Martin boundary theory on inhomogenous fractals
Pith reviewed 2026-05-24 20:20 UTC · model grok-4.3
The pith
Redefining transition probabilities by weights leaves the Martin boundary of fractals unchanged.
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
Core claim
By redefining the transition probabilities of the Markov chain on the word space according to the weights of the probabilistic iterated function scheme, the Martin boundary coincides with the homogeneous case and is homeomorphic to the fractal.
What carries the argument
The Markov chain on the word space whose transition probabilities are redefined proportionally to the weights of each iterated function.
If this is right
- The homeomorphism between Martin boundary and fractal is preserved under the weighted transitions.
- All structural results obtained by Denker and Sato for the homogeneous case apply unchanged to the weighted setting.
- The Martin boundary can be computed without separately tracking the weights once the open set condition is assumed.
Where Pith is reading between the lines
- The same redefinition technique may leave other boundaries, such as the Poisson boundary, invariant under weighting.
- The result suggests the Martin boundary is insensitive to probability weights in any self-similar construction obeying the open set condition.
Load-bearing premise
The open set condition must hold for the probabilistic iterated function scheme so that the weighted transitions still produce a boundary homeomorphic to the fractal.
What would settle it
An explicit example of a probabilistic iterated function scheme satisfying the other hypotheses but violating the open set condition in which the inhomogeneous and homogeneous Martin boundaries differ.
read the original abstract
We want to consider fractals generated by a probabilistic iterated function scheme with open set condition and we want to interpret the probabilities as weights for every part of the fractal. In the homogenous case, where the weights are not taken into account, Denker and Sato introduced in 2001 a Markov chain on the word space and proved, that the Martin boundary is homeomorphic to the fractal set. Our aim is to redefine the transition probability with respect to the weights and to calculate the Martin boundary. As we will see, the inhomogenous Martin boundary coincides with the homogenous case.
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper studies Martin boundaries for fractals arising from probabilistic iterated function systems satisfying the open set condition. It redefines the transition probabilities of the Markov chain on the symbolic word space to incorporate the given weights, then shows that the resulting Martin boundary is homeomorphic to the underlying fractal and coincides with the boundary obtained in the homogeneous (uniform-weight) case previously treated by Denker and Sato.
Significance. If correct, the result establishes that the topological identification of the Martin boundary with the fractal is independent of the choice of positive weights. This supplies a parameter-free extension of the homogeneous theory and confirms that the coding map and the product structure of the Green function along paths remain unchanged under the redefinition.
major comments (2)
- [Section 3 (redefinition of the chain)] The central claim that the Martin kernel evaluated at infinite words yields the same set of minimal harmonic functions relies on the Green function remaining a product along paths after the weight-dependent redefinition of transitions. The manuscript must exhibit the explicit form of the new transition matrix and verify that the resulting Martin kernel is identical to the homogeneous kernel up to a multiplicative factor that does not alter the topology.
- [Theorem 4.2 and its proof] The argument that the Martin boundary is homeomorphic to the fractal via the coding map appears to invoke the open set condition only to guarantee positive measure and dimension; however, the topological identification itself must be shown to hold without OSC, or the manuscript must clarify precisely where OSC enters the proof of the homeomorphism.
minor comments (3)
- The spelling 'inhomogenous' should be corrected to 'inhomogeneous' throughout the text and title.
- Notation for the weight vector and the redefined transition probabilities p_w should be introduced once in a dedicated subsection and used consistently; several passages mix the original and weighted probabilities without explicit warning.
- The reference list should include the original Denker-Sato 2001 paper and any subsequent works on Martin boundaries for self-similar sets that appeared after 2001.
Simulated Author's Rebuttal
We thank the referee for the careful reading and valuable comments on our manuscript. We address each major comment below and will revise the manuscript to incorporate the requested clarifications and explicit details.
read point-by-point responses
-
Referee: [Section 3 (redefinition of the chain)] The central claim that the Martin kernel evaluated at infinite words yields the same set of minimal harmonic functions relies on the Green function remaining a product along paths after the weight-dependent redefinition of transitions. The manuscript must exhibit the explicit form of the new transition matrix and verify that the resulting Martin kernel is identical to the homogeneous kernel up to a multiplicative factor that does not alter the topology.
Authors: We agree that greater explicitness is needed. In the revised manuscript we will display the explicit form of the weighted transition matrix in Section 3, where the new probabilities are obtained by multiplying the homogeneous transitions by the given weights and renormalizing at each step. Because the weights enter multiplicatively along any finite path, the Green function remains a product; consequently the Martin kernel at an infinite word differs from the homogeneous kernel only by a positive multiplicative factor that depends on the weights of the initial segment but is independent of the endpoint. This factor does not affect the topology of the boundary or the set of minimal harmonic functions, which is why the two boundaries coincide. revision: yes
-
Referee: [Theorem 4.2 and its proof] The argument that the Martin boundary is homeomorphic to the fractal via the coding map appears to invoke the open set condition only to guarantee positive measure and dimension; however, the topological identification itself must be shown to hold without OSC, or the manuscript must clarify precisely where OSC enters the proof of the homeomorphism.
Authors: The statement of the theorem is under the open set condition, as required for the weighted IFS to produce a fractal with the expected separation properties. In the proof of Theorem 4.2 the OSC is used at two points: (i) to guarantee that the coding map is injective on the Martin boundary (preventing distinct infinite words from mapping to the same point when overlaps occur), and (ii) to ensure that every point of the fractal is the image of a minimal harmonic function. We will insert a short paragraph immediately after the statement of Theorem 4.2 that isolates these two uses of OSC and notes that the continuity of the coding map itself follows from the contractivity of the IFS alone. revision: yes
Circularity Check
No significant circularity
full rationale
The derivation redefines transition probabilities on the word space of the IFS according to the given weights and shows that the resulting Martin kernel and boundary coincide topologically with the homogeneous case of Denker-Sato 2001. This follows directly from the fact that the underlying graph is still the regular tree, the Green function remains a path product, and the coding map to the fractal is unchanged; the OSC is used only to ensure positive measure, not for the boundary identification itself. No load-bearing step reduces to a self-citation, a fitted parameter renamed as prediction, or an ansatz smuggled from prior work by the same authors. The result is externally grounded in the cited 2001 theorem and the explicit Markov-chain construction.
Axiom & Free-Parameter Ledger
axioms (1)
- domain assumption Open set condition holds for the probabilistic iterated function scheme
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.