Path Integrals
Notes on Path Integals (Draft)
Abstract
These notes survey the theory of Feynman path integrals through the lens of a specific organizing thesis: that path integrals are not uniquely defined without a prescription, and that action principles and geometric structures serve as selection rules to resolve the measure and ordering ambiguities that arise in curved, constrained, and singular settings. The notes are organized around a curated bibliography of 37 primary references spanning 1942–2018, grouped into six thematic sections that trace a deliberate arc from pedagogical foundations through Feynman’s functional integral and Schwinger’s quantum action principle, to technical applications in the hydrogen atom, gauge theories, and path integrals on curved manifolds. The emphasis throughout is on the reciprocal relationship between Schwinger’s variational formulation and Feynman’s sum-over-histories, and on the mathematical structures—Van Vleck–Morette determinants, symplectic two-forms, Laplace–Beltrami operators—that constrain the correct definition of the functional integral in nontrivial settings.
Introduction
The path integral formulation of quantum mechanics, introduced by Feynman in his 1942 doctoral dissertation, provides a third picture of quantum theory complementary to the operator-based formulations of Heisenberg and the wave mechanics of Schrödinger. In this picture, the quantum mechanical propagator is expressed as a “sum over histories”—a functional integral over all paths connecting initial and final configurations, each weighted by $e^{iS/\hbar}$ where $S$ is the classical action. The elegance and physical transparency of this formulation have made it indispensable in quantum field theory, statistical mechanics, and condensed matter physics.
Yet the path integral, for all its intuitive power, conceals a foundational difficulty. In flat Euclidean configuration space with a well-behaved potential, the time-slicing construction yields unambiguous results. But the moment one leaves this setting—introducing singular potentials, constraints, curved manifolds, or gauge redundancies—the definition of the functional integral becomes ambiguous. Different discretization prescriptions, operator orderings, and measure choices can yield different quantum theories, potentially with additional curvature-dependent potential terms or incorrect spectra.
The present notes are organized around a specific thesis regarding this ambiguity:
The correct definition of path integrals in nontrivial settings is inseparable from action principles and geometry. Schwinger’s quantum action principle, geometric semiclassical structures (Van Vleck–Morette determinants, Laplace–Beltrami operators), and symplectic geometry in constrained systems provide the selection rules needed to resolve measure and ordering ambiguities.
This thesis is not original; it is the implicit organizing principle of the bibliography surveyed here, and the explicit conclusion of several key papers within it—most notably Toms (2004), who demonstrates that requiring consistency with the Schwinger action principle uniquely determines the path integral measure in curved space. What follows is a guide to the literature that traces this argument from its foundations to its technical applications.
The notes proceed through six thematic sections that mirror the structure of the bibliography: pedagogical texts that establish the technical ladder (Section 1); the historical development and foundational machinery of Feynman’s functional integral (Section 2); Schwinger’s quantum action principle as the conceptual center of gravity (Section 3); the hydrogen atom as a diagnostic test case where naive definitions fail (Section 4); constrained quantization and gauge theory (Section 5); and path integrals on curved manifolds as the culmination where all threads converge (Section 6). A synthesis section then draws out the cross-cutting themes, and a final section identifies gaps and connections to other parts of the textbook.
1. Pedagogical Texts: A Technical Ladder
The nine texts assembled in the bibliography’s opening section are not a random collection of “good books on path integrals.” Read in sequence, they define a technical ladder that moves the reader from physical intuition to mathematical rigor, and from quantum mechanics to the geometric and measure-theoretic structures that dominate the later sections.
1.1 Entry Points: Physical Intuition and Operator Calculus
The natural starting point is Feynman and Hibbs (1965), the classic pedagogical text that develops the path integral from first principles. The emended Dover edition includes corrections to the original and remains the standard first encounter with the subject. Feynman and Hibbs emphasize physical intuition and computational technique: free particle propagators, the harmonic oscillator, perturbation theory, and the connection to statistical mechanics via Wick rotation. What the text does not address—and what the remaining entries supply—is the question of what happens when the simple time-slicing construction encounters difficulties.
Johnson and Lapidus (2002) provide a bridge between Feynman’s path integral and his operator calculus. The integration of these two formalisms yields tools for handling quantum evolution operators, particularly for systems with time-dependent Hamiltonians where the standard path integral construction requires additional care. The operator-calculus perspective also foreshadows the Schwinger formulation, where the functional integral is intimately connected to operator ordering.
Junker and Inomata (1985) develop transformation techniques for the free propagator, showing how known propagators can be converted into solutions for more complex potentials through systematic coordinate transformations. This early paper in the bibliography signals a theme that will recur prominently in the hydrogen atom section: the idea that the “right” transformation can convert an intractable path integral into a solvable one.
1.2 Mathematical Foundations
Roepstorff (1994) offers a mathematically rigorous treatment of path integrals, addressing the measure-theoretic foundations that physics-oriented texts typically elide. The Wiener measure provides a well-defined probability measure on continuous paths, but the oscillatory Feynman integral does not fit neatly into this framework. Roepstorff develops the functional analysis required to put path integrals on firmer mathematical footing while maintaining physical applicability. For the reader who will later encounter measure ambiguities in curved space, this preparation is essential.
Cartier and DeWitt-Morette (2007) develop functional integration from a modern mathematical perspective, with particular emphasis on the interplay between action principles and symmetries. Cécile DeWitt-Morette’s work on the Van Vleck–Morette determinant and its role in semiclassical approximations is a thread that runs through the entire bibliography. The determinant
\[\Delta(q_2, q_1) = \det\left(-\frac{\partial^2 S}{\partial q_2 \partial q_1}\right)\]encodes the focusing and defocusing of classical trajectories near a stationary path, and appears as the prefactor in the semiclassical approximation to the propagator. Its role in curved-space path integrals—where it enters not merely as a semiclassical correction but as part of the exact measure—is a central concern of Section 6.
1.3 Encyclopedic and Graduate-Level References
Kleinert (2009, 5th edition) is the encyclopedic reference: over 1600 pages covering path integrals in quantum mechanics, statistical mechanics, polymer physics, and financial markets. Key contributions that recur throughout the bibliography include the Duru–Kleinert transformation for the hydrogen atom, a quantum equivalence principle for path integrals in spaces with curvature and torsion, and variational perturbation theory producing convergent (rather than merely asymptotic) expansions. The breadth of applications—from critical phenomena to quantitative finance—makes Kleinert the natural companion volume for a project in scientific computation that bridges these domains.
Baaquie (2014) explores the relationship between path integral and Hamiltonian formulations, with attention to how canonical quantization emerges from the functional integral. This is useful for understanding the equivalence of different quantization schemes, a topic that becomes critical when constraints are present.
Chaichian and Demichev (2018) provide a two-volume graduate-level treatment covering fundamental techniques (Volume I) and advanced applications including gauge fields and statistical mechanics (Volume II). Their systematic development from classical to quantum systems makes these volumes a natural complement to the more specialized treatments in the later sections.
1.4 Implications
The inclusion of DeWitt-Morette and Kleinert, in particular, signals that these notes are not aimed at the reader who merely wants to compute a propagator. The technical ladder is designed to prepare the reader for measure-theoretic subtleties, geometric structures, and the question of what it means to define a path integral correctly—questions that are invisible in flat space with smooth potentials, but that become unavoidable in the settings treated below.
2. Feynman’s Functional Integral: Definition, Approximation, and Environment
The second section of the bibliography traces the historical development and foundational aspects of Feynman’s path integral, organized around three pillars: the original formulation, the semiclassical apparatus, and the extension to open quantum systems.
2.1 The Origin: Feynman’s Dissertation (1942)
Feynman’s Princeton dissertation, written under John Wheeler, established the path integral formulation of quantum mechanics. The work was directly inspired by Dirac’s 1933 paper on the Lagrangian in quantum mechanics, which observed that the quantum transformation function between states at different times “corresponds to” $\exp(iS/\hbar)$. Dirac left the nature of this correspondence unspecified. Feynman’s insight was to take it literally: the quantum propagator is a sum over all possible paths, each weighted by $\exp(iS/\hbar)$, with the sum defined as a limit of finite-dimensional integrals obtained by discretizing the time interval.
The dissertation is worth reading not merely for historical interest but because it reveals the assumptions built into the construction. The time-slicing definition requires: (i) a configuration space with a natural volume element, (ii) a Lagrangian that is at most quadratic in velocities (to make the short-time propagator Gaussian), and (iii) a potential that does not produce singularities in the short-time limit. When any of these conditions fails, the construction must be modified—and the question is how.
2.2 Definition and the Semiclassical Determinant
Morette (1951) provided early rigorous definitions and practical approximation schemes for Feynman’s path integral. Her central contribution is the Van Vleck–Morette determinant, which gives the correct measure factor in the semiclassical (WKB-type) limit. The semiclassical propagator takes the form
\[K(q_2, t_2; q_1, t_1) \approx \sum_{\text{cl. paths}} \left(\frac{1}{2\pi i \hbar}\right)^{n/2} \Delta^{1/2}(q_2, q_1) \, e^{iS_{\text{cl}}/\hbar}\]where the sum runs over classical paths connecting $(q_1, t_1)$ to $(q_2, t_2)$ and $\Delta$ is the Van Vleck–Morette determinant. The determinant is not merely a normalization convenience; it encodes genuine physical information about the density of classical trajectories near the stationary path.
Choquard and Steiner (1996) clarify the relationship between Van Vleck’s original determinant (1928) and the Morette–Van Hove generalizations, establishing precisely when these coincide and when additional corrections are required. The distinction matters because in curved space, the determinant acquires a nontrivial dependence on the geometry that must be handled consistently with the measure.
2.3 Open Systems: The Influence Functional
Feynman and Vernon (1963) extend the path integral framework to open quantum systems through the influence functional formalism. When a system interacts with an environment (“bath”), the environment degrees of freedom can be integrated out exactly (when the bath is linear), leaving an effective action for the system that includes non-local-in-time terms encoding memory effects. The influence functional
\[\mathcal{F}[q, q'] = \int \mathcal{D}[Q] \, \exp \left(\frac{i}{\hbar}\left(S_{\text{env}}[Q; q] - S_{\text{env}}[Q; q']\right)\right)\]where $q, q’$ label the forward and backward paths of the system and $Q$ denotes the bath coordinates, is foundational for the theory of quantum dissipation, decoherence, and the Caldeira–Leggett model of quantum Brownian motion.
The inclusion of Feynman–Vernon in a bibliography otherwise focused on foundational and geometric aspects is noteworthy. It signals that the path integral is not merely a tool for computing propagators in closed systems, but a framework whose real power emerges when degrees of freedom are integrated out —a perspective that connects directly to effective actions and the renormalization group in quantum field theory.
3. Schwinger’s Quantum Action Principle: The Conceptual Center
Julian Schwinger’s variational quantum action principle (QAP) provides an alternative foundation for quantum mechanics that is the conceptual hinge of these notes. While Feynman’s formulation proceeds “globally” through a sum over histories, Schwinger’s proceeds “locally” through a variational principle acting on the transformation function between quantum states. Their equivalence in simple settings is well established, but the real force of the Schwinger formulation emerges precisely where Feynman’s construction becomes ambiguous.
3.1 Statement of the Principle
The Schwinger quantum action principle states that the variation of the transformation function $\langle q_2, t_2 | q_1, t_1 \rangle$ under infinitesimal changes in the dynamics is given by
\[\delta \langle q_2, t_2 | q_1, t_1 \rangle = \frac{i}{\hbar} \langle q_2, t_2 | \delta W_{12} | q_1, t_1 \rangle\]where $W_{12}$ is the action operator. This is a statement about operator matrix elements, not about functional integrals. Its power lies in the fact that it constrains the form of the propagator without requiring an explicit construction via time-slicing.
3.2 Historical and Technical Context
Milton (2007), drawing on his position as Schwinger’s student and biographer, provides a historical and technical overview of Schwinger’s contributions in arXiv:physics/0610054. The development from Dirac’s 1933 observation, through Feynman’s path integral, to Schwinger’s QAP is traced with attention to the distinct conceptual advantages of each formulation. Where Feynman’s approach excels in physical visualization and in perturbative calculations (via Feynman diagrams), Schwinger’s excels in maintaining manifest operator structure and in treating symmetries and boundary conditions.
Milton (2014, 2015), in two companion papers (arXiv:1402.3762 and arXiv:1503.08091) later collected as a Springer Brief, traces the development in greater technical detail. Key topics include:
- The Keldysh–Schwinger time-cycle method for extracting matrix elements in non-equilibrium situations, essential for real-time finite-temperature field theory.
- The connection between the Schwinger action principle and Green’s functions.
- The variational formulation of QED and the development of source theory.
- The treatment of fermionic anticommuting fields within the same formalism as bosonic fields.
The Keldysh–Schwinger formalism is of particular note for computation: the closed-time-path contour provides a systematic way to handle initial-state density matrices and real-time dynamics, complementing the Euclidean (imaginary-time) methods more commonly encountered in textbook treatments.
3.3 From Action Principle to Effective Action
Toms (2009) provides a monograph-length treatment of how the Schwinger action principle connects to the one-particle irreducible (1PI) effective action $\Gamma[\varphi]$, which encodes all quantum corrections in a form directly useful for studying spontaneous symmetry breaking and effective field theories. The connection between Schwinger’s variational approach and the background field method is particularly important for gauge theories, where maintaining gauge invariance order by order in perturbation theory is nontrivial.
3.4 Why Schwinger Matters for These Notes
The role of the Schwinger QAP in this bibliography is not merely historical. As will become clear in Section 6, it is Toms’ application of the Schwinger principle to path integrals in curved space that provides one of the sharpest resolutions of the measure ambiguity. The argument is that the path integral measure must be chosen so that the resulting propagator satisfies the Schwinger variational equation; this requirement uniquely selects a measure involving the Van Vleck–Morette determinant. The Schwinger principle thus acts as a selection rule on functional integrals, constraining the correct definition in settings where multiple definitions are possible.
This is the central thesis of the bibliography, stated in its strongest form: Schwinger’s QAP is not merely “equivalent” to the path integral in simple cases; it provides the principled route to determining what the path integral must be in curved or otherwise ambiguous settings.
4. The Hydrogen Atom: A Diagnostic Test Case
The hydrogen atom occupies a special place in the path integral literature. It is exactly solvable by other methods (the Schrödinger equation in spherical coordinates), which makes it a stringent diagnostic: any proposed definition of the path integral must reproduce the known spectrum and wavefunctions. At the same time, the Coulomb potential’s $1/r$ singularity makes Feynman’s original time-sliced definition divergent, so the hydrogen atom is a “stress test” that exposes the limitations of naive constructions.
4.1 The Problem
The difficulty is straightforward to state. In the time-slicing construction, the short-time propagator involves $\exp(iL\,\Delta t/\hbar)$ where the Lagrangian contains the Coulomb potential $V(r) = -e^2/r$. As any time slice passes through $r = 0$, the potential diverges, and the resulting integral is ill-defined. One cannot simply “regulate” this divergence without making a choice that affects the final answer.
4.2 The Duru–Kleinert Transformation
Duru and Kleinert (1979, 1982) resolved this problem through a transformation that remains one of the most elegant results in the field. The key insight is to perform a combined transformation of both time and space:
- Pseudotime: Introduce a new time parameter $\tau$ related to the physical time by $d\tau = dt/r$, which regularizes the Coulomb singularity by “slowing down” the path as it approaches the origin.
- Kustaanheimo–Stiefel map: Apply a transformation from $\mathbb{R}^3$ to $\mathbb{R}^4$ that converts the Coulomb problem into a constrained four-dimensional harmonic oscillator.
- Gaussian evaluation: The resulting path integral is Gaussian and can be evaluated exactly.
The hydrogen spectrum and wavefunctions are recovered as residues of the energy-dependent Green’s function. This is not merely a technical trick; the Duru–Kleinert transformation reveals a deep structural relationship between the Coulomb and oscillator problems that has implications for the broader theory.
Kleinert (1987) develops refined time-slicing procedures that address subtleties in operator ordering and measure factors arising in the discrete approximation. These refinements are necessary for numerical implementations and illustrate the theme that slicing prescriptions carry physical content.
4.3 Hydrogen in Curved Space
Barut et al. (1990) extend the hydrogen atom path integral to curved-space backgrounds, examining how spacetime geometry affects the atomic spectrum. This provides a testing ground where the methods of Section 6 (curved-space path integrals) can be checked against the exact solvability of Section 4. The combination of singular potential and nontrivial geometry is the most demanding setting for any path integral prescription.
4.4 Jacobi’s Principle and One-Dimensional Quantum Gravity
Fujikawa (1996) (arXiv:hep-th/9602080) establishes a connection between the hydrogen atom path integral and Jacobi’s principle of least action. Jacobi’s principle describes dynamics at fixed energy via a reparametrization-invariant action, so that the choice of time parameter becomes a gauge freedom. Fujikawa shows that this gauge freedom in the hydrogen path integral corresponds to time reparametrization invariance, and that the problem can be viewed as a model of one-dimensional quantum gravity coupled to “matter” (the electron coordinates).
This is a striking reinterpretation. The hydrogen atom—the most elementary bound state in quantum mechanics—becomes a laboratory for ideas about:
- Gauge freedom associated with time reparametrization.
- BRST analysis and the Weyl anomaly.
- The “problem of time” in gravity, treated in its simplest incarnation.
Fujikawa’s paper thus serves as a bridge from “the hydrogen atom as a solvable test case” to the gauge theory and curved manifold themes of the final sections.
5. Gauge Theory and Constrained Quantization
Path integral quantization of gauge theories requires careful treatment of constraints and redundant degrees of freedom. A gauge symmetry means that the classical phase space contains unphysical directions, and the path integral over all configurations naively overcounts physically equivalent states. The methods for handling this overcounting—Faddeev–Popov, Dirac brackets, BRST—constitute a substantial technical apparatus. The references collected here focus less on gauge theory broadly (Yang–Mills quantization, standard model applications) and more on the foundational question: how does constrained quantization determine the path integral measure?
5.1 The Dirac–Faddeev–Popov Framework
Henneaux and Teitelboim (1992) provide the definitive reference on constrained Hamiltonian systems and their quantization. The framework begins with Dirac’s classification of constraints into first class (commuting with all constraints) and second class (not commuting), and proceeds through the Dirac bracket for second-class constraints, the BRST-BV formalism for first-class constraints, the Faddeev–Popov procedure for the path integral, and the introduction of ghost fields. This is the standard technology for quantizing gauge theories from electromagnetism to Yang–Mills to gravity.
The essential point for these notes is that the path integral measure for a constrained system is determined by the constraint structure, not freely chosen. The Faddeev–Popov determinant, the ghost action, and the gauge-fixing condition together specify the functional integral in a way that respects the physical content of the gauge symmetry.
5.2 The Faddeev–Jackiw Approach: Symplectic Geometry
Jackiw (1993) (arXiv:hep-th/9306075) presents the Faddeev–Jackiw approach to constrained quantization, which simplifies the Dirac formalism by working directly with the symplectic structure of phase space. Rather than classifying constraints and then modifying the Poisson bracket, one works with the symplectic two-form $\omega_{ab}$ on the reduced phase space from the outset.
Liao and Huang (2007) reformulate path integral quantization using this symplectic approach. The key result is that the symplectic two-form $\omega_{ab}$ directly determines the measure in the path integral, providing a geometric understanding of the quantization procedure. This is conceptually satisfying: the measure inherits its structure from the geometry of phase space.
Toms (2015) (arXiv:1508.07432) demonstrates equivalence between the Faddeev–Jackiw and Dirac methods for constrained path integrals through explicit examples: a particle constrained to a surface and a Schrödinger field coupled to a vector field. The natural geometric (symplectic) structure of the Faddeev–Jackiw method integrates cleanly into the path integral framework.
5.3 When Constraints Do Not Generate Gauge: A Cautionary Tale
Barbour and Foster (2008) (arXiv:0808.1223) challenge a widely held assumption: that first-class constraints always generate gauge transformations. This identification—sometimes called “Dirac’s theorem”—is central to the standard treatment of constrained systems. Barbour and Foster demonstrate that it fails for reparametrization-invariant systems, precisely the class of systems relevant to Jacobi’s principle and general relativity.
For reparametrization-invariant systems, the Hamiltonian constraint generates genuine physical evolution, not gauge motion. The distinction is crucial: if the constraint is treated as gauge-generating, one concludes that “time is unphysical” (the frozen formalism problem in quantum gravity). If it generates physical evolution, the conclusion is different and potentially more tractable.
The inclusion of this paper signals that the bibliography is not merely cataloguing standard quantization technology, but is attentive to the subtle conceptual issues that arise when these methods are applied to gravitational and cosmological systems. The connection to the “problem of time” is made explicit in Section 6.
5.4 Comprehensive Treatments
Prokhorov and Shabanov (2011) provide a comprehensive Hamiltonian treatment of gauge systems covering both abelian and non-abelian theories. This serves as a reference for readers who need the detailed mathematical framework supporting the conceptual points made by the more focused papers.
6. Path Integrals on Curved Manifolds: The Culmination
The final section of the bibliography is where the threads of the preceding sections converge. Path integrals on curved manifolds present the full constellation of difficulties: the measure is ambiguous, operator ordering matters, curvature-dependent terms may appear in the Schrödinger equation, and the relationship between different prescriptions requires careful analysis. The references assembled here address these issues from multiple angles, and the convergence of their conclusions constitutes the strongest evidence for the bibliography’s organizing thesis.
6.1 The Problem of the Measure
In flat space, the path integral measure $\prod_k d^n q_k$ is unambiguous (up to normalization). In curved space with metric $g_{\mu\nu}(q)$, there are several candidates: $\prod_k d^n q_k$, $\prod_k \sqrt{g(q_k)}\, d^n q_k$, or measures involving the Van Vleck–Morette determinant. Different choices lead to different Schrödinger equations, potentially with additional terms proportional to the scalar curvature $R$. The DeWitt measure, for instance, produces a term $\hbar^2 R/6$ in the effective potential; other prescriptions produce $R/8$ or zero.
This is not an academic curiosity. Different curvature couplings yield different quantum theories—different spectra, different transition amplitudes, different physics. The question of which measure is “correct” cannot be settled by convention; it requires a physical or mathematical selection principle.
6.2 Group Manifolds: Symmetry as a Guide
Marinov (1979) investigates quantum dynamics on group manifolds, where the group structure provides powerful constraints on the path integral. For compact groups like $SU(2)$, the Peter–Weyl theorem and representation theory determine the spectral decomposition of the propagator, and the path integral can be evaluated exactly or semiclassically using symmetry-based methods.
Marinov (1995) extends this analysis to homogeneous manifolds (spaces with transitive group actions), where symmetry properties determine measure choices and enable exact evaluation in many cases. Krausz and Marinov (1997) (arXiv:quant-ph/9709050) treat the technically more demanding case of non-compact group manifolds such as $SL(2,\mathbb{R})$, relevant for hyperbolic geometry and certain cosmological models.
6.3 Constrained Surfaces and Geometric Potentials
Homma et al. (1990) derive the Schrödinger equation for a particle constrained to move on a curved surface embedded in flat space. The result includes geometric potential terms arising from the extrinsic curvature of the constraint surface. This is a clean example of geometry entering the quantum theory through the constraint structure, and it provides a testing ground for the more general curved-space results.
6.4 Kleinert’s Quantum Equivalence Principle
Kleinert (1995) (arXiv:quant-ph/9511020) formulates a quantum equivalence principle: path integrals in metric-affine spaces with curvature and torsion can be obtained from flat-space path integrals via nonholonomic coordinate transformations. This is a direct quantum analog of the classical equivalence principle, which relates physics in curved spacetime to physics in locally flat coordinates.
The key results of Kleinert’s program are:
- The measure ambiguity is resolved by requiring consistency with the nonholonomic mapping procedure.
- No spurious curvature-scalar terms (such as the DeWitt $R/6$) appear in the resulting Schrödinger equation.
- The quantum ordering problem is resolved: the correct operator ordering is determined by the transformation.
- For a particle on a sphere, the correct energy spectrum proportional to $\hat{L}^2$ is obtained.
Kleinert (1996) (arXiv:quant-ph/9606001) develops the nonholonomic mapping procedure in detail for both classical and quantum paths. The connection to the theory of crystal defects—where torsion represents dislocations and curvature represents disclinations—provides concrete physical intuition. The Jacobian of the nonholonomic transformation produces a measure factor that cancels the DeWitt $R/6$ term, illustrating how the transformation procedure determines the measure.
6.5 Schwinger Consistency: Toms’ Resolution
Toms (2004) (arXiv:hep-th/0411233) provides what is arguably the sharpest resolution of the curved-space measure ambiguity. The argument is conceptually simple: require that the path integral propagator satisfy the Schwinger quantum action principle. This consistency requirement uniquely determines the measure, which must include a factor of the Van Vleck–Morette determinant $\Delta(q_2, q_1)$ in addition to the natural volume element $\sqrt{g}$.
The result agrees with the stochastic differential equation approach to path integrals (developed independently in the mathematical literature) and with Kleinert’s quantum equivalence principle, providing three independent lines of argument for the same measure prescription. However, as a cautionary note, the precise sense of “agreement” deserves scrutiny. Toms’ result concerns the exact measure; Kleinert’s concerns the absence of spurious curvature terms; and the stochastic approach concerns the Itô-vs-Stratonovich issue in the continuum limit. That all three select the same answer is evidence of consistency, but the three arguments operate at different levels of the theory and should not be conflated.
6.6 Jacobi’s Principle, Maupertuis, and the Disappearance of Time
The final entries in the bibliography connect path integrals on curved manifolds to fundamental questions about the nature of time in quantum mechanics.
Gryb (2008) (arXiv:0804.2900) examines the “problem of time” arising from Jacobi’s principle. When dynamics is defined via Jacobi’s reparametrization-invariant action, Dirac quantization leads to a time-independent Schrödinger equation—the Wheeler–DeWitt equation in the gravitational context. The path integral perspective clarifies how duration emerges as a classical quantity while being absent from the quantum formalism. This is directly relevant to quantum cosmology, where the absence of an external time parameter forces a reformulation of the quantum theory.
Karamatskou and Kleinert (2011) (arXiv:1102.2486) extend the classical Maupertuis principle to quantum mechanics. The classical Maupertuis principle states that motion in a potential $V(x)$ with energy $E$ is equivalent to geodesic motion in a space with conformally related metric $\tilde{g}{\mu\nu}(x) = 2M[V(x) - E]\,\delta{\mu\nu}$. The quantum generalization requires using the Weyl-invariant Laplace–Beltrami operator, and enables semiclassical expansions via DeWitt’s methods.
This brings the bibliography full circle: the hydrogen atom (Section 4), viewed through Fujikawa’s Jacobi-principle lens, is a special case of motion in a conformally curved space via Maupertuis. The curved-manifold technology of Section 6 applies. And the gauge-theoretic considerations of Section 5 (reparametrization invariance, the status of the Hamiltonian constraint) determine the physical interpretation.
7. Cross-Cutting Themes: A Synthesis
Several themes recur across the six sections and deserve explicit synthesis.
7.1 Schwinger–Feynman Reciprocity as a Selection Principle
The bibliography is organized to support the idea that Schwinger’s quantum action principle is not merely “equivalent” to path integrals in simple cases, but can serve as a constraint on what the correct functional integral must be in curved or otherwise ambiguous settings. Toms (2004) provides the explicit demonstration. The reciprocity runs in both directions: the path integral gives a concrete representation of the abstract transformation function that Schwinger’s principle constrains, while Schwinger’s principle determines which of the many possible path integral definitions is physically correct.
The historical origins of both formulations in Dirac’s 1933 paper make this reciprocity natural in retrospect. Dirac observed a correspondence between the quantum transformation function and the classical action; Feynman made the correspondence explicit through a sum over paths; Schwinger made it variational through the action operator. That their modern synthesis resolves the curved-space measure problem is, in a sense, the completion of the program Dirac initiated.
7.2 Measure and Ordering as Physical Content
A reader encountering path integrals for the first time might regard the measure $\mathcal{D}q$ as a notational convenience—a symbol standing for “the limit of finite-dimensional integrals that we all agree on.” The bibliography systematically dismantles this assumption. In curved space, different measures yield different quantum theories. In gauge theories, the Faddeev–Popov determinant is a nontrivial function of the fields that must be included in the measure. In the hydrogen atom, the choice of time-slicing prescription affects whether the integral converges at all.
The recurring message is that the measure and operator ordering are not bookkeeping; they are part of the definition of the quantum theory. Changing the measure changes the physics. The role of action principles and geometry is to constrain these choices.
7.3 Singular Potentials as Diagnostics
The hydrogen atom serves as a “hard solvable” benchmark: a system where naive constructions fail but where exact results are available by other methods. The Duru–Kleinert transformation resolves the difficulty by revealing hidden structure (the Coulomb–oscillator correspondence), while Fujikawa’s Jacobi-principle analysis reveals that the singularity is intimately connected to time reparametrization invariance. Any proposed path integral prescription must pass this test.
More broadly, singular potentials function as diagnostic tools for path integral definitions in the same way that exactly solvable models function as diagnostic tools for approximation schemes. The ability to reproduce known results in limiting cases is a necessary (if not sufficient) condition for the correctness of a general prescription.
7.4 Constraints, Gauge, and the Problem of Time
The trajectory from Faddeev–Jackiw symplectic quantization (Section 5) through Barbour and Foster’s critique (Section 5) to Gryb’s analysis of Jacobi’s principle (Section 6) traces an argument about the nature of time in reparametrization-invariant systems. The standard identification of first-class constraints with gauge generators leads, in the gravitational case, to the frozen formalism: the Wheeler–DeWitt equation $\hat{H}\Psi = 0$ has no explicit time dependence, and “time” must be recovered as an emergent or internal quantity.
Barbour and Foster’s observation that this identification fails for reparametrization-invariant systems opens an alternative: the Hamiltonian constraint may generate physical evolution, not gauge motion. This does not dissolve the problem of time, but it changes its character. For path integrals, the implication is that boundary conditions and gauge-fixing procedures carry physical content that goes beyond mere technical convenience.
7.5 The Van Vleck–Morette Determinant as a Unifying Structure
The Van Vleck–Morette determinant $\Delta(q_2, q_1)$ appears in nearly every section of the bibliography:
- As the semiclassical prefactor in the WKB approximation (Section 2).
- As part of the exact measure in curved-space path integrals (Section 6).
- As a structure constrained by the Schwinger action principle (Section 3 via Toms 2004).
- Implicitly, in the Duru–Kleinert transformation (Section 4), where the change of variables modifies the determinant structure.
Its ubiquity reflects a deep point: the determinant encodes the relationship between the classical action and the quantum propagator. Where the classical trajectories focus, the determinant diverges (caustics); where they spread, it governs the amplitude. In curved space, it carries information about the geometry that is essential for the correct quantum theory. The Van Vleck–Morette determinant is, in a sense, the mathematical embodiment of the “action-principle-constrains-the-path-integral” thesis.
8. Scope, Gaps, and Connections
8.1 What These Notes Cover—and What They Do Not
The bibliography surveyed here is coherent around a specific thesis: the definition of path integrals in nontrivial settings requires action principles and geometry as selection rules. This focus determines what is included and what is omitted. Several major areas of path integral theory are absent:
Lattice path integrals and Monte Carlo methods. The discretized Euclidean path integral on a spacetime lattice is the foundation of lattice gauge theory and lattice QCD. These methods are entirely computational in character—the path integral is evaluated by importance sampling—and constitute a major application area for scientific computation. Their absence from the present bibliography reflects the focus on analytical and semi-analytical methods, not a judgment about their importance.
Fermionic and Grassmann path integrals. The extension of the path integral to fermionic degrees of freedom, using anticommuting Grassmann variables, is essential for any treatment of the Standard Model or condensed matter systems with fermionic quasiparticles. The formalism is algebraically distinct from the bosonic case and raises its own measure-theoretic questions.
Topological field theories. Chern–Simons theory, topological insulators, and anyonic systems involve path integrals whose value depends only on topological (not metric) properties of the underlying manifold. These connect to knot theory, the Jones polynomial, and topological quantum computation.
Stochastic quantization. The Parisi–Wu approach to quantization via stochastic differential equations in a fictitious time provides yet another perspective on the path integral measure, one that has been particularly fruitful for numerical methods and for establishing rigorous results.
These omissions are deliberate if the scope is “action principle + measure + geometry + constraints.” They would be significant gaps if the scope were “path integrals broadly.” The scope should be understood as the former.
8.2 Connections to Other Chapters
The material surveyed here connects to several other threads in Projects in Scientific Computation:
Quantum field theory. The path integral is the starting point for perturbative and non-perturbative QFT. The Schwinger effective action, the background field method, and the BRST formalism all have their origins in the structures discussed here.
Numerical relativity and quantum gravity. The problem of time, the Wheeler–DeWitt equation, and the status of the Hamiltonian constraint are central issues in canonical quantum gravity. The path integral approach to quantum gravity (whether via Euclidean methods, causal dynamical triangulations, or spin foams) inherits all the measure ambiguities discussed in Section 6, compounded by the infinite-dimensional character of the gravitational field.
Computational finance. Kleinert’s textbook explicitly connects path integrals to quantitative finance via the Fokker–Planck equation and stochastic processes. The influence functional formalism has analogs in the treatment of stochastic volatility models, and the semiclassical approximation maps onto saddle-point methods used in option pricing.
Neural network–quantum field theory correspondence. The path integral provides the natural language for the NNGP (neural network Gaussian process) correspondence, where the width of a neural network plays the role of $1/\hbar$ and the infinite-width limit corresponds to the classical (saddle point) limit of the functional integral.
9. Key ArXiv Resources
The following arXiv preprints are directly accessible and central to the arguments developed in these notes:
- Milton:
physics/0610054,1402.3762,1503.08091 - Toms:
hep-th/0411233,1508.07432 - Barbour & Foster:
0808.1223 - Gryb:
0804.2900 - Kleinert:
quant-ph/9511020,quant-ph/9606001 - Krausz & Marinov:
quant-ph/9709050 - Karamatskou & Kleinert:
1102.2486 - Fujikawa:
hep-th/9602080 - Jackiw:
hep-th/9306075
These preprints provide the primary evidence for the technical claims made in the bibliography and constitute the freely available core of the reading program. They are isolated here for an abreviated and concise reading.
Bibliography
Textbooks
| ID | Link | Notes |
|---|---|---|
| S1 | Feynman and Hibbs “Quantum Mechanics and Path Integrals (Emended)” 1965 | Provides a comprehensive, emended introduction to Feynman’s path integral formulation of quantum mechanics. |
| S2 | Junker and Inomata “Transformation of the free propagator” 1985 | Develops a transformation technique for the free propagator within the path-integral formalism, facilitating analytic solutions. |
| S3 | Roepstorff “Path integral approach to quantum physics” 1994 | Offers a systematic, mathematically rigorous treatment of path integrals applied to quantum physics. |
| S4 | Johnson and Lapidus “Feynman Integral and Operator Calculus” 2002 | Integrates Feynman integral methods with operator calculus to provide a unified approach to quantum evolution operators. |
| S5 | Cartier and Dewitt-Morette “Functional Integration” 2007 | Develops the mathematical foundations of functional integration, emphasizing symmetries and action principles in physics. |
| S6 | Kleinert “Path Integrals in Quantum Mechanics, Statistics, Polymer Physics, and Financial Markets” 2009 | Surveys diverse applications of path integrals across quantum mechanics, statistical physics, polymer theory, and finance. |
| S7 | Baaquie “Path Integrals and Hamiltonians” 2014 | Explores the relationship between path integral formulations and Hamiltonian mechanics, outlining key principles and methods. |
| S8 | Chaichian and Demichev “Path integals in Physics: Vol I” 2018 | Introduces fundamental techniques of path integrals in physics, covering basic methods for classical and quantum systems. |
| S9 | Chaichian and Demichev “Path integrals in Physics: Vol II” 2018 | Continues with advanced topics in path integrals, including gauge fields and applications in statistical mechanics. |
Feynman’s Functional Integral
| ID | Link | Notes |
|---|---|---|
| S10 | Feynman “The Principle of Least Action in Quantum Mechanics” 1942 | Examines how the principle of least action underlies quantum mechanics and leads to Feynman’s path integral approach. |
| S11 | Morette “Definition and Approximation of Feynman’s Path Integral” 1951 | Provides rigorous definitions and practical approximation schemes for evaluating Feynman’s path integral in quantum systems. |
| S12 | Feynman and Vernon “Influence Functionals” 1963 | Introduces the influence functional formalism to model how an environment affects a quantum system within the path integral framework. |
| S13 | Choquard “VanVleck’s and Morette-Van Hove’s Determinant” 1996 | Analyzes the determinants appearing in semiclassical propagator approximations, linking Van Vleck and Morette-Van Hove formulations. |
Schwinger’s Action Principle
| ID | Link | Notes |
|---|---|---|
| S14 | Milton “Julian Schwinger” 2006 | Provides a historical and technical overview of Julian Schwinger’s contributions to the quantum action principle. |
| S15 | Toms “Schwinger Action Principle and Effective Action” 2009 | Details the Schwinger action principle and its use in deriving the one-particle irreducible effective action. |
| S16 | Milton “Schwinger’s Quantum Action Principle I” 2014 | Outlines the foundational aspects of Schwinger’s quantum action principle and its role in quantum dynamics. |
| S17 | Milton “Schwinger’s Quantum Action Principle” 2015 | Presents further developments and modern applications of Schwinger’s quantum action principle in field theory. |
Hydrogen Atom
| ID | Link | Notes |
|---|---|---|
| S18 | Duru and Kleinert “Path integral for the H-atom” 1979 | Derives the exact path integral representation for the hydrogen atom propagator, enabling analytic solution of its dynamics. |
| S19 | Duru and Kleinert “Quantum Mechanics of H-Atom from Path Integrals” | Extends path integral techniques to recover the full spectrum and wavefunctions of the hydrogen atom. |
| S20 | Kleinert “Time sliced path integral of the H Atom” 1987 | Develops time-sliced path integral methods for the hydrogen atom to improve convergence and computational tractability. |
| S21 | Barut et al “Path integarl treatment of hydrogen atom in curved space” 1990 | Examines the hydrogen atom in a curved-space background using path integrals to capture geometric influences on energy levels. |
| S22 | Fujikawa “Path integral of H atom, Jacobi’s principle, and one dimensional gravity” 1996 | Links the hydrogen atom path integral, Jacobi’s principle, and one-dimensional gravity in a unified treatment. |
Gauge Theory
| ID | Link | Notes |
|---|---|---|
| S23 | Henneaux and Teitelboim “Quantization of Gauge Systems” 1992 | Develops canonical and path-integral methods for quantizing constrained gauge systems with detailed mathematical rigor. |
| S24 | Jackiw “Constrained Quantization” 1993 | Reviews methods for quantizing systems with constraints, focusing on Hamiltonian and path integral techniques. |
| S25 | Liao and Huand “Path integral quantization corresponding to Faddeev-Jackiw” | Reformulates path integral quantization using the Faddeev-Jackiw symplectic approach for constrained dynamical systems. |
| S26 | Barbour and Foster “Constraints and gauge transformations” 2008 | Explores how constraints generate gauge transformations in Hamiltonian systems, emphasizing conceptual foundations. |
| S27 | Prokhorov and Shabanov “Hamiltonian Mechanics of Gauge Systems” 2011 | Presents a comprehensive Hamiltonian formulation of gauge systems and their quantization procedures. |
| S28 | Toms “Faddeev-Jackiw quantization and the path integarl” 2015 | Applies Faddeev-Jackiw quantization methods to the path integral framework, highlighting their compatibility and utility. |
Curved Manifolds
| ID | Link | Notes |
|---|---|---|
| S29 | Marinov “Dynamics on Group Manifold and Path Integral” 1979 | Investigates quantum dynamics on group manifolds via path integrals, highlighting symmetry-based integration methods. |
| S30 | Homma et al “Schrödinger equation for the nonrelativistic particle constrained in a curved space” | Derives the Schrödinger equation for a nonrelativistic particle constrained to curved spaces using path-integral techniques. |
| S31 | Kleinert “Path integrals in spaces with curvature and torsion” 1995 | Provides a comprehensive treatment of path integrals in spaces endowed with curvature and torsion, including technical details. |
| S32 | Marinov “Path integals on homogenous manifolds” 1995 | Studies path integrals on homogeneous manifolds, focusing on symmetry properties and measure definitions. |
| S33 | Kleinert “Classical and fluctuating paths in spaces with curvature and torsion” | Analyzes both classical trajectories and quantum fluctuations of paths in curved and torsioned spaces. |
| S34 | Krausz and Marinov “Evolution Operator on Non-compact Group Manifolds” | Develops the evolution operator for quantum systems on non-compact group manifolds through path integral methods. |
| S35 | Toms “Schwinger Action Principle and Path Integral in Curved Space” | Applies the Schwinger action principle to formulate path integrals in curved space-time settings, merging two approaches. |
| S36 | Gryb “Jacobi’s Principle” | Discusses Jacobi’s principle and its implications for both classical mechanics and path-integral quantum formulations. |
| S37 | Karamatskou and Kleinert “Quantum Maupertuis Principle” 2011 | Explores the Quantum Maupertuis principle as a path integral formulation for constrained dynamical systems. |