Skip to content

Coulomb Hamiltonian

The three-dimensional attractive Coulomb potential is singular, but it does not force a new operator domain or an energy unbounded below. The standard Hamiltonian acts on H2(R3)H^2(\mathbb R^3), and a completed-square identity gives its exact lower bound. The operator construction is proved in Domains of Hamiltonians; here we develop the quantitative model estimates and their dependence on the chosen realization.

Required background. Domains of Hamiltonians provides the Coulomb relative-bound estimate and Kato–Rellich theorem. Quadratic Forms explains how lower bounds extend from a form core.

The Coulomb singularity preserves the full-space domain

Section titled “The Coulomb singularity preserves the full-space domain”

Let

Hκ=−αΔ−κ∣x∣,α>0,κ>0,H_\kappa=-\alpha\Delta-\frac{\kappa}{|x|},\qquad \alpha>0,\quad\kappa>0,

on L2(R3)L^2(\mathbb R^3). The Hardy and interpolation estimates in the required background show that multiplication by 1/∣x∣1/|x| has relative operator bound zero with respect to −Δ-\Delta. Therefore

D(Hκ)=H2(R3),Q(Hκ)=H1(R3).D(H_\kappa)=H^2(\mathbb R^3),\qquad Q(H_\kappa)=H^1(\mathbb R^3).

The operator is self-adjoint, and Cc∞(R3)C_c^\infty(\mathbb R^3) is an operator core. No upper bound on this nonrelativistic coupling κ\kappa is needed for these statements. The form is

qκ[ψ]=α∥∇ψ∥2−κ∫R3∣ψ(x)∣2∣x∣ d3x.q_\kappa[\psi]=\alpha\|\nabla\psi\|^2 -\kappa\int_{\mathbb R^3}\frac{|\psi(x)|^2}{|x|}\,d^3x.

These hypotheses are specific to the three-dimensional Laplacian and Coulomb singularity. A different kinetic operator or a more singular potential requires a new estimate.

A completed square gives the sharp lower bound

Section titled “A completed square gives the sharp lower bound”

Put r^=x/∣x∣\widehat r=x/|x| away from zero. For a smooth compactly supported function and c>0c>0, integration by parts gives

∥∇ψ+cr^ψ∥2=∥∇ψ∥2+c2∥ψ∥2+c∫r^⋅∇∣ψ∣2 d3x=∥∇ψ∥2+c2∥ψ∥2−2c∫∣ψ∣2∣x∣ d3x.\begin{aligned} \|\nabla\psi+c\widehat r\psi\|^2 &=\|\nabla\psi\|^2+c^2\|\psi\|^2 +c\int\widehat r\cdot\nabla|\psi|^2\,d^3x\\ &=\|\nabla\psi\|^2+c^2\|\psi\|^2 -2c\int\frac{|\psi|^2}{|x|}\,d^3x. \end{aligned}

Here ∇⋅r^=2/∣x∣\nabla\cdot\widehat r=2/|x| as a distribution in three dimensions. There is no delta term: the flux through a small sphere is proportional to its area and tends to zero. The Coulomb form is continuous in the H1H^1 norm, so density extends the identity to every ψ∈H1\psi\in H^1.

Choose c=κ/(2α)c=\kappa/(2\alpha). Then

qκ[ψ]=α∥∇ψ+cr^ψ∥2−κ24α∥ψ∥2.q_\kappa[\psi]= \alpha\|\nabla\psi+c\widehat r\psi\|^2 -\frac{\kappa^2}{4\alpha}\|\psi\|^2.

It follows that Hκ≥−κ2/(4α)H_\kappa\geq-\kappa^2/(4\alpha). The bound is sharp because the square vanishes for ψ(x)=Ce−c∣x∣\psi(x)=Ce^{-c|x|}, which belongs to the form and operator domains. The full sequence of hydrogen energies and its radial interpretation belong to Hydrogen Atom as a Schrödinger Operator.

The calculation proves more than a variational upper estimate: it gives a lower bound for every state in the form domain and exhibits a state attaining it.

Removing the origin changes the operator seed

Section titled “Removing the origin changes the operator seed”

The core statement uses smooth functions on all of R3\mathbb R^3, including those nonzero at the origin. Replacing that seed by Cc∞(R3∖{0})C_c^\infty(\mathbb R^3\setminus\{0\}) gives a different symmetric operator.

The punctured seed is not essentially self-adjoint. In the s-wave sector the radial endpoint is limit-circle, and additional self-adjoint realizations are possible. The full-space H2H^2 domain selects the ordinary Coulomb realization; it does not leave a free contact parameter.

The punctured smooth functions are nevertheless a form core for the ordinary Coulomb form. A small radial cutoff removes the origin with an H1H^1 error tending to zero, and the Coulomb term is form continuous. Thus its Friedrichs extension recovers the standard Hamiltonian even though its operator closure does not. Operator-core and form-core statements must not be interchanged.

For fixed points aj∈R3a_j\in\mathbb R^3 and finitely many real strengths gjg_j, consider

H=−αΔ+∑j=1Ngj∣x−aj∣.H=-\alpha\Delta+\sum_{j=1}^N\frac{g_j}{|x-a_j|}.

Translation invariance of the relative-bound estimate and finiteness of the sum give self-adjointness on H2H^2. Define K=∑jmax⁡{−gj,0}K=\sum_j\max\{-g_j,0\}. If K>0K>0, assign kinetic coefficients αj=αmax⁡{−gj,0}/K\alpha_j=\alpha\max\{-g_j,0\}/K to the attractive terms. Applying the single-centre lower bound to each such form gives

H≥−∑gj<0gj24αj=−K24α.H\geq-\sum_{g_j<0}\frac{g_j^2}{4\alpha_j} =-\frac{K^2}{4\alpha}.

Repulsive terms are nonnegative and can be dropped for this estimate. For K=0K=0, H≥0H\geq0. The bound is uniform in the positions of finitely many fixed centres, including coincident centres. It is not a many-particle stability theorem with a constant independent of particle number or total charge.

Exercise: localization does not produce Coulomb collapse

Section titled “Exercise: localization does not produce Coulomb collapse”

For normalized smooth ψ\psi, put ψλ(x)=λ3/2ψ(λx)\psi_\lambda(x)=\lambda^{3/2}\psi(\lambda x), λ>0\lambda>0. Determine its energy and minimize over λ\lambda.

Solution

Writing T=α∥∇ψ∥2>0T=\alpha\|\nabla\psi\|^2>0 and C=κ∫∣ψ∣2/∣x∣>0C=\kappa\int|\psi|^2/|x|>0, a change of variables gives qκ[ψλ]=λ2T−λCq_\kappa[\psi_\lambda]=\lambda^2T-\lambda C. Its minimum is −C2/(4T)-C^2/(4T) at λ=C/(2T)\lambda=C/(2T), and its energy grows to positive infinity as λ→∞\lambda\to\infty. Kinetic localization costs one more power of λ\lambda than the Coulomb attraction gains. The sharp form inequality above additionally bounds this minimum by −κ2/(4α)-\kappa^2/(4\alpha), uniformly over the initial shape.

  • Hall, Brian C., Quantum Theory for Mathematicians, Graduate Texts in Mathematics 267, Springer, 2013, doi:10.1007/978-1-4614-7116-5 — Schrödinger operators and the hydrogen ground state.
  • Teschl, Gerald, Mathematical Methods in Quantum Mechanics, 2nd ed., Graduate Studies in Mathematics 157, American Mathematical Society, 2014, author’s text, Chapters 6, 10, and 11 — Coulomb self-adjointness, radial realizations, and atomic Hamiltonians.