pith. sign in

arxiv: 2509.26005 · v4 · pith:U2ZWYLGKnew · submitted 2025-09-30 · 📊 stat.ML · cs.LG

BALLAST: Bayesian Active Learning with Look-ahead Amendment for Sea-drifter Trajectories under Spatio-Temporal Vector Fields

Rui-Yang Zhang , Lachlan Astfalck , Edward Cripps , David S. Leslie , Henry B. Moss This is my paper

Pith reviewed 2026-05-22 12:38 UTC · model grok-4.3

classification 📊 stat.ML cs.LG
keywords active learningGaussian processLagrangian driftersspatio-temporal fieldsoceanographysequential designtrajectory prediction
0
0 comments X

The pith

BALLAST amends active learning with look-ahead trajectory predictions to optimally place drifting observers in spatio-temporal vector fields.

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

The paper develops BALLAST to direct the release of sea-drifters that move with ocean currents while collecting data on the currents themselves. Existing approaches often overlook how these observers will be advected, leading to suboptimal information collection. BALLAST uses a physics-informed Gaussian process to forecast likely future paths and adjusts the selection of placement sites accordingly. Tests on synthetic fields and high-fidelity ocean models demonstrate clearer advantages in learning the vector fields. An auxiliary contribution is VaSE, a faster method for drawing samples from the Gaussian process posterior.

Core claim

The discovery is that accounting for the continuous movement of Lagrangian observers through the time-dependent vector field, by amending standard acquisition functions with simulated future trajectories, produces superior sequential designs for inferring the field.

What carries the argument

BALLAST: Bayesian Active Learning with Look-ahead Amendment, which evaluates candidate placements by simulating how observers would sample the field along their advected paths.

If this is right

  • BALLAST-aided strategies show noticeable benefits on synthetic and high-fidelity ocean current models.
  • It addresses the challenge of observers making measurements at varying locations and times due to advection.
  • VaSE boosts the efficiency of GP posterior sampling as a byproduct.

Where Pith is reading between the lines

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

  • This framework might transfer to placing mobile sensors in other fluid or atmospheric flows.
  • Future work could test robustness when the surrogate model has higher error.

Load-bearing premise

The physics-informed spatio-temporal Gaussian process surrogate is assumed to be accurate enough that its look-ahead trajectory predictions usefully inform placement decisions.

What would settle it

Observing that placements chosen by BALLAST do not lead to faster reduction in uncertainty about the vector field compared to standard space-filling designs in controlled simulations.

Figures

Figures reproduced from arXiv: 2509.26005 by David S. Leslie, Edward Cripps, Henry B. Moss, Lachlan Astfalck, Rui-Yang Zhang.

Figure 1
Figure 1. Figure 1: Illustration of spatio-temporal GP regression of Lagrangian trajectories. The top row shows the [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Deployment comparison under the uniform policy (left), EIG (middle), and our proposed BAL￾LAST (right). Ten Lagrangian observers are placed sequentially, with their placement locations in blue. The observations are plotted with varying brightness according to the observation time (later is brighter). marginal posterior predictive distribution over a suffi￾ciently fine spatio-temporal grid R × T , which mak… view at source ↗
Figure 3
Figure 3. Figure 3: The schematic diagram illustrating the BALLAST algorithm for active learning. Given existing [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Percentage utility gap with 2 standard error [PITH_FULL_IMAGE:figures/full_fig_p007_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Vector fields at selected time slices of the [PITH_FULL_IMAGE:figures/full_fig_p007_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Policy comparison with temporal Helmholtz ground truth. Left is the average policy rank over iterations [PITH_FULL_IMAGE:figures/full_fig_p008_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Policy comparison with SUNTANS ground truth. Left is the average policy rank over iterations at each [PITH_FULL_IMAGE:figures/full_fig_p008_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Combined plot of ablation under synthetic ground truth for decision times [PITH_FULL_IMAGE:figures/full_fig_p020_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Combined plot of ablation under SUNTANS ground truth for decision times [PITH_FULL_IMAGE:figures/full_fig_p020_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Utility gap with 2 standard error bounds of UNIF, EIG, and BALLAST decisions over sample number [PITH_FULL_IMAGE:figures/full_fig_p021_10.png] view at source ↗
read the original abstract

We introduce a formal active learning methodology for guiding the placement of Lagrangian observers to infer time-dependent vector fields -- a key task in oceanography, marine science, and ocean engineering -- using a physics-informed spatio-temporal Gaussian process surrogate model. The majority of existing placement campaigns either follow standard `space-filling' designs or relatively ad-hoc expert opinions. A key challenge to applying principled active learning in this setting is that Lagrangian observers are continuously advected through the vector field, so they make measurements at different locations and times. It is, therefore, important to consider the likely future trajectories of placed observers to account for the utility of candidate placement locations. To this end, we present BALLAST: Bayesian Active Learning with Look-ahead Amendment for Sea-drifter Trajectories. We observe noticeable benefits of BALLAST-aided sequential observer placement strategies on both synthetic and high-fidelity ocean current models. In addition, we developed a novel GP inference method -- the Vanilla SPDE Exchange (VaSE) -- to boost the GP posterior sampling efficiency, which is also of independent interest.

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 / 2 minor

Summary. The manuscript introduces BALLAST, a Bayesian active learning framework with a look-ahead amendment for sequential placement of Lagrangian sea-drifters to infer time-dependent vector fields. It employs a physics-informed spatio-temporal Gaussian process surrogate and proposes the Vanilla SPDE Exchange (VaSE) method to improve GP posterior sampling efficiency. The authors report noticeable performance benefits over standard space-filling and ad-hoc baselines on both synthetic test cases and high-fidelity ocean current models.

Significance. If the central claims are supported by the experiments, the work offers a principled approach to observer placement that accounts for advection under uncertain vector fields, which could improve data efficiency in oceanographic and marine applications. The VaSE sampling technique is presented as potentially useful beyond this setting for scalable GP inference.

major comments (2)
  1. [Experiments (high-fidelity results)] The headline performance gains on high-fidelity models rest on the assumption that the physics-informed spatio-temporal GP produces sufficiently accurate multi-step trajectory forecasts for the look-ahead amendment to add value. The manuscript does not appear to include direct quantitative validation (e.g., held-out predictive error or trajectory forecast skill scores) of the surrogate's forward integration accuracy on the ocean models; without this, the observed benefits could be driven primarily by the base active-learning loop rather than the amendment.
  2. [Section 3 (surrogate model)] The physics-informed constraints and Matérn-type kernel in the GP surrogate may not capture sub-grid turbulence or strong non-stationarity present in real ocean fields. A targeted ablation or sensitivity test showing that look-ahead utility degrades gracefully when surrogate error increases would strengthen the load-bearing claim that the amendment is responsible for the reported improvements.
minor comments (2)
  1. [Section 3] Notation for the look-ahead utility function and the VaSE sampling procedure could be clarified with a small algorithmic pseudocode box to aid reproducibility.
  2. [Figures 4-6] Figure captions for the trajectory visualizations should explicitly state the number of independent runs and whether error bars represent standard deviation or standard error.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their constructive and detailed comments, which have helped us identify opportunities to strengthen the manuscript. We address each major comment below with proposed revisions where appropriate.

read point-by-point responses

  1. Referee: [Experiments (high-fidelity results)] The headline performance gains on high-fidelity models rest on the assumption that the physics-informed spatio-temporal GP produces sufficiently accurate multi-step trajectory forecasts for the look-ahead amendment to add value. The manuscript does not appear to include direct quantitative validation (e.g., held-out predictive error or trajectory forecast skill scores) of the surrogate's forward integration accuracy on the ocean models; without this, the observed benefits could be driven primarily by the base active-learning loop rather than the amendment.

    Authors: We agree that direct quantitative validation of the surrogate's multi-step forecast accuracy on the high-fidelity ocean models would provide stronger evidence isolating the contribution of the look-ahead amendment. In the revised manuscript we will add held-out predictive error metrics and trajectory forecast skill scores for the physics-informed spatio-temporal GP on the ocean current data. These additions will allow readers to assess surrogate reliability and better attribute performance gains to the amendment versus the base active-learning loop. revision: yes

  2. Referee: [Section 3 (surrogate model)] The physics-informed constraints and Matérn-type kernel in the GP surrogate may not capture sub-grid turbulence or strong non-stationarity present in real ocean fields. A targeted ablation or sensitivity test showing that look-ahead utility degrades gracefully when surrogate error increases would strengthen the load-bearing claim that the amendment is responsible for the reported improvements.

    Authors: We acknowledge the potential limitations of the Matérn kernel and physics-informed constraints with respect to sub-grid turbulence and non-stationarity. To address this, we will add a targeted sensitivity analysis in the revised manuscript. This study will introduce controlled increases in surrogate error (e.g., via additive noise to the physics-informed predictions) and demonstrate the resulting degradation in look-ahead utility. While our existing synthetic and high-fidelity experiments already show consistent benefits under realistic conditions, this ablation will provide explicit support for the amendment's role. revision: yes

Circularity Check

0 steps flagged

No significant circularity; derivation is self-contained on standard GP active learning with trajectory look-ahead

full rationale

The paper defines BALLAST as Bayesian active learning augmented by look-ahead amendment that integrates future observer trajectories under the current physics-informed spatio-temporal GP surrogate. This construction uses the surrogate's predictive mean and uncertainty to score candidate placements, which is a standard forward simulation step in sequential design rather than a fitted quantity renamed as a prediction. The VaSE sampling method is presented as an independent efficiency improvement for SPDE-based GPs. No load-bearing claim reduces by the paper's own equations to a self-definition, a self-citation uniqueness theorem, or an ansatz imported from prior author work. Empirical benefits are reported on held-out synthetic and high-fidelity ocean fields, providing an external benchmark that does not collapse into the fitting procedure itself. The central performance claims therefore remain independently falsifiable.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract-only review; no explicit free parameters, axioms, or invented entities are described in sufficient detail to populate the ledger.

pith-pipeline@v0.9.0 · 5740 in / 1130 out tokens · 107011 ms · 2026-05-22T12:38:56.788065+00:00 · methodology

Review history (2 revisions) →

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

53 extracted references · 53 canonical work pages

  1. [1]

    Gaussian processes with linear operator inequality constraints

    Christian Agrell. Gaussian processes with linear operator inequality constraints . Journal of Machine Learning Research, 20 0 (135): 0 1--36, 2019

    work page 2019

  2. [2]

    Kernels for vector-valued functions: A review

    Mauricio A Alvarez, Lorenzo Rosasco, and Neil D Lawrence. Kernels for vector-valued functions: A review . Foundations and Trends in Machine Learning , 4 0 (3): 0 195--266, 2012

    work page 2012

  3. [3]

    A Bayesian approach to Lagrangian data assimilation

    Amit Apte, Christopher KRT Jones, and AM Stuart. A Bayesian approach to Lagrangian data assimilation . Tellus A: Dynamic Meteorology and Oceanography, 60 0 (2): 0 336--347, 2008

    work page 2008

  4. [4]

    Gaussian processes at the Helm (holtz) a more fluid model for ocean currents

    Renato Berlinghieri, Brian L Trippe, David R Burt, Ryan Giordano, Kaushik Srinivasan, Tamay \"O zg \"o kmen, Junfei Xia, and Tamara Broderick. Gaussian processes at the Helm (holtz) a more fluid model for ocean currents . In Proceedings of the 40th International Conference on Machine Learning, pages 2113--2163, 2023

    work page 2023

  5. [5]

    The Helmholtz-Hodge decomposition—a survey

    Harsh Bhatia, Gregory Norgard, Valerio Pascucci, and Peer-Timo Bremer. The Helmholtz-Hodge decomposition—a survey . IEEE Transactions on Visualization and Computer Graphics, 19 0 (8): 0 1386--1404, 2012

    work page 2012

  6. [6]

    A causation-based computationally efficient strategy for deploying Lagrangian drifters to improve real-time state estimation

    Erik Bollt, Nan Chen, and Stephen Wiggins. A causation-based computationally efficient strategy for deploying Lagrangian drifters to improve real-time state estimation . Physica D: Nonlinear Phenomena, 467: 0 134283, 2024

    work page 2024

  7. [7]

    Lagrangian descriptors with uncertainty

    Nan Chen, Evelyn Lunasin, and Stephen Wiggins. Lagrangian descriptors with uncertainty. Physica D: Nonlinear Phenomena, 467: 0 134282, 2024 a

    work page 2024

  8. [8]

    Launching drifter observations in the presence of uncertainty

    Nan Chen, Evelyn Lunasin, and Stephen Wiggins. Launching drifter observations in the presence of uncertainty . Physica D: Nonlinear Phenomena, 460: 0 134086, 2024 b

    work page 2024

  9. [9]

    T. M. Cover and Joy A. Thomas. Elements of Information Theory . Wiley-Interscience, Hoboken, N.J, 2nd ed. edition, 2006. ISBN 9780471241959

    work page 2006

  10. [10]

    Ocean circulation kinetic energy: Reservoirs, sources, and sinks

    Raffaele Ferrari and Carl Wunsch. Ocean circulation kinetic energy: Reservoirs, sources, and sinks . Annual Review of Fluid Mechanics, 41 0 (1): 0 253--282, 2009

    work page 2009

  11. [11]

    Deep adaptive design: Amortizing sequential Bayesian experimental design

    Adam Foster, Desi R Ivanova, Ilyas Malik, and Tom Rainforth. Deep adaptive design: Amortizing sequential Bayesian experimental design . In International conference on machine learning, pages 3384--3395. PMLR, 2021

    work page 2021

  12. [12]

    An unstructured-grid, finite-volume, nonhydrostatic, parallel coastal ocean simulator

    OB Fringer, M Gerritsen, and RL Street. An unstructured-grid, finite-volume, nonhydrostatic, parallel coastal ocean simulator . Ocean modelling, 14 0 (3-4): 0 139--173, 2006

    work page 2006

  13. [13]

    Surrogates: Gaussian Process Modeling, Design, and Optimization for the Applied Sciences

    Robert B Gramacy. Surrogates: Gaussian Process Modeling, Design, and Optimization for the Applied Sciences . Chapman and Hall/CRC, 2020

    work page 2020

  14. [14]

    Lagrangian Analysis and Prediction of Coastal and Ocean Dynamics

    Annalisa Griffa, AD Kirwan Jr, Arthur J Mariano, Tamay \"O zg \"o kmen, and H Thomas Rossby. Lagrangian Analysis and Prediction of Coastal and Ocean Dynamics . Cambridge University Press, 2007

    work page 2007

  15. [15]

    Spatio-temporal variational Gaussian processes

    Oliver Hamelijnck, William Wilkinson, Niki Loppi, Arno Solin, and Theodoros Damoulas. Spatio-temporal variational Gaussian processes . Advances in Neural Information Processing Systems, 34: 0 23621--23633, 2021

    work page 2021

  16. [16]

    Physics-Informed Variational State-Space Gaussian Processes

    Oliver Hamelijnck, Arno Solin, and Theodoros Damoulas. Physics-Informed Variational State-Space Gaussian Processes . In Advances in Neural Information Processing Systems, volume 37, pages 98505--98536. Curran Associates, Inc., 2024

    work page 2024

  17. [17]

    Kalman filtering and smoothing solutions to temporal Gaussian process regression models

    Jouni Hartikainen and Simo S \"a rkk \"a . Kalman filtering and smoothing solutions to temporal Gaussian process regression models . In 2010 IEEE international workshop on machine learning for signal processing, pages 379--384. IEEE, 2010

    work page 2010

  18. [18]

    Nesting particle filters for experimental design in dynamical systems

    Sahel Iqbal, Adrien Corenflos, Simo S \"a rkk \"a , and Hany Abdulsamad. Nesting particle filters for experimental design in dynamical systems . In Proceedings of the 41st International Conference on Machine Learning, pages 21047--21068, 2024

    work page 2024

  19. [19]

    Oil spill modeling: A critical review on current trends, perspectives, and challenges

    Panagiota Keramea, Katerina Spanoudaki, George Zodiatis, Georgios Gikas, and Georgios Sylaios. Oil spill modeling: A critical review on current trends, perspectives, and challenges . Journal of Marine Science and Engineering, 9 0 (2): 0 181, 2021

    work page 2021

  20. [20]

    Robust and Conjugate Spatio-Temporal Gaussian Processes

    William Laplante, Matias Altamirano, Andrew Duncan, Jeremias Knoblauch, and Fran c ois-Xavier Briol. Robust and Conjugate Spatio-Temporal Gaussian Processes . arXiv preprint arXiv:2502.02450, 2025

    work page arXiv 2025

  21. [21]

    A gridded surface current product for the Gulf of Mexico from consolidated drifter measurements

    Jonathan M Lilly and Paula P \'e rez-Brunius. A gridded surface current product for the Gulf of Mexico from consolidated drifter measurements . Earth System Science Data, 13 0 (2): 0 645--669, 2021

    work page 2021

  22. [22]

    Fractional Brownian motion, the Mat \'e rn process, and stochastic modeling of turbulent dispersion

    Jonathan M Lilly, Adam M Sykulski, Jeffrey J Early, and Sofia C Olhede. Fractional Brownian motion, the Mat \'e rn process, and stochastic modeling of turbulent dispersion . Nonlinear Processes in Geophysics, 24 0 (3): 0 481--514, 2017

    work page 2017

  23. [23]

    An explicit link between Gaussian fields and Gaussian Markov random fields: the stochastic partial differential equation approach

    Finn Lindgren, H vard Rue, and Johan Lindstr \"o m. An explicit link between Gaussian fields and Gaussian Markov random fields: the stochastic partial differential equation approach . Journal of the Royal Statistical Society Series B: Statistical Methodology, 73 0 (4): 0 423--498, 2011

    work page 2011

  24. [24]

    The SPDE approach for Gaussian and non-Gaussian fields: 10 years and still running

    Finn Lindgren, David Bolin, and H vard Rue. The SPDE approach for Gaussian and non-Gaussian fields: 10 years and still running . Spatial Statistics, 50: 0 100599, 2022

    work page 2022

  25. [25]

    On a measure of the information provided by an experiment

    Dennis V Lindley. On a measure of the information provided by an experiment . The Annals of Mathematical Statistics, 27 0 (4): 0 986--1005, 1956

    work page 1956

  26. [26]

    Advances in the application of surface drifters

    Rick Lumpkin, Tamay \"O zg \"o kmen, and Luca Centurioni. Advances in the application of surface drifters . Annual Review of Marine Science, 9 0 (1): 0 59--81, 2017

    work page 2017

  27. [27]

    Lagrangian descriptors: A method for revealing phase space structures of general time dependent dynamical systems

    Ana M Mancho, Stephen Wiggins, Jezabel Curbelo, and Carolina Mendoza. Lagrangian descriptors: A method for revealing phase space structures of general time dependent dynamical systems. Communications in Nonlinear Science and Numerical Simulation, 18 0 (12): 0 3530--3557, 2013

    work page 2013

  28. [28]

    GPJax: A Gaussian Process Framework in JAX

    Thomas Pinder and Daniel Dodd. GPJax: A Gaussian Process Framework in JAX . Journal of Open Source Software, 7 0 (75): 0 4455, 2022. doi:10.21105/joss.04455

    work page doi:10.21105/joss.04455 2022

  29. [29]

    Drifter launch strategies based on Lagrangian templates

    AC Poje, M Toner, AD Kirwan Jr, and CKRT Jones. Drifter launch strategies based on Lagrangian templates . Journal of Physical Oceanography, 32 0 (6): 0 1855--1869, 2002

    work page 2002

  30. [30]

    Inferring flow energy, space scales, and timescales: freely drifting vs

    Aurelien Luigi Serge Ponte, Lachlan C Astfalck, Matthew D Rayson, Andrew P Zulberti, and Nicole L Jones. Inferring flow energy, space scales, and timescales: freely drifting vs. fixed-point observations . Nonlinear Processes in Geophysics, 31 0 (4): 0 571--586, 2024

    work page 2024

  31. [31]

    Modern Bayesian experimental design

    Tom Rainforth, Adam Foster, Desi R Ivanova, and Freddie Bickford Smith. Modern Bayesian experimental design . Statistical Science, 39 0 (1): 0 100--114, 2024

    work page 2024

  32. [32]

    A seasonal harmonic model for internal tide amplitude prediction

    Matthew D Rayson, Nicole L Jones, Gregory N Ivey, and Yankun Gong. A seasonal harmonic model for internal tide amplitude prediction . Journal of Geophysical Research: Oceans, 126 0 (10): 0 e2021JC017570, 2021

    work page 2021

  33. [33]

    Monte Carlo Statistical Methods , volume 2

    Christian P Robert and George Casella. Monte Carlo Statistical Methods , volume 2. Springer, 1999

    work page 1999

  34. [34]

    Gaussian Markov Random Fields: Theory and Applications

    Havard Rue and Leonhard Held. Gaussian Markov Random Fields: Theory and Applications . Chapman and Hall/CRC, 2005

    work page 2005

  35. [35]

    A review of modern computational algorithms for Bayesian optimal design

    Elizabeth G Ryan, Christopher C Drovandi, James M McGree, and Anthony N Pettitt. A review of modern computational algorithms for Bayesian optimal design . International Statistical Review, 84 0 (1): 0 128--154, 2016

    work page 2016

  36. [36]

    Scalable inference for structured Gaussian process models

    Yunus Saat c i. Scalable inference for structured Gaussian process models . PhD thesis, University of Cambridge, 2012

    work page 2012

  37. [37]

    Using flow geometry for drifter deployment in Lagrangian data assimilation

    H Salman, K Ide, and Christopher KRT Jones. Using flow geometry for drifter deployment in Lagrangian data assimilation . Tellus A: Dynamic Meteorology and Oceanography, 60 0 (2): 0 321--335, 2008

    work page 2008

  38. [38]

    Applied Stochastic Differential Equations , volume 10

    Simo S \"a rkk \"a and Arno Solin. Applied Stochastic Differential Equations , volume 10. Cambridge University Press, 2019

    work page 2019

  39. [39]

    Spatiotemporal learning via infinite-dimensional Bayesian filtering and smoothing: A look at Gaussian process regression through Kalman filtering

    Simo Sarkka, Arno Solin, and Jouni Hartikainen. Spatiotemporal learning via infinite-dimensional Bayesian filtering and smoothing: A look at Gaussian process regression through Kalman filtering . IEEE Signal Processing Magazine, 30 0 (4): 0 51--61, 2013

    work page 2013

  40. [40]

    Active learning literature survey

    Burr Settles. Active learning literature survey . 2009

    work page 2009

  41. [41]

    Prediction-Centric Uncertainty Quantification via MMD

    Zheyang Shen, Jeremias Knoblauch, Samuel Power, and Chris J Oates. Prediction-Centric Uncertainty Quantification via MMD . In International Conference on Artificial Intelligence and Statistics, pages 649--657. PMLR, 2025

    work page 2025

  42. [42]

    Stochastic differential equation methods for spatio-temporal Gaussian process regression

    Arno Solin. Stochastic differential equation methods for spatio-temporal Gaussian process regression . PhD thesis, Aalto University, 2016

    work page 2016

  43. [43]

    Gaussian Process Optimization in the Bandit Setting: No Regret and Experimental Design

    Niranjan Srinivas, Andreas Krause, Sham Kakade, and Matthias Seeger. Gaussian Process Optimization in the Bandit Setting: No Regret and Experimental Design . In Proceedings of the 27th International Conference on Machine Learning, pages 1015--1022. Omnipress, 2010

    work page 2010

  44. [44]

    An Introduction to Numerical Analysis

    Endre S \"u li and David F Mayers. An Introduction to Numerical Analysis . Cambridge University Press, 2003

    work page 2003

  45. [45]

    Numerical Linear Algebra , volume 50

    Lloyd N Trefethen and David Bau III. Numerical Linear Algebra , volume 50. SIAM, 1997

    work page 1997

  46. [46]

    An efficient drifters deployment strategy to evaluate water current velocity fields

    Murad Tukan, Eli Biton, and Roee Diamant. An efficient drifters deployment strategy to evaluate water current velocity fields . IEEE Journal of Oceanic Engineering, 2024

    work page 2024

  47. [47]

    Dispersion of surface drifters in the tropical Atlantic

    Erik Van Sebille, Erik Zettler, Nicolas Wienders, Linda Amaral-Zettler, Shane Elipot, and Rick Lumpkin. Dispersion of surface drifters in the tropical Atlantic . Frontiers in Marine Science, 7: 0 607426, 2021

    work page 2021

  48. [48]

    Nearshore internal bores and turbulent mixing in southern Monterey Bay

    Ryan K Walter, C Brock Woodson, Robert S Arthur, Oliver B Fringer, and Stephen G Monismith. Nearshore internal bores and turbulent mixing in southern Monterey Bay . Journal of Geophysical Research: Oceans, 117 0 (C7), 2012

    work page 2012

  49. [49]

    On stationary processes in the plane

    Peter Whittle. On stationary processes in the plane. Biometrika, pages 434--449, 1954

    work page 1954

  50. [50]

    Stochastic processes in several dimensions

    Peter Whittle. Stochastic processes in several dimensions. Bulletin of the International Statistical Institute, 40: 0 974--994, 1963

    work page 1963

  51. [51]

    Gaussian Processes for Machine Learning , volume 2

    Christopher KI Williams and Carl Edward Rasmussen. Gaussian Processes for Machine Learning , volume 2. MIT press Cambridge, MA, 2006

    work page 2006

  52. [52]

    HHD-GP: Incorporating Helmholtz-Hodge Decomposition into Gaussian Processes for Learning Dynamical Systems

    Hao Xu and Jia Pan. HHD-GP: Incorporating Helmholtz-Hodge Decomposition into Gaussian Processes for Learning Dynamical Systems . In Advances in Neural Information Processing Systems, volume 37, pages 67282--67318, 2024

    work page 2024

  53. [53]

    BALLAST: Bayesian Active Learning with Look-ahead Amendment for Sea-drifter Trajectories

    Rui-Yang Zhang, Henry Moss, Lachlan Astfalck, Edward Cripps, and David S Leslie. BALLAST: Bayesian Active Learning with Look-ahead Amendment for Sea-drifter Trajectories . In NeurIPS 2024 Workshop on Bayesian Decision-making and Uncertainty, 2024

    work page 2024