Nonperturbative Dynamics

Nonperturbative physics begins whenever the answer to a stated question is not determined by a fixed perturbative expansion about one configuration. The missing information may be a different saddle, a global sector, a reorganized large-$N$ limit, a self-consistent functional equation, an exact factorization structure, or a regulated calculation. None of these routes is reliable merely because it is called nonperturbative: the relevant observable, approximation, control parameter, order of limits, and independent check must all be named.

This volume develops those routes as complementary tools. It starts with vacua and sectors, builds the semiclassical calculus, separates stable solitons from tunneling events and false-vacuum bounces, and then studies controlled strong-coupling models, confinement diagnostics, large-$N$ dynamics, functional equations, resurgence, and integrable relativistic QFT. The final chapter explains how to compare results without treating shared assumptions as independent evidence. Mariño’s treatment of instantons, large-order behavior, sigma models, and large $N$ illustrates why these subjects form a connected body of methods rather than a list of strong-coupling phenomena Mariño 2015, chs. 1–8, pp. 3–273.

Helpful background. Saddles and the semiclassical expansion supplies the Gaussian starting point for Chapters 2–5; asymptotic scales, remainders, and optimal truncation prepares the large-order and transseries route; and effective field theory as a controlled expansion supplies the language of regime, omitted terms, and breakdown. These are route-specific aids, not a single entrance requirement for the volume.

What a nonperturbative statement must specify

Suppose an observable has a formal expansion about one saddle,

$$F(g)\sim\sum_{n=0}^{\infty}a_n g^n.$$

A contribution such as $e^{-A/g^2}$ has a vanishing Taylor series at $g=0$ and is therefore invisible at every finite perturbative order. That is one important meaning of nonperturbative, but not the only one. A strong-coupling numerical result, a $1/N$ expansion, an exact two-dimensional S matrix, and a theorem about Borel summability answer different questions and carry different kinds of justification. The word becomes informative only after the comparison class is stated.

Use the following record for any substantial claim:

Question	What must be stated	Typical failure revealed
Relative to what?	Coupling expansion, saddle, $1/N$, derivative expansion, truncation, regulator, finite volume, or another declared formulation	Calling every difficult or numerical calculation nonperturbative
Which theory data?	Dimension, fields, action, global form, boundary conditions, state, and allowed sectors	Importing a mechanism from a different theory
Which observable?	Correlator, energy, gap, tension, susceptibility, rate, finite-volume level, or operator matrix element	Comparing quantities that share a name but not a definition
What controls the answer?	Small parameter, large parameter, theorem hypotheses, exact algebraic structure, systematic convergence test, or explicitly limited evidence	Treating a formal reorganization as an error estimate
Which limits?	Continuum, volume, infrared, $N\to\infty$, weak coupling, temperature, and their order	Replacing a crossover by a sharp limiting transition
What checks it?	Symmetry identity, normalization, solvable limit, independent method, benchmark, or rigorous bound	Counting two calculations with the same ansatz as independent confirmation
Where does it stop?	First failed hypothesis and the subject that treats the next problem	Turning a controlled example into a universal mechanism

Three recurring mathematical structures organize the volume.

First, global sectors may require a sum

$$Z(\theta)=\sum_Q e^{i\theta Q}Z_Q,$$

where the allowed charge lattice, global form, boundary data, and normalization determine the actual periodicity. Second, a semiclassical contribution is more than its exponential:

$$\mathcal O_\sigma \sim e^{-S_\sigma/\lambda} J_\sigma \left(\frac{\det{}'\mathcal M_\sigma}{\det\mathcal M_0}\right)^{-1/2} \bigl[\text{insertions}\bigr] \bigl(1+O(\lambda)\bigr).$$

The cycle, zero modes, negative modes, collective-coordinate measure, renormalization, and observable all affect the meaning of this expression. Third, reorganized limits replace one expansion by another—large $N$, a closed functional system, a Borel–Laplace representation, or factorized scattering—and must be tested for nonuniformity, missing solutions, or undetermined data. Coleman’s analyses of classical lumps, quantum tunneling, and the fate of the false vacuum remain especially clear demonstrations that boundary conditions and the physical question determine which saddle calculation is appropriate Coleman 1985, ch. 6, pp. 185–264; ch. 7, §§ 2 and 6, pp. 268–339.

Choose a route by the physical question

If you want to…	Start with…	Continue until…
Decide what “nonperturbative” means in a particular problem	Nonperturbative Questions, Vacua, and Sector Dynamics	You can state the expansion, global data, observable, control, and order of limits without conflating theta parameters, states, and branches
Construct a controlled saddle contribution	Semiclassical Expansions and Integration Cycles	The action, determinant, removed modes, collective coordinates, contour, renormalization, and first omitted effect are explicit
Determine whether a localized configuration is stable	Solitons, Defects, and Collective Dynamics	Boundary data, charge, virial test, fluctuation spectrum, and possible lifetime support the stated notion of stability
Compute tunneling or an instanton-induced selection rule	Instantons, Fermion Zero Modes, and Tunneling	The Euclidean boundary problem, moduli measure, zero-mode saturation, ensemble approximation, and infrared limitations are checked
Compute decay of a metastable vacuum	Metastability and Vacuum Decay	Bounce existence, the single negative mode, zero modes, prefactor, gauge and scale consistency, and the flat-space validity boundary are established
Learn from a controlled strongly coupled theory	Strong-Coupling Laboratories and Dual Variables	The model’s dimension, symmetry, controlled limit, observable, exact or approximate method, and non-transferable features are clear
Distinguish confinement, screening, and a mass gap	Confinement, Screening, and Mass Gaps in Controlled Regimes	The probe, matter content, global form, phase diagnostic, mechanism status, and spectral statement are separated
Reorganize a theory at large $N$	Large-N Dynamics	Field and operator normalizations, coupling held fixed, dominant diagrams or saddle, and order of limits have been checked
Turn an exact functional identity into a calculation	Continuum Functional Equations and Controlled Truncations	Closure, renormalization conditions, symmetry residuals, branch choice, numerical stability, and independent benchmarks are recorded
Relate factorial divergence to nonperturbative sectors	Resurgence and Transseries	Borel convention, lateral prescription, Stokes data, transseries parameters, ambiguity cancellation, and the demonstrated domain are explicit
Use exact scattering in a relativistic two-dimensional theory	Integrable Relativistic QFT	Factorization hypotheses, Yang–Baxter consistency, CDD freedom, pole alternatives, finite-volume corrections, and locality limits are visible
Compare claims from unlike methods	Evidence, Validation, and Nonperturbative Status	Observables and conventions have been translated, shared assumptions identified, and the remaining proof or computation obligation named

Readiness checks for different entry points

There is no volume-wide pass score. Try only the task relevant to the route you intend to follow.

Capability	Inspectable task	Sufficient answer	Unlocks	Repair route
Saddle expansion	Given $S_E[x]/\lambda$, expand about a stationary path and identify the operator governing Gaussian fluctuations.	You obtain the Hessian with the stated boundary conditions and explain why zero or negative eigenvalues cannot be included in an ordinary determinant.	Semiclassics, instantons, and vacuum decay	Saddles and the semiclassical expansion and second variation, Hessians, and Jacobi operators
Asymptotic reasoning	Distinguish convergence from an asymptotic expansion and state what optimal truncation estimates.	You can bound the remainder in a declared regime and do not infer a unique function from coefficients alone.	Semiclassics, large $N$, and resurgence	Asymptotic scales, remainders, and optimal truncation
Topological and global data	Explain what fixes the charge lattice and theta periodicity in a gauge theory.	You name the spacetime and boundary conditions, gauge-group global form, allowed bundles or defects, charge normalization, and orientation.	Vacua, instantons, defects, and confinement	Theta terms, periodicity, and vacuum sectors
Controlled approximation	Given a retained expansion, name its dimensionless parameter and the first omitted contribution.	You can state a breakdown condition and distinguish a scale hierarchy from numerical smallness at one point.	Every calculational route	Effective field theory as a controlled expansion
Spectral interpretation	Starting from a two-point function, distinguish an isolated physical pole, a continuum threshold, and a gauge-dependent mass-like scale.	You identify the operator channel, state, positivity assumptions, volume, and analytic continuation required for a spectral claim.	Confinement, functional equations, and integrability	Spectral decomposition of two-point functions
Functional identities	Derive a one-variable integration-by-parts identity and explain why it generates a hierarchy rather than a closed solution.	You separate the exact identity from the ansatz or truncation used to solve it.	Functional equations and cross-method comparison	Schwinger–Dyson identities
Analytic scattering	State the physical strip and crossing relation only after fixing particle masses, rapidity convention, and in/out ordering.	You distinguish an analytic S-matrix assumption from a proof that a local QFT realizes the data.	Integrable relativistic QFT	Analyticity and crossing of amplitudes

The twelve chapters and their roles

The sidebar order makes the broad conceptual progression visible, but readers may enter at any chapter whose preparation they meet.

Chapter	Main task and representative system	What you can carry out	Decisive limitation
1. Nonperturbative Questions, Vacua, and Sector Dynamics	Relate local expansions, global sectors, theta weighting, tunneling, branch structure, susceptibility, and anomaly constraints.	Construct a sector sum and state what a nonanalytic vacuum claim depends on.	An anomaly restricts possible infrared behavior; it does not choose a unique phase.
2. Semiclassical Expansions and Integration Cycles	Build the complete contribution of real or complex saddles, using the double well and finite-dimensional integrals as recurring checks.	Determine loop counting, determinant ratios, collective-coordinate measures, integration cycles, and multi-saddle corrections.	A stationary point contributes only when the boundary problem and integration cycle include it.
3. Solitons, Defects, and Collective Dynamics	Analyze kinks, walls, vortices, monopoles, textures, skyrmions, Q-balls, oscillons, and sphalerons.	Derive charges and energy bounds, apply Derrick scaling, and obtain moduli-space dynamics.	Topological classification, existence, energetic stability, and long lifetime are distinct statements.
4. Instantons, Fermion Zero Modes, and Tunneling	Develop quantum-mechanical and gauge instantons through charge, moduli, determinants, fermion selection rules, size integrals, and fractional constituents.	Assemble an instanton amplitude or correlator insertion and test the dilute approximation.	Finite action alone neither fixes an integer charge nor controls the infrared size integral.
5. Metastability and Vacuum Decay	Formulate the false-vacuum boundary problem, find bounces, extract rates, and test thin-wall, gauge, and scale dependence.	Derive the exponential and prefactor structure of a zero-temperature flat-space decay rate.	Thermal activation and gravitational backreaction require different treatments.
6. Strong-Coupling Laboratories and Dual Variables	Use O($N$), CP($N-1$), principal chiral, Gross–Neveu, and Schwinger models to study generated scales, gaps, topology, screening, and dual variables.	Derive a controlled gap or screening statement and compare alternative descriptions.	Solvability in low dimension is evidence about the stated model, not a mechanism theorem for four-dimensional QCD.
7. Confinement, Screening, and Mass Gaps in Controlled Regimes	Compare line operators, static potentials, flux tubes, effective strings, compact Abelian confinement, small-circle mechanisms, and spectra.	Choose a diagnostic that matches the probes and distinguish a mechanism from evidence for an observable.	Gauge-fixed positivity violation and a propagator scale are not proofs of confinement or a gauge-invariant mass gap.
8. Large-N Dynamics	Develop vector, matrix, gauge, and tensor limits through auxiliary fields, eigenvalues, double lines, factorization, equivalences, and double scaling.	Derive the leading saddle or diagram topology and identify the first subleading or nonuniform effect.	Factorization does not by itself establish a unique classical master field or a string dual.
9. Continuum Functional Equations and Controlled Truncations	Connect Schwinger–Dyson, nPI, Bethe–Salpeter, Faddeev, functional-RG, spectral, and gauge-fixed systems.	Specify a finite closure with renormalization, identities, branch choice, and validation tests.	An exact hierarchy does not make a chosen truncation exact, and shared ansätze correlate comparisons.
10. Resurgence and Transseries	Relate large-order coefficients, Borel singularities, Stokes jumps, transseries sectors, instantons, renormalons, and constructive results.	Test an observable-specific ambiguity cancellation and state its boundary data.	Resurgent structure in one observable or model is not a universal definition of a QFT.
11. Integrable Relativistic QFT	Follow conserved charges through factorized scattering, Yang–Baxter consistency, exact S matrices, poles, Bethe quantization, form factors, and TBA.	Bootstrap and cross-check exact data in a declared $1+1$-dimensional model.	Yang–Baxter consistency is necessary data, not a locality or completeness proof; a pole need not be a bound state.
12. Evidence, Validation, and Nonperturbative Status	Translate observables and assumptions across analytic, exact, regulated, numerical, and rigorous approaches.	Classify a claim, identify correlations between checks, and state what would strengthen it.	Agreement is only as independent as the inputs, approximations, data, and limits that produced it.

Five calculations to follow across the volume

These threads preserve one question while changing the method or level of description. They are routes, not claims that every step is compulsory.

Thread	Progression	Invariant checks and stopping rule
A saddle contribution to an observable	Declare the expansion → choose the saddle and cycle → evaluate the determinant and collective coordinates → sum controlled sectors → renormalize the observable → state the error.	Keep dimensions, boundary conditions, removed modes, normalization, and the exact observable fixed. Stop when no small parameter, contour justification, or density hierarchy remains.
Topological data to infrared constraints	Fix boundary data and global form → identify charge sectors → form the theta sum → include instanton and fermion zero modes → match anomalies → enumerate compatible infrared phases.	Preserve charge and theta normalization throughout. Stop before an anomaly constraint is promoted to a unique dynamical realization.
A strong-dynamics claim	Specify a model and regime → derive a generated scale, gap, or screening observable → compare analytic, exact, or regulated determinations → state the strength of the conclusion.	Translate the same observable and limits before comparison. Stop at the first extrapolation beyond the model’s dimension, matter content, or controlled limit.
A functional-equation result	Derive the exact hierarchy → choose a closure and renormalization conditions → select and solve a branch → translate the output → compare an independent benchmark → quantify sensitivity.	Track symmetry residuals, truncation variation, solver error, and shared ansätze. Stop when these do not support the reported precision.
Exact structure and its boundary	Begin with large-order data or higher conserved charges → impose consistency equations → reconstruct an observable → include ambiguity or finite-volume corrections → compare ultraviolet, infrared, or rigorous information.	Keep the exact object and hypotheses explicit. Stop when uniqueness, locality, constructive existence, or QFT-wide extension becomes an additional conjecture.

Conventions that travel with the calculation

The volume inherits the site’s (+---) Lorentzian metric, natural units, Fourier pair, $+i0$ boundary-value prescription, Lorentzian weight $e^{iS}$, Euclidean weight $e^{-S_E}$, Hermitian gauge generators, and orientation conventions from Conventions. Euclidean pages say so explicitly and never treat Wick rotation as the substitution $t=-i\tau$ without a contour and boundary prescription.

Several additional choices recur:

• a theta calculation states the normalization of $F\widetilde F$, the orientation, allowed charge lattice, global form, and Euclidean phase $e^{i\theta Q}$ before claiming periodicity;
• a saddle formula separates gauge redundancy, normalizable zero modes, collective coordinates, negative directions, and the primed determinant;
• a soliton calculation states spatial dimension, behavior at infinity, target or gauge quotient, charge normalization, and the precise notion of stability;
• a large-$N$ calculation states the index structure, trace and field normalization, coupling held fixed, observable normalization, and order of the $N$, volume, continuum, and infrared limits;
• a functional calculation states formulation, regulator, tensor or operator basis, renormalization conditions, projection, and selected solution branch;
• a Borel calculation states the factorial convention, Borel variable, Laplace ray, lateral direction, branch cuts, and transseries normalization; and
• an integrable-scattering calculation states particle masses, rapidity and in/out ordering, S-matrix normalization, physical strip, crossing map, residue sign, statistics, and CDD convention.

When a source uses another convention, translate the complete package. An integer charge, a decay rate, a mass ratio, a Ward identity, a pole position, or a finite-volume energy provides a convention-independent round-trip check; “up to conventions” does not.

Interfaces with neighboring subjects

Subject	What enters this volume	What remains there	Translation check
Mathematical Methods	Asymptotic expansions, steepest descent, topology, index theory, determinants, spectral methods, and variational analysis	General mathematical definitions and theorem proofs	Preserve domains, orientations, branches, boundary conditions, and determinant normalization; verify a solvable limit.
Foundations of QFT	Vacua, states, spectra, regulated path integrals, Euclidean continuation, effective actions, and exact Schwinger–Dyson identities	Foundational definitions and reconstruction questions	Match state, source, pole, continuation, and functional-derivative conventions before applying them.
Symmetry and Gauge Structure	Global form, theta terms, defects, generalized symmetries, anomalies, BRST, and Gribov data	Primary symmetry and gauge definitions	Check the covariant derivative, generator trace, orientation, charge lattice, background fields, and Ward or Slavnov–Taylor identity.
Perturbative QFT and Scattering	Unitarity, analyticity, poles, crossing, and asymptotic-state conventions	General scattering theory and perturbative amplitudes	Match state and S-matrix normalization, physical sheet, crossing map, and residue before using factorized scattering.
Renormalization and EFT	Running couplings, dimensional transmutation, OPE/renormalon structure, functional RG, truncation tests, and EFT control	General renormalization, matching, and RG construction	Match regulator, scheme, scale, operator basis, and first omitted order; verify scale consistency for the declared observable.
Gauge Theories and the Standard Model	Model data for Yang–Mills, QCD, theta dependence, static sources, and hadronic applications	Model-specific phenomenology and real-QCD synthesis	Preserve gauge group and global form, matter representations, coupling convention, source representation, and observable.
Lattice and Hamiltonian QFT	Regulated observables and numerical benchmarks used in comparisons	Discretizations, algorithms, continuum extrapolation, finite-volume extraction, and numerical uncertainty	Match the renormalized observable and limits; do not equate a finite-volume level with an infinite-volume pole.
Supersymmetry and Duality	Protected comparisons, BPS sectors, holomorphy, and dual descriptions when they test a generic method	Supersymmetric localization, protected instanton counting, and full duality dictionaries	Match charges, global data, operator map, normalization, and protected sector before comparison.
Thermal and Nonequilibrium QFT and QFT in Curved Spacetime	The zero-temperature flat-space result used as input	Thermal activation, real-time rates, gravitational backreaction, Coleman–De Luccia and Hawking–Moss decay	Carry the state, contour, temperature or geometry, boundary conditions, and definition of the rate together.
Holography and Quantum Gravity	Large-$N$ counting and its limitations	Bulk dictionaries, string dynamics, SYK, and gravitational interpretation	Match operator normalization, connected-correlator scaling, limit order, and any gap or sparsity assumption.
Mathematical QFT	Precise hypotheses and status of constructive, spectral, Borel, metastability, factorizing-S-matrix, and Yang–Mills problems	Theorem-first construction and proof	State the theorem’s primitive objects, topology, dimensions, and conclusion exactly; recover the physics claim only inside those hypotheses.

Computation, current evidence, and reference material

The explanatory argument on each page is self-contained for its declared scope.

Dated literature assessments, disputed mechanisms, and changing statements of reach belong in Research. Stable pages here explain enough of the underlying physics to make those assessments intelligible without turning a current judgment into a timeless claim. Reference supplies lookup material, while Conventions remains the common normalization baseline.

What you can carry forward

After completing the route appropriate to your question, you should be able to:

• classify a nonperturbative statement by expansion or formulation, theory data, observable, control, evidence, and order of limits;
• construct a saddle, soliton, instanton, bounce, large-$N$, functional, Borel/transseries, or integrable calculation without hiding its decisive conventions;
• distinguish an exact identity from an exact solution, a controlled approximation from numerical agreement, and evidence from proof;
• identify the first failure of a dilute ensemble, thin-wall limit, large-$N$ equivalence, functional closure, ambiguity cancellation, or factorized-scattering argument;
• translate a physical observable across analytic and regulated methods while identifying shared assumptions; and
• route executable work, mutable status, model-specific applications, and rigorous proof obligations to the subject that develops them.

Check your preparation and synthesis

Choose one claim—an instanton-induced splitting, a string tension, a mass gap, a large-$N$ factorization formula, a functional-equation mass scale, a Borel sum, or an exact finite-volume energy. Write its theory data, observable, expansion or formulation, control parameter or theorem hypotheses, order of limits, two checks, and first known failure. Then ask whether the two checks use independent inputs. If any field is missing, the claim is not yet precise enough to compare across methods; return to the corresponding chapter in the route table above.

References

• Coleman, Sidney. Aspects of Symmetry: Selected Erice Lectures. Cambridge: Cambridge University Press, 1985. DOI.
• Gaiotto, Davide, Anton Kapustin, Nathan Seiberg, and Brian Willett. “Generalized Global Symmetries.” Journal of High Energy Physics 2015, no. 2 (2015): 172. DOI. Open PDF.
• Mariño, Marcos. Instantons and Large N: An Introduction to Non-Perturbative Methods in Quantum Field Theory. Cambridge: Cambridge University Press, 2015. DOI.
• Polyakov, Alexander M. Gauge Fields and Strings. Contemporary Concepts in Physics 3. Chur: Harwood Academic Publishers, 1987. DOI and Open Access.
• Zinn-Justin, Jean. Quantum Field Theory and Critical Phenomena. 5th ed. Oxford: Oxford University Press, 2021. DOI.