[Paper Review] The Functional Determinant and the Partition Function in Geometric Flows

Christopher Lin|arXiv (Cornell University)|Nov 21, 2013

Geometric Analysis and Curvature Flows10 references3 citations

TL;DR

This paper proposes a statistical mechanics framework for geometric flows by defining a partition function using the functional determinant of geometric operators, leading to a monotonic entropy functional along conformal flows on closed surfaces. The approach derives an explicit entropy formula via the Polyakov formula, resolving a foundational ambiguity in Perelman's $χ$-entropy by grounding it in microstate-based statistical mechanics.

ABSTRACT

We propose the use of the functional determinant of geometric operators in constructing an entropy functional associated to geometric flows. Our approach is based on the direct computation of the partition function, with a well-defined set of microstates and macrostates in the canonical ensemble. The approach is motivated by a fundamental enigma in Perelman's derivation of his famous $\mathcal{W}$-entropy. The defining feature of our entropy is that the energy of each microstate in the partition function is invariant along the associated geometric flow - a clue that could be inferred from Perelman's work. Moreover, the monotonicity of our entropy along the associated geometric flow is then a natural result in the statistical mechanics framework. While we will not argue in a completely rigorous manner, we will use the formalism to derive an explicit formula for an entropy associated to conformal flows on a closed surface based on the Polyakov formula for the determinant of the Laplacian. We also discuss possible extensions of our results to more general operators and manifolds.

Motivation & Objective

• To resolve the unresolved statistical mechanics origin of Perelman’s $χ$-entropy, which lacks a defined partition function or microstate structure.
• To construct a well-defined partition function for geometric flows using the functional determinant of geometric operators, ensuring microstates and macrostates are physically meaningful.
• To establish a monotonic entropy functional along geometric flows by ensuring the energy of each microstate is invariant under the flow, consistent with statistical mechanics principles.
• To extend this framework to conformal flows on closed surfaces using the Polyakov formula for the Laplacian determinant.

Proposed method

• Define the partition function as $ Z = ext{Det}(A)^{-1/2} $, where $ A $ is a geometric operator (e.g., Laplacian), using the functional determinant to encode microstate statistics.
• Model microstates as functions $ φ $ in a Hilbert space $ Γ $, with energy $ E(\varphi) = \int_M |\nabla \varphi|^2 dv $, representing Dirichlet energy.
• Treat the Riemannian metric $ g $, function $ f $, and parameter $ \tau $ as macrostates that determine the operator $ A $, thus fixing the energy spectrum.
• Use the finite-dimensional analog of the functional integral $ \int e^{-\langle \varphi, A\varphi \rangle} d\varphi = (\text{Det} A)^{-1/2} $ to justify the infinite-dimensional case.
• Derive the entropy as $ S = \log Z - \beta \partial_\beta \log Z $, with $ \beta = 1/\tau $, ensuring monotonicity via statistical mechanics.
• Apply the Polyakov formula for the determinant of the Laplacian under conformal deformations to compute the partition function explicitly on closed surfaces.

Experimental results

Research questions

• RQ1 How can a partition function be rigorously defined for geometric flows using functional determinants of geometric operators?
• RQ2 Why is Perelman’s $\mathcal{W}$-entropy not derived from a proper partition function, and what is the physical meaning of its missing microstate structure?
• RQ3 Can a monotonic entropy functional be constructed for conformal flows on closed surfaces using the functional determinant and Dirichlet energy?
• RQ4 What is the explicit form of the entropy functional derived from the Polyakov formula for the Laplacian determinant?
• RQ5 How does the invariance of microstate energy along the flow ensure monotonicity of the entropy in the statistical mechanics framework?

Key findings

• The proposed partition function $ Z $ is defined via the functional determinant $ \text{Det}(A)^{-1/2} $, with $ A $ being the Laplacian associated to the metric $ g $, providing a statistical mechanical foundation for geometric flows.
• The energy of each microstate $ \varphi $, given by the Dirichlet energy $ \int_M |\nabla \varphi|^2 dv $, remains invariant under the conformal flow, ensuring consistency with thermodynamic equilibrium.
• The entropy functional $ S = \log Z - \beta \partial_\beta \log Z $ is monotonic along the conformal flow due to the statistical mechanics framework, with monotonicity arising naturally from the invariance of microstate energies.
• An explicit formula for the entropy is derived using the Polyakov formula, linking the determinant of the Laplacian to the conformal factor and yielding a concrete expression for the entropy in terms of the metric and function $ f $.
• The finite-dimensional analog of the functional integral confirms the validity of the measure transformation rule $ \mathcal{D}\varphi_g = J(g_0,g) \mathcal{D}\varphi_{g_0} $, with $ J(g_0,g) = \det(a^{ij}) $, justifying the infinite-dimensional measure dependence on macrostates.
• The framework generalizes to other geometric operators and manifolds, suggesting broader applicability beyond the Laplacian and conformal flows.