pith. sign in

arxiv: 2606.25693 · v1 · pith:BKELN36Gnew · submitted 2026-06-24 · 💻 cs.DC

Dynamic Load Balancing for Uncertainty Quantification with Applications in Bayesian Inversion

C. M. Loi , M. Wille , A. Reinarz This is my paper

Pith reviewed 2026-06-25 20:02 UTC · model grok-4.3

classification 💻 cs.DC
keywords dynamic load balancinguncertainty quantificationBayesian inversionmultilevel samplinghigh performance computingtask schedulingtsunami simulation
0
0 comments X

The pith

A dynamic load balancer distributes heterogeneous sampling tasks in uncertainty quantification without prior workload assumptions.

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

Uncertainty quantification workflows generate many model evaluations whose runtimes vary widely and whose dependencies are loose but real. Traditional static schedulers cannot adapt well to this mix. The paper introduces a dynamic load balancer that assigns tasks as they become ready and measures their observed durations. When applied to multilevel delayed acceptance sampling for a Bayesian inverse problem whose forward models range over orders of magnitude in runtime, the balancer keeps nodes occupied with average idle time near one millisecond. This outcome holds without any pre-supplied model of the expected workload.

Core claim

The load balancer is effective at distributing the sampling requests with an average node idle time of close to a millisecond, while not making any prior assumptions about the workload.

What carries the argument

The dynamic load balancer that assigns tasks on the fly according to observed runtimes and loose dependencies.

If this is right

  • Heterogeneous UQ tasks can be scheduled efficiently on HPC systems without static partitioning.
  • Multilevel simulations whose runtimes span orders of magnitude remain practical under dynamic assignment.
  • Bayesian inversion via multilevel sampling incurs minimal idle time on distributed hardware.
  • Language-agnostic coupling of UQ software to simulations benefits from runtime-only balancing.

Where Pith is reading between the lines

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

  • The same reactive balancing principle could apply to other UQ methods whose task costs are hard to predict in advance.
  • Hardware utilization gains might allow larger ensemble sizes in existing Bayesian inversion studies.
  • Similar on-the-fly assignment could be tested in optimization or ensemble Kalman filter workflows with comparable heterogeneity.

Load-bearing premise

The dependencies between tasks are loose enough that the balancer can handle them by reacting to observed runtimes alone.

What would settle it

Measure average node idle time on the same multilevel sampling workload; if it rises well above one millisecond while the balancer still receives no workload model, the central claim is falsified.

Figures

Figures reproduced from arXiv: 2606.25693 by A. Reinarz, C. M. Loi, M. Wille.

Figure 1
Figure 1. Figure 1: Load balancer configuration for multiple instances of UM-Bridge models and a parallel client. 2.2 SLURM-integrated Load Balancer UQ workflows based on sampling create a distinctive scheduling problem: they submit large numbers of heterogeneous tasks in a short timeframe [10], with runtimes that are often unpredictable and span several orders of magnitude. Traditional SLURM-based approaches handle this poor… view at source ↗
Figure 2
Figure 2. Figure 2: Schematic of the a posteriori subcell limiter. A rejected ADER-DG candidate is replaced by an FV recomputation on subcell averages and then projected back to the DG polynomial space. 3.2 2011 T¯ohoku Tsunami The tsunami is modelled with the depth-averaged shallow-water equations over a variable bathymetry b(x). We use the basic shallow-water equations with bathymetry source terms and neglect bottom frictio… view at source ↗
Figure 3
Figure 3. Figure 3: T¯ohoku tsunami snapshots at t = 340.9 s, 1363.9 s, and 3068.7 s. Top: SSHA at these times. Bottom: cells treated by the finite-volume limiter for the same times. For this scenario, the limiter is deliberately geometry dominated. A DG cell remains on the ADER-DG layer only if the state is finite, the water column exceeds the wet/dry threshold, the cell lies in deep water, and the cell-local variation of η … view at source ↗
Figure 4
Figure 4. Figure 4: T¯ohoku computational domain used for the uncertainty-quantification workflow. The red box marks the displacement translation window [−200, 200] × [−200, 200] km. In this work, we are concerned with the 2011 T¯ohoku tsunami, which resulted from an earthquake in the Japan trench. Our inverse problem is recovering [PITH_FULL_IMAGE:figures/full_fig_p009_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Two level accept-reject scheme of the MLDA algorithm [PITH_FULL_IMAGE:figures/full_fig_p011_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: shows raw data from probe 21418 overlaid with level 0 draws from the prior and posterior distribution, using a separate GP trained to reconstruct the full time series [PITH_FULL_IMAGE:figures/full_fig_p013_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Density of posterior samples from each MLDA levels. The dashed lines indicate the sample mean, and the red cross provides a known reference at (0, 0) [PITH_FULL_IMAGE:figures/full_fig_p014_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: The execution of client requests in the 5-element job array is indicated by the y-axis labels. The x-axis is the elapsed time. The colour of the bars indicates different model fidelities: green is level 2, orange is level 1, and blue is level 0. Arrows indicate the request dependencies within one MLDA chain; the numbers are reduced to avoid clutter. A higher-resolution image is available on the linked auth… view at source ↗
Figure 9
Figure 9. Figure 9: Boxplot of idle time between sampling requests. The idle times, shown in [PITH_FULL_IMAGE:figures/full_fig_p015_9.png] view at source ↗
read the original abstract

Uncertainty Quantification (UQ) workflows present a particular scheduling challenge in high performance computing environments, as they typically generate large numbers of heterogeneous model evaluations with loose but non-trivial dependencies between tasks. A static one-size-fits-all approach in traditional schedulers is inadequate to handle heterogeneous tasks optimally. We introduce an improved load balancer in the UQ and Modelling Bridge (UM-Bridge) framework aimed at mitigating these issues; UM-Bridge is a language-agnostic interface developed to couple UQ software with numerical simulation. As a realistic example, we test the load balancer with a Bayesian inverse problem solved via multilevel delayed acceptance sampling. The underlying forward problem is a hierarchy of tsunami simulations enabled through ExaHyPE, whose runtimes span several orders of magnitude and loose dependencies between levels make the workload particularly challenging to schedule. Our results indicate the load balancer is effective at distributing the sampling requests with an average node idle time of close to a millisecond, while not making any prior assumptions about the workload.

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.

Referee Report

2 major / 0 minor

Summary. The manuscript introduces an improved dynamic load balancer within the UM-Bridge framework for uncertainty quantification workflows. These workflows involve large numbers of heterogeneous model evaluations with loose dependencies, which static schedulers handle poorly. The balancer is tested on a Bayesian inverse problem solved via multilevel delayed acceptance sampling, where the forward model is a hierarchy of tsunami simulations in ExaHyPE with runtimes spanning orders of magnitude. The central empirical result is that the balancer distributes sampling requests effectively, achieving an average node idle time of close to one millisecond while making no prior assumptions about the workload.

Significance. If the central empirical result holds under scrutiny, the work offers a practical demonstration of runtime-adaptive scheduling for UQ applications on HPC systems without requiring a priori workload models. The choice of a multilevel tsunami simulation with loose inter-level dependencies as the test case provides a concrete, challenging example that strengthens the applicability claim. The language-agnostic nature of UM-Bridge is a further asset for coupling diverse UQ and simulation codes.

major comments (2)
  1. [Abstract] Abstract: The reported average node idle time of close to a millisecond is presented as the primary quantitative evidence of effectiveness, yet the abstract supplies no error bars, standard deviation, number of nodes or runs, baseline comparisons against static or other dynamic schedulers, or definition of how idle time was measured. This measurement detail is load-bearing for the central claim.
  2. [Abstract (final paragraph)] The manuscript states that the balancer manages loose but non-trivial dependencies between tasks in the multilevel hierarchy on the fly without any workload model. However, no section details the exact runtime information (e.g., task completion signals, queue monitoring) used by the balancer or how dependency resolution is performed dynamically; this information is required to substantiate that no implicit assumptions are encoded.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback. We address the two major comments below and will revise the manuscript accordingly to strengthen the presentation of our results.

read point-by-point responses

  1. Referee: [Abstract] Abstract: The reported average node idle time of close to a millisecond is presented as the primary quantitative evidence of effectiveness, yet the abstract supplies no error bars, standard deviation, number of nodes or runs, baseline comparisons against static or other dynamic schedulers, or definition of how idle time was measured. This measurement detail is load-bearing for the central claim.

    Authors: We agree that the abstract should be more self-contained to support the central empirical claim. In the revised version we will expand the abstract to report the number of nodes and independent runs, include error bars or standard deviation on the idle-time figure, provide a concise definition of the idle-time metric, and briefly reference the static-scheduler baseline comparison that appears in the results section. revision: yes

  2. Referee: [Abstract (final paragraph)] The manuscript states that the balancer manages loose but non-trivial dependencies between tasks in the multilevel hierarchy on the fly without any workload model. However, no section details the exact runtime information (e.g., task completion signals, queue monitoring) used by the balancer or how dependency resolution is performed dynamically; this information is required to substantiate that no implicit assumptions are encoded.

    Authors: We acknowledge that the current manuscript does not provide an explicit description of the runtime signals and dependency-resolution mechanism. We will add a dedicated subsection (likely in the Implementation or Methods section) that details the task-completion signals, queue-monitoring approach, and on-the-fly dependency handling used by the dynamic balancer, thereby clarifying that no a-priori workload model is required. revision: yes

Circularity Check

0 steps flagged

No significant circularity; empirical measurement of scheduler performance

full rationale

The paper introduces a dynamic load balancer in the UM-Bridge framework and evaluates it experimentally on a multilevel tsunami simulation for Bayesian inversion. The central claim rests on runtime measurements (average node idle time near 1 ms) obtained without prior workload assumptions. No derivation chain, equations, or predictions are presented that reduce by construction to fitted parameters, self-definitions, or self-citation chains. The result is a direct empirical observation on a concrete application and does not invoke uniqueness theorems, ansatzes smuggled via citation, or renaming of known results.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The work rests on the existence and behavior of the UM-Bridge and ExaHyPE frameworks plus standard HPC scheduling primitives; no free parameters, new axioms, or invented entities are introduced.

axioms (1)
  • domain assumption UM-Bridge supplies a language-agnostic interface that correctly couples UQ software to numerical simulators
    Invoked as the platform for the load balancer (abstract).

pith-pipeline@v0.9.1-grok · 5702 in / 1089 out tokens · 18498 ms · 2026-06-25T20:02:23.657750+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

37 extracted references · 27 canonical work pages

  1. [1]

    In: 2018 IEEE/ACM Workflows in Support of Large-Scale Science (WORKS)

    Ahn, D.H., Bass, N., Chu, A., Garlick, J., Grondona, M., Herbein, S., Koning, J., Patki, T., Scogland, T.R.W., Springmeyer, B., Taufer, M.: Flux: Overcoming scheduling challenges for exascale workflows. In: 2018 IEEE/ACM Workflows in Support of Large-Scale Science (WORKS). pp. 10–19 (2018). https://doi.org/10. 1109/WORKS.2018.00007

    arXiv 2018

  2. [2]

    SoftwareX27, 101814 (2024)

    Beránek, J., Böhm, A., Palermo, G., Martinovič, J., Jansík, B.: Hyperqueue: Efficient and ergonomic task graphs on hpc clusters. SoftwareX27, 101814 (2024). https: //doi.org/https://doi.org/10.1016/j.softx.2024.101814, https://www.sciencedirect. com/science/article/pii/S2352711024001857

    work page doi:10.1016/j.softx.2024.101814 2024

  3. [3]

    In: Paszynski, M., Kranzlmüller, D., Krzhizhanovskaya, V.V., Dongarra, J.J., Sloot, P.M

    Bosak, B., Piontek,T., Karlshoefer, P.,Raffin, E., Lakhlili, J., Kopta, P.: Verification, validationanduncertaintyquantificationoflarge-scaleapplicationswithqcg-pilotjob. In: Paszynski, M., Kranzlmüller, D., Krzhizhanovskaya, V.V., Dongarra, J.J., Sloot, P.M. (eds.) Computational Science – ICCS 2021. pp. 495–501. Springer International Publishing, Cham (2021)

    2021

  4. [4]

    Journal of Computational and Graphical Statistics14(4), 795–810 (2005)

    Christen, J.A., Fox, C.: Markov chain monte carlo using an approximation. Journal of Computational and Graphical Statistics14(4), 795–810 (2005). https://doi.org/ 10.1198/106186005X76983, https://doi.org/10.1198/106186005X76983

    work page doi:10.1198/106186005x76983 2005

  5. [5]

    SIAM Review61(3), 509–545 (2019)

    Dodwell, T.J., Ketelsen, C., Scheichl, R., Teckentrup, A.L.: Multilevel markov chain monte carlo. SIAM Review61(3), 509–545 (2019). https://doi.org/10.1137/ 19M126966X, https://doi.org/10.1137/19M126966X

    work page doi:10.1137/19m126966x 2019

  6. [6]

    Physics Letters B195(2), 216–222 (Sep 1987)

    Duane, S., Kennedy, A., Pendleton, B.J., Roweth, D.: Hybrid Monte Carlo. Physics Letters B195(2), 216–222 (Sep 1987). https://doi.org/10.1016/0370-2693(87) 91197-X, https://linkinghub.elsevier.com/retrieve/pii/037026938791197X

    work page doi:10.1016/0370-2693(87 1987

  7. [7]

    Journal of Computational Physics227, 8209–8253 (2008)

    Dumbser, M., Balsara, D.S., Toro, E.F., Munz, C.D.: A unified framework for the construction of one-step finite volume and discontinuous Galerkin schemes on unstructured meshes. Journal of Computational Physics227, 8209–8253 (2008)

    2008

  8. [8]

    Axioms7(3), 63 (2018)

    Dumbser, M., Fambri, F., Tavelli, M., Bader, M., Weinzierl, T.: Efficient imple- mentation of ADER discontinuous Galerkin schemes for a scalable hyperbolic PDE engine. Axioms7(3), 63 (2018). https://doi.org/10.3390/axioms7030063

    work page doi:10.3390/axioms7030063 2018

  9. [9]

    Journal of Computational Physics319, 163–199 (2016)

    Dumbser, M., Loubère, R.: A simple robust and accurate a posteriori sub-cell finite volume limiter for the discontinuous Galerkin method on unstructured meshes. Journal of Computational Physics319, 163–199 (2016)

    2016

  10. [10]

    Communications Engineering1(2022), https://api.semanticscholar.org/ CorpusID:246652492

    Farcas, I.G., Merlo, G., Jenko, F.: A general framework for quantifying uncertainty at scale. Communications Engineering1(2022), https://api.semanticscholar.org/ CorpusID:246652492

    2022

  11. [11]

    Geophysical Journal International Dynamic Load Balancing for Uncertainty Quantification 17 198(2), 1222–1240 (08 2014)

    Galvez, P., Ampuero, J.P., Dalguer, L.A., Somala, S.N., Nissen-Meyer, T.: Dynamic earthquake rupture modelled with an unstructured 3-d spectral element method applied to the 2011 m9 tohoku earthquake. Geophysical Journal International Dynamic Load Balancing for Uncertainty Quantification 17 198(2), 1222–1240 (08 2014). https://doi.org/10.1093/gji/ggu203, ...

    work page doi:10.1093/gji/ggu203 2011

  12. [12]

    Journal of Computational Physics227(6), 3089–3113 (2008)

    George, D.L.: Augmented Riemann solvers for the shallow water equations over variable topography with steady states and inundation. Journal of Computational Physics227(6), 3089–3113 (2008). https://doi.org/10.1016/j.jcp.2007.10.027

    work page doi:10.1016/j.jcp.2007.10.027 2008

  13. [13]

    Frontiers in Earth ScienceV olume 8 - 2020(2020)

    Gibbons, S.J., Lorito, S., Macías, J., Løvholt, F., Selva, J., Volpe, M., Sánchez- Linares, C., Babeyko, A., Brizuela, B., Cirella, A., Castro, M.J., de la Asunción, M., Lanucara, P., Glimsdal, S., Lorenzino, M.C., Nazaria, M., Pizzimenti, L., Romano, F., Scala, A., Tonini, R., Manuel González Vida, J., Vöge, M.: Probabilistic tsunami hazard analysis: Hig...

    work page doi:10.3389/feart.2020.591549 2020

  14. [14]

    Operations research56(3), 607–617 (2008)

    Giles, M.B.: Multilevel monte carlo path simulation. Operations research56(3), 607–617 (2008)

    2008

  15. [15]

    Acta numerica24, 259–328 (2015)

    Giles, M.B.: Multilevel monte carlo methods. Acta numerica24, 259–328 (2015)

    2015

  16. [16]

    Philosophi- cal Transactions of the Royal Society A: Mathematical, Physical and Engineer- ing Sciences379(2197), 20200221 (2021)

    Groen, D., Arabnejad, H., Jancauskas, V., Edeling, W.N., Jansson, F., Richard- son, R.A., Lakhlili, J., Veen, L., Bosak, B., Kopta, P., Wright, D.W., Monnier, N., Karlshoefer, P., Suleimenova, D., Sinclair, R., Vassaux, M., Nikishova, A., Bieniek, M., Luk, O.O., Kulczewski, M., Raffin, E., Crommelin, D., Hoenen, O., Coster, D.P., Piontek, T., Coveney, P.V...

    work page doi:10.1098/rsta.2020.0221 2021

  17. [17]

    Biometrika77(2), 401–403 (1990).https://doi.org/10.1093/biomet/ 77.2.401

    Hastings, W.K.: Monte carlo sampling methods using markov chains and their applications. Biometrika57(1), 97–109 (04 1970). https://doi.org/10.1093/biomet/ 57.1.97, https://doi.org/10.1093/biomet/57.1.97

    work page doi:10.1093/biomet/ 1970

  18. [18]

    Real-time bayesian inference at extreme scale: A digital twin for tsunami early warning applied to the cascadia subduction zone,

    Henneking, S., Venkat, S., Dobrev, V., Camier, J., Kolev, T., Fernando, M., Gabriel, A.A., Ghattas, O.: Real-time bayesian inference at extreme scale: A digital twin for tsunami early warning applied to the cascadia subduction zone. In: Proceedings of the International Conference for High Performance Computing, Networking, Storage and Analysis. p. 60–71. ...

    work page doi:10.1145/3712285.3771787 2025

  19. [19]

    Journal of Computational Physics552, 114682 (2026)

    Henneking,S.,Venkat,S.,Ghattas,O.:Goal-orientedreal-timebayesianinferencefor linear autonomous dynamical systems with application to digital twins for tsunami early warning. Journal of Computational Physics552, 114682 (2026). https:// doi.org/https://doi.org/10.1016/j.jcp.2026.114682, https://www.sciencedirect.com/ science/article/pii/S002199912600032X

    work page doi:10.1016/j.jcp.2026.114682 2026

  20. [20]

    The International Journal of High Performance Computing Applications36(3), 289–306 (2022)

    Herbein, S., Patki, T., Ahn, D.H., Mobo, S., Hathaway, C., Caíno-Lores, S., Corbett, J., Domyancic, D., Scogland, T.R., de Supinski, B.R., Taufer, M.: An analytical performance model of generalized hierarchical scheduling. The International Journal of High Performance Computing Applications36(3), 289–306 (2022). https://doi. org/10.1177/10943420211051039,...

    work page doi:10.1177/10943420211051039 2022

  21. [21]

    Hoffman, M.D., Gelman, A., et al.: The No-U-Turn sampler: adaptively setting path lengths in Hamiltonian Monte Carlo. J. Mach. Learn. Res.15(1), 1593–1623 (2014)

    2014

  22. [22]

    SIAM Review65(3), 831–865 (2023)

    Latz, J.: Bayesian inverse problems are usually well-posed. SIAM Review65(3), 831–865 (2023). https://doi.org/10.1137/23M1556435, https://doi.org/10.1137/ 23M1556435 18 C. M. Loi et al

    work page doi:10.1137/23m1556435 2023

  23. [23]

    In: ISC High Performance 2025 Research Paper Proceedings (40th International Conference)

    Loi, C.M., Reinarz, A., Seelinger, L., Hornsby, W., Buchanan, J., Lykkegaard, M.: A performance analysis of task scheduling for uq workflows on hpc systems. In: ISC High Performance 2025 Research Paper Proceedings (40th International Conference). pp. 1–14 (2025). https://doi.org/10.23919/ISC.2025.11018268

    work page doi:10.23919/isc.2025.11018268 2025

  24. [24]

    SIAM/ASA Journal on Uncertainty Quantification 11(1), 1–30 (2023)

    Lykkegaard, M.B., Dodwell, T.J., Fox, C., Mingas, G., Scheichl, R.: Multilevel delayed acceptance mcmc. SIAM/ASA Journal on Uncertainty Quantification 11(1), 1–30 (2023). https://doi.org/10.1137/22M1476770, https://doi.org/10.1137/ 22M1476770

    work page doi:10.1137/22m1476770 2023

  25. [26]

    Rosenbluth, Marshall N

    Metropolis, N., Rosenbluth, A.W., Rosenbluth, M.N., Teller, A.H., Teller, E.: Equation of state calculations by fast computing machines. The Journal of Chemical Physics21(6), 1087–1092 (06 1953). https://doi.org/10.1063/1.1699114, https: //doi.org/10.1063/1.1699114

    work page doi:10.1063/1.1699114 1953

  26. [27]

    Adaptative computation and machine learning series, University Press Group Limited (2006)

    Rasmussen, C., Williams, C.: Gaussian Processes for Machine Learning. Adaptative computation and machine learning series, University Press Group Limited (2006)

    2006

  27. [28]

    Computer Physics Communications254, 107251 (2020)

    Reinarz, A., Charrier, D.E., Bader, M., Bovard, L., Dumbser, M., Duru, K., Fambri, F., Gabriel, A.A., Gallard, J.M., Köppel, S., Krenz, L., Rannabauer, L., Rezzolla, L., Samfass, P., Tavelli, M., Weinzierl, T.: ExaHyPE: An engine for parallel dynamically adaptive simulations of wave problems. Computer Physics Communications254, 107251 (2020). https://doi....

    work page doi:10.1016/j.cpc.2020.107251 2020

  28. [29]

    In: Kathryn Huff, James Bergstra (eds.) Proceedings of the 14th Python in Science Conference

    Matthew Rocklin: Dask: Parallel Computation with Blocked algorithms and Task Scheduling. In: Kathryn Huff, James Bergstra (eds.) Proceedings of the 14th Python in Science Conference. pp. 126 – 132 (2015). https://doi.org/10.25080/ Majora-7b98e3ed-013

    2015

  29. [30]

    Natural Hazards and Earth System Sciences 12(6), 2003–2018 (2012)

    Sarri, A., Guillas, S., Dias, F.: Statistical emulation of a tsunami model for sensitivity analysis and uncertainty quantification. Natural Hazards and Earth System Sciences 12(6), 2003–2018 (2012). https://doi.org/10.5194/nhess-12-2003-2012, https:// nhess.copernicus.org/articles/12/2003/2012/

    work page doi:10.5194/nhess-12-2003-2012 2003

  30. [31]

    Journal of Open Source Software 8(83), 4748 (2023)

    Seelinger, L., Cheng-Seelinger, V., Davis, A., Parno, M., Reinarz, A.: UM-Bridge: Uncertainty quantification and modeling bridge. Journal of Open Source Software 8(83), 4748 (2023). https://doi.org/10.21105/joss.04748, https://doi.org/10.21105/ joss.04748

    work page doi:10.21105/joss.04748 2023

  31. [32]

    Journal of Computational Physics p

    Seelinger, L., Reinarz, A., Lykkegaard, M.B., Akers, R., Alghamdi, A.M., Aristoff, D., Bangerth, W., Bénézech, J., Diez, M., Frey, K., Jakeman, J.D., Jørgensen, J.S., Kim, K.T., Kent, B.M., Martinelli, M., Parno, M., Pellegrini, R., Petra, N., Riis, N.A., Rosenfeld, K., Serani, A., Tamellini, L., Villa, U., Dodwell, T.J., Scheichl, R.: Democratizing uncer...

    work page doi:10.1016/j.jcp.2024.113542 2024

  32. [33]

    In: Proceedings of the International Conference for High Performance Computing, Networking, Storage and Analysis

    Seelinger, L., Reinarz, A., Rannabauer, L., Bader, M., Bastian, P., Scheichl, R.: High performance uncertainty quantification with parallelized multilevel markov chain monte carlo. In: Proceedings of the International Conference for High Performance Computing, Networking, Storage and Analysis. SC ’21, Association for Computing Machinery, New York, NY, USA...

    work page doi:10.1145/3458817.3476150 2021

  33. [34]

    Ocean Modelling83, 82–97 (2014)

    Sraj, I., Mandli, K.T., Knio, O.M., Dawson, C.N., Hoteit, I.: Uncertainty quantifi- cation and inference of manning’s friction coefficients using dart buoy data during Dynamic Load Balancing for Uncertainty Quantification 19 the t¯ ohoku tsunami. Ocean Modelling83, 82–97 (2014). https://doi.org/https: //doi.org/10.1016/j.ocemod.2014.09.001, https://www.sc...

    work page doi:10.1016/j.ocemod.2014.09.001 2014

  34. [35]

    Inverse problems: A Bayesian perspective

    Stuart, A.M.: Inverse problems: A Bayesian perspective. Acta Numerica19, 451–559 (2010). https://doi.org/10.1017/S0962492910000061

    work page doi:10.1017/s0962492910000061 2010

  35. [36]

    In: Proceedings of the International Conference for High Performance Computing, Networking, Storage and Analysis

    Uphoff, C., Rettenberger, S., Bader, M., Madden, E.H., Ulrich, T., Wollherr, S., Gabriel, A.A.: Extreme scale multi-physics simulations of the tsunamigenic 2004 sumatra megathrust earthquake. In: Proceedings of the International Conference for High Performance Computing, Networking, Storage and Analysis. SC ’17, Association for Computing Machinery, New Yo...

    work page doi:10.1145/3126908.3126948 2004

  36. [37]

    ACM Trans

    Weinzierl,T.:Thepeanosoftware—parallel,automaton-based,dynamicallyadaptive grid traversals. ACM Trans. Math. Softw.45(2) (Apr 2019). https://doi.org/10. 1145/3319797, https://doi.org/10.1145/3319797

    work page doi:10.1145/3319797 2019

  37. [38]

    Willoughby, R.A.: Solutions of ill-posed problems (a. n. tikhonov and v. y. arsenin). SIAM Review21(2), 266–267 (1979). https://doi.org/10.1137/1021044, https:// doi.org/10.1137/1021044

    work page doi:10.1137/1021044 1979