State Reconstruction and Reduced Coordinates from Sparse Sensors

Definitions and mathematical statement

The supplied record poses a research programme rather than a single quantified conjecture. The following notation explains its terms without selecting hypotheses absent from the source.

Let $X$ denote an unspecified state space, $T$ an unspecified time set, and $\mathcal{A}\subseteq X^T$ a class of admissible trajectories of a dynamical system. A sensor or measurement map is a map $h_i:X\to Y_i$, where $i$ indexes sensors or measurement modalities and $Y_i$ is its output space. For a collection $I$ of sensor--time pairs, define the ideal observation map by \[ \mathcal{O}_I:\mathcal{A}\longrightarrow\prod_{(i,t)\in I}Y_i, \qquad \mathcal{O}_I(x)=\left(h_i(x(t))\right)_{(i,t)\in I}. \] Here sensor placement concerns the choice of available sensors or their locations; measurement-map discovery concerns identifying or choosing the maps $h_i$; time-delay methods use observations at multiple appropriately spaced times; and multimodal fusion combines outputs from different measurement modalities. The source specifies neither the admissible choices nor their constraints.

For a specified target time $t_*\in T$, exact noiseless full-state reconstruction would mean finding a map \[ R:\mathcal{O}_I(\mathcal{A})\to X \quad\text{such that}\quad R(\mathcal{O}_I(x))=x(t_*) \quad\text{for every }x\in\mathcal{A}. \] This is a definition of one possible reconstruction target, not an assertion that such a map exists or that the source intends this particular target. Reduced coordinates may be represented abstractly by a map $r:X\to Z$, but the source supplies no dimension, regularity, dynamical fidelity, or other criterion making $Z$ reduced or the coordinates useful. The background also includes noisy measurements; it gives no noise model, error metric, stability requirement, or probabilistic quantifier, so the displayed noiseless definition does not formalize that additional requirement.

Source question. What are the mathematical limits and possibilities of sensor placement, measurement-map discovery, time-delay methods, and multimodal data fusion for reconstructing the full state of a complex dynamical system and producing useful reduced coordinates?

A precise theorem or conjecture cannot be extracted uniquely without specifying the dynamical-system class, available information, observation constraints, reconstruction target, noise assumptions, and meaning of useful reduced coordinates. These qualifications remain unresolved.

Short English statement

Determine when sensor placement, learned measurement maps, delayed observations, and multimodal measurements permit full-state reconstruction and useful reduced coordinates. The supplied question leaves the system class, noise assumptions, admissible methods, and success criteria unspecified, so it remains a research programme requiring further formulation.

Sources

• Local UnsolvedMath registry: supplied complete record for permanent ID $30004685$, titled State Reconstruction and Reduced Coordinates from Sparse Sensors, in the Oberwolfach Reports Open Problems collection; exact source code catalogue entry. No individual proposer is supplied.
• Supplied report citation: Mini-Workshop: Analysis of Data-driven Optimal Control (2022), Oberwolfach Reports, pp. 1209--1257, DOI 10.4171/owr/2021/23. https://doi.org/10.4171/owr/2021/23. The registry records proposed year 2021; its relationship to the citation's 2022 date is unresolved. No specific problem page or workshop contributor is identified.
• Casdagli et al., state reconstruction with noise: https://doi.org/10.1016/0167-2789(91)90222-U.
• Serpas et al., sensor location for nonlinear systems: https://doi.org/10.1016/j.compchemeng.2012.07.014.
• Sendiña-Nadal and Letellier, nonlinear network reconstruction: https://doi.org/10.1063/5.0090239.
• Supplied literature assessment, checked 2026-08-22: the record labels the problem partially solved and reports restricted rigorous or constructive results involving Takens-type embeddings, observability-Gramian placement, and nonlinear network reconstruction, while describing general limits of full-state recovery from multimodal noisy measurements as open. This is supplied dated curation derived from the OWR record, not an independently verified assessment; its status and cited claims require independent verification.