🔬 physics

Topological Signatures of Imperial Collapse and Fragmentation: Administrative Dissolution, Territorial Reorganization and Early-Warning Observables in the Han Dynasty Network (206~BCE\,--\,220~CE)

This paper applies persistent homology and network analysis to Han Dynasty data, revealing that the divergence between collapsing administrative and expanding geographic networks, alongside specific topological cycles and early-warning indicators, successfully generalizes the Roman collapse threshold to predict imperial fragmentation decades before formal dissolution.

Original authors: Jose de Jesus Bernal-Alvarado, David Delepine, Carlos Pinedo Guadarrama

Published 2026-07-13

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Jose de Jesus Bernal-Alvarado, David Delepine, Carlos Pinedo Guadarrama

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine the Han Dynasty not as a dusty list of emperors and battles, but as a giant, living web of roads connecting thousands of cities. Now, imagine trying to measure how "tangled" or "connected" that web is using a special kind of mathematical magic called Topological Data Analysis (TDA). This paper does exactly that, turning 420 years of Chinese history into a set of numbers that tell a story of how an empire unraveled from the inside out.

Here is the story of the Han Dynasty's collapse, told through the lens of three different maps.

The Three Maps of the Empire

To understand what happened, the authors didn't just look at one map. They built three different networks, like three different lenses on a camera:

The "Official" Map ( $H_{Admin}$ ): This tracks only the places the Emperor actually controlled. If a city was lost to a warlord or a rebellion, it disappears from this map. It's the map of legal power.
The "Physical" Map ( $H_{Geo}$ ): This tracks the actual land, cities, and roads, regardless of who was in charge. Even if the Emperor lost control, the city of Chang'an was still there, and the roads still existed. This map tracks the physical territory.
The "Trade" Map ( $H_{Trade}$ ): This follows the Silk Road. It's the network of merchants and oasis cities that kept moving goods across the desert, mostly ignoring the political squabbles happening in the capital.

The Great Divergence: When the Map Split

For most of the Han Dynasty, the "Official" map and the "Physical" map looked identical. The Emperor's power matched the land he held. But then, something strange happened around 184 CE, during the Yellow Turban Rebellion.

Think of it like a spiderweb. If you cut the center, the whole thing might fall apart. But in this case, the "Official" web started shrinking rapidly, while the "Physical" web actually got more complex.

The Official Map collapsed. By 220 CE, the number of connected administrative routes dropped to almost zero. The Emperor's formal reach was gone.
The Physical Map, however, didn't vanish. Instead, it reorganized. The land didn't disappear; it just split into three distinct, self-sustaining loops.

This is the paper's biggest discovery: The empire didn't just break; it rearranged itself. While the central government dissolved, the land naturally formed three strong, independent clusters. The math shows that the "Physical" map gained 3.065 units of topological complexity (a measure of how many independent loops exist) while the "Official" map dropped to 0.

The Early Warning System: Seeing the Crash Before It Happens

One of the coolest parts of this study is that the math predicted the collapse decades before historians say it officially ended. The authors used three "early-warning sensors" to spot trouble:

The Speedometer ( $\dot{W}_2$ ): This measures how fast the network's shape is changing. In 175 CE, 45 years before the final collapse, the speed of change spiked to 4.3 times the normal baseline. The web was vibrating violently.
The Distance Meter ( $W_\times$ ): This measures the gap between the "Official" map and the "Physical" map. For centuries, they were identical (distance = 0). But in 190 CE, just one decade after the rebellion, the distance suddenly jumped to 737. It was a sudden, sharp break, not a slow drift.
The "Integrated Change Tracker" (ICT): This is a score that combines all the warning signs. It first crossed its "early warning" threshold in 170 CE, a full 50 years before the dynasty officially ended. At that moment, the official map looked fine, but the underlying math knew the system was about to snap.

The Three Kingdoms: A Mathematical Surprise

The paper argues that the famous "Three Kingdoms" (Wei, Shu, and Wu) weren't just a political decision made by generals. They were a topological inevitability.

The math shows that around 190 CE, the physical network spontaneously formed three long-lasting loops. These loops corresponded perfectly to the three future kingdoms:

Wei: The north.
Shu: The southwest (protected by mountains).
Wu: The southeast.

These three loops appeared 30 years before the kingdoms were officially declared. The mountains and rivers naturally created three separate "rooms" in the house, and the math saw them forming long before the politicians moved in. The signal for these three kingdoms was 10.9 times stronger than the background noise of smaller local roads.

What Didn't Change? The Silk Road

While the empire's internal government fell apart, the Silk Road ( $H_{Trade}$ ) remained stubbornly stable. The trade network didn't care about the political chaos. It kept its shape, its loops, and its connections intact for the entire 420 years. This suggests that long-distance trade is a "structural invariant"—a pattern that survives even when the political system holding it together collapses.

The Bottom Line

This paper suggests that we can use math to see the "shape" of history. It shows that the Han Dynasty didn't just fade away; it underwent a phase transition.

The Official power vanished (dropping to 0).
The Physical land reorganized into three strong, independent domains.
The Trade routes kept humming along, ignoring the politics.

The authors are confident in these numbers, having run 50 different simulations (bootstrap runs) to make sure the results weren't just a fluke. They found that the collapse was 45 to 50 years in the making, detectable by math long before the final date of 220 CE. The Three Kingdoms weren't just a political accident; they were the natural, mathematical shape of the land once the central authority was removed.

Technical Summary: Topological Signatures of Imperial Collapse and Fragmentation in the Han Dynasty Network

Problem Statement
The collapse of large-scale political systems remains a complex phenomenon often analyzed through qualitative models (e.g., Tainter's diminishing returns or Turchin's structural-demographic cycles) or system-specific proxies. There is a need for a quantitative, mechanism-agnostic measure of systemic disintegration that allows for cross-civilizational comparison. This paper addresses the collapse of the Han Dynasty (206 BCE – 220 CE), an event characterized by internal fragmentation following the Yellow Turban Rebellion (184 CE) rather than external conquest, leading to the partition of authority into the Three Kingdoms (Wei, Shu, Wu). The study tests whether topological signatures of collapse previously identified in the Roman–Byzantine system generalize to this distinct imperial context.

Methodology
The authors extend the persistent homology formalism to the Han Dynasty using a three-network decomposition approach. The methodology involves constructing three complementary weighted graphs derived from geospatial and historical data, modulated by a five-channel friction model.

Data Sources and Network Construction:
- Administrative Network ( $H_{Admin}$ ): Nodes represent 392 active prefectures (Guo and Houguo) from the China Historical GIS (CHGIS v6). Nodes are active only during their documented administrative tenure.
- Geographic Network ( $H_{Geo}$ ): Uses the same node catalog but extends node persistence beyond administrative dissolution (post-184 CE) to reflect the physical continuity of the territory under successor states.
- Trade Network ( $H_{Trade}$ ): An 81-node Silk Road network representing exterior commercial connectivity, treated as a structural control invariant to internal political fragmentation.
- Edge Costs: Derived via Least-Cost Path (LCP) analysis on a 1 km/pixel Digital Elevation Model (DEM), incorporating a Tobler hiking function and altitude penalties.
Five-Channel Friction Model:
Edge weights are adjusted by five multiplicative channels to account for historical dynamics:
- $C_1$ : Historical events (e.g., wars, peace treaties).
- $C_2$ : Stochastic noise (unobserved variability in road conditions).
- $C_3$ : Xiongnu frontier pressure.
- $C_4$ : Maintenance Friction Index (MFI), linking population decline to reduced road maintenance.
- $C_5$ : Territorial fragmentation (active only in $H_{Geo}$ post-184 CE), penalizing inter-domain routes to simulate the loss of central authority.
Topological Data Analysis (TDA):
The study employs persistent homology to compute persistent entropy ( $H$ ), a scalar measure of network redundancy. The analysis includes:
- Phase Regressions: Ordinary Least Squares (OLS) to determine trends in $H$ across historical phases.
- Chow Tests: To identify structural breaks in the time series.
- Early-Warning Observables: Wasserstein velocity ( $\dot{W}_2$ ), cross-network Wasserstein distance ( $W_\times$ ), correlation length ( $\xi$ ), and an Integrated Change Tracker (ICT).

Key Contributions and Results

Three-Network Decomposition and Divergence:
The study demonstrates that a single network metric is insufficient. $H_{Admin}$ (formal authority) collapses to zero by 220 CE with a slope of $-0.0261 \text{ yr}^{-1}$ ( $R^2 = 0.689$ ). Conversely, $H_{Geo}$ (territorial persistence) increases to $3.065$ during the same period. The divergence $\Delta H = H_{Geo} - H_{Admin}$ rises from zero to $3.065$, serving as a topological signature of internal fragmentation where the territory reorganizes even as the state dissolves.
Topological Identification of the Three Kingdoms:
The Three Kingdoms emerge as three long-lived $\beta_1$ cycles in the geographic network's persistence diagram. These cycles appear at $d = 190$ CE (thirty years before the formal partition) with a signal-to-noise gap ratio ( $G_3$ ) of $10.9\times$ . These cycles correspond to the geographically coherent domains of Wei, Shu, and Wu, driven by topographic barriers (e.g., Qinling mountains) and the fragmentation friction model.
Early-Warning Chronology:
Three independent indicators signal the onset of collapse 45–50 years before formal dissolution:
- Integrated Change Tracker (ICT): First exceeds its threshold ($0.1$) at 170 CE, 50 years prior to the end, while $H_{Admin}$ shows no significant trend.
- Wasserstein Velocity ( $\dot{W}_2$ ): Exceeds $4\times$ baseline at 175 CE.
- Correlation Length ( $\xi$ ): The geographic correlation length diverges monotonically, reaching $55\times$ the baseline by 220 CE.
- Cross-Network Distance: The Wasserstein distance between $H_{Admin}$ and $H_{Geo}$ jumps from 0 to 737 at 190 CE, marking an abrupt topological separation.
Structural Invariance of Trade:
The Silk Road network ( $H_{Trade}$ ) remains topologically stable throughout the 420-year period, showing no significant structural breaks. This suggests that long-distance trade circuits constitute a topological pattern invariant to the political systems facilitating them.
Combined Resilience Index:
A combined index $H_{comb}$ , weighting administrative and trade resilience, is calibrated to reach the collapse threshold $H^* = 0.5241$ exactly at 220 CE, validating the model's ability to synthesize different network dimensions.

Significance
The paper claims that Topological Data Analysis (TDA), when applied to decomposed historical networks, can:

Distinguish between mechanistically distinct modes of imperial collapse (internal fragmentation vs. external conquest).
Identify topological precursors of collapse decades before formal political dissolution occurs.
Predict the spatial structure of successor polities (the Three Kingdoms) purely from topographic data and historically motivated friction models, without prior knowledge of the eventual political outcome.
Provide an objective, quantitative basis for comparing the disintegration of different civilizations.

The study concludes that the Han collapse was not a sudden disintegration but an accelerating process of topological reorganization, where the physical territory developed new, coherent sub-networks (the Three Kingdoms) while the central administrative network condensed and vanished.