Home > Research > Publications & Outputs > Efficient Parameter Estimation of Sampled Rando...

Research

Associated organisational units

Links

Keywords

View graph of relations

Efficient Parameter Estimation of Sampled Random Fields

Research output: Contribution to Journal/Magazine › Journal article

Published

Standard

Efficient Parameter Estimation of Sampled Random Fields. / Guillaumin, Arthur P.; Sykulski, Adam M.; Olhede, Sofia C. et al.
In: arXiv, 04.07.2019.

Research output: Contribution to Journal/Magazine › Journal article

Harvard

Guillaumin, AP, Sykulski, AM, Olhede, SC & Simons, FJ 2019, 'Efficient Parameter Estimation of Sampled Random Fields', arXiv. <https://arxiv.org/abs/1907.02447>

APA

Guillaumin, A. P., Sykulski, A. M., Olhede, S. C., & Simons, F. J. (2019). Efficient Parameter Estimation of Sampled Random Fields. arXiv. https://arxiv.org/abs/1907.02447

Vancouver

Guillaumin AP, Sykulski AM, Olhede SC, Simons FJ. Efficient Parameter Estimation of Sampled Random Fields. arXiv. 2019 Jul 4.

Author

Guillaumin, Arthur P. ; Sykulski, Adam M. ; Olhede, Sofia C. et al. / Efficient Parameter Estimation of Sampled Random Fields. In: arXiv. 2019.

Bibtex

@article{4229d9e9049f44a4afdbb851e5f0fb5c,
title = "Efficient Parameter Estimation of Sampled Random Fields",
abstract = "We provide a computationally and statistically efficient method for estimating the parameters of a stochastic Gaussian model observed on a spatial grid, which need not be rectangular. Standard methods are plagued by computational intractability, where designing methods that can be implemented for realistically sized problems has been an issue for a long time. This has motivated the use of the Fourier Transform and the Whittle likelihood approximation. The challenge of frequency-domain methods is to determine and account for observational boundary effects, missing data, and the shape of the observed spatial grid. In this paper we address these effects explicitly by proposing a new quasi-likelihood estimator. We prove consistency and asymptotic normality of our estimator in settings that include irregularly shaped grids. Our simulations show that the proposed method solves boundary issues with Whittle estimation for random fields, yielding parameter estimates with significantly reduced bias and error. We demonstrate the effectiveness of our method for incomplete lattices, in comparison to other recent methods. Finally, we apply our method to estimate the parameters of a Mat\'ern process used to model data from Venus' topography. ",
keywords = "stat.ME, stat.AP, stat.CO, stat.ML",
author = "Guillaumin, {Arthur P.} and Sykulski, {Adam M.} and Olhede, {Sofia C.} and Simons, {Frederik J.}",
year = "2019",
month = jul,
day = "4",
language = "English",
journal = "arXiv",

}

RIS

TY - JOUR

T1 - Efficient Parameter Estimation of Sampled Random Fields

AU - Guillaumin, Arthur P.

AU - Sykulski, Adam M.

AU - Olhede, Sofia C.

AU - Simons, Frederik J.

PY - 2019/7/4

Y1 - 2019/7/4

N2 - We provide a computationally and statistically efficient method for estimating the parameters of a stochastic Gaussian model observed on a spatial grid, which need not be rectangular. Standard methods are plagued by computational intractability, where designing methods that can be implemented for realistically sized problems has been an issue for a long time. This has motivated the use of the Fourier Transform and the Whittle likelihood approximation. The challenge of frequency-domain methods is to determine and account for observational boundary effects, missing data, and the shape of the observed spatial grid. In this paper we address these effects explicitly by proposing a new quasi-likelihood estimator. We prove consistency and asymptotic normality of our estimator in settings that include irregularly shaped grids. Our simulations show that the proposed method solves boundary issues with Whittle estimation for random fields, yielding parameter estimates with significantly reduced bias and error. We demonstrate the effectiveness of our method for incomplete lattices, in comparison to other recent methods. Finally, we apply our method to estimate the parameters of a Mat\'ern process used to model data from Venus' topography.

AB - We provide a computationally and statistically efficient method for estimating the parameters of a stochastic Gaussian model observed on a spatial grid, which need not be rectangular. Standard methods are plagued by computational intractability, where designing methods that can be implemented for realistically sized problems has been an issue for a long time. This has motivated the use of the Fourier Transform and the Whittle likelihood approximation. The challenge of frequency-domain methods is to determine and account for observational boundary effects, missing data, and the shape of the observed spatial grid. In this paper we address these effects explicitly by proposing a new quasi-likelihood estimator. We prove consistency and asymptotic normality of our estimator in settings that include irregularly shaped grids. Our simulations show that the proposed method solves boundary issues with Whittle estimation for random fields, yielding parameter estimates with significantly reduced bias and error. We demonstrate the effectiveness of our method for incomplete lattices, in comparison to other recent methods. Finally, we apply our method to estimate the parameters of a Mat\'ern process used to model data from Venus' topography.

KW - stat.ME

KW - stat.AP

KW - stat.CO

KW - stat.ML

M3 - Journal article

JO - arXiv

JF - arXiv

ER -