The Technical University of Crete (TUC), proud member of the European University on Responsible Consumption and Production (EURECA-PRO), organizes a series of open scientific lectures to facilitate collaboration among students and faculty from all EURECA-PRO partners.
The next lecture of the series will be delivered by our own Prof. Dionissios T. Hristopulos, Professor with the School of Electrical and Computer Engineering (ECE) at the Technical University of Crete (TUC). Prof. Hristopulos will deliver a celebrated lecture, following the prestigious 2024 Georges Matheron Lectureship Award he received from the International Association for Mathematical Geosciences (IAMG), which recognized his outstanding scientific contributions to spatial statistics and mathematical morphology.
Prof. Hristopulos' lecture is titled:
"From Particles to Patterns: An Odyssey from Physics to Geostatistics and Beyond"
and is scheduled to take place on:
Tuesday, June 9, 2026 at 16:30 EET (15:30 CET)
and can be attended either physically or virtually:
Room 145.P58 (School of ECE @ TUC Campus)
https://tuc-gr.zoom.us/j/96511690683?pwd=BF3BVAhr7YpNSuHBrRf5FWGdrw3EwF.1
The lecture is open to all interested members of TUC and EURECA-PRO. An official, electronic certificate of attendance will be issued to all actual attendees who will register their participation during the lecture.
Abstract
Understanding and managing our natural resources requires reliable ways to analyze spatial data—from climate variables and soil properties to pollution levels and ecosystem indicators. This talk explores how ideas from physics, geostatistics, and artificial intelligence (AI) can be combined to support sustainable development through efficient and transparent data modeling. We begin by showing that traditional geostatistical methods, such as kriging, are closely related to modern AI tools known as Gaussian process regression. This connection helps bridge established spatial analysis techniques with newer data‑driven approaches used in environmental and sustainability studies. However, applying these methods to the large datasets now common in Earth observation, sensor networks, and climate science can be computationally demanding. To address this challenge, we introduce an alternative perspective inspired by statistical physics. By formulating spatial dependence through local interactions—rather than large covariance matrices—we obtain models that are both physically interpretable and computationally efficient. These stochastic local interaction (SLI) models function as Markov random fields and are particularly well suited for large‑scale spatial prediction and uncertainty assessment, which are essential for sustainable planning and risk analysis. We further discuss how these ideas extend naturally from discrete data to continuous spatial processes, enabling smooth representations of environmental variables. We also briefly explore how concepts from complex systems and statistical physics, such as the Ising model, can help describe collective spatial behavior, for example in land‑use patterns or ecosystem transitions. We conclude by highlighting how physics‑inspired, computationally efficient spatial models can contribute to more informed, scalable, and sustainable decision‑making in environmental and resource management.