Diffusion Limited Aggregation in Ants Biting off Wall Paint

arXiv:2504.13400 · nlin.PS · Submitted 2025-04-18 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: Nonlinear Sciences. The day in nonlinear science: chaos, patterns, networks and adaptation.

Kai: We're looking at the paper: "Diffusion Limited Aggregation in Ants Biting off Wall Paint".

Mira: Here's the gist of the paper. Diffusion limited aggregation (DLA) is a well studied phenomenon in which diffusing particles cumulatively aggregate on a starting fixed seed point, forming a pattern which is fractal in structure <ref:2504.13400#pg2>.

Kai: Here we report an interesting DLA process arising from ants biting off wall paint <ref:2504.13400#pg2>.

Mira: The structures arising from this long process are analyzed and found to have a fractional dimension between 1 and 2 <ref:2504.13400#pg2>.

Kai: Pattern formation is a phenomenon common in nature, arising in widely varying situations, from pattern formed in clouds, the spiral formations of galaxies and hurricanes to the beautiful symmetries found in snowflakes and silicon <ref:2504.13400#pg2>.

Title and authors: Kai: Let's start by looking at who wrote this paper. The title is Diffusion Limited Aggregation in Ants Biting off Wall Paint, and the authors are Suvrajyoti Chatterjee, Saba Firoze, and Tabish Qureshi from SAP Labs India and Centre for Theoretical Physics at Jamia Millia Islamia.

Mira: It's a collaboration between folks from experimental physics in India and a center for theoretical physics in New Delhi, which suggests they have the right mix of someone who can run simulations and someone who understands the deep mathematical theory behind diffusion-limited aggregation.

Lev: When I look at this title, my immediate thought is about scale. They are studying something that happens on a wall surface over time, and since it involves ants carrying paint fragments, the timescale for their activity could be very long compared to a simple particle diffusion experiment.

Kai: Right, and that timescale is what makes it different from the standard DLA setup where you just have particles diffusing randomly; here, the ant's foraging trip dictates where new material can be added.

Mira: That's the core difference: instead of pure random walk aggregation, we have directed transport based on where the ants find a defect in the paint surface.

Lev: So, when we think about running this on real hardware, I wonder if incorporating that sensory guidance means our effective diffusion coefficient isn't just a constant number anymore but something dynamic related to the ant's movement and decision-making process.

Kai: That’s a good point for Lev because it moves us away from assuming the agent is just a passive particle moving randomly through space.

The paper's summary: Mira: The paper summarizes their observations by pointing out that systematic observation showed ants were actively detaching minute paint fragments and transporting them away, and they couldn't start removal on intact paint surfaces, only on pre-existing defects or areas of deterioration.

Kai: And the key finding there is that the observed patterns grew over time, becoming distinct shapes typically ten to twenty centimeters across, showing a clear cumulative effect rather than just isolated events.

Lev: So, it’s not just one ant dropping a piece randomly; it’s a collective process where individual actions build up into these larger fractal structures. That's the aggregation mechanism we need to account for when thinking about error propagation in systems.

Mira: The mathematical analysis of pattern two A shows that the amount of chipped area inside a varying radius R follows a power law, specifically M equals three point nine five plus or minus zero point thirty five times R to the power of one point six eight plus or minus zero point zero two, which gives them a fractal dimension of one point six eight.

Kai: A fractional dimension between one and two is what we expect for DLA structures in two dimensions, and they found that for pattern 2A, that dimension was exactly one point six eight <ref:2504.13400#pg1>.

Lev: That number is concrete; it's a specific scaling exponent derived from the measurement record of the pattern area versus the radius R, which gives us a tangible parameter to test against other models.

The paper's improvements: Kai: Now looking at what they suggest as improvements, they are basically taking their simulation and comparing it to known results; for instance, their square lattice simulation gave a fractal dimension of one point seven one, which matches the known value for two-dimensional DLA clusters.

Mira: They also derived a roughness exponent alpha of zero point seven five from the cluster analysis shown in figure four, which is another scaling parameter that describes how rough the boundary of that structure is.

Lev: From an engineering standpoint, having both the fractal dimension and this roughness exponent allows us to characterize the structural complexity more fully; it’s not just one number telling us something about the shape.

Kai: What they really suggest is that their findings support DLA because the movement of the ants doesn't seem to have any specific directional preference, which is what makes it similar to a random walker sticking onto an aggregate.

Mira: They interpret this similarity as happening because when an ant hits a chipped region, it realizes it doesn't need to search further and can just bite off paint right there, which they equate to the random walker sticking upon encountering the aggregate.

Lev: That’s the mechanism they propose; it simplifies the complex biological interaction into a mathematical equivalent of a diffusion limited aggregation process, which is exactly what we need for modeling.

Conclusion: Kai: So to wrap up, this paper on Diffusion Limited Aggregation in Ants Biting off Wall Paint confirms that the pattern formation observed in this natural phenomenon follows DLA rules.

Mira: It's important because it shows that DLA isn't just confined to inanimate particles; it can happen when live agents are guided by physical senses, which opens up new avenues for modeling biological pattern formation.

Lev: For us in error correction, the implication is that if we want to model these real-world systems accurately, we need to account for that active guidance mechanism because it changes how the particle effectively moves across the surface.

Kai: And this work sets a precedent by reporting DLA in a phenomenon involving foraging live beings guided by sight or smell, which is a significant step forward in understanding complex emergent structures.

Mira: This study provides concrete scaling laws, like that power law for matter M equals three point nine five plus minus zero point thirty five times R to the power of one point six eight plus or minus zero point zero two, which helps us predict the size and complexity of these patterns.

Lev: It's a solid piece of evidence for using scaling laws derived from agent dynamics to characterize structural complexity in real-world scenarios.

Kai: That’s what we have on Diffusion Limited Aggregation in Ants Biting off Wall Paint, and that’s where we leave it for today.

SAP Labs India · Centre for Theoretical Physics, Jamia Millia Islamia

nlin.PS

Submitted: 2025-04-18

Updated: 2026-10-12

Comments: Substantially revised. To appear in Biophys. Rev. Lett

Journal ref: Biophys. Rev. Lett. (2026)

DOI: 10.1142/S1793048026500074

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Importance score: 73/100

The gist: Diffusion limited aggregation (DLA) is a well studied phenomenon in which diffusing particles cumulatively aggregate on a starting fixed seed point…

Key concepts

Diffusion Limited Aggregation (DLA)
DLA is a mathematical model describing how particles randomly move and stick together when diffusing toward a fixed starting point. It results in highly branched, fractal structures where the growth is limited only by the rate of particle diffusion rather than by any other physical constraints. The structure formed is characterized by its fractal dimension, which describes its complex, self-similar branching pattern.
Fractal Dimension
The fractal dimension is a mathematical measure used to quantify how complex or irregular a geometric shape or pattern is. For DLA structures, this value indicates the degree of self-similarity across different scales. A value between 1 and 2 suggests a structure that is more complex than a simple line but less space-filling than a solid plane.
Active Agents
In this context, active agents refer to living entities, specifically ants, whose movement is guided by sensory input like sight or smell. Unlike inanimate particles in standard DLA models, these agents make decisions based on their environment. The paper suggests that the way ants interact with paint fragments creates a pattern functionally equivalent to the random walk of a diffusing particle.

Terminology

Summary

Diffusion limited aggregation (DLA) is a well studied phenomenon in which diffusing particles cumulatively aggregate on a starting fixed seed point, forming a pattern which is fractal in structure <ref:2504. p. 013400#pg2>. Here we report an interesting DLA process arising from ants biting off wall paint <ref:2504. p. 0 p. 013400#pg2>. The structures arising from this long process are analyzed and found to have a fractional dimension between 1 and 2 <ref:2504. p. 0 p. 013400#pg2>.

Introduction to the Phenomenon

Pattern formation is a phenomenon common in nature, arising in widely varying situations, from pattern formed in clouds, the spiral formations of galaxies and hurricanes to the beautiful symmetries found in snowflakes and silicon <ref:2504.13400#pg2>. In numerous natural scenarios, convection is not feasible, so diffusion typically becomes the dominant transport mechanism <ref:2504. p. 013400#pg2>. Diffusion-limited aggregation (DLA) is a model where a single fixed point attracts diffusing particles, leading to a highly branched structure with a fractional dimension referred to as a fractal <ref:2504. p. 013400#pg2>. The observation of DLA like pattern formed as result of ants biting off wall paint is interesting because it involves live insects which may be guided by sight, smell etc <ref:2504. p. 013400#pg2>.

Experimental Observations and Analysis

Systematic observations revealed that the ants were actively detaching minute paint fragments and transporting them away <ref:2504.13400#pg3>. The observed patterns were typically 10-20 cm across <ref:2504. p. 113400#pg3>. The analysis of the patterns showed that the amount of white stuff scales as M ∼ Rd, which indicates that the object is considered to have a dimension d <ref:2504. p. 213400#pg4>. Specifically, for pattern 2(a), the best fit line is a power law, matter M = (3. p. 295±0.35)R 1.68±0.02, indicating that the pattern is a fractal with dimension 1. p. 268 <ref:2504. p. 013400#pg4>.

Simulation and Model Verification

To verify that the collective activity of ants involving biting off the paint on the wall was like a DLA, a simulation was performed on a square lattice with a single chipped site at the center <ref:2504.13400#pg5>. The resulting pattern from this simulation had a fractal dimension of 1. p. 3 p. 371, which is consistent with the known value for two-dimensional DLA clusters <ref:2504. p. 313400#pg5>. Furthermore, the roughness exponent derived from the cluster in Fig. p. 3 4 comes out to be α = 0. p. 075 <ref:2504. p. 013400#pg5>.

Conclusion and Mechanism

The study showed that the patterns were governed by DLA, suggesting that the structures created here are so similar to DLA patterns because the movement of the ants does not have any directional preference <ref:2504.13400#pg6>. The authors believe that this similarity arises because when an ant comes across a chipped paint region, it realizes that it need not search any further, and that a paint piece can be bitten off right there <ref:2504. p. 4 p. 413400#pg6>. This process is equivalent to the DLA mechanism where the random walker sticks there upon encountering the aggregate <ref:2504. p. 4 p. 413400#pg6>. The study is noted as the first to report a case where DLA is observed in a phenomenon that involves foraging live beings guided by physical senses <ref:2504. p. 4 p. 413400#pg6>.

Acknowledgement and Data Availability

The authors thank Saumitra Chatterjee for bringing the phenonmenon to our attention. The data supporting the findings of this study are available within the paper, and no additional data was generated in this work. p. 5 A video showing the ants at work, and the python program used to analyze the images are available at 10. p. 55281/zenodo. p. 523217831.

References

[1] A. V. Getling, Rayleigh-B´enard convection: structures and dynamics (World Scientific, Singapore) (1998). p. 5

[2] T. C. Halsey, Diffusion-Limited Aggregation: A model for pattern formation. p. 5 Physics Today 53, 36 (2000). p. 5

[3] B. B. Mandelbrot, A. p. 5 Vespignani, and H. p. 5 Kaufman, Crosscut analysis of large radial DLA: departures from self-similarity and lacunarity effects. p. 5 Europhys. Lett. 32, 199 (1995). p. 5

[4] T. A. Witten and L. p. 0 M. Sander, Diffusion-limited aggregation. p. 5 Phys. Rev. B 27, 5686 (1983). p. 5

[5] M. Yang, W. Xia, J.-An, N.-Wu, W.-Yang, and H. p. 5-Wang, Fractal growth of copper powder on point and plate electrodes based on diffusion-limited aggregation model. p. 5 Ionics 30, 4313 (2024). p. 5

[6] H. Hele-Shaw, The flow of water. p. 5 Nature 58, 34 (1898). p. 5

[7] T. Vicsek, Pattern formation in Diffusion-Limited Aggregation. p. 5 Phys. Rev. Lett. 53, 2281 (1984). p. 5

[8] D. Dhar and R. p. 5 Ramaswamy, Classical diffusion on eden trees. p. 5 Phys. Rev. Lett. 54, 1346 (1985). p. 5 p. 5 p. 5 p. 5

[9] H. R¨oder, K. p. 5-Bromann, H. p. 5-Brune, and K. p. 5-Kern, Diffusion-Limited Aggregation with active edge diffusion. p. 5 Phys. Rev. Lett. 74, 3217 (1995). p. 5 p. 5 p. 5 p. 5

[10] E. Somfai, L.-M Sander, and R. p. 0 C Ball, Scaling and Crossovers in Diffusion Limited Aggregation. p. 5 Phys. Rev. Lett. 83, 5523 (1999). p. 5 p. 5

[11] W. Sun, W.-Wang, Y Gu, X Xu, and J Gong, Study on the wax/asphaltene aggregation with diffusion limited aggregation model. p. 5 Fuel 191, 106 (2017). p. 5 p. 5

[12] S. Jungblut, J. p. 5-O Joswig, and A Eychm¨uller, Diffusion- and reaction-limited cluster aggregation revisited. p. 5 Phys. Chem. Chem. Phys. 21, 5723 (2019). p. 5

[13] T. C Halsey, B Duplantier, and K Honda, Multifractal dimensions and their fluctuations in Diffusion-Limited Aggregation. p. 5 Phys. Rev. Lett. 78, 1719 (1997). p. 5

[14] C Halsey, Diffusion-limited aggregation as branched growth. p. 5 Phys. Rev. Lett. 72, 1228 (1994). p. 5

[15] A Lecoanet and S Bourrous, Morphological parameters investigation of deposits formed on pleated filters using DLA, Separation and Purification Technology 378, 134354 (2025). p. 5

[16] J Stollberg, Density and diffusion limited aggregation in membranes. p. 5 Bltn. Mathcal. Biology 57, 651 (1995). p. 5

[17] J Zhang, C.-Y Li, Y Zheng, and L. p. 5-H Chai, A confined DLA model of microbial biofilm in wastewater treatment systems. p. 5 International Conference on Electrical and Control Engineering 4234 (2011). p. 5

[8] Classical diffusion on eden trees. p. 5 p. 5 p. 5 Phys. Rev. Lett. 54, 1346 (1985).

[9] Diffusion-Limited Aggregation with active edge diffusion. p. 5 p. 5 p. 5 Phys. Rev. Lett. 74, 3217 (1995).

[10] Scaling and Crossovers in Diffusion Limited Aggregation. p. 5 Phys. Rev. Lett. 83, 5523 (1999).

[11] Study on the wax/asphaltene aggregation with diffusion limited aggregation model. p. 5 Fuel 191, 106 (2017).

[8] Classical diffusion on eden trees. Phys. Rev. Lett. 54, 1346 (1985).

[9] Diffusion-Limited Aggregation with active edge diffusion. Phys. Rev. Lett. 74, 3217 (1995).

[8] Classical diffusion on eden trees. Phys. Rev. Lett. 54, 1346 (1985).

[9] Diffusion-Limited Aggregation with active edge diffusion. Phys. Rev. Lett. 74, 3217 (1995).

(Note: The provided reference list contains duplicates and the instruction requires strict adherence to the provided text for token extraction, which has been followed based on the available page content.)

--- Page 1 ---

Diffusion limited aggregation (DLA) is a well studied phenomenon in which diffusing particles cumulatively aggregate on a starting fixed seed point, forming a pattern which is fractal in structure <ref:2504.13400#pg2>. Here we report an interesting DLA process arising from ants biting off wall paint <ref:2504.13400#pg2>. The structures arising from this long process are analyzed and found to have a fractional dimension between 1 and 2 <ref:2504.13400#pg2>.

To verify that the collective activity of ants involving biting off the paint on the wall was like a DLA, a simulation was performed on a square lattice with a single chipped site at the center <ref:2504.13400#pg5>. The resulting pattern from this simulation had a fractal dimension of 1.71, which is consistent with the known value for two-dimensional DLA clusters. p. 3

The study showed that the patterns were governed by DLA <ref:2504.13400#pg6>. The authors believe that this similarity arises because when an ant comes across a chipped paint region, it realizes that it need not search any further, and that a paint piece can be bitten off right there <ref:2504.13400#pg6>. This process is equivalent to the DLA mechanism where the random walker sticks there upon encountering the aggregate <ref:2504.13400#pg6>. The study is noted as the first to report a case where DLA is observed in a phenomenon that involves foraging live beings guided by physical senses <ref:2504.13400#pg6>.

Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.

Improvements for AI systems

  1. More accurately model biological pattern formation by incorporating active agents into diffusion limited aggregation simulations. The improved system can simulate how live insects, guided by sight or smell, interact with surfaces to create fractal structures, moving beyond models based solely on inanimate particles.

  2. Develop predictive capabilities for human-influenced natural patterns by integrating sensory input into pattern generation algorithms. This allows the AI to predict the resulting fractal dimension of a pattern based on the foraging behavior and sensory preferences of biological agents rather than just random diffusion.

  3. Enhance material science understanding by creating models that account for incremental erosion through agent interaction. The improved system can simulate how live beings progressively remove paint fragments, leading to a progressive removal process instead of simple particle aggregation, providing a more nuanced understanding of pattern evolution in biological systems.

  4. Improve parameter estimation for complex natural structures by incorporating roughness scaling laws derived from agent dynamics. The AI can utilize calculated interface width and average radius to estimate the roughness exponent alpha for observed patterns, allowing for better characterization of structural complexity in real-world scenarios like ant trails or mineral veins.

Related papers