Continuous and Optimally Complete Description of Chemical Environments Using Spherical Bessel Descriptors
Pith reviewed 2026-05-25 09:09 UTC · model grok-4.3
The pith
Spherical Bessel descriptors provide a continuous, twice-differentiable, and optimally complete description of local atomic environments.
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
Core claim
An updated version of the Spherical Bessel descriptors satisfies continuity throughout the local atomic environment, twice-differentiability with respect to atomic positions, and completeness in the sense of containing all possible information about the neighborhood; moreover it is optimally complete by encoding all configurational information with the smallest possible number of descriptors. The Smooth Overlap of Atomic Position descriptors and Zernike descriptors are shown to be incapable of satisfying the full list of core requirements.
What carries the argument
Spherical Bessel descriptors, a set of basis functions for encoding local atomic environments that achieve continuity, differentiability, and optimal completeness through specific combination and truncation.
If this is right
- Machine learning potentials built on these descriptors achieve comparable accuracy while requiring roughly an order of magnitude less computation time per evaluation.
- All configurational information about an atomic neighborhood is captured without redundancy from extra terms.
- Twice-differentiability supports direct force calculations in simulations.
- The descriptors outperform Smooth Overlap of Atomic Position and Zernike alternatives on both mathematical and practical grounds.
Where Pith is reading between the lines
- The same basis could be tested for use in classical interatomic potentials that require smooth forces.
- Optimality might allow smaller training sets for machine learning models in large-scale materials studies.
- Application to systems with long-range interactions could check whether truncation still preserves completeness.
Load-bearing premise
The Spherical Bessel basis functions can be combined and truncated to achieve both mathematical completeness and optimality without introducing discontinuities or losing differentiability.
What would settle it
A demonstration that some distinct atomic configuration produces identical descriptor values under the proposed truncation, or that the descriptors change discontinuously under an arbitrarily small atomic displacement.
read the original abstract
Recently, machine learning potentials have been advanced as candidates to combine the high-accuracy of quantum mechanical simulations with the speed of classical interatomic potentials. A crucial component of a machine learning potential is the description of local atomic environments by some set of descriptors. These should ideally be continuous throughout the specified local atomic environment, twice-differentiable with respect to atomic positions and complete in the sense of containing all possible information about the neighborhood. An updated version of the recently proposed Spherical Bessel descriptors satisfies all three of these properties, and moreover is optimally complete in the sense of encoding all configurational information with the smallest possible number of descriptors. The Smooth Overlap of Atomic Position descriptors that are frequently visited in the literature and the Zernike descriptors that are built upon a similar basis are included into the discussion as being the natural counterparts of the Spherical Bessel descriptors, and shown to be incapable of satisfying the full list of core requirements for an accurate description. Aside being mathematically and physically superior, the Spherical Bessel descriptors have also the advantage of allowing machine learning potentials of comparable accuracy that require roughly an order of magnitude less computation time per evaluation than the Smooth Overlap of Atomic Position descriptors, which appear to be the common choice of descriptors in recent studies.
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The manuscript presents an updated version of Spherical Bessel descriptors for local atomic environments in machine learning interatomic potentials. It asserts that these descriptors achieve continuity, twice-differentiability, mathematical completeness, and optimality (encoding all configurational information with the minimal number of descriptors), while SOAP and Zernike descriptors are shown to be incapable of satisfying the full set of requirements simultaneously. Computational advantages, including roughly an order of magnitude faster evaluation than SOAP, are also claimed.
Significance. If the claims of simultaneous C^2 continuity, completeness, and optimality for a finite descriptor set are rigorously established with explicit constructions and verifications, the work would offer a more efficient and mathematically grounded alternative to existing descriptors, potentially improving the speed-accuracy tradeoff in ML potentials.
major comments (2)
- [Abstract and the section defining the updated Spherical Bessel descriptors] The central claim requires a finite linear combination of spherical Bessel functions (plus angular factors) that is simultaneously complete (injective on local environments up to rotation), twice differentiable everywhere including the cutoff, and minimal in descriptor count. Completeness arguments typically hold only in the infinite-basis limit; the explicit construction and proof that any radial cutoff smoothing or truncation enforces C^2 without collapsing distinct environments or introducing linear dependence must be provided (this is the single step whose failure would falsify the headline superiority over SOAP/Zernike).
- [The discussion/comparison section with SOAP and Zernike] The assertion that SOAP and Zernike descriptors cannot satisfy the full list of core requirements (continuity, C^2 differentiability, completeness, and optimality) needs to be tied to concrete limitations in their formulations, such as a specific discontinuity or non-minimal basis size, rather than stated as a general incapacity.
minor comments (2)
- [Abstract] The abstract states mathematical properties and comparisons but provides no derivation details, error analysis, or explicit verification steps; the full text must include these to allow confirmation of the central claims.
- [Abstract] The claim of 'roughly an order of magnitude less computation time' requires specification of the benchmark systems, descriptor counts, and evaluation conditions used for the timing comparison.
Simulated Author's Rebuttal
Thank you for the constructive feedback. We appreciate the opportunity to strengthen the manuscript and address the concerns regarding the rigor of our claims on the properties of the Spherical Bessel descriptors.
read point-by-point responses
-
Referee: [Abstract and the section defining the updated Spherical Bessel descriptors] The central claim requires a finite linear combination of spherical Bessel functions (plus angular factors) that is simultaneously complete (injective on local environments up to rotation), twice differentiable everywhere including the cutoff, and minimal in descriptor count. Completeness arguments typically hold only in the infinite-basis limit; the explicit construction and proof that any radial cutoff smoothing or truncation enforces C^2 without collapsing distinct environments or introducing linear dependence must be provided (this is the single step whose failure would falsify the headline superiority over SOAP/Zernike).
Authors: We agree that an explicit construction and proof is essential for the central claim. The manuscript outlines the use of a finite set of spherical Bessel functions with a smoothing function to ensure C^2 continuity at the cutoff. However, to address the referee's concern, we will expand the relevant section with a detailed mathematical proof demonstrating that the finite combination remains injective for distinct environments (up to rotation) and that the smoothing does not introduce linear dependence or collapse configurations. This will include verification through explicit examples and mathematical arguments based on the properties of the basis functions. revision: yes
-
Referee: [The discussion/comparison section with SOAP and Zernike] The assertion that SOAP and Zernike descriptors cannot satisfy the full list of core requirements (continuity, C^2 differentiability, completeness, and optimality) needs to be tied to concrete limitations in their formulations, such as a specific discontinuity or non-minimal basis size, rather than stated as a general incapacity.
Authors: We will revise the comparison section to provide concrete examples of the limitations in SOAP and Zernike descriptors. For instance, we will highlight specific cases where SOAP exhibits discontinuities in higher-order derivatives or requires a larger basis set that is not minimal, and similarly for Zernike descriptors regarding their inability to achieve optimality in descriptor count while maintaining completeness and differentiability. This will make the assertions more specific and tied to the formulations. revision: yes
Circularity Check
No circularity; claims rest on explicit mathematical construction and comparison.
full rationale
The paper constructs an updated Spherical Bessel descriptor set and derives its continuity, C^2 differentiability, and completeness properties directly from the functional form and truncation scheme. Completeness and optimality are argued via explicit basis properties and comparison to SOAP/Zernike, without reducing to fitted parameters, self-definitional loops, or load-bearing self-citations. The derivation chain is self-contained against external mathematical benchmarks and does not rename known results or smuggle ansatzes via prior work by the same authors.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.