Gel'fand-Yaglom Theorem Overview

• Gel'fand-Yaglom theorem is a fundamental result in spectral analysis that reformulates functional determinants as the solution to an initial-value problem.
• The theorem extends to discrete (lattice) systems and Hamiltonian setups with Lagrangian boundary conditions, enabling both analytic and numerical applications.
• It streamlines semiclassical path integral computations by bypassing explicit spectral sums, thereby improving efficiency in fluctuation determinant evaluations.

The Gelʹfand–Yaglom theorem is a foundational result in spectral analysis of differential operators, yielding a practical means of computing functional determinants central to quantum mechanics, statistical physics, and field theory. The theorem bypasses explicit spectral summation, reformulating the determinant as the solution to an initial-value problem. Significant generalizations include its lattice (discrete) counterpart and its extension to Hamiltonian systems with Lagrangian boundary conditions, as captured in "Generalized Gelfand-Yaglom Formula for a Discretized Quantum Mechanic System" (Shea, 2020). The Gelʹfand–Yaglom methodology informs both analytic and numerical calculations across diverse settings.

1. Classical Formulation for Sturm–Liouville Operators

Let $L = -\frac{d^2}{dt^2} + U(t)$ act on $y(t)$ with Dirichlet boundary conditions $y(0)=0$, $y(T)=0$. The $\zeta$-regularized determinant, $\det_\zeta L = \exp[-\zeta_L'(0)]$, where $\zeta_L(s) = \sum_{\lambda_n} \lambda_n^{-s}$ over nonzero eigenvalues, is not computed via spectral summation but via an ODE:

• Define $y(t)$ by $y''(t) = U(t) y(t)$, $y(0)=0$, $y'(0)=1$.
• Then $\det_\zeta L = y(T)$.

This approach appears in semiclassical quantum mechanics, often with $L$ replaced by $A = -\frac{d^2}{dt^2} - \frac{1}{m} V''(q_c(t))$, and prefactor calculations such as

$$\partial_p q_c(T) = \frac{1}{2m} \det_\zeta \left(-\frac{d^2}{dt^2} - \frac{1}{m} V''(q_c)\right)$$

(Shea, 2020).

2. Hamiltonian and Lagrangian Boundary Condition Generalization

Given a Hamiltonian system with phase-space paths $(p(t), q(t))$ and action

$$\tilde S[\tilde\gamma] = \int_0^T \bigl(p\,\dot q - H(p,q)\bigr) dt + f_1(q(0), b_1) - f_2(q(T), b_2)$$

with $f_1$, $f_2$ enforcing boundary conditions, the critical path solves Hamilton’s equations and boundary constraints

$$\dot q = \partial_p H, \quad \dot p = -\partial_q H,$$

$$p(0) = \partial_q f_1(q(0), b_1), \quad p(T) = \partial_{q'} f_2(q(T), b_2).$$

The second variation yields a first-order block operator

$$\tilde A = \begin{pmatrix} -H_{pp} & \frac{d}{dt} - H_{qp} \ - \frac{d}{dt} - H_{pq} & -H_{qq} \end{pmatrix}_{(p_c, q_c)},$$

acting on $x = (\delta p, \delta q)^T$ (Shea, 2020). Reducing further, one obtains a second-order operator acting on $\delta q$ with mixed (Robin) conditions:

$$A = -\frac{d^2}{dt^2} - \frac{1}{m} V''(q_c(t)), \quad \delta q'(0) = \frac{1}{m} f''_1(q(0)) \delta q(0), \quad \delta q'(T) = \frac{1}{m} f''_2(q(T)) \delta q(T).$$

The generalized Gelʹfand–Yaglom formula relates the regularized determinant of $A$ to derivatives of the principal function,

$$\frac{\partial^2 \tilde S_{\tilde\gamma_c}}{\partial b_1 \partial b_2} = 2 \frac{\partial^2 f_1}{\partial b_1 \partial q} \frac{\partial^2 f_2}{\partial b_2 \partial q'} m \det_\zeta A.$$

For scalar $f''_1 = a_1, f''_2 = a_2$:

$$\det_\zeta A = \frac{1}{2m} (\partial^2_{b_1 b_2} \tilde S_{\tilde\gamma_c}) \left( \frac{\partial^2 f_1}{\partial b_1 \partial q} \frac{\partial^2 f_2}{\partial b_2 \partial q'} \right)^{-1}.$$

3. Discrete (Lattice) Gelʹfand–Yaglom Formulation

Discretization divides $[0,T]$ into $N$ steps ($\varepsilon = T/(N-1)$), $q_i$, $p_i$ label positions and momenta. The discrete action is

$$\tilde S_d = \sum_{i=1}^{N-1} [p_i (q_{i+1} - q_i) - \varepsilon \mathcal{H}(p_i, q_i)] + f_1(q_1, b_1) - f_2(q_N, b_2).$$

Discrete Hamilton's equations yield difference equations,

$$q_{i+1} - q_i = \varepsilon \partial_p H(p_i, q_i), \quad p_i - p_{i-1} = \varepsilon \partial_q H(p_i, q_i),$$

with discrete boundary analogues.

Second variation gives a finite block-tridiagonal matrix $\tilde A_N$, with blocks involving Hessians $H_{pp}$, $H_{pq}$, $H_{qq}$. The discrete Gelʹfand–Yaglom formula is

$$\det \left[ \partial_{b_1} \partial_{b_2} \tilde S_{d, \tilde\gamma_c} \right] = \prod_{i=1}^{N-1} \det(- H_{pq}(p_i, q_i) - I) \frac{ \det(\partial^2_{q_1 b_1} f_1) \det( \partial^2_{q_N b_2} f_2 ) } { \det \tilde A_N }$$

(see Theorem II.1 in (Shea, 2020)). For pure kinetic-plus-potential systems,

$$\det \left[ \partial^2_{b_1 b_2} \tilde S_d \right] = \frac{ \partial^2_{q_1 b_1} f_1 \, \partial^2_{q_N b_2} f_2 } { \det \tilde A_N }$$

For the corresponding discretized second-order operator $A_N$, with boundary encoding $f''_1$, $f''_2$ as in Robin conditions, one finds

$\det \tilde A_N = (-1)^{N-1} m \det A_N$

(Theorem II.3). This links discrete and continuous determinant constructions.

4. Continuum Limit, Regularization, and Convergence

Weak convergence from discrete to continuous operators is established by mapping discrete sums to integrals:

$$Y_N^T \tilde A_N X_N \to \int_0^T Y^T(t) \tilde A X(t) dt$$

Boundary terms vanish as $N \to \infty$ under proper conditions. The lattice-regularized determinant is defined as

$$\det_{\text{reg}} \tilde A := \lim_{N\to\infty} \det \tilde A_N, \qquad \det_{\text{reg}} A := \lim_{N\to\infty} \varepsilon^{N-1} \det A_N$$

The continuum limit returns, for pure kinetic-plus-potential cases,

$$\det_{\text{reg}} A = \frac{ \partial^2_{q_1 b_1} f_1 \, \partial^2_{q_N b_2} f_2 } { m \, \partial^2_{b_1 b_2} \tilde S_{\tilde\gamma_c} }$$

(Shea, 2020).

5. Path Integral and Semiclassical Applications

In the case $f_1=f_2=0$ (Neumann-type or Dirichlet after swapping $p \leftrightarrow q$), the classical Gelʹfand–Yaglom formula is recovered for fluctuation determinants in path-integral semiclassical prefactors. The lattice regularization supplies an algebraic path to calculate these determinants without analytic continuation of $\zeta$-functions, valuable for

• Semi-classical quantization in 1D quantum mechanics
• Computation of path integral prefactors
• Prospective extension to higher-dimensional quantum field theories via lattice regularized fluctuation determinants

The flexibility in Hamiltonian structure (general $H(p,q)$ including mixed $H_{pq}$ terms) is retained (Shea, 2020).

6. Comparison with Other Analytic and Numerical Methods

The determinant expressions obtained via Gelʹfand–Yaglom have known exact or closed forms in certain cases. The direct differentiation and block-Laplace expansions simplify computational workflows, handling finite tridiagonal matrices and block matrices efficiently. These approaches bypass the explicit eigenvalue product, yielding significant computational speedup and analytic clarity compared to traditional spectral sum regularizations.

A tabular summary for the main operator/determinant constructions:

Setting	Operator/Matrix	GY Determinant Formula
Continuum (Dirichlet)	$L = -d^2/dt^2 + U(t)$	$\det_\zeta L = y(T)$
Phase space (general)	$\tilde A$/$A$	$\det_\zeta A \propto \partial^2 \tilde S$
Discrete (lattice)	$\tilde A_N$, $A_N$	$\det [\partial_{b_1} \partial_{b_2} \tilde S_d]$, etc.

7. Impact, Extensions, and Future Directions

The generalized Gelʹfand–Yaglom formula provides both theoretical insight and a concrete computational tool for quantum mechanical and field theoretic path integrals. Its discrete version is particularly promising for numerical evaluation in systems with nontrivial Hamiltonians and boundary conditions. Prospective applications include systematic computation of fluctuation determinants in lattice field theories and rigorous foundation for higher-dimensional generalizations.

Extensions not covered here but motivated by the algebraic structure include Gelʹfand–Yaglom approaches for coupled systems, non-Hermitian dynamics, and fully non-commutative matrix operator formulations. Adapting lattice regularization for more complex quantum graphs, theory-space models, and gauge/fermion systems is a plausible direction for future research (Shea, 2020).