Skip to content

Hamiltonian Lattice Field Theory

Hamiltonian lattice field theory regulates space while retaining continuous real time, as in the canonical lattice-gauge construction of Kogut and Susskind 1975, pp. 395–408. It makes states, constraints, and unitary evolution explicit, but it also introduces several independent approximations: spatial spacing and volume, local Hilbert dimension, basis truncation, constraint enforcement, time evolution, and finite observation windows. This chapter develops the formulation and shows how its spectra and observables are cross-validated against Euclidean lattice results.

QuestionPageDecisive evidence
What exactly defines the regulated theory?Regulated Hamiltonian field theoryAlgebra, representation, domain, state, constraints, observables, target limits
How are gauge dynamics and Gauss’s law built?Lattice gauge Hamiltonians and Gauss’s law[H,Gx]=0[H,G_x]=0 including matter and boundary charge
When does a Euclidean action define HH?Transfer matrices between Euclidean and Hamiltonian QFTReflection positivity and a positive transfer operator
How may each site or link be made finite?Local Hilbert-space regulatorsExact algebraic relations plus convergence in local dimension
How is the physical gauge sector represented?Physical gauge Hilbert spacesIndependent state count and quantified constraint leakage
How are response or scattering data extracted?Real-time evolution and observable extractionPreparation, time-step, window, boundary, and asymptotic controls
Do Hamiltonian and Euclidean regulators share a target?Continuum limits and Euclidean cross-validationMatched dimensionless observables and a discrepancy decomposition

The order is important. A unitary finite matrix can be simulated exactly and still be the wrong QFT because its physical sector, counterterms, or regulator-removal path is wrong.

A Hamiltonian calculation should state

CH={Aa,Ha,d,D(H),Ha,d(g),Gx,ρ0,{Oi},L,boundaries,limit order}.\mathcal C_H= \{\mathcal A_a,\mathcal H_{a,d},\mathcal D(H),H_{a,d}(\mathbf g), G_x,\rho_0,\{O_i\},L,\text{boundaries},\text{limit order}\}.

Aa\mathcal A_a is the regulated local algebra, Ha,d\mathcal H_{a,d} its representation with local dimension dd when finite, D(H)\mathcal D(H) the operator domain, GxG_x any constraints, and ρ0\rho_0 the prepared state. The bare parameters g\mathbf g are tuned by renormalized conditions rather than held fixed mechanically as aa or dd changes.

For gauge theories, the physical space is not the full link-by-link tensor product:

Hphys={ψ:Gxψ=qxψ for all x}.\mathcal H_{\mathrm{phys}}= \{\lvert\psi\rangle:G_x\lvert\psi\rangle=q_x\lvert\psi\rangle \ \text{for all }x\}.

The prescribed charges qxq_x include dynamical matter and boundary flux. Exact projection, gauge-invariant bases, energy penalties, and postselection have different guarantees and must not share one label.

A positive Euclidean transfer matrix TT can yield H=at1logTH=-a_t^{-1}\log T, so Euclidean exponential decay and Hamiltonian energies can be compared at the same finite regulator. Direct real-time evolution gives unequal-time and retarded observables unavailable from noisy analytic continuation. The two branches meet only after their conventions, states, operators, and renormalized parameters are matched.

The chapter’s shared physical-sector diagram shows where exact statements end. In particular, a penalty Hamiltonian does not turn a leaky state into a physical state, and a Euclidean action without reflection positivity does not furnish a positive-metric Hamiltonian by this route.

A schematic target observable has the dependence

O=O(a,L,d,ΛB,δt,T,ϵG;g),O=O(a,L,d,\Lambda_B,\delta t,T,\epsilon_G;\mathbf g),

where ΛB\Lambda_B is a basis cutoff, δt\delta t an evolution step, TT a measurement window, and ϵG\epsilon_G a constraint-violation diagnostic. A credible calculation varies each relevant axis while the others are controlled. Typical ordering is

δt0,T,d,ΛB,L,a0,\delta t\to0,\quad T\to\infty,\quad d,\Lambda_B\to\infty, \quad L\to\infty,\quad a\to0,

but nonuniform convergence can require joint fits. The order is part of the scientific claim, especially when a local cutoff changes symmetry, when long-time errors grow, or when finite-volume scattering precedes an infinite-volume limit.

For every spectrum or time-dependent observable, report the finite regulated Hamiltonian, physical-sector definition, state preparation, operator matching, time and boundary conditions, all cutoff axes, and at least one independent benchmark. Label a result “regulator level” until the required axes are actually extrapolated.

  • Given a proposed Hamiltonian calculation, identify which of the seven chapter stages is missing and name an observable test that would expose the omission.
  • Write a complete regulated calculation specification and separate its spatial, local-Hilbert, basis, constraint, time-evolution, observation-window, and continuum limits.
  • Kogut, J. B., and Susskind, L. (1975). Hamiltonian formulation of Wilson’s lattice gauge theories. Physical Review D, 11, 395–408. DOI.
  • Banks, T., Susskind, L., and Kogut, J. (1976). Strong-coupling calculations of lattice gauge theories: (1+1)(1+1)-dimensional exercises. Physical Review D, 13, 1043–1053. DOI.
  • Rothe, H. J. (2012). Lattice Gauge Theories: An Introduction, 4th ed. World Scientific. DOI.