News

Terra Quantum and Evonik Successfully Complete Geometry Optimization Project

Written by Terra Quantum AG | Jan 13, 2025 4:22:33 PM
  • Terra Quantum successfully deployed its TQOptimaX multi-objective optimization solver, incorporating TetraOpt, as a tensor networks-based solver for black-box optimization to address Evonik's complex geometry optimization tasks within simulation processes.
  • The performance of Terra Quantum's solutions was largely comparable to Evonik's state-of-the-art production solver, demonstrating the viability of quantum-inspired technologies in real-world industrial applications.
  • This project showcases a practical, production-ready application of quantum-inspired technologies in industrial settings, moving beyond theoretical use cases and highlighting the potential for wider adoption in various industries.

We are pleased to announce the successful completion of a collaborative project between Terra Quantum and Evonik, focusing on geometry optimization tasks modelled with CFD-simulation. This project showcases the practical application of quantum-inspired technologies in addressing complex industrial challenges.

Innovative Optimization Approach

The project utilized Terra Quantum’s innovative multi-objective optimization solver, leveraging tensor networks-based techniques.

 TQOptimaX: Multi-objective Optimization Solver

At the heart of our geometry optimization approach lies TQOptimaX, our advanced multi-objective optimization solver, designed to tackle complex industrial design problems. It efficiently maximizes the hypervolume in solution-space while utilizing as little as possible number of target function evaluations, crucial for reducing computational costs. TQOptimaX handles non-linear constraints effectively and produces a well-distributed Pareto front, providing a comprehensive view of optimal trade-offs. Its parallelizable structure allows for simultaneous evaluation of multiple solution candidates, accelerating the optimization process. By integrating our black-box tensor network-based solver, TetraOpt, we deliver robust performance in real-world engineering scenarios.

 

 TetraOpt: Tensor Networks-based Solver for Universal Black-Box Optimization

Complementing TQOptimaX is TetraOpt, our proprietary tensor train-based optimizer, designed for complex, high-dimensional problems where target function evaluations are limited. It efficiently reaches lower cost function values compared to alternative algorithms, making it particularly valuable in industrial settings with time and resource constraints. Built on tensor network mathematics, TetraOpt navigates high-dimensional spaces more effectively than traditional methods. Its highly parallelizable nature and straightforward implementation make it a powerful tool for demanding industrial optimization tasks, enabling solutions to problems previously considered intractable.

Tensor networks, a crucial component in the quantum computing stack, offer efficient methods for representing and processing complex, high-dimensional data. This approach not only provides powerful capabilities with current technology but also positions the solution to leverage future advancements in quantum computing hardware, as they can also be mapped onto quantum circuits, enabling execution and improved performance, leveraging large scale, fault tolerant quantum computing hardware.

 

Real-World Performance

The true test of any technology is its performance in real-world scenarios. As part of the project, TQOptimaX was deployed in the customer’s environment to benchmark an internal use case provided by Evonik. The results were significant, with our solution performing largely comparably with Evonik's state-of-the-art production solver. This achievement demonstrates that tensor network-powered solutions can effectively address industrial, production-grade challenges, moving beyond theoretical applications. 

Markus Pflitsch, CEO and Founder of Terra Quantum, commented on the project's significance: "This collaboration showcases the practical potential of quantum technologies for industrial applications. With TQOptimaX, we are challenging limitations in optimization and demonstrating the value of quantum computing for real-world use cases."

 

Advancing Industrial Optimization 

The successful implementation at Evonik represents a notable step in applying quantum-inspired technologies to industry challenges. Henrik Hahn, CDO of Evonik, shared his perspective: "Our work with Terra Quantum has proven that quantum computing solutions, such as TQOptimaX, can compete with the best solvers available. We look forward to exploring further applications of quantum computing in our operations as the technology continues to evolve."

 

Future Perspectives 

This collaboration between Terra Quantum and Evonik opens up new possibilities for industrial optimization. As quantum technologies advance, we anticipate further improvements in solving complex, multi-objective optimization problems across various industries.

We extend our appreciation to both the Terra Quantum and Evonik teams for their hard work and contributions to this project. This achievement not only demonstrates the current capabilities of quantum-inspired technologies but also sets the stage for future innovations in industrial optimization.

Stay tuned for more updates as we continue our work in applying quantum-inspired solutions to real-world industrial challenges. To learn more about Terra Quantum's tensor networks-based approach to advanced simulation, explore our TetraBox page.