Verifying and Validating Mission-Critical Systems Using Quantum Computing
Created March 2019
As software engineering challenges become more complex, we are finding that existing computing paradigms are unable to offer the solutions we’re looking for. At the SEI, we are studying quantum computing to see if it can offer some of these solutions. Among our aims is to make components such as aircraft flight controls function in mission-critical time.
The Need for New Computing Capabilities
Over the past five decades, the integrated circuit computing paradigm has powered many technological breakthroughs. However, the computing power we’re able to fit into computer chips is beginning to reach a limit, and the hardest problems in fields like software verification and validation (VV), or machine learning and artificial intelligence (ML and AI), can’t be solved by existing technology.
The challenging software engineering problems these fields face are combinatorial optimization problems. For many combinatorial optimization problems, finding the exact optimal solution is non-deterministic polynomial hard (NP-hard), which means that each of these problems could take billions of years to solve using classical computing paradigms.
Some examples of these problems involve complicated software components of aircraft flight controls. For example, Lockheed Martin recently published code that is used to control an aircraft safety mechanism. The purpose of the code is to automate evasive maneuvers if the aircraft is in danger. However, because of the complexity of the code and the maneuvers it controls, classical computers are unable to verify and validate that the code is safe. To solve these problems in mission-critical time for the sake of safety and mission success, we need a new paradigm in optimization algorithms and computational capabilities.
Achieving Quantum Advantage
At the SEI, we are investigating whether quantum algorithms and computers can serve as this next paradigm for optimization in applications like software VV. To do so, we are working toward quantum advantage: a clear demonstration of a quantum computer solving a problem of practical interest faster than a classical computer.
We are investigating near-term optimization algorithms that can run effectively on NISQ QPUs, like Variational Quantum Eigensolver (VQE) and Quantum Approximation Optimization Algorithm (QAOA). Currently, we are focusing on
- benchmarking Variational Quantum Optimization techniques, such as QAOA, and their ability to tolerate NISQ-era QPUs
- improving circuit generation for NISQ-era QPUs
- analyzing the hierarchy of the problems of interest and identifying which parts can be mapped effectively to QPUs
- addressing the challenges of scaling up to O(102-103) qubits as well as predicting and projecting quantum advantage
- developing software tools to help data scientists and engineers use quantum computers
As we continute to explore these problems, we plan to extend our work to new applications, including studying quantum machine learning involving quantum algorithms to perform machine learning and artificial intelligence tasks. In addition, we plan to work on quantum interactive proof systems, using QPUs to form interactive proof systems, and verifying and validating quantum computation.
We hope that this work will give us better ways to verify the solution to complex problems more quickly, and to be able to confidently confirm that problems that are outsourced to quantum computing services were actually solved using quantum computers. Our work will also make it possible to use a quantum computer as a training kernel or classification model for machine learning so that we can train algorithms and classifications much more quickly than previously possible.
March 05, 2019 Fact Sheet
In 2018, the SEI began work to understand and demonstrate how quantum computing can be applied to DoD missions such as machine learning and artificial intelligence and software verification and validation.read