---
title: Pre‑Born‑Oppenheimer Molecular Hamiltonian
url: https://www.emergentmind.com/topics/pre-born-oppenheimer-molecular-hamiltonian
type: topic
---

# Pre‑Born‑Oppenheimer Molecular Hamiltonian

The pre‑Born–Oppenheimer molecular Hamiltonian is the full non‑relativistic many‑particle Coulomb Hamiltonian for a molecular system, acting on a single wavefunction of all electrons and nuclei without any a priori electronic–nuclear separation. In this formulation, electrons and nuclei are treated on equal footing as quantum particles with their physical masses and charges, and the Hamiltonian contains all kinetic terms and all Coulomb interactions in one operator. In laboratory‑fixed Cartesian coordinates, a standard form is  
$$
\hat H
=
-\sum_i \frac{1}{2m_i}\nabla_{\mathbf r_i}^2
+
\sum_{i<j}\frac{q_i q_j}{|\mathbf r_i-\mathbf r_j|},
$$
or, with explicit electronic and nuclear partitions,
$$
\hat H
=
\hat T_e+\hat T_n+\hat V_{ee}+\hat V_{nn}+\hat V_{en}.
$$
This is the Hamiltonian that underlies pre‑BO molecular structure theory, explicitly correlated Gaussian implementations, multicomponent orbital methods such as NOMO, and recent first‑quantized quantum algorithms for direct electron–nuclear simulation [1801.05885] [1807.11727] [2602.11272].

## 1. Definition and relation to the Born–Oppenheimer framework

The defining feature of the pre‑BO Hamiltonian is the absence of any separation into “electronic” and “nuclear” problems. The exact stationary equation is
$$
\hat H \Psi = E \Psi,
$$
with $\Psi$ depending on all electronic and nuclear coordinates simultaneously. In the formulation used for rigorous large‑mass analysis, the same structure is written as
$$
\mathcal H(x,X)=\mathcal V(x,X)-\frac{1}{2M}\sum_{n=1}^N \Delta_{X^n},
$$
where $\mathcal V(\cdot,X)$ is the electronic operator at fixed nuclear configuration and the nuclear kinetic energy appears explicitly with the nuclear mass parameter $M$ [1104.0953].

This differs fundamentally from the standard Born–Oppenheimer construction. In BO theory one first solves, for fixed nuclear positions,
$$
\hat H_{\rm el}(\{\mathbf R_A\})
=
\hat T_e+\hat V_{ee}+\hat V_{en}(\{\mathbf R_A\})+V_{nn}(\{\mathbf R_A\}),
$$
and only afterwards solves a nuclear problem on the resulting potential energy surfaces. By contrast, in the pre‑BO Hamiltonian the nuclear degrees of freedom are dynamic at the Hamiltonian level, electron–nucleus coupling is explicit, and there is no potential‑energy‑surface reduction built into the operator itself [1507.03271] [1801.05885].

A useful consequence is conceptual completeness: rovibrational, electronic, and non‑adiabatic effects are not added in layers but are contained in a single spectrum. A corresponding difficulty is that the full Hamiltonian acts on a very high‑dimensional Hilbert space and must respect translation, rotation, inversion, and permutation symmetries at the many‑particle level [1807.11727] [1801.05885].

## 2. Symmetry, coordinates, and internal motion

The pre‑BO Hamiltonian is invariant under overall translations and rotations, and, in the absence of external fields, under inversion. This allows eigenstates to be classified by total spatial angular momentum $N$, its projection $M_N$, parity $p$, and spin and permutation quantum numbers appropriate to the particle content [1807.11727].

Because the center‑of‑mass motion is physically irrelevant for internal molecular structure, one introduces translationally invariant Cartesian coordinates together with the center‑of‑mass coordinate, so that
$$
\hat T = \hat T_{\rm int}+\hat T_{\rm CM},
\qquad
\hat H_{\rm int}=\hat H-\hat T_{\rm CM}.
$$
In coordinate‑transformed approaches this yields an exact internal Hamiltonian. In laboratory‑fixed Cartesian coordinate implementations, one may instead retain the simple laboratory form of the operator and remove translational contamination analytically from matrix elements. In the explicitly correlated Gaussian treatment of the singlet hydrogen molecule, the translational contamination term was identified as
$$
\delta_{\rm Tr}=\frac{3c_A}{4m_{\rm tot}},
$$
which is subtracted to recover translationally invariant energies while preserving the simple LFCC form of the Hamiltonian and basis functions [1212.3700].

Rotation is more delicate. In explicitly correlated Gaussian approaches, total angular momentum is enforced at the basis level, so the Hamiltonian need not be rewritten in body‑fixed form. In the NOMO translation‑ and rotation‑free construction, by contrast, the working Hamiltonian is
$$
\hat H_{\rm TRF}
=
\hat H-\hat T_{\rm T}-\hat T_{\rm R},
$$
where the center‑of‑mass kinetic energy is “simply subtracted,” while rotational subtraction is only approximate because molecular rotations and vibrations are coupled. In the numerical study of H$_2$ and HT, $\hat T_{\rm R}$ was treated by a zeroth‑order Taylor expansion, which was adequate for the lowest rotationless states considered [1507.03271].

These coordinate and symmetry issues are a common source of misunderstanding. Pre‑BO theory does not require internal coordinates or body‑fixed coordinates as a matter of principle; laboratory‑fixed Cartesian formulations are viable provided translational contamination is removed and rotational symmetry is enforced correctly [1212.3700] [1406.3785].

## 3. Variational realizations in explicitly correlated bases

A central realization of the pre‑BO Hamiltonian is variational expansion in explicitly correlated Gaussian basis functions. In one widely used form, the spatial basis function is
$$
\phi(\mathbf r;A)
=
\exp\!\left(
-\frac{1}{2}\mathbf r^{T}(A\otimes I_3)\mathbf r
\right),
$$
and the angular‑momentum adapted global‑vector representation augments this with a polynomial and spherical harmonic,
$$
\phi^{[\lambda]}(\mathbf r;A,u,K)
=
|\mathbf v|^{2K+N}
Y_N^{M_N}(\hat{\mathbf v})
\exp\!\left(
-\frac{1}{2}\mathbf r^{T}(A\otimes I_3)\mathbf r
\right),
$$
with $\mathbf v=\sum_i u_i \mathbf r_i$ [1801.05885].

The total wavefunction is then expanded as a linear combination of symmetrized products of spatial and spin functions,
$$
\Psi(\mathbf r)=\sum_I c_I\,\boldsymbol{\chi}_I^{S,M_S}\,\hat Y\,\phi_I^{\rm FECG}(\mathbf r;\{\omega_I\}),
$$
or, in review notation,
$$
\Psi^{[\lambda,\varsigma]}=\sum_{I=1}^{N_b} c_I \Phi_I^{[\lambda,\varsigma]}.
$$
This permits direct variational solution of the full electron–nuclear Schrödinger equation while enforcing the Pauli principle and the permutation symmetry of identical nuclei [1807.11727] [1801.05885].

Shifted‑center or floating ECGs increase flexibility but are not, in general, eigenfunctions of total angular momentum and parity. One remedy is numerical projection onto irreducible representations of the rotation–inversion group,
$$
\hat P^{[N]}_{M_1M_2}
=
\int \frac{d\Omega}{4\pi^3}
D^{[N]}_{M_1M_2}(\Omega)^*\,\hat R(\Omega),
$$
combined with parity projection. In the five‑particle pre‑BO treatment of H$_3^+$, shifted FECGs plus projection yielded a variational upper bound that substantially improved earlier ECG results [1807.11727].

The same Hamiltonian supports property calculations and resonance theory. Electric transition dipole moments have been evaluated directly from the full pre‑BO Hamiltonian in laboratory‑fixed Cartesian coordinates, in both length and velocity forms, without relying on clamped‑nuclei transition dipole surfaces [1406.3785]. Resonances have been accessed by complex coordinate rotation, with the transformed Hamiltonian
$$
\hat{\mathcal H}(\theta)=e^{-2i\theta}\hat T+e^{-i\theta}\hat V,
$$
allowing rovibronic resonances of H$_2$, Ps$^-$, and Ps$_2$ to appear as discrete complex eigenvalues [1810.05493].

## 4. Orbital, second‑quantized, and quantum‑algorithmic representations

A distinct practical realization is the nuclear‑orbital plus molecular‑orbital framework. In NOMO, electrons and nuclei are represented by one‑particle orbitals, and the translation‑ and rotation‑free Hamiltonian is written in second quantization as
$$
\begin{aligned}
\hat H_{\rm TRF}
&=
\sum_{pq} h_{pq}^{\rm ee} \hat a_p^\dagger \hat a_q
+
\sum_{PQ} h_{PQ}^{\rm nn} \hat a_P^\dagger \hat a_Q \\
&\quad+
\frac12\sum_{pqrs} V_{pqrs}^{\rm ee}\hat a_p^\dagger \hat a_q^\dagger \hat a_s \hat a_r
+
\frac12\sum_{PQRS} V_{PQRS}^{\rm nn}\hat a_P^\dagger \hat a_Q^\dagger \hat a_S \hat a_R \\
&\quad+
\sum_{pQrS} V_{pQrS}^{\rm en}\hat a_p^\dagger \hat a_Q^\dagger \hat a_S \hat a_r .
\end{aligned}
$$
The indices $p,q,r,s$ run over electronic spin orbitals and $P,Q,R,S$ over nuclear spin orbitals. In the complete‑basis and complete‑configuration limit, the authors state that “the NOMO/FCI theory for a complete configuration space is an exact theory” within the TRF framework [1507.03271].

This multicomponent second‑quantized form can be mapped to qubits. Fermionic particles use Jordan–Wigner or Bravyi–Kitaev mappings; bosonic nuclei admit either a direct boson mapping or the compact boson mapping proposed in that work; and distinguishable nuclei can be treated with one‑qubit occupation variables. In the proof‑of‑principle H$_2$ and HT calculations, the nuclei were treated as distinguishable, which the authors state is justified because exchange interaction between nuclei is negligibly small for the states studied [1507.03271].

Recent first‑quantized quantum algorithms instead represent each particle on a real‑space grid and simulate the full Hamiltonian
$$
H=T+V,
\qquad
T=\sum_{i=1}^{\eta}\frac{-\nabla_i^2}{2m_i},
\qquad
V=\sum_{i\neq j}(-1)^{\sigma_i+\sigma_j}\frac{\zeta_i\zeta_j}{2r_{ij}}.
$$
In this setting, the Hamiltonian is block‑encoded using swap networks and an alternating‑sign implementation of the Coulomb interaction. For the $\rm NH_3+BF_3$ reaction, the reported resource estimate is a Toffoli cost of $8.72\times10^9$ per femtosecond with $1362$ logical qubits, illustrating that the full pre‑BO Hamiltonian has become an explicit target for fault‑tolerant quantum simulation [2602.11272].

## 5. Controlled reductions and alternative beyond‑BO Hamiltonians

Although the pre‑BO Hamiltonian is the fundamental operator, much work analyzes controlled reductions derived from it. One rigorous route starts from the full molecular Schrödinger operator and shows that, under a uniform electronic spectral gap and in the large nuclear mass limit, Born–Oppenheimer molecular dynamics approximates stationary Schrödinger observables with
$$
\langle g\rangle_{\rm BO}
=
\langle g\rangle_{\rm Sch}
+
O(M^{-1+\delta})
$$
for any bounded nuclear observable $g(X)$ and any $\delta>0$ [1104.0953]. This does not redefine the pre‑BO Hamiltonian; it quantifies when a reduced BO description is accurate for specific observables.

A second route derives a second‑order effective nuclear Hamiltonian associated with one isolated electronic state. The resulting operator contains the BO potential energy surface, the diagonal Born–Oppenheimer correction, and a mass‑correction tensor. In compact form,
$$
\hat H^{(2)}_{\rm n}
=
\hat T^{(2)} + V(\mathbf R)+U(\mathbf R),
$$
with
$$
\hat T^{(2)}
=
-\frac{1}{2n}\nabla_{\mathbf R}^{T}\boldsymbol{\mathcal M}(\mathbf R)\nabla_{\mathbf R},
\qquad
\mathcal M_{ia,jb}(\mathbf R)
=
\delta_{ia,jb}-\frac{1}{n}A_{ia,jb}(\mathbf R).
$$
This Hamiltonian is explicitly presented as a reduction of the full electron–nuclear Hamiltonian, not as a replacement for pre‑BO theory itself [1811.01186].

More recently, phase‑space electronic Hamiltonians have been proposed that depend on both nuclear positions and momenta,
$$
\hat H_{\rm PS}(\mathbf X,\mathbf P)
=
\sum_A \frac{1}{2M_A}
\left(
\mathbf P_A-i\hbar\hat{\boldsymbol\Gamma}_A
\right)^2
+
\hat H_{\rm el}.
$$
These constructions are designed to recover beyond‑BO electronic momentum, current density, and vibrational circular dichroism while conserving total linear and angular momentum along quantum‑classical trajectories [2407.16918] [2407.19257]. They are not, however, fully quantum pre‑BO Hamiltonians: in the authors’ own characterization, nuclei enter as classical phase‑space parameters, so the method is best viewed as a mixed quantum–classical approximation motivated by pre‑BO structure rather than a full pre‑BO treatment.

## 6. Spectral consequences, observables, and conceptual significance

The spectrum of the pre‑BO Hamiltonian contains bound states, rovibronic excitations, and resonances without invoking potential energy surfaces. This has been demonstrated for systems as different as H$_2^+$, H$_2$, H$_3^+$, Ps$^-$, Ps$_2$, and hydrogen isotopomers such as HT [1801.05885] [1810.05493] [1507.03271]. In H$_2$ and HT, for example, vibrational excitation appears in the NOMO picture as excitation among nuclear orbitals, with dominant configurations
$$
1\sigma_g^2 1\sigma_{n1} 1\sigma_{n2}
\quad\text{and}\quad
1\sigma_g^2 2\sigma_{n1} 2\sigma_{n2},
$$
for the ground and first excited rotationless vibrational states, respectively [1507.03271].

Pre‑BO calculations also serve as benchmarks for approximate non‑adiabatic models. For H$_3^+$, shifted FECGs with projection produced an extrapolated ground‑state energy close to a non‑adiabatic estimate based on BO, DBOC, and effective‑mass corrections, showing that well‑constructed BO plus non‑adiabatic schemes can approach full pre‑BO results while also clarifying their residual error [1807.11727].

At the level of wavefunction structure, the pre‑BO Hamiltonian defines a genuine bipartite electron–nuclear problem. Schmidt decompositions and reduced density matrices therefore provide a natural measure of electronic–nuclear entanglement. In the one‑dimensional H$_2^+$ and Shin–Metiu analyses, the ground BO vibronic state could be “almost separable and non‑entangled,” whereas vibrational excitation increased the entanglement monotonically, avoided crossings strongly enhanced entanglement in the adiabatic BO picture, and Born–Huang superpositions could be more entangled than any individual BO component [2508.21578]. A plausible implication is that the pre‑BO Hamiltonian is not only a route to non‑adiabatic spectra but also a direct object for quantifying electron–nuclear correlation.

A persistent conceptual issue is the status of molecular structure. Because exact eigenstates of the full Hamiltonian respect overall symmetry, familiar localized structures do not appear as fixed classical geometries in the operator itself. Instead, one extracts structural information from marginal probability densities and angle distributions; for H$_2^+$ and H$_2$D$^+$ these reveal shell structures and characteristic angular correlations [1801.05885]. This suggests that classical molecular structure is not encoded by adding geometry parameters to the Hamiltonian, but emerges from symmetry‑resolved quantum states, measurement, and, in the broader view of the review literature, decoherence [1801.05885].

The pre‑Born–Oppenheimer molecular Hamiltonian is therefore both a specific operator and a framework. As an operator, it is the full many‑particle Coulomb Hamiltonian with all electrons and nuclei quantum mechanical. As a framework, it encompasses exact internal‑coordinate and laboratory‑coordinate formulations, explicitly correlated variational methods, multicomponent orbital theories, resonance formalisms, and modern quantum algorithms, while also providing the reference point from which controlled BO and beyond‑BO reductions are derived [1801.05885] [1212.3700] [2602.11272].

Source: https://www.emergentmind.com/topics/pre-born-oppenheimer-molecular-hamiltonian