Powered by OpenAIRE graph
Found an issue? Give us feedback
image/svg+xml art designer at PLoS, modified by Wikipedia users Nina, Beao, JakobVoss, and AnonMoos Open Access logo, converted into svg, designed by PLoS. This version with transparent background. http://commons.wikimedia.org/wiki/File:Open_Access_logo_PLoS_white.svg art designer at PLoS, modified by Wikipedia users Nina, Beao, JakobVoss, and AnonMoos http://www.plos.org/ Journal of Computati...arrow_drop_down
image/svg+xml art designer at PLoS, modified by Wikipedia users Nina, Beao, JakobVoss, and AnonMoos Open Access logo, converted into svg, designed by PLoS. This version with transparent background. http://commons.wikimedia.org/wiki/File:Open_Access_logo_PLoS_white.svg art designer at PLoS, modified by Wikipedia users Nina, Beao, JakobVoss, and AnonMoos http://www.plos.org/
Journal of Computational Physics
Article . 2019 . Peer-reviewed
License: CC BY
Data sources: Crossref
image/svg+xml art designer at PLoS, modified by Wikipedia users Nina, Beao, JakobVoss, and AnonMoos Open Access logo, converted into svg, designed by PLoS. This version with transparent background. http://commons.wikimedia.org/wiki/File:Open_Access_logo_PLoS_white.svg art designer at PLoS, modified by Wikipedia users Nina, Beao, JakobVoss, and AnonMoos http://www.plos.org/
Journal of Computational Physics
Article
License: CC BY
Data sources: UnpayWall
image/svg+xml art designer at PLoS, modified by Wikipedia users Nina, Beao, JakobVoss, and AnonMoos Open Access logo, converted into svg, designed by PLoS. This version with transparent background. http://commons.wikimedia.org/wiki/File:Open_Access_logo_PLoS_white.svg art designer at PLoS, modified by Wikipedia users Nina, Beao, JakobVoss, and AnonMoos http://www.plos.org/
image/svg+xml art designer at PLoS, modified by Wikipedia users Nina, Beao, JakobVoss, and AnonMoos Open Access logo, converted into svg, designed by PLoS. This version with transparent background. http://commons.wikimedia.org/wiki/File:Open_Access_logo_PLoS_white.svg art designer at PLoS, modified by Wikipedia users Nina, Beao, JakobVoss, and AnonMoos http://www.plos.org/
CONICET Digital
Article . 2019
License: CC BY
Data sources: CONICET Digital
image/svg+xml Jakob Voss, based on art designer at PLoS, modified by Wikipedia users Nina and Beao Closed Access logo, derived from PLoS Open Access logo. This version with transparent background. http://commons.wikimedia.org/wiki/File:Closed_Access_logo_transparent.svg Jakob Voss, based on art designer at PLoS, modified by Wikipedia users Nina and Beao
zbMATH Open
Article . 2019
Data sources: zbMATH Open
https://dx.doi.org/10.60692/jr...
Other literature type . 2019
Data sources: Datacite
https://dx.doi.org/10.60692/8s...
Other literature type . 2019
Data sources: Datacite
versions View all 7 versions
addClaim

This Research product is the result of merged Research products in OpenAIRE.

You have already added 0 works in your ORCID record related to the merged Research product.

Sequential Monte Carlo with kernel embedded mappings: The mapping particle filter

مونتي كارلو المتسلسل مع تعيينات مضمنة في النواة: مرشح جسيمات التعيين
Authors: Manuel Pulido; Peter Jan van Leeuwen;

Sequential Monte Carlo with kernel embedded mappings: The mapping particle filter

Abstract

Dans ce travail, un nouveau filtre de Monte Carlo séquentiel est introduit qui vise à un échantillonnage efficace de l'espace d'état. Les particules sont poussées de la prédiction à la densité postérieure à l'aide d'une séquence de mappages qui minimise la divergence de Kullback-Leibler entre la densité postérieure et la séquence de densités intermédiaires. La séquence de mappages représente un flux de gradient basé sur les principes du transport optimal local. Un ingrédient clé des mappages est qu'ils sont intégrés dans un espace de Hilbert du noyau de reproduction, ce qui permet un algorithme de Monte Carlo pratique et efficace. L'intégration du noyau fournit un moyen direct de calculer le gradient de la divergence de Kullback-Leibler conduisant à une convergence rapide à l'aide d'algorithmes d'optimisation stochastique basés sur le gradient bien connus. L'évaluation de la méthode est menée dans le système chaotique Lorenz-63, le système Lorenz-96, qui est un prototype grossier de la dynamique atmosphérique, et un modèle épidémique qui décrit la dynamique du choléra. Aucun rééchantillonnage n'est nécessaire dans le filtre à particules de cartographie, même pour les longues séquences récursives. Le nombre de particules efficaces reste proche du nombre total de particules dans toute la séquence. Par conséquent, le filtre à particules de cartographie ne souffre pas d'appauvrissement de l'échantillon.

En este trabajo, se introduce un nuevo filtro secuencial de Monte Carlo que tiene como objetivo un muestreo eficiente del espacio de estado. Las partículas se empujan hacia adelante desde la predicción hasta la densidad posterior utilizando una secuencia de mapeos que minimiza la divergencia de Kullback-Leibler entre la posterior y la secuencia de densidades intermedias. La secuencia de mapeos representa un flujo de gradiente basado en los principios del transporte óptimo local. Un ingrediente clave de los mapeos es que están incrustados en un espacio Hilbert del núcleo de reproducción, lo que permite un algoritmo Monte Carlo práctico y eficiente. La incrustación del kernel proporciona un medio directo para calcular el gradiente de la divergencia de Kullback-Leibler que conduce a una convergencia rápida utilizando algoritmos de optimización estocástica basados en gradientes bien conocidos. La evaluación del método se realiza en el sistema caótico Lorenz-63, el sistema Lorenz-96, que es un prototipo grueso de dinámica atmosférica, y un modelo epidémico que describe la dinámica del cólera. No se requiere remuestreo en el filtro de partículas de mapeo, incluso para secuencias recursivas largas. El número de partículas efectivas permanece cercano al número total de partículas en toda la secuencia. Por lo tanto, el filtro de partículas de mapeo no sufre de empobrecimiento de la muestra.

In this work, a novel sequential Monte Carlo filter is introduced which aims at an efficient sampling of the state space. Particles are pushed forward from the prediction to the posterior density using a sequence of mappings that minimizes the Kullback-Leibler divergence between the posterior and the sequence of intermediate densities. The sequence of mappings represents a gradient flow based on the principles of local optimal transport. A key ingredient of the mappings is that they are embedded in a reproducing kernel Hilbert space, which allows for a practical and efficient Monte Carlo algorithm. The kernel embedding provides a direct means to calculate the gradient of the Kullback-Leibler divergence leading to quick convergence using well-known gradient-based stochastic optimization algorithms. Evaluation of the method is conducted in the chaotic Lorenz-63 system, the Lorenz-96 system, which is a coarse prototype of atmospheric dynamics, and an epidemic model that describes cholera dynamics. No resampling is required in the mapping particle filter even for long recursive sequences. The number of effective particles remains close to the total number of particles in all the sequence. Hence, the mapping particle filter does not suffer from sample impoverishment.

في هذا العمل، يتم تقديم مرشح مونت كارلو التسلسلي الجديد الذي يهدف إلى أخذ عينات فعالة من مساحة الولاية. يتم دفع الجسيمات إلى الأمام من التنبؤ إلى الكثافة الخلفية باستخدام سلسلة من التعيينات التي تقلل من تباعد Kullback - Leibler بين الخلفي وتسلسل الكثافات المتوسطة. يمثل تسلسل التعيينات تدفقًا متدرجًا يعتمد على مبادئ النقل الأمثل المحلي. أحد المكونات الرئيسية للتعيينات هو أنها مضمنة في مساحة نواة هيلبرت المستنسخة، والتي تسمح بخوارزمية مونت كارلو العملية والفعالة. يوفر تضمين النواة وسيلة مباشرة لحساب تدرج تباعد Kullback - Leibler مما يؤدي إلى تقارب سريع باستخدام خوارزميات تحسين عشوائية معروفة تعتمد على التدرج. يتم إجراء تقييم الطريقة في نظام Lorenz -63 الفوضوي، ونظام Lorenz -96، وهو نموذج أولي خشن لديناميكيات الغلاف الجوي، ونموذج وبائي يصف ديناميكيات الكوليرا. لا يلزم إعادة أخذ العينات في مرشح جسيمات التعيين حتى بالنسبة للتسلسلات التكرارية الطويلة. يظل عدد الجسيمات الفعالة قريبًا من العدد الإجمالي للجسيمات في كل التسلسل. وبالتالي، فإن مرشح جسيمات رسم الخرائط لا يعاني من إفقار العينة.

Keywords

Atmospheric Science, Resampling, Artificial intelligence, Stein variational gradient descent, kernel embedding, Filter (signal processing), Gaussian Processes in Machine Learning, Divergence (linguistics), https://purl.org/becyt/ford/1.2, Sequential Monte Carlo, PARTICLE FLOWS, particle flows, Mathematical optimization, Statistics, Monte Carlo methods, Hybrid Monte Carlo, SWAM OPTIMIZATION, Extended Kalman filter, FOS: Philosophy, ethics and religion, Earth and Planetary Sciences, Monte Carlo method, Algorithm, optimal transport, Sequential estimation, Physical Sciences, STEIN GRADIENT DESCENT, Kalman filter, OPTIMAL TRANSPORT, Numerical optimization and variational techniques, Importance sampling, Particle Filtering and Nonlinear Estimation Methods, SEQUENTIAL BAYES, Artificial Intelligence, Particle filter, FOS: Mathematics, Genetics, KERNEL EMBEDDING, https://purl.org/becyt/ford/1, Biology, Linguistics, Numerical Weather Prediction Models, Applied mathematics, Computer science, Markov chain Monte Carlo, Philosophy, Combinatorics, FOS: Biological sciences, Computer Science, Kernel (algebra), Ensemble Kalman filter, FOS: Languages and literature, Computer vision, Mathematics, Sequence (biology), Quasi-Monte Carlo method

  • BIP!
    Impact byBIP!
    selected citations
    These citations are derived from selected sources.
    This is an alternative to the "Influence" indicator, which also reflects the overall/total impact of an article in the research community at large, based on the underlying citation network (diachronically).
    36
    popularity
    This indicator reflects the "current" impact/attention (the "hype") of an article in the research community at large, based on the underlying citation network.
    Top 10%
    influence
    This indicator reflects the overall/total impact of an article in the research community at large, based on the underlying citation network (diachronically).
    Top 10%
    impulse
    This indicator reflects the initial momentum of an article directly after its publication, based on the underlying citation network.
    Top 10%
Powered by OpenAIRE graph
Found an issue? Give us feedback
selected citations
These citations are derived from selected sources.
This is an alternative to the "Influence" indicator, which also reflects the overall/total impact of an article in the research community at large, based on the underlying citation network (diachronically).
BIP!Citations provided by BIP!
popularity
This indicator reflects the "current" impact/attention (the "hype") of an article in the research community at large, based on the underlying citation network.
BIP!Popularity provided by BIP!
influence
This indicator reflects the overall/total impact of an article in the research community at large, based on the underlying citation network (diachronically).
BIP!Influence provided by BIP!
impulse
This indicator reflects the initial momentum of an article directly after its publication, based on the underlying citation network.
BIP!Impulse provided by BIP!
36
Top 10%
Top 10%
Top 10%
Green
hybrid
Funded by
Related to Research communities