Pith. sign in

REVIEW 5 minor 4 references

TUNA: A streamlined quantum chemistry program for atoms and diatomics

T0 review · 0 major / 5 minor · reviewed 2026-07-14 · grok-4.5

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read One principle turns any energy method into structures, vibrations and response properties for atoms and diatomics.

desk verdict Clean, open diatomic-only QC package whose real novelty is the uniform numerical-diff architecture and one-line CLI, not new theory. read the letter →

arxiv 2604.01471 v2 pith:UIRL2UO2 submitted 2026-04-01 physics.chem-ph physics.atm-clusphysics.comp-phphysics.ed-ph

Harry Brough This is my paper · ORCID
classification physics.chem-phphysics.atm-clusphysics.comp-phphysics.ed-ph
keywords quantumchemistrysoftwarediatomicmoleculesnumericaldifferentiationelectronicstructuremethodscoupledclusterdensityfunctionaltheoryteachingplatformresponseproperties
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

TUNA is an open-source quantum chemistry program restricted to atoms and diatomic molecules. The author argues that this narrow scope, combined with a single design rule, produces a program that is simultaneously a teaching tool, a benchmarking environment and a development sandbox. The rule is simple: once a method can evaluate the energy, every structural, vibrational and field-response property is obtained by numerical differentiation on the one-dimensional bond coordinate. Because diatomic symmetry reduces many observables to one or two independent components, carefully chosen finite-difference stencils become practical rather than a last resort. The result is a uniform command-line interface that runs Hartree–Fock, density-functional, Møller–Plesset and high-order coupled-cluster methods, and immediately yields optimisations, anharmonic frequencies, bond-dissociation energies, polarisabilities and molecular-dynamics trajectories for every one of them. Reference implementations and detailed theory documentation make the code an accessible starting point for students and for researchers who want to prototype new algorithms on the simplest instructive molecules.

What carries the argument

Numerical differentiation of the energy under diatomic symmetry: central differences for gradients, five- to nine-point stencils for higher derivatives, with step sizes and convergence thresholds chosen once for the whole program so that every implemented electronic-structure method inherits the full suite of properties automatically.

What would settle it

Run the same high-order coupled-cluster or double-hybrid calculation for a polarisability or anharmonic frequency both with TUNA’s default finite-difference scheme and with a well-validated analytic or much denser numerical reference; a systematic discrepancy larger than chemical or spectroscopic accuracy would falsify the numerical-stability premise.

Watch

Extended reading notes

Core claim

The paper establishes that a deliberately restricted molecular domain (atoms and diatomics) plus the uniform principle “energy evaluation implies all properties by numerical differentiation” yields a transparent, consistent electronic-structure program. Once any method supplies an energy, optimised geometries, harmonic and non-perturbative anharmonic frequencies, finite-field response properties, scans and ab initio trajectories become available without method-specific analytic derivatives.

Load-bearing premise

The claim rests on the premise that a single set of carefully chosen finite-difference step sizes and tightened energy thresholds remains numerically stable and accurate for every method and every property across the diatomics the program targets.

Editorial extensions

If this is right

  • Any newly implemented energy method automatically gains geometry optimisations, frequencies, response properties and dynamics without additional analytic-derivative code.
  • Students can compare restricted Hartree–Fock, DFT, MP2–MP4 and CCSDTQ on the same diatomic with identical syntax and immediately plot orbitals, densities and potential-energy curves.
  • Benchmarking campaigns on bond breaking, spin contamination and basis-set extrapolation become uniform across a wide ladder of methods.
  • Developers can prototype improved algorithms or new functionals on the simplest instructive systems and obtain full property suites at once.
  • Basis-set extrapolation applied to the energy automatically propagates to approximate complete-basis response properties.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The same numerical-differentiation philosophy could be stress-tested on a small set of linear triatomics to measure how quickly the symmetry advantage erodes.
  • Because every property is a numerical derivative, TUNA provides a ready-made platform for quantifying finite-difference error bars as a function of method and basis set—an analysis rarely published for production codes.
  • The planned addition of magnetic response and CASSCF will test whether the single-principle design continues to scale once analytic derivatives become far more complicated than the energy itself.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

0 major / 5 minor

Summary. TUNA is an open-source Python quantum chemistry program restricted to atoms and diatomics. Its design principle is that once an electronic-structure method can evaluate the energy, structural, vibrational, response and dynamical properties are obtained uniformly by numerical differentiation, exploiting diatomic symmetry. The manuscript describes a compact CLI covering SPE, OPT, FREQ, ANHARM, SCAN, MD, BDE, IP and EA; a broad method set (HF, DFT through double hybrids, MP2–MP4 and variants, CCD through CCSDTQ, QCI, CEPA, CIS); many Gaussian basis sets; finite-difference stencils for gradients and higher derivatives; and a concrete H2 BDE example (CCSD/cc-pVQZ + ZPE/VPT2 + extrapolation) that matches experiment to 0.05 kcal mol−1. The code is MIT-licensed, installable from PyPI, and accompanied by theoretical documentation.

Significance. If the released code and documentation match the description, TUNA fills a genuine niche: a transparent, uniform teaching and prototyping environment for electronic structure on the simplest chemically interesting systems. The architectural claim—energy evaluation plus numerical differentiation under diatomic symmetry—is sound and is supported by open code, explicit stencils (§2), and a falsifiable numerical demonstration. Strengths that should be credited include the MIT-licensed Python/NumPy/Cython implementation, the consistent CLI, reference implementations of high-order CC and double-hybrid DFT, and the integrated plotting of orbitals, densities and vibrational wavefunctions. Residual numerical caveats about finite-difference step sizes do not undermine the pedagogical and developmental value of the contribution.

minor comments (5)
  1. §2 states that h has been optimised and that tightened convergence is used for higher derivatives, but no numerical values of h, stencil widths, or convergence thresholds are given. A short table or appendix listing the defaults would improve reproducibility without changing the architectural claim.
  2. The H2 BDE example is persuasive but solitary. One additional sentence or table entry for a second diatomic (e.g. N2 or HF frequency or BDE) would better illustrate breadth across methods.
  3. Figure 2 captions and axis labels appear partially corrupted by Unicode/encoding artefacts in the manuscript source; these should be cleaned for the published version.
  4. §5 asserts competitiveness with established codes via symmetry and Cython integrals; a single wall-time comparison (even for H2 or N2 at a fixed level) would make that claim concrete.
  5. Table 2 lists exchange and correlation functionals separately; a brief note on which combinations are actually exposed as named XC functionals (and any double-hybrid defaults) would help users.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: TUNA is a software architecture paper whose central claim is that numerical differentiation of any energy evaluator yields all properties for diatomics; nothing is fitted or predicted from itself.

full rationale

The paper presents an open-source program (TUNA) whose design principle is architectural and pedagogical: once any electronic-structure method can evaluate the energy, diatomic symmetry makes numerical differentiation a uniform route to geometries, frequencies, response properties, scans and MD (Abstract, §2). The finite-difference formulae (central differences for gradients, five- to nine-point stencils for higher derivatives) are textbook numerical analysis, not derived from the target observables. The single quantitative illustration (CCSD/cc-pVQZ BDE of H2 = 103.32 kcal mol^{-1} vs experiment 103.27) is an external experimental comparison, not a fit of a parameter that is then re-predicted. All methods (HF, DFT, MPn, CC, CIS, etc.) and basis sets are standard literature implementations; citations are to independent prior work (Psi4NumPy, ORCA, Dunning, Grimme, etc.) and do not form a self-citation chain that forces the result. There is no uniqueness theorem, no ansatz smuggled via self-citation, and no renaming of a known empirical pattern. The contribution is therefore self-contained software infrastructure with open code; residual numerical-stability caveats about step size h do not constitute circularity. Score 0 is the correct, proportionate finding.

Assumptions & free parameters 1 free parameters · 3 assumptions · 0 invented entities

As a software paper the load-bearing content is the implementation itself. The only free parameters are the finite-difference step sizes chosen by the author; all electronic-structure methods and basis sets are taken from the established literature. No new physical entities are postulated.

free parameters (1)
  • finite-difference step size h (and stencil widths) = optimised per derivative order (exact values in source)
    Author-optimised values used for first through fourth derivatives of the energy; stated to maximise precision but not derived from first principles or published as a universal constant.
assumptions (3)
  • domain assumption Standard electronic-structure methods (HF, DFT, MPn, CC, CIS, …) correctly evaluate the electronic energy for a given nuclear geometry and basis set.
    Invoked throughout §§2–4; the entire numerical-differentiation framework rests on the energy being available from these methods.
  • standard math Central finite-difference formulas of the stated orders converge to the analytic derivatives for sufficiently small h and tight energy convergence.
    §2; classical numerical analysis used without further proof.
  • domain assumption Diatomic symmetry reduces all nuclear and field responses to one or two independent components, making pure numerical differentiation practical.
    Stated in Introduction and §2 as the justification for the program’s architecture.

how reviews work

0 comments
Cite this review

Pith. "Pith review of TUNA: A streamlined quantum chemistry program for atoms and diatomics." pith.science (2026). https://pith.science/paper/UIRL2UO2

@misc{pith2026260401471,
  author       = {Pith},
  title        = {Pith review of: TUNA: A streamlined quantum chemistry program for atoms and diatomics},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/UIRL2UO2}},
  note         = {Machine review of arXiv:2604.01471}
}
abstract

We present TUNA, an open-source quantum chemistry program specifically designed for atoms and diatomic molecules. Within this narrow molecular domain, a broad and consistent set of electronic structure methods and calculation types is available. Energies, optimisations, vibrational frequencies, response properties, coordinate scans and ab initio molecular dynamics trajectories can all be accessed through an intuitive command-line interface. A single principle underlies TUNA: once a method can be used to evaluate the energy, all properties follow from numerical differentiation. This makes the program both a transparent teaching platform and a compact environment for benchmarking methods on diatomic molecules $\unicode{x2014}$ among the most simple yet instructive systems in quantum chemistry. Reference implementations including density functional theory, many-body perturbation theory and coupled cluster theory, supported by detailed theoretical documentation, make TUNA an accessible foundation for developing improved methods and algorithms in electronic structure.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

4 extracted references · 1 canonical work pages

  1. [1]

    (1) J. D. Hunter, Comput. Sci. Eng. , 2007, 9, 90–95. (2) F. Neese and E. F. Valeev, J. Chem. Theory Comput. , 2011, 7, 33–43. 7 (3) M. Piccardo, J. Bloino and V. Barone, Int. J. Quantum Chem. , 2015, 115, 948–982. (4) J. Liu, E. J. Salumbides, U. Hollenstein, J. C. J. Koelemeij, K. S. E. Eikema, W. Ubachs and F. Merkt, J. Chem. Phys. , 2009, 130, 174306....

    work page doi:10.26434/chemrxiv-2025-qgc1q 2007

  2. [2]

    Jensen, J

    (30) F. Jensen, J. Chem. Phys. , 2001, 115, 9113–9125. (31) F. Jensen, J. Chem. Phys. , 2002, 117, 9234–9240. (32) F. Jensen, J. Chem. Phys. , 2002, 116, 7372–7379. 8 (33) F. Jensen, J. Chem. Theory Comput. , 2014, 10, 1074–1085. (34) F. Neese and E. F. Valeev, J. Chem. Theory Comput. , 2011, 7, 33–43. (35) J. C. Slater, Phys. Rev. , 1951, 81, 385–390. (3...

    2001

  3. [3]

    (48) T. D. Crawford, C++ Programming Tutorial in Chemistry , https : / / github . com / CrawfordGroup/ProgrammingProjects, Accessed: 2026-03-28,

    2026

  4. [4]

    Neese, WIREs Comput

    (50) F. Neese, WIREs Comput. Mol. Sci , 2012, 2, 73–78. (51) F. Neese, WIREs Comput. Mol. Sci. , 2022, 12, e1606. 9

    2012

Pith tools

Reviewed July 14, 2026 · model on record in the stance chip above.