Beyond energy bounds: comparing Schrödinger eigenstates

AI summary

Overview

Research area: Mathematical quantum physics — spectral theory of Schrödinger operators, comparison theorems for eigenstates, and bound-state models (Yukawa and Hulthén potentials).

Technical level: Advanced (graduate-level quantum mechanics, spectral theory and min–max principles assumed).

Scope (one sentence): The paper shows how the ordering of three Schrödinger operators, plus knowledge of comparison spectra and gaps, can be converted into quantitative statements about ground-state overlaps, spatial probability distributions, excited-state nodes, and the binding range of Yukawa states.

What This Paper Is About

If two quantum Hamiltonians differ only by a potential that is larger for one of them, the min–max principle already says the energies are ordered. The paper asks what else that ordering buys: how much the ground states of the two systems must overlap, how their probability densities must be arranged in space, and where the nodes of excited states must sit. The authors then apply the machinery to the Yukawa potential, using exactly solvable Hulthén potentials as comparison systems to bound overlaps, radial moments, the density at the nucleus, and the critical screening at which bound states disappear.

Key Contributions

1. Overlap bounds from spectral data alone. The authors prove that three spectral numbers — the lower operator's first two min–max values and the upper operator's ground-state energy — bound all three squared ground-state overlaps F₁₂, F₁₃ and F₂₃ by a single quantity κ, and guarantee that the middle operator has a simple discrete ground state.
2. Ground-state distribution order from potential shape. A single-crossing condition on the potential difference relative to the ground-state energy shift is shown to order likelihood ratios, and hence every monotone observable, covering asymmetric intervals, a shortened right Dirichlet wall, and radially symmetric problems in any dimension d ≥ 2.
3. Coordinatewise comparison without reflection symmetry. For a submodular reference potential and a coordinatewise nondecreasing perturbation, the reference ground state is multivariate totally positive of order two (MTP₂) and the ratio ψ₂/ψ₁ is coordinatewise nonincreasing, together with a counterexample showing the conclusion fails without submodularity.
4. Node ordering and its limits. A two-sided Sturm argument orders corresponding internal nodes for nondecreasing potential differences, with an explicit derivative formula for node motion, plus an example where ordered nodes coexist with exactly orthogonal excited states.
5. Yukawa–Hulthén application. Task-specific Hulthén comparisons yield explicit overlap, moment and contact-density bounds, and an optimized guaranteed binding range n²δ_c(n,0) ≥ 1.13611…, which the authors state is optimal within the two-parameter scaled Hulthén family.

Main Findings

• Overlap lower bound via κ: With g₁ = μ₂⁽¹⁾ − E₁⁽¹⁾ and κ = (μ₂⁽¹⁾ − E₁⁽³⁾)/g₁ > 0, the middle operator has a simple discrete ground state with μ₂⁽²⁾ − E₁⁽²⁾ ≥ μ₂⁽¹⁾ − E₁⁽³⁾, and F₁₂, F₁₃, F₂₃ ≥ κ. The ratio of the energy-bracket width to the reference gap bounds the infidelity 1 − F.
• Shrinking box example: For zero potentials on (0, Lᵢ) with (L₁, L₂, L₃) = (1, 0.9, 0.8), κ = [4 − (L₁/L₃)²]/3 = 0.8125, while the exact overlaps are approximately (0.96485, 0.86434, 0.95672), all above κ.
• Low-energy subspaces: Subspace leakage obeys n − tr(P_n⁽²⁾P_n⁽¹⁾) = Σₖ sin²ϑₖ ≤ ξₙ with ξₙ = (Sₙ⁽³⁾ − Sₙ⁽¹⁾)/(μ_{n+1}⁽¹⁾ − μ_n⁽¹⁾); for n ≥ 2, individual excited-state overlaps satisfy |⟨ψ_n⁽¹⁾, ψ_n⁽²⁾⟩|² ≥ 1 − ξₙ − ξ_{n−1}, ruling out orthogonality when the right side is positive.
• Single-crossing order: If D − (E₁⁽²⁾ − E₁⁽¹⁾) changes sign at most once from negative to positive, then ψ₂/ψ₁ is nonincreasing; a nondecreasing D suffices, regardless of its sign. Applying the result twice gives p₃ ≤_lr p₂ ≤_lr p₁ and ⟨f⟩₃ ≤ ⟨f⟩₂ ≤ ⟨f⟩₁ for nondecreasing f.
• Direction matters: For V₁ = x²/2, V₂ = V₁ + a e^(−x²/s²) and V₃ = V₁ + a with a, s > 0, the outer ground states coincide, yet the middle state has larger strictly increasing moments of |x|, because V₁ − V₂ increases in |x| and the reversed pair is ordered.
• Coordinatewise result: Under submodularity of V₁ and coordinatewise monotonicity of D, densities satisfy p₂(x)p₁(y) ≤ p₂(x∧y)p₁(x∨y) and ⟨A⟩₂ ≤ ⟨A⟩₁ for every bounded coordinatewise nondecreasing observable A.
• Failure without submodularity: For V₁ = (x² + y²)/2 + cxy with 0 < c < 1 and D = t(1 + tanh x)/2 with t > 0, stationarity gives ⟨y⟩₂ = tc⟨((1 + tanh x)′/2)⟩₂/(1 − c²) > 0 = ⟨y⟩₁, violating the coordinatewise conclusion.
• Node ordering: For the same interval and nondecreasing D, the k-th internal nodes satisfy z_k⁽²⁾ ≤ z_k⁽¹⁾ for k = 1, …, n − 1; along H(t) = H₁ + tD the node derivative is dz_k/dt = (2m/(ħ²ψ_t′(z_k)²)) ∫a^{z_k}(D − ⟨D⟩{ψ_t})ψ_t² dx ≤ 0. A uniform field D = eFx with eF > 0 shifts every internal node to the left.
• Nodes do not imply overlap: In the unit box with V₁ = 0 and V₂ = h·1_{(c,1)}, h ≥ 0, 0 < c < 1/2, the second-state overlap equals 1 at h = 0 and tends to −√c sin(2πc)/(π(1 − c²)) < 0 as h → ∞, so exact orthogonality occurs at a finite h despite a monotone potential difference.
• Yukawa overlap bound: Using V_H^(2δ) − cδ ≤ V_Y ≤ V_H^(2δ) with c = max_{x>0} e^(−x)(1/x − 1/sinh x) = 0.055810…, one gets F_{Y,H} ≥ 1 − cδ/g_H for 0 < δ < 1, with g_H = μ₂^H(2δ) − E₁^H(2δ). The positive guarantee ends at δ₀ = 1 + c − √(c(2+c)) ≈ 0.7171.
• Seelmann improvement: Because the potential difference is bounded, F_{Y,H} ≥ (1 + √(1 − t²))/2 with t = cδ/g_H < 1. At δ = 0.5 the two bounds are approximately 0.987382 and 0.776760, against a numerical overlap of 0.999759.
• Spatial Yukawa bounds: For 0 < δ < 2/√6, the chain V_C ≤ V_H^(δ) ≤ V_Y ≤ V_H^(√6 δ) gives p_C ≤lr p{H,δ} ≤lr p_Y ≤lr p{H,√6δ}, hence ⟨r^k⟩{H,δ} ≤ ⟨r^k⟩Y ≤ ⟨r^k⟩{H,√6δ} for every k > 0 and (4 − 6δ²)/(4π) ≤ |ψ_Y(0)|² ≤ (4 − δ²)/(4π). The factor √6 is stated to be the smallest upper screening factor giving a nonincreasing potential difference.
• Moment inequalities reverse: For −3 < k < 0 the moment ordering reverses, with moments defined by the convergent radial integral to avoid removable singularities at k = −1, −2. For stronger screening, p_{H,δ} ≤_lr p_Y survives whenever the Yukawa ground state exists, but the lower contact bound drops to zero.
• Binding range: With m* ∈ (1, 2) the root of (2 − m)e^m = 2 and b* = [e(2 − m*)]⁻¹, V_Y ≤ bV_H^(mδ) gives n²δ_c(n,0) ≥ 2b*/m* = 1.13611… for every s-level index n ≥ 1, improving the elementary bound n²δ_c(n,0) ≥ 1. This is the largest bound from any pointwise majorant bV_H^(mδ), attained only at (m, b) = (m*, b*). The authors note Brau and Calogero's counting bound becomes stronger at sufficiently large n, approaching 4/π.

Methodology in Plain English

The strategy is to sandwich a difficult Hamiltonian between two tractable ones. Because the three operators share the same kinetic energy and are form-ordered, the min–max principle orders their eigenvalues. Writing a target state in the reference operator's eigenbasis turns an unknown expectation value into a weighted average of known comparison eigenvalues, and the leftover weight is exactly the squared overlap to be bounded. That converts energy gaps into overlap guarantees without any spatial input.

For spatial statements, the authors study the logarithmic derivative of the ratio of two ground states through the cross-product w = ψ₁ψ₂′ − ψ₁′ψ₂, whose derivative is proportional to ψ₁ψ₂(D − ΔE). A single sign change in D − ΔE forces w to decrease and then increase, and endpoint control (via Hardy's inequality, or an argument at a shortened wall) forces w ≤ 0, giving the likelihood-ratio order. For coordinatewise results, they discretize imaginary-time paths, use total positivity of the heat kernel and of submodular potential factors, and project back onto the ground state. Node statements use a two-sided Sturm comparison of the coefficient q_i = 2m(E_n⁽ⁱ⁾ − V_i)/ħ², combined with equal node counts from the left and right.

The Yukawa analysis selects different Hulthén comparison potentials for different questions: V_H^(2δ) for energy and overlap, and the pair V_H^(δ), V_H^(√6δ) for radial moments and contact density, since these make the potential differences monotone. Optimizing over the strength b and screening factor m of a majorant bV_H^(mδ) yields the improved binding constant. Numerical checks in the figures are stated to be verified against the Airy-function solution and against direct wavefunction calculations.

Why This Matters

Energy bounds are cheap but say little about the states themselves. This work turns operator ordering into certified statements about state preparation fidelity, about where a particle is likely to be found, and about where excited-state wavefunctions vanish — quantities that matter when a comparison potential is solvable but the target is not. It also delineates sharp limits: monotone potentials order nodes but can produce exactly orthogonal excited states, so node order is not a substitute for overlap control. The Yukawa results give rigorous, closed-form guarantees in a model where no exact solution exists.

Real-world applications (as motivations for the methods, not demonstrated in the paper):

• Atomic and molecular structure: bounds on wavefunctions and contact densities for screened Coulomb (Yukawa/Hulthén) models of plasmas and impurities.
• Nuclear and particle phenomenology: the Yukawa potential is the classic meson-exchange model, where binding thresholds and node positions of excited levels are of direct interest.
• Quantum simulation and algorithm design: overlap lower bounds translate into success probabilities for phase-estimation-style ground-state energy estimation when a comparison state can be prepared.
• Statistical mechanics and probability: likelihood-ratio and MTP₂ orderings connect to stochastic domination results for diffusions and to concentration inequalities for interacting particle systems.

Industry relevance: The results are foundational rather than device-oriented. The most direct relevance is to groups building quantum algorithms that need certified ground-state preparation and to modellers of screened Coulomb systems who need error bars rather than point estimates.

Future Directions

• Extending the overlap bounds beyond the ground state and low-energy subspaces to individual highly excited states, where the paper's conditions require additional gap assumptions.
• Characterizing when node ordering can be upgraded to a nonzero overlap guarantee for excited states, given the explicit counterexample with orthogonal second states.
• Going beyond the two-parameter scaled Hulthén majorant family, since the 1.13611… constant is optimal only within that family; the authors point to Brau–Calogero counting bounds as stronger at sufficiently large n.
• Developing analogue comparison machinery for non-submodular or non-radial multidimensional potentials, where the coordinatewise conclusions currently fail without submodularity.

Target Audience

Mathematical physicists and spectral theorists working on Schrödinger operators, comparison theorems and Sturm–Liouville theory; quantum-information researchers interested in certified ground-state preparation; and atomic, nuclear or plasma physicists who model screened Coulomb systems and need rigorous bounds on wavefunctions, moments and binding thresholds. Readers should be comfortable with min–max principles, form domains and essential spectra.

Authors’ abstract

We derive eigenstate comparisons from ordered Schrödinger operators, extending energy bounds to overlaps, spatial distributions and nodes. For three form-ordered operators with discrete comparison ground states, suppose the upper ground-state energy lies in the lower operator's ground-state gap. Three comparison spectral values then bound all three squared ground-state overlaps and guarantee a simple discrete intermediate ground state. Related gap estimates control low-energy subspaces and individual excited-state overlaps under additional gap conditions. In one dimension and radially, a single crossing of the potential difference relative to the energy shift orders ground-state likelihood ratios and monotone observables; a nondecreasing difference suffices. The results cover asymmetric wells and shortened right Dirichlet walls, with coordinatewise comparisons for submodular reference potentials and coordinatewise nondecreasing perturbations. A two-sided Sturm argument orders corresponding internal nodes for nondecreasing potential differences, even between orthogonal one-dimensional excited states. For Yukawa ground states, Hulthén comparisons bound overlaps, radial moments and the density at the nucleus; the latter two-sided bounds hold for screening $0&lt;δ&lt;2/\sqrt6$. Optimizing scaled Hulthén majorants gives a lower bound of approximately $1.13611$ for $n^2δ_c(n,0)$, where $δ_c(n,0)$ is the critical screening for the $n$th $s$-wave bound state. This bound is optimal within that two-parameter comparison family.