A Time-State Representation for Multidimensional Data

This paper introduces a multi-axis time-state representation for interpreting multidimensional data as evolving state structures rather than static collections of values.

Building on the author's previous work on the t + icos⁡t unfolding representation, time is treated not as an external index but as a common progression axis embedded within the state representation itself. Individual state components are arranged on separate structure-preserving axes, allowing observations at each moment to be represented as multi-axis states and interpreted as continuous structures of state evolution.

The framework incorporates first- and second-order changes to analyze directions of change, structural persistence, restoration, and transitions. It also provides a basis for AI systems to learn patterns of evolving state structures, enabling the interpretation of anomaly locations, propagation paths of changes, and the maintenance of stable structures.

Possible applications include process control, sensor systems, medical data analysis, urban systems, financial systems, AI language understanding, generative AI, and digital twins.

This work proposes an additional layer of interpretation beyond conventional result-centered analysis by focusing on the underlying structures and their evolution over time.

Correspondence and discussion regarding this work are welcome:

J. San Park

Email: heyj.research@gmail.com