Assessing the impact of nodal surface optimization in fixed-node diffusion Monte Carlo on non-covalent interactions

Andrea Zen; Benjamin X. Shi; Dario Alf\`e; Kousuke Nakano

arxiv: 2604.04329 · v1 · submitted 2026-04-06 · ⚛️ physics.chem-ph · cond-mat.str-el· physics.comp-ph

Assessing the impact of nodal surface optimization in fixed-node diffusion Monte Carlo on non-covalent interactions

Kousuke Nakano , Benjamin X. Shi , Dario Alf\`e , Andrea Zen This is my paper

Pith reviewed 2026-05-10 20:15 UTC · model grok-4.3

classification ⚛️ physics.chem-ph cond-mat.str-elphysics.comp-ph

keywords non-covalent interactionsdiffusion Monte Carlofixed-node approximationnodal surface optimizationantisymmetrized geminal powerhydrogen bondingdispersion interactionsCCSD(T)

0 comments

The pith

Optimizing the nodal surface in fixed-node diffusion Monte Carlo brings results closer to CCSD(T) for hydrogen-bonded non-covalent interactions.

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

The paper examines whether refining the nodes that constrain the wavefunction in diffusion Monte Carlo can resolve known mismatches with coupled-cluster benchmarks when calculating non-covalent interactions. It applies a recently introduced antisymmetrized geminal power ansatz built from natural orbitals to generate trial nodes for twelve representative systems. The change narrows the gap with CCSD(T) reference energies in hydrogen-bonded cases but leaves dispersion-dominated cases essentially unaltered. A sympathetic reader would see this as a practical test of whether nodal error is the main source of discrepancy in one important class of interactions.

Core claim

Adopting an antisymmetrized geminal power ansatz with natural orbitals to optimize the nodal surface in fixed-node diffusion Monte Carlo produces improved agreement with CCSD(T) reference values for hydrogen-bonded systems, while the same optimization has negligible effect on dispersion-dominated systems.

What carries the argument

Antisymmetrized geminal power (AGP) ansatz with natural orbitals, used to generate an improved nodal surface for fixed-node diffusion Monte Carlo.

If this is right

Fixed-node DMC can achieve closer agreement with CCSD(T) for hydrogen-bonded non-covalent interactions by using the AGP nodal surface.
The approach supplies a computationally modest route to reducing discrepancies specifically in hydrogen-bonded systems.
Remaining differences between DMC and CCSD(T) in dispersion-dominated systems are unlikely to be removed by further nodal optimization alone.
The pattern of improvement versus non-improvement distinguishes the dominant error source in the two classes of interactions.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

Nodal quality appears more limiting for hydrogen-bonded complexes than for dispersion-bound ones, suggesting electron correlation near the nodes is stronger in the former.
The same AGP nodal construction could be applied to larger clusters where CCSD(T) becomes prohibitive, providing an independent check on interaction energies.
Dispersion systems may require improvements in the trial wavefunction beyond the nodes, such as better treatment of long-range correlation or finite-size effects.

Load-bearing premise

The AGP nodal surface lies close enough to the true nodes that any remaining fixed-node bias is not masked by compensating errors introduced by the optimization itself.

What would settle it

Repeating the calculations on the same hydrogen-bonded test set with an independent nodal-surface method that differs from both the original Slater determinant and the AGP surface and finding no systematic improvement over the original DMC results would falsify the central claim.

Figures

Figures reproduced from arXiv: 2604.04329 by Andrea Zen, Benjamin X. Shi, Dario Alf\`e, Kousuke Nakano.

**Figure 2.** Figure 2: Pairwise MAEs of binding energies obtained by CCSD(T), CCSD(cT)-fit, FNSD-DMC, and FN-AGPn-DMC for hydrogen bonds (H-bond), dispersion-dominated bonds (Dispersion), and all of them (All). The unit is kcal/mol. thus, FN-AGPn-DMC gives more accurate energies than FN-SD-DMC does, allowing one to discuss whether the discrepancy arises from poor nodal surfaces in FN-DMC. On the one hand, our calculations rev… view at source ↗

read the original abstract

Diffusion quantum Monte Carlo (DMC) and coupled cluster theory [CCSD(T)] are widely-employed benchmark methods for noncovalent interactions (NCIs). However, recent studies have reported notable discrepancies across several hydrogen-bonded and dispersion-dominated systems, raising questions on the accuracy of the approximations underlying each approach. In DMC, the dominant error is expected to stem from the fixed-node approximation, where the nodal surface is typically taken from a single Slater determinant derived from a density functional theory or Hartree-Fock calculation. In this work, we assess the impact of nodal surface optimization on DMC predictions for 12 compounds spanning diverse NCIs, using a recently proposed antisymmetrized geminal power ansatz with natural orbitals. We find improved agreement with CCSD(T) for hydrogen-bonded systems, while having negligible effect for dispersion-dominated systems. These results provide a practical and computationally efficient route to resolving discrepancies in hydrogen-bonded interactions, while offering insight into the remaining differences in dispersion-dominated systems.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

AGP nodal optimization in DMC narrows the gap to CCSD(T) for hydrogen bonds but not dispersion, though the data do not separate real node improvement from possible error cancellation.

read the letter

The key point here is that using an antisymmetrized geminal power ansatz for the nodes in diffusion Monte Carlo reduces the difference from CCSD(T) on hydrogen-bonded non-covalent interactions but leaves dispersion cases largely unchanged. This comes from tests on 12 small systems. The work applies a recent nodal optimization method to a set of molecules where the two high-accuracy approaches have shown discrepancies. It demonstrates that the choice of nodal surface can selectively affect different types of interactions. The paper reports the changes in a straightforward way and compares directly to an external coupled-cluster reference without introducing new parameters. One soft spot is the interpretation of why the agreement improves. The paper assumes the AGP nodes are closer to the true ones, but since CCSD(T) is also approximate, the better match could result from the AGP introducing a bias that offsets the remaining error in CCSD(T) for hydrogen bonds. No additional test, like a variational Monte Carlo energy or nodal overlap with a more accurate wavefunction, is provided to confirm genuine node improvement over cancellation. This makes the claim about resolving the discrepancy through better nodes less secure than the raw numbers. The paper targets researchers who use or develop benchmark data for non-covalent interactions, particularly those concerned with method consistency in small-molecule models relevant to biochemistry. A reader focused on quantum Monte Carlo applications or high-level reference calculations will find the specific results on nodal effects informative. It deserves peer review because it tackles a practical issue with concrete calculations on a defined test set, and the findings can guide further work even if more supporting diagnostics are needed. I recommend sending it to referees with a request for those extra checks on the nodal surfaces to clarify the source of the improvement.

Referee Report

1 major / 0 minor

Summary. The manuscript evaluates the effect of nodal surface optimization in fixed-node diffusion Monte Carlo using an antisymmetrized geminal power (AGP) ansatz constructed from natural orbitals. For a test set of 12 compounds spanning hydrogen-bonded and dispersion-dominated non-covalent interactions, the optimized nodes yield improved agreement with CCSD(T) reference values for the hydrogen-bonded subset while producing negligible changes for dispersion-dominated cases.

Significance. If the observed improvements reflect genuine reduction of fixed-node error rather than compensatory cancellation, the approach supplies a computationally tractable route to higher-accuracy DMC energies for hydrogen-bonded NCIs and clarifies the origin of remaining DMC–CCSD(T) discrepancies in dispersion systems. The work directly addresses a known source of benchmark inconsistency in the field.

major comments (1)

[Abstract and Results] The central claim that AGP nodal optimization reduces fixed-node error (Abstract; Results section) rests on the untested premise that the AGP surface lies closer to the exact nodes. Because both DMC and CCSD(T) are approximate, the reported reduction in discrepancy for hydrogen-bonded systems could arise from a new AGP bias whose sign aligns with residual CCSD(T) error. No independent diagnostic—such as pure variational Monte Carlo energies, nodal overlap integrals, or comparison against a higher-level nodal reference—is presented to separate genuine node improvement from error cancellation.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for the careful review and for recognizing the potential significance of the work. The major comment is addressed below with proposed revisions to strengthen the manuscript.

read point-by-point responses

Referee: [Abstract and Results] The central claim that AGP nodal optimization reduces fixed-node error (Abstract; Results section) rests on the untested premise that the AGP surface lies closer to the exact nodes. Because both DMC and CCSD(T) are approximate, the reported reduction in discrepancy for hydrogen-bonded systems could arise from a new AGP bias whose sign aligns with residual CCSD(T) error. No independent diagnostic—such as pure variational Monte Carlo energies, nodal overlap integrals, or comparison against a higher-level nodal reference—is presented to separate genuine node improvement from error cancellation.

Authors: We agree that the manuscript does not present independent diagnostics (such as VMC energies or nodal overlaps) that would directly confirm the AGP nodes are closer to the exact nodes. Our primary finding is the observed improvement in agreement with CCSD(T) for the hydrogen-bonded subset and the lack of change for dispersion-dominated cases. We acknowledge that this improvement could in principle arise from a compensatory bias rather than genuine nodal improvement. To address this, we will revise the abstract and results section to phrase the conclusions strictly in terms of the change in DMC–CCSD(T) discrepancy rather than claiming a reduction in fixed-node error. We will also add a concise discussion paragraph noting the limitation and the value of future diagnostics. The system-dependent pattern (improvement only for hydrogen-bonded systems) is consistent with the physical expectation that nodal quality matters more when correlation is dominated by short-range effects, but we recognize this does not constitute proof. revision: yes

Circularity Check

0 steps flagged

No significant circularity; computational assessment relies on external benchmarks

full rationale

The manuscript conducts a numerical assessment of AGP-based nodal optimization in fixed-node DMC by direct comparison of computed interaction energies against independent CCSD(T) reference values for 12 systems. No load-bearing derivation, equation, or fitted parameter reduces the reported improvements (or lack thereof) to the input data by construction. The AGP ansatz is invoked as a recently proposed tool rather than derived within the paper, and the central claim rests on external falsifiable benchmarks rather than self-citation chains or self-definitional loops. This is the expected non-circular outcome for a benchmark-style computational study.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The work rests on standard quantum Monte Carlo assumptions plus the practical choice of AGP form; no new particles or forces are introduced.

axioms (2)

domain assumption Fixed-node approximation error dominates DMC inaccuracies for these systems
Stated in the abstract as the expected dominant error source.
domain assumption Natural orbitals from a prior calculation provide a good starting point for AGP nodal optimization
Implicit in the choice of ansatz.

pith-pipeline@v0.9.0 · 5489 in / 1260 out tokens · 45373 ms · 2026-05-10T20:15:11.774598+00:00 · methodology

discussion (0)

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

IndisputableMonolith/Foundation/AlexanderDuality.lean alexander_duality_circle_linking unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

the dominant error is expected to stem from the fixed-node approximation, where the nodal surface is typically taken from a single Slater determinant derived from a density functional theory or Hartree-Fock calculation
IndisputableMonolith/Cost/FunctionalEquation.lean washburn_uniqueness_aczel unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

FN-AGPn-DMC yields smaller binding energies than the FN-SD-DMC does; thus, provides bonding energies very close to those of CCSD(T)

What do these tags mean?

matches: The paper's claim is directly supported by a theorem in the formal canon.
supports: The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends: The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses: The paper appears to rely on the theorem as machinery.
contradicts: The paper's claim conflicts with a theorem or certificate in the canon.
unclear: Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

11 extracted references · 11 canonical work pages

[1]

X.; Della Pia, F.; Al-Hamdani, Y

Shi, B. X.; Della Pia, F.; Al-Hamdani, Y. S.; Michaelides, A.; Alf` e, D.; Zen, A. System- atic discrepancies between reference methods for noncovalent interactions within the S66 dataset.J. Chem. Phys.2025,162, 144107

work page 2025
[2]

Noncovalent Interactions by Fixed-Node Diffusion Monte Carlo: Conver- gence of Nodes and Energy Differences vs Gaussian Basis-Set Size.J

Dubeck´ y, M. Noncovalent Interactions by Fixed-Node Diffusion Monte Carlo: Conver- gence of Nodes and Energy Differences vs Gaussian Basis-Set Size.J. Chem. Theory Comput.2017,13, 3626–3635

work page 2017
[3]

X.; Alf` e, D.; Zen, A

Nakano, K.; Shi, B. X.; Alf` e, D.; Zen, A. Basis Set Incompleteness Errors in Fixed- Node Diffusion Monte Carlo Calculations on Noncovalent Interactions.J. Chem. Theory Comput.2025,21, 4426–4434

work page 2025
[4]

C.; Melton, C

Bennett, M. C.; Melton, C. A.; Annaberdiyev, A.; Wang, G.; Shulenburger, L.; Mitas, L. A new generation of effective core potentials for correlated calculations.J. Chem. Phys. 2017,147, 224106

work page 2017
[5]

G.; Michaelides, A.; Alf` e, D

Zen, A.; Brandenburg, J. G.; Michaelides, A.; Alf` e, D. A new scheme for fixed node diffusion quantum Monte Carlo with pseudopotentials: Improving reproducibility and reducing the trial-wave-function bias.J. Chem. Phys.2019,151, 134105

work page 2019
[6]

TurboRVB: A many-body toolkit for ab initio electronic simulations by quantum Monte Carlo.J

Nakano, K.; Attaccalite, C.; Barborini, M.; Capriotti, L.; Casula, M.; Coccia, E.; Dagrada, M.; Genovese, C.; Luo, Y.; Mazzola, G.; Zen, A.; Sorella, S. TurboRVB: A many-body toolkit for ab initio electronic simulations by quantum Monte Carlo.J. Chem. Phys.2020,152, 204121

work page 2020
[7]

Energy-consistent pseudopotentials for quantum Monte Carlo calculations.J

Burkatzki, M.; Filippi, C.; Dolg, M. Energy-consistent pseudopotentials for quantum Monte Carlo calculations.J. Chem. Phys.2007,126, 234105

work page 2007
[8]

Beyond the locality approximation in the standard diffusion Monte Carlo method.Phys

Casula, M. Beyond the locality approximation in the standard diffusion Monte Carlo method.Phys. Rev. B2006,74, 161102. s8

work page
[9]

K.; Bajdich, M.; Mitas, L

Wagner, L. K.; Bajdich, M.; Mitas, L. QWalk: A quantum Monte Carlo program for electronic structure.J. Comput. Phys.2009,228, 3390–3404

work page 2009
[10]

R.; Needs, R

Trail, J. R.; Needs, R. J. Shape and energy consistent pseudopotentials for correlated electron systems.The Journal of Chemical Physics2017,146, 204107

work page
[11]

J.; Towler, M

Needs, R. J.; Towler, M. D.; Drummond, N. D.; L´ opez R´ ıos, P.; Trail, J. R. Variational and diffusion quantum Monte Carlo calculations with the CASINO code.J. Chem. Phys. 2020,152, 154106. s9

work page 2020

[1] [1]

X.; Della Pia, F.; Al-Hamdani, Y

Shi, B. X.; Della Pia, F.; Al-Hamdani, Y. S.; Michaelides, A.; Alf` e, D.; Zen, A. System- atic discrepancies between reference methods for noncovalent interactions within the S66 dataset.J. Chem. Phys.2025,162, 144107

work page 2025

[2] [2]

Noncovalent Interactions by Fixed-Node Diffusion Monte Carlo: Conver- gence of Nodes and Energy Differences vs Gaussian Basis-Set Size.J

Dubeck´ y, M. Noncovalent Interactions by Fixed-Node Diffusion Monte Carlo: Conver- gence of Nodes and Energy Differences vs Gaussian Basis-Set Size.J. Chem. Theory Comput.2017,13, 3626–3635

work page 2017

[3] [3]

X.; Alf` e, D.; Zen, A

Nakano, K.; Shi, B. X.; Alf` e, D.; Zen, A. Basis Set Incompleteness Errors in Fixed- Node Diffusion Monte Carlo Calculations on Noncovalent Interactions.J. Chem. Theory Comput.2025,21, 4426–4434

work page 2025

[4] [4]

C.; Melton, C

Bennett, M. C.; Melton, C. A.; Annaberdiyev, A.; Wang, G.; Shulenburger, L.; Mitas, L. A new generation of effective core potentials for correlated calculations.J. Chem. Phys. 2017,147, 224106

work page 2017

[5] [5]

G.; Michaelides, A.; Alf` e, D

Zen, A.; Brandenburg, J. G.; Michaelides, A.; Alf` e, D. A new scheme for fixed node diffusion quantum Monte Carlo with pseudopotentials: Improving reproducibility and reducing the trial-wave-function bias.J. Chem. Phys.2019,151, 134105

work page 2019

[6] [6]

TurboRVB: A many-body toolkit for ab initio electronic simulations by quantum Monte Carlo.J

Nakano, K.; Attaccalite, C.; Barborini, M.; Capriotti, L.; Casula, M.; Coccia, E.; Dagrada, M.; Genovese, C.; Luo, Y.; Mazzola, G.; Zen, A.; Sorella, S. TurboRVB: A many-body toolkit for ab initio electronic simulations by quantum Monte Carlo.J. Chem. Phys.2020,152, 204121

work page 2020

[7] [7]

Energy-consistent pseudopotentials for quantum Monte Carlo calculations.J

Burkatzki, M.; Filippi, C.; Dolg, M. Energy-consistent pseudopotentials for quantum Monte Carlo calculations.J. Chem. Phys.2007,126, 234105

work page 2007

[8] [8]

Beyond the locality approximation in the standard diffusion Monte Carlo method.Phys

Casula, M. Beyond the locality approximation in the standard diffusion Monte Carlo method.Phys. Rev. B2006,74, 161102. s8

work page

[9] [9]

K.; Bajdich, M.; Mitas, L

Wagner, L. K.; Bajdich, M.; Mitas, L. QWalk: A quantum Monte Carlo program for electronic structure.J. Comput. Phys.2009,228, 3390–3404

work page 2009

[10] [10]

R.; Needs, R

Trail, J. R.; Needs, R. J. Shape and energy consistent pseudopotentials for correlated electron systems.The Journal of Chemical Physics2017,146, 204107

work page

[11] [11]

J.; Towler, M

Needs, R. J.; Towler, M. D.; Drummond, N. D.; L´ opez R´ ıos, P.; Trail, J. R. Variational and diffusion quantum Monte Carlo calculations with the CASINO code.J. Chem. Phys. 2020,152, 154106. s9

work page 2020