Reweighting metric measure spaces and Onsager-Machlup

Basic Information

• Paper ID: 2510.10591
• Title: Reweighting metric measure spaces and Onsager-Machlup
• Author: Zachary Selk (Florida State University)
• Classification: math.PR (Probability Theory), math.MG (Metric Geometry)
• Publication Date: October 14, 2025 (arXiv preprint)
• Paper Link: https://arxiv.org/abs/2510.10591

Abstract

Given a metric measure space $M:=(X,d,\mu)$, the Onsager-Machlup (OM) functional is a real-valued function viewed as a generalization of the concept of probability density functions. While the effects of measure reweighting on the OM functional have been studied, the analogous reweighting problem for metrics remains open to the author's knowledge. This paper establishes transformation formulas for the OM functional when simultaneously reweighting both the metric and measure on geodesic metric measure spaces.

Research Background and Motivation

Problem Background

1. Difficulties in infinite-dimensional probability: In the natural setting of stochastic processes—infinite-dimensional probability—the absence of a Lebesgue measure to define probability density functions is a fundamental difficulty.
2. Significance of the OM functional: The Onsager-Machlup functional is a generalization of the probability density function, defined through the limit of small ball measure ratios: $\lim_{r \to 0^+} \frac{\mu(B(r,x))}{\mu(B(r,y))} = \exp(OM(y) - OM(x))$
3. Limitations of existing research: Although the transformation of the OM functional under measure reweighting has been thoroughly studied, the effects of metric reweighting remain unknown.

Research Motivation

• OM theory can be placed within the research framework of general metric measure spaces, making it natural to study the effects of metric changes
• It is known that even equivalent metrics can lead to different OM functionals, but transformation formulas for joint metric and measure reweighting are lacking
• Filling this theoretical gap is important for understanding the geometric properties of the OM functional

Core Contributions

1. Established transformation formulas for joint metric-measure reweighting: For the metric measure space $(X, e^{-U}d_0, e^{-V}\mu_0)$, explicit expressions for the OM functional are provided
2. Distinguished finite-dimensional and infinite-dimensional cases: Proved that in finite-dimensional spaces satisfying small ball estimates, the OM functional exists with explicit formulas; in infinite-dimensional spaces, non-constant metric reweighting destroys the existence of the OM functional
3. Provided concrete applications: Including treatment of important examples such as Riemannian manifolds and Gaussian processes
4. Revealed deep connections between geometry and probability: Through an analogy with "chart projections," demonstrated the role of geometric deformation in probability theory

Methodology Details

Problem Formulation

Study how the OM functional changes on a metric measure space $M_0 = (X, d_0, \mu_0)$ when simultaneously performing the following transformations:

• Metric transformation: $d_0 \mapsto e^{-U}d_0$
• Measure transformation: $\mu_0 \mapsto e^{-V}\mu_0$

where $U, V: X \to \mathbb{R}$ are locally uniformly continuous functions.

Core Theorem

Theorem 1.3 Let $M_0 := (X, d_0, \mu_0)$ be a geodesic metric measure space with OM functional $OM_0$.

(a) Constant metric reweighting: If $U$ is constant, then $OM = OM_0 + V$

(b) Finite-dimensional case: If there exists a point $q \in Z$ such that $\lim_{r \to 0^+} \frac{\mu_0(B_0(Cr, q))}{\mu_0(B_0(r, q))} = C^p$ then $OM = OM_0 - pU + V$

(c) Infinite-dimensional case: If there exists a small ball estimate $\lim_{r \to 0^+} r^\alpha \log \mu_0(B_0(r, q)) = -C$ and $U$ is non-constant, then the OM functional does not exist on $Z$.

Technical Innovations

1. Geometric treatment of metric reweighting: The new metric is defined through integration of absolutely continuous paths: $d(x,y) = \inf\left\{\int_0^1 e^{-U(\gamma(t))}|\dot{\gamma}(t)|dt : \gamma \in AC([0,1], X), \gamma(0)=x, \gamma(1)=y\right\}$
2. Utilization of local uniform continuity: Key use of local uniform continuity of $U, V$ to control ball inclusions and measure estimates
3. Classification of small ball estimates: Distinguishes finite-dimensional and infinite-dimensional cases based on different small ball behaviors (power law vs. exponential decay)

Experimental Setup

Theoretical Verification

This work is primarily theoretical, with results verified through:

1. Concrete example verification:
   • $n$-dimensional Riemannian manifolds satisfy condition (2) with $p = n$
   • Fractional Brownian motion, SPDE solutions, etc. satisfy condition (3)
2. Limiting case verification: Verification that when $U$ is constant, the result reduces to known measure reweighting results

Application Examples

1. Euclidean space: $M_0 = (\mathbb{R}^n, d_0, e^{-f}\lambda)$
2. Riemannian manifolds: $(M, g, \text{vol}_g)$
3. Infinite-dimensional Gaussian measure spaces

Experimental Results

Main Results

1. Flexibility in finite dimensions: In finite-dimensional spaces, the OM functional can approximate arbitrary functions through metric adjustment (Corollary 3.1)
2. Rigidity in infinite dimensions: In infinite-dimensional spaces, conformally equivalent metrics yield the same OM functional (Corollary 3.2)
3. Uniformization result: One can choose a "uniformizing" metric such that the OM functional is identically zero (Corollary 3.3)

Key Findings

1. Dimensional effects: Finite-dimensional and infinite-dimensional spaces exhibit fundamentally different behavior under metric reweighting
2. Geometric-probabilistic duality: In finite dimensions, measure reweighting is equivalent to metric reweighting; in infinite dimensions, they are not equivalent
3. Importance of regularity: Local uniform continuity is the key technical condition ensuring the validity of the results

Related Work

Main Research Directions

1. Development of OM theory: From Onsager-Machlup's original work to modern metric measure space theory
2. Measure transformations: Existing results on measure reweighting by Ayanbayev et al. (Theorem 1.2)
3. Small ball probabilities: Work on small ball estimates in Gaussian processes by Kuelbs, Li, and others

Position of This Work

• First systematic study of the effects of metric reweighting on the OM functional
• Unified treatment framework for finite and infinite dimensions
• Related to Han's work on conformal transformations of metric measure spaces

Conclusions and Discussion

Main Conclusions

1. Established a complete transformation theory for the OM functional under joint metric-measure reweighting
2. Revealed the essential differences between finite-dimensional and infinite-dimensional spaces
3. Proved that the OM functional possesses stronger "regularity" in infinite-dimensional spaces

Limitations

1. Requires local uniform continuity of functions $U, V$, which may be too restrictive in some applications
2. The assumption of geodesic spaces limits the scope of applicability
3. Small ball estimate conditions in the infinite-dimensional case require specific verification

Future Directions

1. Generalization to non-geodesic spaces
2. Study of more general metric transformations
3. Exploration of applications in machine learning and Bayesian inverse problems

In-Depth Evaluation

Strengths

1. Theoretical completeness: Provides a complete solution to the metric reweighting problem, filling an important theoretical gap
2. Technical depth: Cleverly combines techniques from metric geometry, measure theory, and probability theory
3. Profound results: Reveals essential differences between finite-dimensional and infinite-dimensional spaces
4. Broad applicability: Encompasses numerous examples from Riemannian geometry to stochastic analysis

Weaknesses

1. Technical conditions: Local uniform continuity and other conditions may be difficult to verify in practical applications
2. Constructiveness: While existence results are provided, concrete construction methods are lacking
3. Computational complexity: The definition of the new metric involves variational problems, making computation difficult

Impact

1. Theoretical contribution: Provides important geometric perspective for OM theory
2. Cross-disciplinary value: Connects metric geometry and probability theory, potentially inspiring new research directions
3. Application potential: Has potential applications in Bayesian statistics, machine learning, and related fields

Applicable Scenarios

1. Theoretical research: Metric measure space theory, geometric probability
2. Applied mathematics: Bayesian inverse problems, stochastic analysis
3. Data science: Geometric data analysis, manifold learning

References

The paper cites 12 important references covering classical works in core areas including OM theory, small ball probabilities, and metric geometry, providing a solid theoretical foundation for the research. Particularly noteworthy are Onsager-Machlup's original work 10 and recent related research by Ayanbayev et al. 1.