pith. machine review for the scientific record. sign in

arxiv: 2605.09775 · v1 · submitted 2026-05-10 · 💻 cs.LG · math.OC

Recognition: 2 theorem links

· Lean Theorem

Bayesian Optimization with Structured Measurements: A Vector-Valued RKHS Framework

Wenbin Wang , Colin N. Jones
Authors on Pith no claims yet

Pith reviewed 2026-05-12 02:02 UTC · model grok-4.3

classification 💻 cs.LG math.OC
keywords bayesian optimizationvector-valued RKHSstructured measurementskernel ridge regressionregret boundsupper confidence boundsample efficiency
0
0 comments X

The pith

Bayesian optimization can use vector-valued measurements of trajectories or fields to learn faster than with single scalar values.

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

The paper extends Bayesian optimization from scalar objectives to cases where each query returns rich structured outputs such as multidimensional data or functions. It models the unknown system as an operator belonging to a vector-valued reproducing kernel Hilbert space and derives concentration bounds for kernel ridge regression directly in that measurement space. These bounds support an upper confidence bound acquisition function that carries regret guarantees recovering sublinear rates for standard kernels. A reader would care because each costly evaluation now supplies far more information, enabling information transfer across objectives and better handling of time-varying problems.

Core claim

We study Bayesian optimization over a vector-valued operator with structured measurements, where each measurement observes multidimensional or functional outputs rather than a single scalar value and the objective is defined as a linear functional of these measurements. Assuming the unknown operator lies in a vector-valued reproducing kernel Hilbert space, we derive high-probability concentration bounds for the kernel ridge regression estimator directly in the measurement space. Building on these results, we propose an algorithm based on the upper confidence bound acquisition function with regret guarantees under mild assumptions, recovering sublinear rates for common kernels.

What carries the argument

Vector-valued reproducing kernel Hilbert space for the unknown operator, together with high-probability concentration bounds on kernel ridge regression performed in the space of structured measurements.

If this is right

  • The UCB acquisition function yields sublinear regret for common kernel choices.
  • Structured measurements improve sample efficiency by transferring information across related objectives.
  • The method supports adaptation when the underlying system varies over time.
  • Uncertainty quantification works directly in general Hilbert spaces of measurements.

Where Pith is reading between the lines

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

  • Similar structured-output modeling could be applied to other acquisition strategies or to non-Bayesian optimization routines that currently collapse outputs to scalars.
  • Domains such as control or robotics might optimize directly over trajectory outputs without intermediate scalar reduction.
  • The dependence of convergence rates on the choice of linear functional defining the objective could be studied further to tighten practical performance.

Load-bearing premise

The unknown operator that produces the structured vector or functional measurements belongs to a vector-valued reproducing kernel Hilbert space.

What would settle it

An experiment in which the true mapping from inputs to structured outputs lies outside any vector-valued RKHS and the observed cumulative regret grows faster than the sublinear rate predicted by the analysis.

Figures

Figures reproduced from arXiv: 2605.09775 by Colin N. Jones, Wenbin Wang.

Figure 1
Figure 1. Figure 1: Vector-valued BO framework. Red box: classical BO framework. Blue box: vector-valued BO framework introduced in this work. Practical implication. This formulation gen￾eralizes different observation regimes, ranging from full observation (M = I) to scalar ob￾servation (M is a linear functional). The inner product ⟨m, Mf(x)⟩M naturally arises in many applications where the objective depends on vector-valued … view at source ↗
Figure 2
Figure 2. Figure 2: Comparison of simple regret and cumulative regret for three test operators with different [PITH_FULL_IMAGE:figures/full_fig_p008_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Real-world tuning of an MPC controller for building climate control. [PITH_FULL_IMAGE:figures/full_fig_p009_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Comparison of simple regret and cumulative regret for additional test operators under full [PITH_FULL_IMAGE:figures/full_fig_p026_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Comparison of simple regret and cumulative regret for three test operators under partial [PITH_FULL_IMAGE:figures/full_fig_p027_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Comparison of simple regret and cumulative regret for additional test operators under [PITH_FULL_IMAGE:figures/full_fig_p028_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Comparison between vvBO and CTBO for the eggholder test operator for different contexts. [PITH_FULL_IMAGE:figures/full_fig_p029_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Comparison of the confidence bounds of CTBO for the eggholder test operator for different [PITH_FULL_IMAGE:figures/full_fig_p029_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Daily cost during learning and validation phases. [PITH_FULL_IMAGE:figures/full_fig_p030_9.png] view at source ↗
read the original abstract

Bayesian optimization (BO) is an efficient framework for optimizing expensive black-box functions. However, it is typically formulated as learning an end-to-end mapping from inputs to scalar objectives, thereby discarding the potentially rich information whenever a structured system output is available. In this work, we study Bayesian optimization over a vector-valued operator with structured measurements, where each measurement observes multidimensional or functional outputs, e.g., trajectories or spatial fields, rather than a single scalar value. The objective is then defined as a linear functional of these measurements. This allows each observation to reveal substantially richer information about the underlying system compared to scalar observations. Assuming the unknown operator lies in a vector-valued reproducing kernel Hilbert space (RKHS), we derive high-probability concentration bounds for the kernel ridge regression (KRR) estimator directly in the measurement space, characterizing uncertainty in a general Hilbert space. Building on these results, we propose an algorithm based on the upper confidence bound (UCB) acquisition function with regret guarantees under mild assumptions, recovering sublinear rates for common kernels. Empirically, we demonstrate that leveraging structured measurements leads to improved sample efficiency by enabling efficient transfer of information across objectives and adaptation to time-varying settings.

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

0 major / 3 minor

Summary. The paper extends Bayesian optimization to vector-valued operators observed via structured multidimensional or functional measurements (e.g., trajectories), rather than scalar outputs. The objective is a linear functional of these measurements. Under the assumption that the unknown operator lies in a vector-valued RKHS, the authors derive high-probability concentration bounds for the kernel ridge regression estimator directly in the measurement space. They then propose a UCB acquisition function algorithm with regret guarantees that recover sublinear rates for common kernels, and provide empirical evidence of improved sample efficiency through information transfer across objectives and adaptation to time-varying settings.

Significance. If the derived concentration bounds and regret guarantees hold, the work offers a principled extension of scalar BO to richer observation models, enabling more efficient optimization when structured outputs are available. The recovery of standard sublinear regret rates for common kernels is a strength, as is the explicit use of vector-valued RKHS to handle functional data. This could benefit applications in control, robotics, and scientific computing where measurements yield trajectories or fields rather than scalars.

minor comments (3)
  1. §2 (Preliminaries): the notation for the vector-valued RKHS and the associated operator-valued kernel could be clarified with an explicit example of how the linear functional defining the objective is represented in the measurement space.
  2. §4 (Algorithm): the transition from the KRR estimator to the UCB acquisition function is described at a high level; a short pseudocode block or explicit formula for the acquisition function in terms of the vector-valued posterior would improve readability.
  3. §5 (Experiments): the time-varying setting experiment would benefit from a clearer description of how the operator drift is modeled and whether the regret analysis extends directly or requires additional assumptions.

Simulated Author's Rebuttal

0 responses · 0 unresolved

We thank the referee for the positive summary of our work and the recommendation for minor revision. The referee accurately captures the core contribution: extending Bayesian optimization to vector-valued operators observed through structured measurements in a vector-valued RKHS, with concentration bounds and UCB regret guarantees that recover standard sublinear rates.

Circularity Check

0 steps flagged

No significant circularity; derivation self-contained under stated RKHS assumptions

full rationale

The paper states the core assumption that the unknown operator lies in a vector-valued RKHS, derives high-probability KRR concentration bounds directly in the measurement space from that assumption, and then constructs a UCB acquisition function whose regret analysis recovers sublinear rates for common kernels under mild conditions. These steps are presented as standard derivations in the Hilbert-space setting with no reduction of a 'prediction' to a fitted quantity by construction, no load-bearing self-citation chain, and no ansatz or uniqueness claim imported from prior author work. The central claims therefore remain independent of the inputs they are derived from.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

Review performed on abstract only; full paper details on additional parameters or axioms unavailable. The central modeling assumption is stated explicitly.

axioms (1)
  • domain assumption The unknown operator lies in a vector-valued reproducing kernel Hilbert space (RKHS)
    Invoked in the abstract to derive concentration bounds for the KRR estimator.

pith-pipeline@v0.9.0 · 5507 in / 1165 out tokens · 50593 ms · 2026-05-12T02:02:56.373366+00:00 · methodology

discussion (0)

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

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

52 extracted references · 52 canonical work pages · 3 internal anchors

  1. [1]

    Pitman Publishing, 1981

    Naum Ilich Akhiezer.Theory of linear operators in Hilbert space. Pitman Publishing, 1981

    work page 1981

  2. [2]

    David Blum, Javier Arroyo, Sen Huang, Ján Drgoˇna, Filip Jorissen, Harald Taxt Walnum, Yan Chen, Kyle Benne, Draguna Vrabie, Michael Wetter, et al. Building optimization testing framework (BOPTEST) for simulation-based benchmarking of control strategies in buildings.Journal of Building Performance Simulation, 14(5):586–610, 2021

    work page 2021

  3. [3]

    Multi-task Gaussian process prediction.Advances in Neural Information Processing Systems, 20, 2007

    Edwin V Bonilla, Kian Chai, and Christopher Williams. Multi-task Gaussian process prediction.Advances in Neural Information Processing Systems, 20, 2007

    work page 2007

  4. [4]

    Infinite task learning in RKHSs

    Romain Brault, Alex Lambert, Zoltán Szabó, Maxime Sangnier, and Florence d’Alché Buc. Infinite task learning in RKHSs. InThe 22nd International Conference on Artificial Intelligence and Statistics, pages 1294–1302. PMLR, 2019

    work page 2019

  5. [5]

    On controller tuning with time-varying Bayesian optimization

    Paul Brunzema, Alexander V on Rohr, and Sebastian Trimpe. On controller tuning with time-varying Bayesian optimization. In2022 IEEE 61st Conference on Decision and Control, pages 4046–4052. IEEE, 2022

    work page 2022

  6. [6]

    Universal multi-task kernels.The Journal of Machine Learning Research, 9:1615–1646, 2008

    Andrea Caponnetto, Charles A Micchelli, Massimiliano Pontil, and Yiming Ying. Universal multi-task kernels.The Journal of Machine Learning Research, 9:1615–1646, 2008

    work page 2008

  7. [7]

    Vector valued reproducing kernel Hilbert spaces of integrable functions and Mercer theorem.Analysis and Applications, 4(04):377–408, 2006

    Claudio Carmeli, Ernesto De Vito, and Alessandro Toigo. Vector valued reproducing kernel Hilbert spaces of integrable functions and Mercer theorem.Analysis and Applications, 4(04):377–408, 2006

    work page 2006

  8. [8]

    Vector valued reproducing kernel Hilbert spaces and universality.Analysis and Applications, 8(01):19–61, 2010

    Claudio Carmeli, Ernesto De Vito, Alessandro Toigo, and Veronica Umanitá. Vector valued reproducing kernel Hilbert spaces and universality.Analysis and Applications, 8(01):19–61, 2010

    work page 2010

  9. [9]

    Model predictive control for indoor thermal comfort and energy optimization using occupant feedback.Energy and Buildings, 102:357–369, 2015

    Xiao Chen, Qian Wang, and Jelena Srebric. Model predictive control for indoor thermal comfort and energy optimization using occupant feedback.Energy and Buildings, 102:357–369, 2015

    work page 2015

  10. [10]

    On kernelized multi-armed bandits

    Sayak Ray Chowdhury and Aditya Gopalan. On kernelized multi-armed bandits. InInternational Conference on Machine Learning, pages 844–853. PMLR, 2017

    work page 2017

  11. [11]

    No-regret algorithms for multi-task Bayesian optimization

    Sayak Ray Chowdhury and Aditya Gopalan. No-regret algorithms for multi-task Bayesian optimization. InInternational Conference on Artificial Intelligence and Statistics, pages 1873–1881. PMLR, 2021

    work page 2021

  12. [12]

    Efficient human-in-the-loop MPC tuning with multi-task preferential Bayesian optimization.Control Engineering Practice, 173:106972, 2026

    João PL Coutinho, You Peng, Ricardo Rendall, Kaiwen Ma, Swee-Teng Chin, Ivan Castillo, and Marco S Reis. Efficient human-in-the-loop MPC tuning with multi-task preferential Bayesian optimization.Control Engineering Practice, 173:106972, 2026

    work page 2026

  13. [13]

    Multi-objective Bayesian optimization over high-dimensional search spaces

    Samuel Daulton, David Eriksson, Maximilian Balandat, and Eytan Bakshy. Multi-objective Bayesian optimization over high-dimensional search spaces. InUncertainty in Artificial Intelligence, pages 507–517. PMLR, 2022

    work page 2022

  14. [14]

    Regularized multi–task learning

    Theodoros Evgeniou and Massimiliano Pontil. Regularized multi–task learning. InProceedings of the 10th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pages 109–117, 2004

    work page 2004

  15. [15]

    A Tutorial on Bayesian Optimization

    Peter I Frazier. A tutorial on Bayesian optimization.arXiv preprint arXiv:1807.02811, 2018

    work page internal anchor Pith review arXiv 2018

  16. [16]

    Cambridge University Press, 2023

    Roman Garnett.Bayesian optimization. Cambridge University Press, 2023

    work page 2023

  17. [17]

    Linas Gelazanskas and Kelum A.A. Gamage. Demand side management in smart grid: A review and proposals for future direction.Sustainable Cities and Society, 11:22–30, 2014

    work page 2014

  18. [18]

    Functional data analysis: An introduction and recent developments.Biometrical Journal, 66(7):e202300363, 2024

    Jan Gertheiss, David Rügamer, Bernard XW Liew, and Sonja Greven. Functional data analysis: An introduction and recent developments.Biometrical Journal, 66(7):e202300363, 2024

    work page 2024

  19. [19]

    Fine-tuning of neural network approximate MPC without retraining via Bayesian optimization

    Henrik Hose, Paul Brunzema, Alexander V on Rohr, Alexander Gräfe, Angela P Schoellig, and Sebastian Trimpe. Fine-tuning of neural network approximate MPC without retraining via Bayesian optimization. arXiv preprint arXiv:2512.14350, 2025

    work page arXiv 2025

  20. [20]

    Function-on-function Bayesian optimization

    Jingru Huang, Haijie Xu, Manrui Jiang, and Chen Zhang. Function-on-function Bayesian optimization. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 40, pages 21994–22002, 2026

    work page 2026

  21. [21]

    Korotyaev

    Hiroshi Isozaki and Evgeny L. Korotyaev. Trace formulas for Schrödinger operators – from the view point of complex analysis. InRIMS Kôkyûroku, pages 16–32, 2011. 10

    work page 2011

  22. [22]

    PhD thesis, INRIA, 2009

    Hachem Kadri, Emmanuel Duflos, Manuel Davy, Philippe Preux, and Stephane Canu.General framework for nonlinear functional regression with reproducing kernel Hilbert spaces. PhD thesis, INRIA, 2009

    work page 2009

  23. [23]

    Nonlinear functional regression: a functional RKHS approach

    Hachem Kadri, Emmanuel Duflos, Philippe Preux, Stéphane Canu, and Manuel Davy. Nonlinear functional regression: a functional RKHS approach. InProceedings of the Thirteenth International Conference on Artificial Intelligence and Statistics, pages 374–380. JMLR Workshop and Conference Proceedings, 2010

    work page 2010

  24. [24]

    Operator-valued kernels for learning from functional response data.Journal of Machine Learning Research, 17(20):1–54, 2016

    Hachem Kadri, Emmanuel Duflos, Philippe Preux, Stéphane Canu, Alain Rakotomamonjy, and Julien Audiffren. Operator-valued kernels for learning from functional response data.Journal of Machine Learning Research, 17(20):1–54, 2016

    work page 2016

  25. [25]

    Multi-objective Bayesian optimization algorithm

    Nazan Khan, David E Goldberg, and Martin Pelikan. Multi-objective Bayesian optimization algorithm. In Proceedings of the 4th Annual Conference on Genetic and Evolutionary Computation, pages 684–684, 2002

    work page 2002

  26. [26]

    An overview of demand side management control schemes for buildings in smart grids

    Anna Magdalena Kosek, Giuseppe Tommaso Costanzo, Henrik W Bindner, and Oliver Gehrke. An overview of demand side management control schemes for buildings in smart grids. In2013 IEEE International Conference on Smart Energy Grid Engineering, pages 1–9. IEEE, 2013

    work page 2013

  27. [27]

    Neural operator: Learning maps between function spaces with applications to PDEs.Journal of Machine Learning Research, 24(89):1–97, 2023

    Nikola Kovachki, Zongyi Li, Burigede Liu, Kamyar Azizzadenesheli, Kaushik Bhattacharya, Andrew Stuart, and Anima Anandkumar. Neural operator: Learning maps between function spaces with applications to PDEs.Journal of Machine Learning Research, 24(89):1–97, 2023

    work page 2023

  28. [28]

    Contextual Gaussian process bandit optimization.Advances in Neural Information Processing Systems, 24, 2011

    Andreas Krause and Cheng Ong. Contextual Gaussian process bandit optimization.Advances in Neural Information Processing Systems, 24, 2011

    work page 2011

  29. [29]

    Akshay Kudva and Joel A. Paulson. Bonsai: Structure-exploiting robust Bayesian optimization for networked black-box systems under uncertainty.Computers & Chemical Engineering, 204:109393, Jan 2026

    work page 2026

  30. [30]

    Optimal uncertainty bounds for multivariate kernel regression under bounded noise: A Gaussian process-based dual function.arXiv preprint arXiv:2603.16481, 2026

    Amon Lahr, Anna Scampicchio, Johannes Köhler, and Melanie N Zeilinger. Optimal uncertainty bounds for multivariate kernel regression under bounded noise: A Gaussian process-based dual function.arXiv preprint arXiv:2603.16481, 2026

    work page arXiv 2026

  31. [31]

    Neural Operator: Graph Kernel Network for Partial Differential Equations

    Zongyi Li, Nikola Kovachki, Kamyar Azizzadenesheli, Burigede Liu, Kaushik Bhattacharya, Andrew Stuart, and Anima Anandkumar. Neural operator: Graph kernel network for partial differential equations. arXiv preprint arXiv:2003.03485, 2020

    work page internal anchor Pith review arXiv 2003

  32. [32]

    Preferential Bayesian optimization with crash feedback.IEEE Robotics and Automation Letters, 2026

    Johanna Menn, David Stenger, and Sebastian Trimpe. Preferential Bayesian optimization with crash feedback.IEEE Robotics and Automation Letters, 2026

    work page 2026

  33. [33]

    On learning vector-valued functions.Neural computation, 17(1):177–204, 2005

    Charles A Micchelli and Massimiliano Pontil. On learning vector-valued functions.Neural computation, 17(1):177–204, 2005

    work page 2005

  34. [34]

    Universal kernels.Journal of Machine Learning Research, 7(12), 2006

    Charles A Micchelli, Yuesheng Xu, and Haizhang Zhang. Universal kernels.Journal of Machine Learning Research, 7(12), 2006

    work page 2006

  35. [35]

    Infinite-dimensional log-determinant divergences between positive definite trace class operators.Linear Algebra and Its Applications, 528:331–383, 2017

    Ha Quang Minh. Infinite-dimensional log-determinant divergences between positive definite trace class operators.Linear Algebra and Its Applications, 528:331–383, 2017

    work page 2017

  36. [36]

    Springer, 1997

    James O Ramsay and Bernard W Silverman.Functional data analysis. Springer, 1997

    work page 1997

  37. [37]

    A survey on kernel-based multi-task learning

    Carlos Ruiz, Carlos M Alaíz, and José R Dorronsoro. A survey on kernel-based multi-task learning. Neurocomputing, 577:127255, 2024

    work page 2024

  38. [38]

    Multitask learning with no regret: from improved confidence bounds to active learning.Advances in Neural Information Processing Systems, 36:6770–6781, 2023

    Pier Giuseppe Sessa, Pierre Laforgue, Nicolò Cesa-Bianchi, and Andreas Krause. Multitask learning with no regret: from improved confidence bounds to active learning.Advances in Neural Information Processing Systems, 36:6770–6781, 2023

    work page 2023

  39. [39]

    Disturbance-adaptive data-driven predictive control: Trading comfort violation for savings in building climate control.IEEE Transactions on Control Systems Technology, 2026

    Jicheng Shi, Christophe Salzmann, and Colin N Jones. Disturbance-adaptive data-driven predictive control: Trading comfort violation for savings in building climate control.IEEE Transactions on Control Systems Technology, 2026

    work page 2026

  40. [40]

    Information-theoretic regret bounds for Gaussian process optimization in the bandit setting.IEEE Transactions on Information Theory, 58(5):3250–3265, 2012

    Niranjan Srinivas, Andreas Krause, Sham M Kakade, and Matthias W Seeger. Information-theoretic regret bounds for Gaussian process optimization in the bandit setting.IEEE Transactions on Information Theory, 58(5):3250–3265, 2012

    work page 2012

  41. [41]

    Multi-task Bayesian optimization.Advances in Neural Information Processing Systems, 26, 2013

    Kevin Swersky, Jasper Snoek, and Ryan P Adams. Multi-task Bayesian optimization.Advances in Neural Information Processing Systems, 26, 2013. 11

    work page 2013

  42. [42]

    Bayesian functional optimization

    Ngo Anh Vien, Heiko Zimmermann, and Marc Toussaint. Bayesian functional optimization. InProceedings of the AAAI Conference on Artificial Intelligence, volume 32, 2018

    work page 2018

  43. [43]

    Decentralized feedback optimization via sensitivity decoupling: Stability and sub-optimality

    Wenbin Wang, Zhiyu He, Giuseppe Belgioioso, Saverio Bolognani, and Florian Dorfler. Decentralized feedback optimization via sensitivity decoupling: Stability and sub-optimality. In2024 European Control Conference, pages 3201–3206. IEEE, 2024

    work page 2024

  44. [44]

    Personalized building climate control with contextual preferential Bayesian optimization.arXiv preprint arXiv:2512.09481, 2025

    Wenbin Wang, Jicheng Shi, and Colin N Jones. Personalized building climate control with contextual preferential Bayesian optimization.arXiv preprint arXiv:2512.09481, 2025

    work page arXiv 2025

  45. [45]

    Data-driven adaptive building thermal controller tuning with constraints: A primal–dual contextual Bayesian optimization approach.Applied Energy, 358:122493, 2024

    Wenjie Xu, Bratislav Svetozarevic, Loris Di Natale, Philipp Heer, and Colin N Jones. Data-driven adaptive building thermal controller tuning with constraints: A primal–dual contextual Bayesian optimization approach.Applied Energy, 358:122493, 2024

    work page 2024

  46. [46]

    Wenjie Xu, Bratislav Svetozarevic, Loris Di Natale, Philipp Heer, and Colin N. Jones. Data-driven adaptive building thermal controller tuning with constraints: A primal–dual contextual Bayesian optimization approach.Applied Energy, 358:122493, 2024

    work page 2024

  47. [47]

    Implementation and performance analysis of a multi-energy building emulator

    Tao Yang, Konstantin Filonenko, Krzysztof Arendt, and Christain Veje. Implementation and performance analysis of a multi-energy building emulator. In2020 6th IEEE International Energy Conference, pages 451–456. IEEE, 2020

    work page 2020

  48. [48]

    What price to pay? Auto-tuning a building MPC controller for optimal economic cost

    Jiarui Yu, Jicheng Shi, Wenjie Xu, and Colin N Jones. Which price to pay? auto-tuning building mpc controller for optimal economic cost.arXiv preprint arXiv:2501.10859, 2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  49. [49]

    Tony Cai

    Ming Yuan and T. Tony Cai. A reproducing kernel hilbert space approach to functional linear regression. The Annals of Statistics, 38(6), December 2010. A Common linear functionals In this section, we provide examples of linear functionals in common Hilbert spaces. In all cases, continuous linear functionals admit a representation via the inner product, as...

    work page 2010

  50. [50]

    Linear kernel:R T ≤ O logT √ T ,

    work page

  51. [51]

    Gaussian kernel:R T ≤ O (logT) d+1√ T),

    work page

  52. [52]

    5X i=1 icos (i+ 1) u1 2 +i #

    Matérn kernel:R T ≤ O T d(d+1)/(2ν+d(d+1)) logT √ T . F Proof for Lemma 4 For the first part, according to the Fredholm determinant for positive definite trace class operators [35], we observe that log det I+λ −1GXT XT ⊗B M = TX t=1 ∞X i=1 log(1 + 1 λ λG t λBM i )≤ 1 λ TX t=1 ∞X i=1 λG t λBM i <∞. Hence, applying Fubini’s theorem for infinite series (swit...

    work page