Nikolaus Hansen
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Nikolaus Hansen has a total of 91 co-authors. Here are the top co-authors by number of shared publications:
Anne Auger
54 shared publicationsDimo Brockhoff
28 shared publicationsYouhei Akimoto
University of Tsukuba, Japan
9 shared publications- Petros Koumoutsakos
School of Engineering and Applied Sciences, Harvard University, Cambridge, USA
8 shared publications - Youhei Akimoto
University of Tsukuba
8 shared publications - Anne Auger
Institute for Neural Computation CMAP INRIA Saclay Ecole Polytechnique Route de Saclay Ile-de-France France
7 shared publications
Latest Publications
- 25 August 2026
- Published by Springer Nature
- p. 136-151
Reinforcement Learning for Solving Multimodal Functions with CMA-ES
- 25 August 2026
- Published by Springer Nature
- p. 136-151
Reinforcement Learning for Solving Multimodal Functions with CMA-ES
- Open Access
- 13 August 2026
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Evolutionary Learning and Optimization
dd-RCMA-ES: CMA-ES with Projection-Based Restricted Covariance Matrix Adaptation for High-Dimensional Problems
This paper addresses the computational challenges of the covariance matrix adaptation evolution strategy (CMA-ES) in high-dimensional optimization of an objective function \(f:\mathbb{R}^{n}\to\mathbb{R}\) where \(n\gg 100\) , particularly for problems exhibiting sparse-sensitivity structures where only a small number of directions are highly sensitive. To overcome the bottleneck associated with the covariance matrix, we propose a novel restricted covariance matrix model, reducing storage requirements from quadratic to linear in \(n\) . An efficient adaptation mechanism for this model is introduced, with linear time complexity per iteration, providing a substantial improvement over the standard CMA-ES. We also present two-point step-size adaptation (TPA) as an alternative to the cumulative step-size adaptation (CSA), particularly suitable for high-dimensional optimization. In contrast to CSA, TPA is not negatively affected by ineffective subspaces, prevalent in higher dimensions where non-relevant design variables may be introduced. The proposed algorithm, dd-RCMA-ES, determines its hyper-parameters based on the problem dimension \(n\) and the model complexity \(k\) , thereby maintaining the out-of-the-box usability of the standard CMA-ES, while offering the flexibility to upper bound the model complexity if desired. Experimental evaluations across 18 test problems in dimensions varying from 80 to 10240 demonstrate the efficiency of dd-RCMA-ES in terms of the number of function calls required to achieve a given target precision. In comparison with other CMA-ES variants, the dd-RCMA-ES excels on problems with sparse structures that comprise a small number of both, highly sensitive and insensitive directions.- 13 August 2026
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Evolutionary Learning and Optimization
Open Accessdd-RCMA-ES: CMA-ES with Projection-Based Restricted Covariance Matrix Adaptation for High-Dimensional Problems
This paper addresses the computational challenges of the covariance matrix adaptation evolution strategy (CMA-ES) in high-dimensional optimization of an objective function \(f:\mathbb{R}^{n}\to\mathbb{R}\) where \(n\gg 100\) , particularly for problems exhibiting sparse-sensitivity structures where only a small number of directions are highly sensitive. To overcome the bottleneck associated with the covariance matrix, we propose a novel restricted covariance matrix model, reducing storage requirements from quadratic to linear in \(n\) . An efficient adaptation mechanism for this model is introduced, with linear time complexity per iteration, providing a substantial improvement over the standard CMA-ES. We also present two-point step-size adaptation (TPA) as an alternative to the cumulative step-size adaptation (CSA), particularly suitable for high-dimensional optimization. In contrast to CSA, TPA is not negatively affected by ineffective subspaces, prevalent in higher dimensions where non-relevant design variables may be introduced. The proposed algorithm, dd-RCMA-ES, determines its hyper-parameters based on the problem dimension \(n\) and the model complexity \(k\) , thereby maintaining the out-of-the-box usability of the standard CMA-ES, while offering the flexibility to upper bound the model complexity if desired. Experimental evaluations across 18 test problems in dimensions varying from 80 to 10240 demonstrate the efficiency of dd-RCMA-ES in terms of the number of function calls required to achieve a given target precision. In comparison with other CMA-ES variants, the dd-RCMA-ES excels on problems with sparse structures that comprise a small number of both, highly sensitive and insensitive directions. - Open Access
- 14 July 2025
- Published by Association for Computing Machinery (ACM)
- p. 679-682
Rank-based Linear-Quadratic Surrogate Assisted CMA-ES
- 14 July 2025
- Published by Association for Computing Machinery (ACM)
- p. 679-682
Open AccessRank-based Linear-Quadratic Surrogate Assisted CMA-ES
- Open Access
- 13 July 2025
- Published by Association for Computing Machinery (ACM)
- p. 728-736
Classification-Based Linear Surrogate Modeling of Constraints for AL-CMA-ES
- 13 July 2025
- Published by Association for Computing Machinery (ACM)
- p. 728-736
Open AccessClassification-Based Linear Surrogate Modeling of Constraints for AL-CMA-ES
- 7 September 2024
- Published by Springer Nature
- p. 284-299
LB+IC-CMA-ES: Two Simple Modifications of CMA-ES to Handle Mixed-Integer Problems
- 7 September 2024
- Published by Springer Nature
- p. 284-299
LB+IC-CMA-ES: Two Simple Modifications of CMA-ES to Handle Mixed-Integer Problems
- 2 December 2023
- Published by Elsevier in Applied Mathematics Letters
- Vol. 150
Asymptotic estimations of a perturbed symmetric eigenproblem
- 2 December 2023
- Published by Elsevier in Applied Mathematics Letters
- Vol. 150
Asymptotic estimations of a perturbed symmetric eigenproblem
- Open Access
- 15 July 2023
- Published by Association for Computing Machinery (ACM)
- p. 854-877
An Introduction to Scientific Experimentation and Benchmarking
- 15 July 2023
- Published by Association for Computing Machinery (ACM)
- p. 854-877
Open AccessAn Introduction to Scientific Experimentation and Benchmarking
- Open Access
- 15 July 2023
- Published by Association for Computing Machinery (ACM)
- p. 1157-1182
CMA-ES and Advanced Adaptation Mechanisms
- 15 July 2023
- Published by Association for Computing Machinery (ACM)
- p. 1157-1182
Open AccessCMA-ES and Advanced Adaptation Mechanisms
- Open Access
- 15 July 2023
- Published by Association for Computing Machinery (ACM)
- p. 1628-1635
Benchmarking CMA-ES with Basic Integer Handling on a Mixed-Integer Test Problem Suite
We compare the performances of one implementation of CMA-ES (pycma version 3.3.0) for optimizing functions with both continuous and integer variables. The implementation incorporates a lower bound on the variance along the integer coordinates to keep the optimization from stalling. This benchmark will serve as a baseline for further works on pycma. Results show substantial improvement since the last benchmarked version of pycma. Also this implementation is competitive with other mixed integer algorithms.
- 15 July 2023
- Published by Association for Computing Machinery (ACM)
- p. 1628-1635
Open AccessBenchmarking CMA-ES with Basic Integer Handling on a Mixed-Integer Test Problem Suite
We compare the performances of one implementation of CMA-ES (pycma version 3.3.0) for optimizing functions with both continuous and integer variables. The implementation incorporates a lower bound on the variance along the integer coordinates to keep the optimization from stalling. This benchmark will serve as a baseline for further works on pycma. Results show substantial improvement since the last benchmarked version of pycma. Also this implementation is competitive with other mixed integer algorithms.
- Open Access
- 12 July 2023
- Published by Association for Computing Machinery (ACM)
- p. 652-660
Multiobjective Optimization with a Quadratic Surrogate-assisted CMA-ES
We present a surrogate-assisted multiobjective optimization algorithm. The aggregation of the objectives relies on the Uncrowded Hypervolume Improvement (UHVI) which is partly replaced by a linear-quadratic surrogate that is integrated into the CMA-ES algorithm. Surrogating the UHVI poses two challenges. First, the UHVI is a dynamic function, changing with the empirical Pareto set. Second, it is a composite function, defined differently for dominated and nondominated points. The presented algorithm is thought to be used with expensive functions of moderate dimension (up to about 50) with a quadratic surrogate which is updated based on its ranking ability. We report numerical experiments which include tests on the COCO benchmark. The algorithm shows in particular linear convergence on the double sphere function with a convergence rate that is 6--20 times faster than without surrogate assistance.
- 12 July 2023
- Published by Association for Computing Machinery (ACM)
- p. 652-660
Open AccessMultiobjective Optimization with a Quadratic Surrogate-assisted CMA-ES
We present a surrogate-assisted multiobjective optimization algorithm. The aggregation of the objectives relies on the Uncrowded Hypervolume Improvement (UHVI) which is partly replaced by a linear-quadratic surrogate that is integrated into the CMA-ES algorithm. Surrogating the UHVI poses two challenges. First, the UHVI is a dynamic function, changing with the empirical Pareto set. Second, it is a composite function, defined differently for dominated and nondominated points. The presented algorithm is thought to be used with expensive functions of moderate dimension (up to about 50) with a quadratic surrogate which is updated based on its ranking ability. We report numerical experiments which include tests on the COCO benchmark. The algorithm shows in particular linear convergence on the double sphere function with a convergence rate that is 6--20 times faster than without surrogate assistance.