example for molecular simulations or optimization problems, will become available.
Fraunhofer FOKUS (project lead), eleQtron GmbH, Universität Siegen, TU Berlin, Deutsches Institut für Normung e.V. (subcontract)
January 2022 – December 2024
Total costs: € 2,4 million
Funding volume: € 1,9 million