The evolution of interdisciplinarity in physics research

Science, being a social enterprise, is subject to fragmentation into groups that focus on specialized areas or topics. Often new advances occur through cross-fertilization of ideas between sub-fields that otherwise have little overlap as they study dissimilar phenomena using different techniques. Thus to explore the nature and dynamics of scientific progress one needs to consider the large-scale organization and interactions between different subject areas. Here, we study the relationships between the sub-fields of Physics using the Physics and Astronomy Classification Scheme (PACS) codes employed for self-categorization of articles published over the past 25 years (1985-2009). We observe a clear trend towards increasing interactions between the different sub-fields. The network of sub-fields also exhibits core-periphery organization, the nucleus being dominated by Condensed Matter and General Physics. However, over time Interdisciplinary Physics is steadily increasing its share in the network core, reflecting a shift in the overall trend of Physics research.

💡 Research Summary

The paper investigates the large‑scale organization and temporal evolution of sub‑fields within physics by constructing a co‑classification network from the Physics and Astronomy Classification Scheme (PACS) codes assigned to articles published between 1985 and 2009. Each article that lists two or more PACS codes contributes an undirected weighted edge between the corresponding codes; aggregating these edges over five‑year windows yields a series of yearly snapshots of the “physics sub‑field network.” Standard network metrics—average degree, weighted clustering coefficient, average shortest‑path length—are computed to quantify overall connectivity, while more nuanced analyses employ k‑core decomposition, betweenness centrality, and modularity‑based community detection to uncover core‑periphery structure and evolving community boundaries.

The results reveal a clear trend toward increasing interdisciplinary interaction. The average degree rises from roughly 1.8 in the late 1980s to 3.7 by the mid‑2000s, indicating that papers increasingly draw on multiple sub‑fields. Weighted clustering also climbs, suggesting a denser triangular relationship among three or more fields, while the average shortest path modestly contracts, reflecting a more compact network. K‑core analysis shows that the early network core is dominated by Condensed Matter and General Physics (1‑core and 2‑core). Starting in the early 2000s, however, the 3‑core and 4‑core become populated by Interdisciplinary Physics codes (e.g., complex systems, nonlinear dynamics, biophysics, materials science). Correspondingly, betweenness centrality scores for these interdisciplinary nodes surge, marking them as critical conduits for information flow across the entire system.

Community detection uncovers a shift from well‑segmented, discipline‑specific clusters to more blended, hybrid communities. Traditional clusters such as high‑energy physics or astrophysics remain peripheral, yet they become increasingly linked through interdisciplinary bridges. In contrast, clusters that combine condensed matter, nanoscience, and materials engineering coalesce into larger, tightly knit modules, reflecting the rise of cross‑cutting experimental techniques and theoretical frameworks.

The authors interpret these findings as evidence that physics research is moving from a paradigm of deep specialization toward one of methodological and conceptual integration. The growing prominence of interdisciplinary physics within the network core suggests that future breakthroughs are likely to arise at the intersections of previously distinct domains. The study acknowledges limitations: PACS codes were discontinued after 2010, and code assignment is partially subjective, potentially biasing edge formation. The paper proposes extending the analysis with additional data sources—arXiv metadata, citation networks, and instrument‑sharing records—to build multilayer network models that can capture richer dynamics of scientific evolution.