Bram Wouters, Cees G. H. Diks · 2026 · arXiv (Cornell University)
DOI: 10.48550/arxiv.2609.27614 · cited by 0 works (OpenAlex) · open-access PDF
Desarrollamos un marco de datos modelo-agnostico para la reducción del ruido en las series de tiempo de alta dimensión que se dirige explícitamente a la recuperación óptima de un componente dinámico latente de baja dimensión contaminado por el ruido blanco observacional. Bajo la suposición de que la dinámica latente vive en un subespacio dinámico lineal de baja dimensión, caracterizamos la proyección lineal óptima en el subespacio dinámico y proporcionamos una descripción geométrica del error residual en términos de la orientación relativa de la señal y los espacios de ruido. Proponemos estimadores para el subespacio dinámico y la proyección óptima basados en las matrices de covario retrasado, la selección de dimensiones de bootstrap y una representación de bajo rango del ruido estructur
Machine translation (Meta M2M100); the English original follows.
We develop a model-agnostic framework for noise reduction in high-dimensional time series that explicitly targets optimal recovery of a low-dimensional latent dynamic component contaminated by observational white noise. Under the assumption that the latent dynamics live in a low-dimensional linear dynamic subspace, we characterize the optimal linear projection onto the dynamic subspace and provide a geometric description of the residual error in terms of the relative orientation of the signal and noise spaces. We propose estimators for the dynamic subspace and the optimal projection based on lagged covariance matrices, bootstrap dimension selection, and a low-rank representation of the structured noise. Under mild conditions, the resulting denoised series is shown to converge to its population target at the usual parametric rate. Simulations show that the proposed method can substantially improve subspace estimation, reconstruction error, and one-step-ahead forecast accuracy compared with both orthogonal projection-based denoising and the raw data. The approach is illustrated by empirical applications to high-dimensional stock returns and to a 20-variate time series of macroeconomic indicators.
Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese
Topics: Mathematics, Subspace topology, Estimator, Algorithm, Series (stratigraphy), Projection (relational algebra)
GET https://store.agentexchange.work/gw/papers/verify?title=...&year=... (0.003 USD), fetch this record as JSON with /gw/papers/paper?id=2609.27614, or list what it cites with /gw/papers/references?id=.... Pay per call in USDC (x402) or with a 19 USD card key for 1,000 calls. Catalogue · llms.txtSources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange