[Paper Review] Detecting level crossings without looking at the spectrum

M. Bhattacharya, Chandra Raman|arXiv (Cornell University)|Apr 26, 2006

History and advancements in chemistry3 citations

TL;DR

This paper presents an algebraic method to detect energy level crossings in quantum systems without computing eigenvalues, by mapping the problem to finding roots of a discriminant polynomial derived from the Hamiltonian's coefficients. The method enables detection of curve crossings in atoms and molecules in magnetic fields—critical for identifying Feshbach resonances—using only matrix elements and atomic constants, revealing two critical magnetic fields (0 G and 502.2 G) where the Born-Oppenheimer approximation breaks down.

ABSTRACT

In many physical systems it is important to be aware of the crossings and avoided crossings which occur when eigenvalues of a physical observable are varied using an external parameter. We have discovered a powerful algebraic method of finding such crossings via a mapping to the problem of locating the roots of a polynomial in that parameter. We demonstrate our method on atoms and molecules in a magnetic field, where it has implications in the search for Feshbach resonances. In the atomic case our method allows us to point out a new class of invariants of the Breit-Rabi Hamiltonian of magnetic resonance. In the case of molecules, it enables us to find curve crossings with practically no knowledge of the corresponding Born-Oppenheimer potentials.

Motivation & Objective

• To develop a method for detecting energy level crossings in quantum systems without computing eigenvalues or requiring full knowledge of the Hamiltonian.
• To provide a rigorous, algebraic alternative to spectral analysis for identifying curve crossings in atomic and molecular systems under external magnetic fields.
• To establish a foundation for validating the Born-Oppenheimer approximation in diatomic molecules by detecting breakdown points without detailed knowledge of potential energy surfaces.
• To identify Feshbach resonances in ultracold atomic gases by locating magnetic fields where energy level crossings occur, using minimal input data.

Proposed method

• The method uses the discriminant of the characteristic polynomial of a parameter-dependent Hamiltonian matrix as a proxy for level crossings.
• The discriminant D[H(P)] is computed from the coefficients of the characteristic polynomial using the Sylvester matrix determinant, avoiding explicit eigenvalue computation.
• For systems with multiple parameters, such as magnetic field B and exchange energy X, the discriminant is expressed as a multivariate polynomial D[H(P)] = ∑p_n(B)X^n.
• Roots of D[H(P)] = 0 correspond to curve crossings, and their existence is determined via Sturm-Habicht sequences to count real roots in parameter intervals.
• The method is invariant under global spectral shifts, allowing simplification by removing constant diagonal offsets from the Hamiltonian.
• The approach is generalized to any matrix-based physical model, including spin-Hamiltonians and stability matrices in engineering and game theory.

Experimental results

Research questions

• RQ1 Can level crossings in quantum systems be detected without computing the spectrum or eigenvalues?
• RQ2 What magnetic fields cause breakdown of the Born-Oppenheimer approximation in diatomic molecules like 23Na-85Rb?
• RQ3 How can Feshbach resonances be located using minimal information about molecular potential energy surfaces?
• RQ4 What invariants of the Breit-Rabi Hamiltonian emerge from this algebraic approach?
• RQ5 Can the method be applied to other physical systems described by parameter-dependent matrices?

Key findings

• The method successfully identifies two magnetic fields—0 G and 502.2 G—where curve crossings occur in 23Na-85Rb, indicating breakdown of the Born-Oppenheimer approximation.
• The discriminant D[H_BO] is a polynomial of degree 6 in the exchange energy X, with coefficients depending on the magnetic field B, enabling systematic root analysis.
• Only two real roots of D[H_BO] = 0 exist in the experimentally accessible range [0, 1000 G], both occurring at X = 0.
• The approach detects crossings without requiring detailed knowledge of the Born-Oppenheimer potentials V_S and V_T, relying only on atomic constants and matrix structure.
• A new class of invariants for the Breit-Rabi Hamiltonian is identified through the algebraic structure of the discriminant.
• The method is generalizable to other spin-Hamiltonians and matrix models in physics, engineering, and game theory, enabling curve-crossing detection with minimal input.