arxiv: 2604.22057 · v1 · submitted 2026-04-23 · 🧮 math.NA · cs.NA

Recognition: unknown

A-priori error estimation for space-time Galerkin POD for linear evolution problems

Carmen Gr\"a{\ss}le, Jan Heiland, Jannis Marquardt

Authors on Pith no claims yet

Pith reviewed 2026-05-09 20:28 UTC · model grok-4.3

classification 🧮 math.NA cs.NA

keywords a priori error estimatespace-time PODmodel order reductionGalerkin PODlinear parabolic PDEproper orthogonal decompositionevolution problems

0 comments

The pith

A new a-priori error estimate bounds the difference between the full numerical solution of a linear parabolic PDE and its space-time POD reduced-order approximation.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

This paper develops an a-priori error estimate for the space-time proper orthogonal decomposition method applied to linear parabolic partial differential equations. The estimate quantifies how much the reduced solution deviates from the full numerical solution obtained via Galerkin discretization. A reader would care because it offers a way to guarantee the accuracy of the reduced model based on the POD truncation without solving the full problem repeatedly. The approach combines space and time reduction in the POD basis for efficiency in evolution problems. Numerical tests validate the theoretical predictions against observed errors.

Core claim

The authors propose and derive an a-priori error estimate for the space-time Galerkin POD reduced solution of linear evolution problems. This estimate applies to linear parabolic PDEs and provides a bound on the error between the high-fidelity numerical solution and the reduced-order model obtained by projecting onto a POD basis computed from snapshot data in both space and time.

What carries the argument

The a-priori error estimate for the space-time POD Galerkin approximation, which uses the orthogonality properties of the POD basis and the stability of the underlying parabolic problem to bound the approximation error.

If this is right

The error bound depends on the decay of the POD singular values and the approximation quality of the snapshots.
If the POD basis captures most of the energy, the reduced solution error remains small uniformly in time and space.
The method allows simultaneous reduction in space and time dimensions for linear parabolic equations.
Comparison with numerical examples shows the estimate is realistic and not overly conservative.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

This framework might be adaptable to other evolution equations by modifying the stability estimates.
Such error estimates could guide the selection of snapshot data for better POD performance in practice.
Extending the analysis to include discretization errors in the full model would strengthen applicability to real computations.

Load-bearing premise

The derivation assumes the underlying PDE is linear and parabolic, with the space-time POD basis generated from suitable snapshot data using the Galerkin method.

What would settle it

A specific linear parabolic PDE, such as the heat equation, where the actual L2 error between the full solution and the space-time POD approximation exceeds the derived a-priori bound.

read the original abstract

In this paper, we propose an a-priori error estimate for the model order reduction (MOR) method of space-time proper orthogonal decomposition (space-time POD). The original space-time POD approach extends standard POD by reducing not only the space dimension but simultaneously the time dimension as well. The proposed a-priori error estimate is developed for a linear parabolic partial differential equation and estimates the error between the numerical solution to a linear parabolic partial differential equation (PDE) and its space-time POD reduced solution. Numerical examples illustrate the occurring errors and analyze them in comparison to the theoretical bounds.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

The paper derives an a-priori error bound for space-time POD on linear parabolic PDEs using standard Galerkin stability arguments.

read the letter

This paper proposes an a-priori error estimate for space-time Galerkin POD on linear parabolic PDEs. It bounds the error between the full numerical solution and the reduced space-time POD solution. The contribution is the extension of POD error analysis to the space-time reduction setting. For linear problems, the proof likely combines the POD's optimality in the space-time L2 inner product with the coercivity of the parabolic operator to control the Galerkin error. This is a natural step and the abstract gives no indication of problems with the derivation. The paper does well by including numerical examples that compare the actual errors to the theoretical bounds. This helps assess the sharpness of the estimate in practice. Soft spots are limited. The result applies only to linear parabolic problems, as stated, which is a reasonable scope but restricts broader use. The bound depends on the quality of the snapshot data for the POD basis, which is standard but not explored in detail. The numerics are illustrative, not a comprehensive validation across many cases or discretizations. The central claim holds up based on the described approach, with no circularity or unsupported steps apparent. This paper is for specialists in model order reduction for time-dependent PDEs who want theoretical error control for space-time reduced models. It would be of interest to readers following work on POD for evolution problems. I recommend sending it for peer review, as it provides a useful theoretical tool with supporting evidence in a focused area.

Referee Report

1 major / 3 minor

Summary. The manuscript proposes an a-priori error estimate for space-time Galerkin POD applied to linear parabolic PDEs. It derives a bound on the difference between the full-order finite-element solution and the reduced space-time POD Galerkin solution, relying on POD optimality in the space-time L2 norm together with stability of the linear evolution operator, and illustrates the bound on two model problems.

Significance. A rigorous a-priori bound for space-time POD would be a useful addition to the MOR literature, since most existing POD error analyses are either purely spatial or rely on a posteriori residuals. The numerical examples show that the predicted scaling with the POD truncation tolerance is observed in practice, which is a positive indicator that the derivation captures the dominant terms.

major comments (1)

[§4, Theorem 4.1] §4, Theorem 4.1: the constant C in the final bound depends on the coercivity constant of the bilinear form and on the POD truncation error; however, the proof sketch does not explicitly track how the time-stepping stability constant enters when the full-order solution is itself a discrete-in-time approximation. This makes it unclear whether the estimate remains uniform with respect to the temporal mesh size.

minor comments (3)

[Eq. (2.7)] The definition of the space-time POD inner product (Eq. (2.7)) uses a weighted L2 norm whose weight is never stated explicitly; please add the precise weight function.
[Figure 3] Figure 3 caption refers to 'relative error in the energy norm' but the y-axis label is simply 'error'; please align caption and axis label.
[abstract and §1] The statement that the estimate is 'parameter-free' appears in the abstract and §1 but is qualified in §4.2 by dependence on the stability constant; please remove or qualify the claim in the abstract.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for the positive assessment of the manuscript, the recommendation for minor revision, and the constructive comment on Theorem 4.1. We address the major comment below.

read point-by-point responses

Referee: [§4, Theorem 4.1] §4, Theorem 4.1: the constant C in the final bound depends on the coercivity constant of the bilinear form and on the POD truncation error; however, the proof sketch does not explicitly track how the time-stepping stability constant enters when the full-order solution is itself a discrete-in-time approximation. This makes it unclear whether the estimate remains uniform with respect to the temporal mesh size.

Authors: We agree that the proof sketch in the current version does not explicitly track the stability constant arising from the time discretization of the full-order model. In the revised manuscript we will expand the proof of Theorem 4.1 to derive an explicit bound on this constant. Under the standard coercivity and continuity assumptions on the bilinear form, and for unconditionally stable implicit time-stepping schemes (such as backward Euler), the discrete stability constant remains bounded independently of the time-step size. This establishes uniformity of the a-priori error bound with respect to the temporal mesh size. A clarifying remark will be added after the theorem statement. revision: yes

Circularity Check

0 steps flagged

Derivation is self-contained; no circular steps identified

full rationale

The paper presents a theoretical a-priori error bound for space-time Galerkin POD on linear parabolic PDEs. The derivation proceeds from POD optimality in the space-time L2 inner product, Galerkin orthogonality, and standard coercivity/stability estimates for the linear evolution operator. No equation reduces by construction to a fitted parameter or self-defined quantity, no load-bearing step collapses to a self-citation chain, and the central estimate is not a renaming of an empirical pattern. The result remains independent of its own inputs under the stated linearity and snapshot assumptions.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The claim rests on standard assumptions for linear parabolic PDEs and the definition of the space-time POD reduced solution; no free parameters or invented entities are mentioned.

axioms (2)

domain assumption The problem is a linear parabolic partial differential equation
Explicitly stated as the setting for which the error estimate is developed.
standard math Existence of a numerical solution and a space-time POD reduced solution
Implicit requirement for defining the error between them.

pith-pipeline@v0.9.0 · 5400 in / 1187 out tokens · 28083 ms · 2026-05-09T20:28:02.180487+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

32 extracted references · 22 canonical work pages

[1]

Arian, E., Fahl, M., Sachs, E.W.: Trust-Region Proper Orthogonal Decomposition for Flow Control. Tech. Rep. ICASE Report No. 2000-25, Institute for Computer 23 Applications in Science and Engineering (ICASE), Hampton, VA, USA (2000). https://apps.dtic.mil/sti/citations/ADA377382

2000
[2]

Ballarin, F., Rozza, G., Strazzullo, M.: Chapter 9 - Space-time POD-Galerkin approach for parametric flow control. In: E. Tr´ elat, E. Zuazua, eds., Numerical Control: Part A, vol. 23 ofHandbook of Numerical Analysis, pp. 307–338. Elsevier (2022). https://doi.org/10.1016/bs.hna.2021.12.009

work page doi:10.1016/bs.hna.2021.12.009 2022
[3]

https://doi.org/10

Banholzer, S., Beermann, D., Mechelli, L., Volkwein, S.: POD Suboptimal Control of Evolution Problems : Theory and Applications (2024). https://doi.org/10. 48550/arXiv.2404.07015

work page arXiv 2024
[4]

Model Reduction for Parametrized Systems

Baumann, M., Benner, P., Heiland, J.: A generalized POD space-time Galerkin scheme for parameter dependent dynamical systems. ScienceOpen Posters (2015). https://doi.org/10.14293/P2199-8442.1.SOP-MATH.P8ECXQ.v1. Poster at MoRePaS III — Workshop on “Model Reduction for Parametrized Systems”, Trieste, Italy

work page doi:10.14293/p2199-8442.1.sop-math.p8ecxq.v1 2015
[5]

SIAM Journal on Scientific Computing 40(3), A1611–A1641 (2018)

Baumann, M., Benner, P., Heiland, J.: Space-Time Galerkin POD with Appli- cation in Optimal Control of Semilinear Partial Differential Equations. SIAM Journal on Scientific Computing 40(3), A1611–A1641 (2018). https://doi.org/10. 1137/17M1135281

2018
[6]

Baumann, M., Heiland, J., Schmidt, M.: Discrete Input/Output Maps and their Relation to Proper Orthogonal Decomposition, pp. 585–608. Springer Interna- tional Publishing, Cham (2015). ISBN 978-3-319-15260-8. https://doi.org/10. 1007/978-3-319-15260-8 21

2015
[7]

Computers & Fluids 106, 19–32 (2015)

Behzad, F., Helenbrook, B.T., Ahmadi, G.: On the sensitivity and accuracy of proper-orthogonal-decomposition-based reduced order models for Burgers equation. Computers & Fluids 106, 19–32 (2015). ISSN 0045-7930. https: //doi.org/10.1016/j.compfluid.2014.09.041

work page doi:10.1016/j.compfluid.2014.09.041 2015
[8]

Chinesta, F., Ammar, A., Leygue, A., Keunings, R.: An overview of the proper generalized decomposition with applications in computational rheology. J. Non- Newtonian Fluid Mech. 166(11), 578–592 (2011). https://doi.org/10.1016/j. jnnfm.2010.12.012

work page doi:10.1016/j 2011
[10]

SIAM Journal on Scientific Computing 41(1), A26–A58 (2019)

Choi, Y., Carlberg, K.: Space–Time Least-Squares Petrov–Galerkin Projection for Nonlinear Model Reduction. SIAM Journal on Scientific Computing 41(1), A26–A58 (2019). https://doi.org/10.1137/17M1120531

work page doi:10.1137/17m1120531 2019
[11]

De Lathauwer, L., De Moor, B., Vandewalle, J.: A multilinear singular value decomposition. SIAM J. Matrix Anal. Appl. 21(4), 1253–1278 (2000). https: //doi.org/10.1137/S0895479896305696

work page doi:10.1137/s0895479896305696 2000
[12]

PLOS ONE 18(8), 1–31 (2023)

Frame, P., Towne, A.: Space-time POD and the Hankel matrix. PLOS ONE 18(8), 1–31 (2023). https://doi.org/10.1371/journal.pone.0289637

work page doi:10.1371/journal.pone.0289637 2023
[13]

Gr¨ aßle, C., Hinze, M., Volkwein, S.: Model order reduction by proper orthogonal decomposition, pp. 47–96. De Gruyter, Berlin, Boston (2021). https://doi.org/ 10.1515/9783110671490-002 24

work page doi:10.1515/9783110671490-002 2021
[14]

Gubisch, M., Volkwein, S.: Proper Orthogonal Decomposition for Linear- Quadratic Optimal Control, chap. 1, pp. 3–63. SIAM (2017). https://doi.org/10. 1137/1.9781611974829.ch1

2017
[15]

Flow, Turbulence and Combustion 65(3), 273–298 (2000)

Hinze, M., Kunisch, K.: Three Control Methods for Time-Dependent Fluid Flow. Flow, Turbulence and Combustion 65(3), 273–298 (2000). https://doi.org/10. 1023/A:1011417305739

2000
[16]

Hinze, M., Volkwein, S.: Proper Orthogonal Decomposition Surrogate Models for Nonlinear Dynamical Systems: Error Estimates and Suboptimal Control. In: P. Benner, D.C. Sorensen, V. Mehrmann, eds., Dimension Reduction of Large- Scale Systems, pp. 261–306. Springer Berlin Heidelberg, Berlin, Heidelberg (2005). ISBN 978-3-540-27909-9. https://doi.org/10.1007/...

work page doi:10.1007/3-540-27909-1 2005
[17]

Computational Op- timization and Applications 39(3), 319–345 (2008)

Hinze, M., Volkwein, S.: Error estimates for abstract linear–quadratic optimal control problems using proper orthogonal decomposition. Computational Op- timization and Applications 39(3), 319–345 (2008). https://doi.org/10.1007/ s10589-007-9058-4

2008
[18]

ArXiv abs/2111.06435 (2021)

Jin, R., Rizzi, F., Parish, E.J.: Space-time reduced-order modeling for uncertainty quantification. ArXiv abs/2111.06435 (2021). https://api.semanticscholar.org/ CorpusID:244102752

work page arXiv 2021
[19]

Random forests.Machine Learning, 45(1):5–32, 2001

Kunisch, K., Volkwein, S.: Control of the Burgers Equation by a Reduced-Order Approach Using Proper Orthogonal Decomposition. Journal of Optimization Theory and Applications 102(2), 345–371 (1999). https://doi.org/10.1023/A: 1021732508059

work page doi:10.1023/a: 1999
[20]

Numerische Mathematik 90(1), 117–148 (2001)

Kunisch, K., Volkwein, S.: Galerkin proper orthogonal decomposition methods for parabolic problems. Numerische Mathematik 90(1), 117–148 (2001). https: //doi.org/10.1007/s002110100282

work page doi:10.1007/s002110100282 2001
[21]

SIAM Journal on Numerical Analysis 40(2), 492–515 (2002)

Kunisch, K., Volkwein, S.: Galerkin Proper Orthogonal Decomposition Methods for a General Equation in Fluid Dynamics. SIAM Journal on Numerical Analysis 40(2), 492–515 (2002). https://doi.org/10.1137/S0036142900382612

work page doi:10.1137/s0036142900382612 2002
[22]

Lassila, T., Manzoni, A., Quarteroni, A., Rozza, G.: Model Order Reduction in Fluid Dynamics: Challenges and Perspectives, pp. 235–273. Springer International Publishing, Cham (2014). https://doi.org/10.1007/978-3-319-02090-7 9

work page doi:10.1007/978-3-319-02090-7 2014
[23]

https://arxiv.org/abs/2508.00638

Li, X., Lasagna, D.: Space-time nonlinear reduced-order modelling for unsteady flows (2026). https://arxiv.org/abs/2508.00638

work page arXiv 2026
[24]

L.: The Structure of Inhomogeneous Turbulent Flows

Lumley, J. L.: The Structure of Inhomogeneous Turbulent Flows. In: M. Yaglom A. I. Tatarski V. eds., Atmospheric Turbulence and Radio Wave Propagation, pp. 166–178. Nauka, Moscow (1967)

1967
[25]

Pinnau, R.: Model Reduction via Proper Orthogonal Decomposition, pp. 95–109. Springer Berlin Heidelberg, Berlin, Heidelberg (2008). https://doi.org/10.1007/ 978-3-540-78841-6 5

2008
[26]

Journal of Fluid Mechanics 867, R2 (2019)

Schmidt, O.T., Schmid, P.J.: A conditional space–time POD formalism for inter- mittent and rare events: example of acoustic bursts in turbulent jets. Journal of Fluid Mechanics 867, R2 (2019). https://doi.org/10.1017/jfm.2019.200

work page doi:10.1017/jfm.2019.200 2019
[27]

Computer Methods in Applied Mechanics and Engineering 386, 114050 (2021)

Shimizu, Y.S., Parish, E.J.: Windowed space–time least-squares Petrov–Galerkin model order reduction for nonlinear dynamical systems. Computer Methods in Applied Mechanics and Engineering 386, 114050 (2021). ISSN 0045-7825. https: 25 //doi.org/10.1016/j.cma.2021.114050

work page doi:10.1016/j.cma.2021.114050 2021
[28]

SIAM Journal on Numerical Analysis 52(2), 852–876 (2014)

Singler, J.R.: New POD Error Expressions, Error Bounds, and Asymptotic Re- sults for Reduced Order Models of Parabolic PDEs. SIAM Journal on Numerical Analysis 52(2), 852–876 (2014). https://doi.org/10.1137/120886947

work page doi:10.1137/120886947 2014
[29]

PhD thesis, Universit¨ at Ulm (2014)

Steih, K.: Reduced Basis Methods for Time-Periodic Parametric Partial Dif- ferential Equations. PhD thesis, Universit¨ at Ulm (2014). Available at https: //vts.uni-ulm.de/query/longview.meta.asp?document id=9093

2014
[30]

IFAC Proceedings Volumes 45(2), 710–715 (2012)

Steih, K., Urban, K.: Space-Time Reduced Basis Methods for Time-Periodic Partial Differential Equations. IFAC Proceedings Volumes 45(2), 710–715 (2012). ISSN 1474-6670. https://doi.org/10.3182/20120215-3-AT-3016.00126. 7th Vienna International Conference on Mathematical Modelling

work page doi:10.3182/20120215-3-at-3016.00126 2012
[31]

SIAM Journal on Scientific Computing 46(1), B1–B32 (2024)

Tenderini, R., Mueller, N., Deparis, S.: Space-Time Reduced Basis Methods for Parametrized Unsteady Stokes Equations. SIAM Journal on Scientific Computing 46(1), B1–B32 (2024). ISSN 1095-7197. https://doi.org/10.1137/22m1509114

work page doi:10.1137/22m1509114 2024
[32]

25 of Springer Series in Computational Mathematics

Thom´ ee, V.: Galerkin Finite Element Methods for Parabolic Problems, vol. 25 of Springer Series in Computational Mathematics . Springer Berlin Heidelberg, 2 edn. (2006). https://doi.org/10.1007/3-540-33122-0

work page doi:10.1007/3-540-33122-0 2006
[33]

Proceedings of the 17th International Symposium on Mathematical Theory of Networks and Systems, Kyoto, Japan (2006) 26

Volkwein, S., Weiland, S.: An Algorithm for Galerkin Projections in both Time and Spatial Coordinates. Proceedings of the 17th International Symposium on Mathematical Theory of Networks and Systems, Kyoto, Japan (2006) 26

2006