Skip to content

Hamiltonian Operator

The capital Latin letter HH, read “Hamiltonian,” denotes the operator that generates time translations. For a closed system it also represents total energy. Authors write either HH or H^\hat H; the hat is a typography convention, not a mathematical operation.

In the Schrödinger picture, the state obeys

iℏddt∣ψ(t)⟩=H(t)∣ψ(t)⟩.i\hbar\frac{d}{dt} \lvert\psi(t)\rangle = H(t)\lvert\psi(t)\rangle.

For a time-independent Hamiltonian,

U(t,t0)=exp⁡[−iℏH(t−t0)].U(t,t_0) = \exp\left[ -\frac{i}{\hbar} H(t-t_0) \right].

For a general time-dependent Hamiltonian,

U(t,t0)=Texp⁡[−iℏ∫t0tH(s) ds],U(t,t_0) = \mathcal T \exp\left[ -\frac{i}{\hbar} \int_{t_0}^{t}H(s)\,ds \right],

where T\mathcal T orders later-time operators to the left. The time-ordering symbol can be omitted only when the relevant Hamiltonians commute at different times, or when the exponential is otherwise understood as shorthand for the evolution equation.

For closed-system quantum mechanics, HH is a self-adjoint linear operator on the Hilbert space:

H:D(H)⊆H⟶H,H=H†.H:\mathcal D(H) \subseteq\mathcal H \longrightarrow\mathcal H, \qquad H=H^\dagger.

In finite dimensions, this means that the Hamiltonian is represented by a Hermitian matrix. In infinite dimensions, the domain D(H)\mathcal D(H) is part of the operator’s definition. A formally Hermitian differential expression need not determine a self-adjoint operator until boundary conditions and the domain have been specified.

Self-adjointness gives real spectral values and, under the standard hypotheses, unitary time evolution. Physical Hamiltonians are also commonly bounded below, but boundedness below is a stability property rather than the definition of a Hamiltonian.

The dimensions are

[H]=[energy]=ML2T−2.[H]=[\text{energy}] =ML^2T^{-2}.

In SI units, Hamiltonian matrix elements and eigenvalues are quoted in joules or electronvolts. Some communities divide by ℏ\hbar and report H/ℏH/\hbar in angular-frequency units. An equation written with ℏ=1\hbar=1 can therefore hide an energy-versus-frequency convention that must be restored before numerical work.

An energy eigenstate satisfies

H∣En⟩=En∣En⟩.H\lvert E_n\rangle =E_n\lvert E_n\rangle.

For discrete, nondegenerate notation one often writes

H=∑nEn∣En⟩⟨En∣.H = \sum_n E_n\lvert E_n\rangle \langle E_n\rvert.

Degenerate eigenspaces require an additional label or spectral projectors. Continuous spectra require a spectral measure rather than an ordinary countable sum. The compact eigenket formula should not be read as a universal spectral decomposition.

In an orthonormal basis {∣n⟩}\{\lvert n\rangle\},

Hmn=⟨m∣H∣n⟩.H_{mn} = \langle m\rvert H\lvert n\rangle.

The same abstract operator may look very different in another basis. A basis change by a unitary matrix VV gives

H′=VHV†.H' =VHV^\dagger.

For a one-dimensional particle with a local potential,

(Hψ)(x)=[−ℏ22md2dx2+V(x)]ψ(x).(H\psi)(x) = \left[ -\frac{\hbar^2}{2m} \frac{d^2}{dx^2} +V(x) \right]\psi(x).

This line states the coordinate representation, not the complete operator. The allowed wavefunctions and their boundary conditions still determine the domain.

A nonrelativistic particle in a scalar potential is commonly written

H=p22m+V(r,t).H = \frac{\mathbf p^2}{2m} +V(\mathbf r,t).

For a particle of charge qq in electromagnetic potentials,

H=[p−qA(r,t)]22m+qϕ(r,t).H = \frac{ \left[ \mathbf p-q\mathbf A(\mathbf r,t) \right]^2 }{2m} +q\phi(\mathbf r,t).

Here p\mathbf p is canonical momentum. The kinetic momentum is p−qA\mathbf p-q\mathbf A, so replacing one by the other changes the physical meaning.

Every two-level Hamiltonian can be written

H=cI+12h⋅σ,H =cI +\frac12 \mathbf h\cdot\boldsymbol{\sigma},

where cc and the components of h\mathbf h have energy units. The identity term shifts both energies equally, while h\mathbf h determines the level splitting and eigenvectors.

SymbolUsual meaningImportant qualification
H0H_0Free or unperturbed HamiltonianThe split into “free” and “interaction” parts is chosen for the problem
H1H_1 or VVPerturbationVV may also mean a potential-energy function
HintH_{\mathrm{int}}Interaction HamiltonianMay mean coupling between subsystems or the interaction-picture operator
H(t)H(t)Explicitly time-dependent HamiltonianEnergy need not be conserved
HAH_A, HBH_BSubsystem HamiltoniansTensor identity factors are often suppressed
HtotH_{\mathrm{tot}}Total HamiltonianOften includes subsystem and coupling terms
HeffH_{\mathrm{eff}}Effective HamiltonianIts validity depends on scale, subspace, and approximation order
HI(t)H_I(t)Interaction-picture HamiltonianDefined relative to a selected H0H_0
HFH_FFloquet HamiltonianQuasienergies are branch-dependent modulo the drive frequency
HclH_{\mathrm{cl}}Classical Hamiltonian functionA function on phase space, not a Hilbert-space operator

The decomposition

H=H0+VH=H_0+V

is not unique. It is a bookkeeping choice whose usefulness depends on whether H0H_0 is solvable and VV is controlled by an approximation.

For HA⊗HB\mathcal H_A\otimes\mathcal H_B, a noninteracting composite Hamiltonian is

Htot=HA⊗IB+IA⊗HB.H_{\mathrm{tot}} = H_A\otimes I_B +I_A\otimes H_B.

With coupling,

Htot=HA⊗IB+IA⊗HB+HAB.H_{\mathrm{tot}} = H_A\otimes I_B +I_A\otimes H_B +H_{AB}.

Authors often abbreviate the first two terms as HA+HBH_A+H_B. The omitted identity operators must be restored when checking dimensions, commutators, or matrix representations.

Suppose states are transformed according to

∣ψ′(t)⟩=V(t)∣ψ(t)⟩.\lvert\psi'(t)\rangle = V(t)\lvert\psi(t)\rangle.

The Hamiltonian in the transformed frame is

H′(t)=V(t)H(t)V†(t)+iℏV˙(t)V†(t).H'(t) = V(t)H(t)V^\dagger(t) +i\hbar \dot V(t)V^\dagger(t).

The second term is essential for a time-dependent transformation. Omitting it is a common source of incorrect rotating-frame and interaction-picture Hamiltonians.

For H=H0+VH=H_0+V and

U0(t,t0)=exp⁡[−iℏH0(t−t0)],U_0(t,t_0) = \exp\left[ -\frac{i}{\hbar} H_0(t-t_0) \right],

the interaction-picture perturbation is

HI(t)=U0†(t,t0)V(t)U0(t,t0).H_I(t) = U_0^\dagger(t,t_0) V(t) U_0(t,t_0).

Thus HintH_{\mathrm{int}} in the Schrödinger picture and HI(t)H_I(t) in the interaction picture are related but generally not identical expressions.

Replacing the Hamiltonian by

H′(t)=H(t)+c(t)IH'(t)=H(t)+c(t)I

multiplies every state by the same phase:

U′(t,t0)=exp⁡[−iℏ∫t0tc(s) ds]U(t,t0).U'(t,t_0) = \exp\left[ -\frac{i}{\hbar} \int_{t_0}^{t}c(s)\,ds \right] U(t,t_0).

For an isolated nonrelativistic system this global phase leaves expectation values and transition probabilities unchanged. Energy differences remain invariant. The statement does not license arbitrary shifts when energy couples to additional dynamical structures or when comparing Hamiltonians across differently defined effective descriptions.

For an operator A(t)A(t) in the Schrödinger picture,

ddt⟨A⟩=iℏ⟨[H,A]⟩+⟨∂A∂t⟩.\frac{d}{dt} \langle A\rangle = \frac{i}{\hbar} \langle[H,A]\rangle +\left\langle \frac{\partial A}{\partial t} \right\rangle.

Setting A=HA=H gives

ddt⟨H⟩=⟨∂H∂t⟩.\frac{d}{dt} \langle H\rangle = \left\langle \frac{\partial H}{\partial t} \right\rangle.

Therefore the expected energy is conserved when HH has no explicit time dependence, subject to the regularity and domain assumptions needed for the derivatives and products to exist. A state need not be an energy eigenstate for its mean energy to be conserved.

Form or contextMeaning
HH or H^\hat H in dynamicsHamiltonian operator
H\mathcal HHilbert space, not the Hamiltonian
HmnH_{mn}Hamiltonian matrix element in a chosen basis
Hn(x)H_n(x)Hermite polynomial
HH in electromagnetismMagnetic-field intensity
HH in quantum circuitsHadamard gate
Hcl(q,p)H_{\mathrm{cl}}(q,p)Classical Hamiltonian function

Context usually distinguishes the Hamiltonian from the Hadamard gate: the latter is a particular dimensionless 2×22\times2 unitary matrix,

HHad=12(111−1).H_{\mathrm{Had}} = \frac{1}{\sqrt2} \begin{pmatrix} 1&1\\ 1&-1 \end{pmatrix}.

Writing HHadH_{\mathrm{Had}} on a notation-sensitive page avoids overloading HH twice in the same calculation.

Both

H∣ψ⟩andH^∣ψ⟩H\lvert\psi\rangle \qquad\text{and}\qquad \hat H\lvert\psi\rangle

are standard. A text that reserves hats for operators may contrast H^\hat H with the classical function H(q,p)H(q,p). A text that omits hats relies on context and typography. Do not insert or remove hats selectively inside one derivation unless the notation policy is being changed consistently.

Hamiltonians owns the operator’s physical and mathematical role. Time-Evolution Operator develops the propagator, and Time-Dependent Hamiltonians develops time ordering.

Hermitian versus Self-Adjoint explains the domain distinction. Coordinate Representation and Hamiltonians in Coordinate Space own the wave-mechanical realizations. The Hamiltonian glossary entry gives a shorter conceptual lookup.

  • HH has energy units; H/ℏH/\hbar has angular-frequency units.
  • Hermitian matrix and self-adjoint operator are interchangeable only in finite-dimensional settings.
  • A differential expression without a domain and boundary conditions is not a complete Hamiltonian.
  • H0+VH_0+V is a chosen decomposition, not an invariant separation.
  • Subsystem Hamiltonians often carry suppressed tensor-product identities.
  • A time-dependent frame change adds iℏV˙V†i\hbar\dot V V^\dagger.
  • HintH_{\mathrm{int}} and HI(t)H_I(t) can denote operators in different pictures.
  • Explicit time dependence generally prevents conservation of ⟨H⟩\langle H\rangle.
  • Adding a scalar multiple of the identity changes global phase but not isolated-system energy differences.
  • H\mathcal H conventionally denotes Hilbert space, while HH denotes the Hamiltonian.
  • J. J. Sakurai and J. Napolitano, Modern Quantum Mechanics, 3rd ed., Cambridge University Press, 2020, chs. 1 and 2.
  • R. Shankar, Principles of Quantum Mechanics, 2nd ed., Springer, 1994, chs. 4 and 7.
  • A. Messiah, Quantum Mechanics, Dover, 1999, vol. I, chs. IV and VIII.
  • M. Reed and B. Simon, Methods of Modern Mathematical Physics I: Functional Analysis, rev. ed., Academic Press, 1980, secs. VIII.4 and VIII.7.