Gustaf Söderlind
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Gustaf Söderlind has a total of 23 co-authors. Here are the top co-authors by number of shared publications:
- Imre Fekete
MTA-ELTE Numerical Analysis and Large Networks Research Group, Budapest, Hungary
2 shared publications - Imre Fekete
Department of Applied Analysis and Computational Mathematics, Eötvös Loránd University, Budapest, Hungary
2 shared publications - Klas Modin
Department of Mathematical Sciences, Chalmers University of Technology and University of Gothenburg, Gothenburg, Sweden
2 shared publications - András Molnár
Department of Applied Analysis and Computational Mathematics, Institute of Mathematics, ELTE Eötvös Loránd University, Budapest, Hungary
1 shared publication - Carmen Arévalo
Centre for Mathematical Sciences, Lund University, Box 118, SE-221 00 Lund, Sweden
1 shared publication - Carmen Arévalo
Universidad Simón Bolívar
1 shared publication
Latest Publications
- Open Access
- 5 August 2025
- Published by MDPI AG in Algorithms
- Vol. 18 (8) , 484
Tolerance Proportionality and Computational Stability in Adaptive Parallel-in-Time Runge–Kutta Methods
In this paper, we investigate how adaptive time-integration strategies can be effectively combined with parallel-in-time numerical methods for solving systems of ordinary differential equations. Our focus is particularly on their influence on tolerance proportionality. We examine various grid-refinement strategies within the multigrid reduction-in-time (MGRIT) framework. Our results show that a simple adjustment to the original refinement factor can substantially improve computational stability and reliability. Through numerical experiments on standard test problems using the XBraid library, we demonstrate that parallel-in-time solutions closely match their sequential counterparts. Moreover, with the use of multiple processors, computing time can be significantly reduced.
- 5 August 2025
- Published by MDPI AG in Algorithms
- Vol. 18 (8) , 484
Open AccessTolerance Proportionality and Computational Stability in Adaptive Parallel-in-Time Runge–Kutta Methods
In this paper, we investigate how adaptive time-integration strategies can be effectively combined with parallel-in-time numerical methods for solving systems of ordinary differential equations. Our focus is particularly on their influence on tolerance proportionality. We examine various grid-refinement strategies within the multigrid reduction-in-time (MGRIT) framework. Our results show that a simple adjustment to the original refinement factor can substantially improve computational stability and reliability. Through numerical experiments on standard test problems using the XBraid library, we demonstrate that parallel-in-time solutions closely match their sequential counterparts. Moreover, with the use of multiple processors, computing time can be significantly reduced.
- Open Access
- 5 January 2023
- Published by Springer Nature in Periodica Mathematica Hungarica
- Vol. 87 (1) , 167-181
Runge–Kutta–Möbius methods
In the numerical integration of nonlinear autonomous initial value problems, the computational process depends on the step size scaled vector field hf as a distinct entity. This paper considers a parameterized transformation $$\begin{aligned} hf \mapsto hf \circ (I-\gamma hf)^{-1}, \end{aligned}$$ and its role in the finite step size stability of singly diagonally implicit Runge—Kutta (SDIRK) methods. For a suitably chosen $$\gamma > 0$$ , the transformed map is Lipschitz continuous with a reasonably small constant, whenever hf is negative monotone. With this transformation, an SDIRK method is equivalent to an explicit Runge–Kutta (ERK) method applied to the transformed vector field. Through this mapping, the SDIRK methods’ A-stability, and linear order conditions are investigated. The latter are closely related to approximations of the exponential function $$\textrm{e}^z$$ that are polynomial in z, when considering ERK methods, and polynomial in terms of the transformed variable $$z(1-\gamma z)^{-1}$$ , in case of SDIRK methods. Considering the second family of methods, and expanding the exponential function in terms of this transformed variable, the coefficients can be expressed in terms of Laguerre polynomials. Lastly, a family of methods is constructed using the transformed vector field, and its order conditions, A-stability, and B-stability are investigated.- 5 January 2023
- Published by Springer Nature in Periodica Mathematica Hungarica
- Vol. 87 (1) , 167-181
Open AccessRunge–Kutta–Möbius methods
In the numerical integration of nonlinear autonomous initial value problems, the computational process depends on the step size scaled vector field hf as a distinct entity. This paper considers a parameterized transformation $$\begin{aligned} hf \mapsto hf \circ (I-\gamma hf)^{-1}, \end{aligned}$$ and its role in the finite step size stability of singly diagonally implicit Runge—Kutta (SDIRK) methods. For a suitably chosen $$\gamma > 0$$ , the transformed map is Lipschitz continuous with a reasonably small constant, whenever hf is negative monotone. With this transformation, an SDIRK method is equivalent to an explicit Runge–Kutta (ERK) method applied to the transformed vector field. Through this mapping, the SDIRK methods’ A-stability, and linear order conditions are investigated. The latter are closely related to approximations of the exponential function $$\textrm{e}^z$$ that are polynomial in z, when considering ERK methods, and polynomial in terms of the transformed variable $$z(1-\gamma z)^{-1}$$ , in case of SDIRK methods. Considering the second family of methods, and expanding the exponential function in terms of this transformed variable, the coefficients can be expressed in terms of Laguerre polynomials. Lastly, a family of methods is constructed using the transformed vector field, and its order conditions, A-stability, and B-stability are investigated. - Open Access
- 5 June 2020
- Published by Springer Nature in Numerical Algorithms
- Vol. 86 (2) , 537-563
Local error estimation and step size control in adaptive linear multistep methods
In a k-step adaptive linear multistep methods the coefficients depend on the k − 1 most recent step size ratios. In a similar way, both the actual and the estimated local error will depend on these step ratios. The classical error model has been the asymptotic model, chp+ 1y(p+ 1)(t), based on the constant step size analysis, where all past step sizes simultaneously go to zero. This does not reflect actual computations with multistep methods, where the step size control selects the next step, based on error information from previously accepted steps and the recent step size history. In variable step size implementations the error model must therefore be dynamic and include past step ratios, even in the asymptotic regime. In this paper we derive dynamic asymptotic models of the local error and its estimator, and show how to use dynamically compensated step size controllers that keep the asymptotic local error near a prescribed tolerance tol. The new error models enable the use of controllers with enhanced stability, producing more regular step size sequences. Numerical examples illustrate the impact of dynamically compensated control, and that the proper choice of error estimator affects efficiency.
- 5 June 2020
- Published by Springer Nature in Numerical Algorithms
- Vol. 86 (2) , 537-563
Open AccessLocal error estimation and step size control in adaptive linear multistep methods
In a k-step adaptive linear multistep methods the coefficients depend on the k − 1 most recent step size ratios. In a similar way, both the actual and the estimated local error will depend on these step ratios. The classical error model has been the asymptotic model, chp+ 1y(p+ 1)(t), based on the constant step size analysis, where all past step sizes simultaneously go to zero. This does not reflect actual computations with multistep methods, where the step size control selects the next step, based on error information from previously accepted steps and the recent step size history. In variable step size implementations the error model must therefore be dynamic and include past step ratios, even in the asymptotic regime. In this paper we derive dynamic asymptotic models of the local error and its estimator, and show how to use dynamically compensated step size controllers that keep the asymptotic local error near a prescribed tolerance tol. The new error models enable the use of controllers with enhanced stability, producing more regular step size sequences. Numerical examples illustrate the impact of dynamically compensated control, and that the proper choice of error estimator affects efficiency.
- Open Access
- 10 July 2018
- Published by Springer Nature in BIT Numerical Mathematics
- Vol. 58 (4) , 1125-1143
On the zero-stability of multistep methods on smooth nonuniform grids
- 10 July 2018
- Published by Springer Nature in BIT Numerical Mathematics
- Vol. 58 (4) , 1125-1143
Open AccessOn the zero-stability of multistep methods on smooth nonuniform grids
- Open Access
On the zero-stability of multistep methods on smooth nonuniform grids
In order to be convergent, linear multistep methods must be zero stable. While constant step size theory was established in the 1950's, zero stability on nonuniform grids is less well understood. Here we investigate zero stability on compact intervals and smooth nonuniform grids. In practical computations, step size control can be implemented using smooth (small) step size changes. The resulting grid $\{t_n\}_{n=0}^N$ can be modeled as the image of an equidistant grid under a smooth deformation map, i.e., $t_n = \Phi(\tau_n)$, where $\tau_n = n/N$ and the map $\Phi$ is monotonically increasing with $\Phi(0)=0$ and $\Phi(1)=1$. The model is justified for any fixed order method operating in its asymptotic regime when applied to smooth problems, since the step size is then determined by the (smooth) principal error function which determines $\Phi$, and a tolerance requirement which determines $N$. Given any strongly stable multistep method, there is an $N^*$ such that the method is zero stable for $N>N^*$, provided that $\Phi \in C^2[0,1]$. Thus zero stability holds on all nonuniform grids such that adjacent step sizes satisfy $h_n/h_{n-1} = 1 + \mathrm O(N^{-1})$ as $N\rightarrow\infty$. The results are exemplified for BDF-type methods.Open AccessOn the zero-stability of multistep methods on smooth nonuniform grids
In order to be convergent, linear multistep methods must be zero stable. While constant step size theory was established in the 1950's, zero stability on nonuniform grids is less well understood. Here we investigate zero stability on compact intervals and smooth nonuniform grids. In practical computations, step size control can be implemented using smooth (small) step size changes. The resulting grid $\{t_n\}_{n=0}^N$ can be modeled as the image of an equidistant grid under a smooth deformation map, i.e., $t_n = \Phi(\tau_n)$, where $\tau_n = n/N$ and the map $\Phi$ is monotonically increasing with $\Phi(0)=0$ and $\Phi(1)=1$. The model is justified for any fixed order method operating in its asymptotic regime when applied to smooth problems, since the step size is then determined by the (smooth) principal error function which determines $\Phi$, and a tolerance requirement which determines $N$. Given any strongly stable multistep method, there is an $N^*$ such that the method is zero stable for $N>N^*$, provided that $\Phi \in C^2[0,1]$. Thus zero stability holds on all nonuniform grids such that adjacent step sizes satisfy $h_n/h_{n-1} = 1 + \mathrm O(N^{-1})$ as $N\rightarrow\infty$. The results are exemplified for BDF-type methods. - 21 November 2015
- Published by Springer Nature
- p. 1007-1014
Multistep Methods
- 21 November 2015
- Published by Springer Nature
- p. 826-830
Logarithmic Norms
- 13 June 2014
- Published by Springer Nature in BIT Numerical Mathematics
- Vol. 55 (2) , 531-558
Stiffness 1952–2012: Sixty years in search of a definition
- 13 June 2014
- Published by Springer Nature in BIT Numerical Mathematics
- Vol. 55 (2) , 531-558
Stiffness 1952–2012: Sixty years in search of a definition
- 28 July 2011
- Published by Springer Nature in BIT Numerical Mathematics
- Vol. 51 (4) , 977-1007
Geometric integration of Hamiltonian systems perturbed by Rayleigh damping
- 28 July 2011
- Published by Springer Nature in BIT Numerical Mathematics
- Vol. 51 (4) , 977-1007
Geometric integration of Hamiltonian systems perturbed by Rayleigh damping
- Open Access
Geometric Integration of Hamiltonian Systems Perturbed by Rayleigh Damping
Explicit and semi-explicit geometric integration schemes for dissipative perturbations of Hamiltonian systems are analyzed. The dissipation is characterized by a small parameter $\epsilon$, and the schemes under study preserve the symplectic structure in the case $\epsilon=0$. In the case $0<\epsilon\ll 1$ the energy dissipation rate is shown to be asymptotically correct by backward error analysis. Theoretical results on monotone decrease of the modified Hamiltonian function for small enough step sizes are given. Further, an analysis proving near conservation of relative equilibria for small enough step sizes is conducted. Numerical examples, verifying the analyses, are given for a planar pendulum and an elastic 3--D pendulum. The results are superior in comparison with a conventional explicit Runge-Kutta method of the same order.Open AccessGeometric Integration of Hamiltonian Systems Perturbed by Rayleigh Damping
Explicit and semi-explicit geometric integration schemes for dissipative perturbations of Hamiltonian systems are analyzed. The dissipation is characterized by a small parameter $\epsilon$, and the schemes under study preserve the symplectic structure in the case $\epsilon=0$. In the case $0<\epsilon\ll 1$ the energy dissipation rate is shown to be asymptotically correct by backward error analysis. Theoretical results on monotone decrease of the modified Hamiltonian function for small enough step sizes are given. Further, an analysis proving near conservation of relative equilibria for small enough step sizes is conducted. Numerical examples, verifying the analyses, are given for a planar pendulum and an elastic 3--D pendulum. The results are superior in comparison with a conventional explicit Runge-Kutta method of the same order.