🔬 atomic physics

Preparing Small Gaussian Basis for Highly Accurate Ab Initio Description of Lithium Rydberg States

This paper introduces a new, highly accurate, and compact Gaussian basis set with a specific optimization protocol for describing lithium Rydberg states up to $n=7$ , which achieves superior excitation energy accuracy compared to universal basis sets while enabling routine *ab initio* investigations of Rydberg states in more complex systems.

Original authors: Jan Šmydke, Benjamín Andreides, Simona Dubcová

Published 2026-07-30

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Jan Šmydke, Benjamín Andreides, Simona Dubcová

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine the atom not as a tiny, solid marble, but as a bustling solar system. In the center sits the nucleus, a heavy sun, orbited by electrons. Usually, these electrons stick close to home, huddling in the "valence" neighborhood just a few angstroms away, doing the heavy lifting of chemical bonding. But sometimes, an electron gets a massive energy boost and zooms out, traveling tens, hundreds, or even thousands of angstroms away. At this extreme distance, the electron barely feels the messy details of the other electrons; it only sees the nucleus as a simple, point-like charge. This is the world of Rydberg states. It's a realm where the electron behaves almost exactly like it does in a hydrogen atom, following neat, predictable rules.

Why do we care about these distant, lonely electrons? Because they are the gatekeepers of how atoms react and how they absorb light. If we want to predict how a molecule will behave in a laser beam or a chemical reaction, we have to understand these Rydberg states. However, describing them is a nightmare for computer scientists. The standard tools used to model atoms are like flashlights that shine brightly for a few feet but fade to black immediately after. They are great for the "valence" neighborhood but useless for the vast, empty distances where Rydberg electrons live. To study these states, scientists need a new kind of "flashlight" that can reach out into the dark without losing its shape or becoming too heavy to carry.

This paper introduces a new, super-efficient "flashlight" designed specifically for lithium atoms. The researchers, Jan ˇSmydke, Benjam´ın Andreides, and Simona Dubcov´a, have developed a tiny, highly accurate mathematical toolkit called a Gaussian basis set. Think of a basis set as a set of building blocks used to construct the shape of an electron's path. The old building blocks were either too short to reach the Rydberg electron or so numerous and clunky that they made the computer calculations crash. The team created a new set of blocks using a clever mathematical trick called "exponentially tempered Gaussians." These new blocks are flexible enough to stretch out to distances of tens of angstroms (and even further in previous tests) while remaining small and lightweight.

The results are impressive. When the team used their new, compact toolkit to simulate lithium atoms excited to high energy levels (up to the principal quantum number $n = 7$ ), the computer predicted the energy of these states with an accuracy better than $10^{-2}$ eV. To put that in perspective, this is a precision of a few thousandths of an electronvolt, which is incredibly sharp for such a small set of functions. They found that their new method works better than older, "universal" sets that try to fit every possible atom, and it does so using far fewer building blocks. For instance, to calculate specific high-energy states, their new method only needed to add two S-type and two P-type functions to a standard set, whereas older methods required dozens of extra functions.

Perhaps the most mind-bending part of the study is what the simulations revealed about the size of these excited atoms. By plotting the paths of the electrons, the researchers saw a regular, rhythmic pattern of "nodes" (places where the electron is never found) stretching out along a logarithmic scale. For the highest states they calculated ( $n=7$ ), these patterns reached tens of angstroms from the nucleus. Even more striking, when they looked at data from a previous, larger simulation of a $25S$ state, the electron's wave function stretched out to nearly 100,000 angstroms (or 10 micrometers). That is a distance comparable to the size of a bacterium! It means that a single, excited lithium atom can become as large as a microscopic living creature, yet the researchers were able to describe this enormous, diffuse cloud of probability using a surprisingly small and simple set of mathematical functions.

The paper suggests that this new, streamlined approach is a major step forward. It proves that we don't need massive, unwieldy computer models to understand these giant, diffuse atoms. Instead, with the right "exponentially tempered" building blocks, we can achieve high-precision results with a minimal amount of computational effort. This opens the door for scientists to routinely study Rydberg states in more complex systems, like large atoms and polyatomic molecules, which were previously too difficult to model accurately. While the paper focuses on lithium, the methodology suggests a path toward solving similar puzzles in the wider world of quantum chemistry.

Technical Summary: Preparing Small Gaussian Basis for Highly Accurate Ab Initio Description of Lithium Rydberg States

Problem Statement
Accurate ab initio modeling of Rydberg states in atoms and molecules presents a significant challenge in computational chemistry. Unlike valence electrons, which are confined to distances of a few Ångströms from the nucleus, Rydberg electrons occupy diffuse orbitals extending tens to thousands of Ångströms. Standard Gaussian-type basis sets, which are ubiquitous in quantum chemistry codes, dissipate rapidly and fail to describe these long-range electronic motions. While augmenting standard bases with diffuse functions can improve descriptions, unoptimized additions often lead to near-linear dependence, numerical instability, and qualitative results limited to only the lowest few states. Previous attempts to generate systematic Gaussian bases for Rydberg states (e.g., Kaufmann et al.) utilized universal, system-independent formulas that often resulted in excessively large basis sets or suboptimal performance. A tailored approach that balances high accuracy with a manageable basis set size for routine calculations remains a critical need.

Methodology
The authors developed a new, compact Gaussian basis set specifically optimized for the lithium atom ( $^2$ Li), targeting Rydberg excited states with principal quantum numbers $n=2$ to $7$ and angular momenta $S, P, D, F,$ and $G$ . The methodology relies on a tailored optimization protocol grounded in Koopmans' theorem, where virtual orbitals of the $^1$ Li $^+$ cation are optimized to approximate the Rydberg orbitals of the neutral $^2$ Li atom.

Key methodological components include:

Parameterization: Unlike the authors' previous work which utilized Even-Tempered Gaussians (ETG), the new basis employs Exponentially Tempered Gaussians (ExTG). The exponents follow the form $\zeta_k = \zeta_0 \alpha^{[(e^{\beta k}-1)/\beta]}$ , offering greater flexibility in the distribution of exponents compared to the linear logarithmic spacing of ETG.
Optimization Protocol:
1. A standard aug-ANO-pVQZ basis is selected as the core.
2. A primitive S-series is tuned to reach the Hartree-Fock energy limit of the $^1$ Li $^+$ ion, ensuring the core field remains invariant during subsequent Rydberg optimizations.
3. Rydberg orbitals are optimized sequentially (from $2S$ up to $7S$ , and similarly for $P, D, F, G$ ) by minimizing orbital energies while maintaining the Hartree-Fock limit.
4. The resulting optimized primitives are contracted into final basis functions.
State-Selectivity: The protocol allows for the generation of a "minimal" basis where only specific Rydberg functions (e.g., for a specific $nL$ state) are added to the core, rather than including the full set of functions for all $n$ .

Key Contributions

Development of AAQZ-ExTG-Ry7G: A new, highly accurate Gaussian basis set for lithium Rydberg states ( $n \le 7$ ) utilizing ExTG primitives.
Reduction in Basis Set Size: The new basis significantly reduces the number of primitive and contracted functions compared to the authors' previous large ETG-based basis (2023Sm) and universal Gaussian sets.
- The minimal state-selective variant, AAQZ-ExTG-Ry3P, requires adding only two $S$ and two $P$ functions to the standard aug-ANO-pVQZ basis to achieve high accuracy for higher states.
- In contrast, the previous minimal variant (2023Sm-Ry7D) required 51 extra functions.
Improved Numerical Stability: The ExTG parameterization was found to produce a smoother optimization process with fewer near-linear dependence issues compared to ETG.

Results
The performance of the new basis sets was evaluated at the Equation-of-Motion Coupled Cluster Singles and Doubles (EOM-CCSD) level of theory:

Excitation Energies: The new AAQZ-ExTG-Ry7G and the minimal AAQZ-ExTG-Ry3P + nL bases achieved excitation energies with errors generally below $10^{-2}$ $1 0^{- 2}$ eV (specifically a few meV) when compared to high-level non-relativistic reference values.
- The results were virtually identical between the full and minimal bases (differences $\le 0.1$ meV).
- The new tailored basis outperformed the universal Gaussian basis (AAQZ-1989KaBaJu-Ry7G), particularly for higher angular momenta ( $D, F, G$ ), where the universal basis showed fluctuating errors.
Ionization Potential (IP): The calculated IP for $^2$ Li was 5.38766 eV, differing from the experimental value by only 4 meV, consistent with the accuracy of the excitation energies.
Orbital Structure: Analysis of one-dimensional orbital cuts revealed a regular nodal structure extending to tens of Ångströms for $n=7$ states. The authors noted that the previous 2023Sm basis could describe a $25S$ state extending to $\approx 100,000$ Å ($10$ $\mu$ m), comparable to the size of a bacterium, demonstrating the capability of Gaussian bases to model such diffuse systems when properly optimized.

Significance
The paper claims that the developed methodology represents a significant step toward routine ab initio investigations of Rydberg states in more complex systems, such as large atoms and polyatomic molecules. By demonstrating that a small, tailored Gaussian basis can achieve high accuracy (better than $10^{-2}$ eV) while utilizing fewer functions than universal sets or previous tailored sets, the authors argue that the computational cost of describing diffuse systems can be substantially reduced. The work highlights that the ExTG parameterization is highly appropriate for Rydberg orbitals and that tailored basis optimization is superior to the use of universal basis sets for these specific applications. The authors conclude that their improved, simple generation methodology makes accurate descriptions of diffuse lithium Rydberg states possible with minimal Gaussian requirements.

More like this