At the dawn of the quantum era, we are seeing more and more discussions as well as publications on the new and insightful demonstrations of quantum algorithms and quantum solutions, that have the potential to outperform classical methods. Quantum computing in particular has a lot of untapped potential but is expected to be the “late bloomer” out of the 3 main quantum domains of computing, communication, and sensing, having realistic industrial perspectives in 5 to 10 years’ time. Nevertheless, it can be potentially game-changing or even disruptive for certain Intelligent Industries.
The journey starts in the Capgemini Quantum Lab
Recognising the potentially disruptive impacts of quantum technologies on certain industrial and IT domains, Capgemini officially launched its own Quantum Lab earlier this year. The Lab brings together different local teams, research facilities and projects in quantum fields into a virtual global research laboratory. This aggregation of experts and specialists from various industry domains as well as different technologies within the Group – all of whom are dedicated to quantum – provides Capgemini with the means to help our clients jumpstart their quantum journey.
The vision of Capgemini’s Quantum Lab is:
- To become a leading partner to our clients for applied quantum research and application of quantum technologies.
- To accelerate quantum readiness for our clients.
- To achieve industry leadership with timely investments and through our ever-growing partnership network.
- To create intellectual capital within the group by training and attracting talent.
The Lab is focused on all 3 quantum technology application fields:
- Quantum Computing: leveraging Capgemini’s partner ecosystem as well as Capgemini’s strong computational experience and capabilities.
- Quantum Communication: leveraging our cybersecurity center in Utrecht as well as two world-class quantum communication laboratories in Portugal and in Cambridge.
- Quantum Sensing: leveraging Capgemini’s strong global research and engineering capabilities as well as software competencies.
The potential of Quantum Computing
Quantum computing technology is on the verge of reaching sufficient maturity for consideration in more complex algorithmic use cases. We are working with our clients in discovery sessions to pinpoint high-value use cases for their respective sectors. Clients are enthusiastic about working on complex optimisation problems where quantum could, we believe, bring a huge computational advantage.
However, as is often the case, the devil is in the details. A deeper understanding of the different use cases is essential in determining the proper extent of the quantum domain, as well as taking into consideration its limitations and potential drawbacks.
Current key opportunities are concentrated on 3 particular fields: optimisation, simulation, and machine learning.
1. Complex optimisation
As our recent Capgemini Research Institute report Quantum technologies: How to prepare your organisation for a quantum advantage now revealed, a broad scope of problems and business cases fall under this category:
- Energy: smart grids, oil well and wind farm optimisation, supply chain optimisation, etc.
- Financial services: risk management, dynamic portfolio management, derivative pricing, fraud identification, etc.
- Life Sciences: clinical trial optimisation, personalized medication, radiotherapy optimisation, medical diagnosis, etc.
- Manufacturing: traffic simulation, flow management, control, design optimisation, routing, scheduling, search algorithms, fleet/crew/resource optimisation, cargo loading management, supply chain optimisation, etc.
Optimisation problems (be it combinatorial or continuous) are typically difficult problems from a theoretical point of view. They are either challenging – or in some cases unsolvable – with the available (or even future) classical hardware. High complexity and high dimensional optimisation problems are typically good candidates for hybrid, quantum-inspired and purely quantum approaches.
As Richard Feynman once said: “Nature isn’t classical, dammit, and if you want to make a simulation of nature, you’d better make it quantum mechanical, and by golly it’s a wonderful problem, because it doesn’t look so easy.”
Life sciences with the modelling of chemical reactions, molecular simulations, protein folding, drug discovery challenges are, by definition, problems on the quantum scale. Therefore, their implementation and resolution on a quantum computer provides for a scientifically more natural choice, perhaps you could even say a more natural language, and – in a lot of cases – computationally, it is a more beneficial solution.
Other fields such as signal treatment, photonics and sensing algorithms can also benefit from a more precise resolution of certain mathematical equations through quantum algorithms, opening up the playing field to some more unexpected relevant use cases.
3.Quantum Machine Learning
Current machine learning methods are solvable but generally computationally intensive. Problems stemming from deep learning and computer vision, instances of resource allocation and product design, are some typical examples where quantum can provide a significant gain in speed-up and performance as well.
It has been hypothesized for the past few years that quantum provides the means of more expressible data, especially in higher-dimensional or more convoluted datasets.
A spotlight on the QUASAR project
According to Airbus, the conception, development, and production of current airplanes is the result of decades of multidisciplinary investigation and countless improvements. From a shape and structure optimisation point of view, the granularity of the problem and the extensive number of backend numerical and structural computations push even the highest performing classical machines to their limits. This can cause long design lead times, convoluted processes and conservative assessments. Quantum computing offers an alternative path to explore a wider design space by evaluating different parameters simultaneously, thus preserving structural integrity while optimising weight.
To address 5 flight physics problems, Airbus launched in 2019 the Airbus Quantum Challenge, asking academia and technology industries to propose specific solutions with the help of quantum technology. Partnering with quantum technology experts, such as their investment in IBM Quantum Network and QC Ware, Airbus is at the forefront in ushering in new possible applications for quantum computing in the aerospace domain. The 5 distinct flight physics problems were:
- Aircraft Climb Optimisation
- Computational Fluid Dynamics for aerodynamics simulations
- Quantum Neural Networks for Solving Partial Differential Equations for aerodynamics
- Aircraft Loading Optimisation
- Wingbox Design Optimisation
The proposed solution
Given the current technological state of quantum computing, the most viable route for considering such complex and potentially high-dimensional problems is by means of hybrid algorithms. This entails replacing a computationally intensive, iterated step of the whole optimisation process with a quantum iteration instead, therefore potentially improving the overall performance.
On the French R&D platform of Capgemini Engineering Research & Development, the current ongoing QUASAR R&D project, short for Quantum Algorithms for Structural Analysis Research, aims to address specifically this solution for finding the optimal design of the wings of an airplane, where weight optimisation is key to low operating costs and reduced environmental impact.
The wingbox optimisation problem can be summarized in the following 5 step process:
- Model definition and data preparation: discretization of the wingbox model, description of the physical model of the current wingbox design, identification of the physical parameter space,
- Structural analysis of a given design: it comes down to solving the linear system of static equilibrium on a Finite Element Grid, a process that we implemented in a quantum computer via a novel reinforced Harrow-Hassidim-Lloyd (HHL) algorithm as well as an adapted variational quantum linear solver (VQLS),
- Classification of a given design based on the structural integrity criteria,
- Exploring the parameter space by means of a trained quantum-enhanced support vector machine (QSVM) to identify other potential structurally sound designs,
- Searching for the minimal weight among successful designs.
In order to optimise the performance, a thorough analysis and a priori implementation is necessary for the main quantum algorithms (HHL/VQLS and QSVM) in the process. A key element in a hybrid approach is the transmission of information between a classical and a quantum component of the problem.
Some initial results
As of today, hybrid algorithms are relatively underrepresented in the scientific literature even though they are one of the most promising candidates in terms of implementation and applicability. The question whether they can lead to an actual quantum advantage is yet to be definitively answered. Nevertheless, with the current state of the technology, they are the main means of exploitation of more complex quantum algorithms in industrial use cases. Our preliminary investigations and performance analysis on the implementation of the two key quantum algorithm groups have already highlighted some key factors.
An a priori eigenvalue analysis (by means of a Variational Quantum Eigensolver, VQE) allows for a close to optimal parametrization of the Quantum Phase Estimation necessary for the HHL algorithm, thus improving considerably its performance. On the other hand, the VQLS algorithm, albeit being already well optimised, is disappointingly underperforming due to the heavy emphasis on the complex (classical) optimisation steps.
Quantum circuit representation: an example of a structured quantum circuit (top) vs. an example of a random, unstructured quantum circuit (bottom)
QSVM performance analysis: performance of a QSVM for random quantum circuits of different size and depth (left) vs. performance of a QSVM for simple structured quantum circuits
It has been observed on multiple data structures that a certain degree of entanglement in the quantum circuit is positively beneficial to the training performance of a QSVM. However, the influence of the structure of the quantum circuit representing the feature map on the performance is not clearly defined and is yet to be properly quantified.
Quantum technologies are developing at a significant rate and are predicted to be a highly disruptive and ultimately beneficial in helping to solve previously intractable problems for industry, including aerospace, and even society. It is our belief that despite the nascent state of the technology, now is the best time:
- To identify the relevant use cases and prove their feasibility.
- To gather the relevant skills and talents.
- To deploy a dedicated organisation capable of leveraging all the quantum value.
To read more blogs in the Intelligent Industry: Journey to Farnborough International Airshow series, see quick links below:
A Quantum of Intelligent Industry – Mike Dwyer considers the potential impact that the world of quantum computing, sensing and communication could have on our ability to create new intelligent products and services.
Innovation at Speed: What Intelligent Industry can learn from Formula One’s data driven innovation – Ashish Padhi delves into the data driven rapid innovation process of Formula One aerodynamic design to prise out lessons for Intelligent Industry.
Enabling Digital Twins with Systems Engineering – Adam Lancaster & Scott Reid explore how to enable Digital Twins across the full lifecycle, using Systems Engineering techniques.
How the advent of advanced air mobility will pave the way for more connected and sustainable aviation – Gianmarco Scalabrin explores how advanced air mobility is ready for prime time and will play a crucial role in connecting communities while helping aviation drastically reduce its CO2 emissions.
R&D Project Lead, Capgemini Engineering Research & Development France
Quantum Lab Deputy Lead France
Krisztian is the project lead of a quantum computing project on the R&D Platform of Capgemini Engineering. He has a Ph.D. in mathematical physics, and he is quantum deputy lead in France for the Capgemini Quantum Lab, as well as a global contact point for quantum related matters in the aerospace and defense industries.