Hermite spectral approximation for functions with endpoint singularities using exponential transforms

Haiyong Wang

Hermite spectral approximation with exponential maps turns endpoint singularities into root-exponential convergence, further sharpened by optimal scaling.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · grok-4.5

2026-07-15 04:26 UTC pith:NGG44SS4

load-bearing objection Abstract-only Hermite + SE/DE/EF maps for endpoint singularities: plausible useful NA contribution, but rates and sinc comparison cannot be checked yet. the 2 major comments →

arxiv 2607.12648 v1 pith:NGG44SS4 submitted 2026-07-14 math.NA cs.NA

Hermite spectral approximation for functions with endpoint singularities using exponential transforms

Haiyong Wang This is my paper

classification math.NA cs.NA MSC 65N3565D1541A25

keywords Hermite spectral approximationendpoint singularitiessingle exponential transformdouble exponential transformerror function transformoptimal scalingroot-exponential convergencesinc method

verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

The paper shows that Hermite spectral approximation, after a single-exponential, double-exponential or error-function change of variable, can approximate functions that blow up or lose smoothness only at the endpoints of an interval. Without any extra scaling the resulting series still converges at a root-exponential rate; with carefully chosen scaling factors the same series becomes markedly more accurate for a given number of terms. The author supplies explicit optimal scalings for each of the three transforms and proves the corresponding rates. Side-by-side numerical tests indicate that the scaled Hermite expansions match or beat the classical sinc method when both are truncated after the same number of terms. The same machinery is then reused for numerical quadrature and for a spectral root-finding algorithm on intervals with endpoint singularities.

Core claim

Hermite spectral approximation composed with an exponential (SE, DE or EF) transform converges root-exponentially for functions with endpoint singularities; the rate is further improved by optimal scaling factors derived for each transform, yielding accuracy that is comparable or superior to the sinc method at equal term count.

What carries the argument

The composition of a Hermite spectral expansion with an exponential change of variable (SE, DE or EF), together with the optimal scaling parameters that minimise the asymptotic size of the Hermite coefficients after the map.

Load-bearing premise

The asymptotic analysis of Hermite coefficients after the exponential map, and the optimality of the derived scalings, must hold for the precise singularity classes that arise in the target applications.

What would settle it

Compute the scaled Hermite coefficients for a model function with a known endpoint singularity (for example (1-x)^α on [-1,1]) and check whether the observed decay rate matches the predicted root-exponential rate under the claimed optimal scaling; any systematic slower decay falsifies the analysis.

Watch this falsifier — get emailed when new claim-graph text bears on it.

Editorial analysis

A structured set of objections, weighed in public.

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

Desk Editor's Note

Abstract-only Hermite + SE/DE/EF maps for endpoint singularities: plausible useful NA contribution, but rates and sinc comparison cannot be checked yet.

read the letter

This is an abstract-only look at a math.NA paper, so treat everything below as provisional. The punchline is that Wang proposes Hermite spectral approximation after SE, DE, and EF exponential maps for endpoint singularities, claims root-exponential rates without scaling, derives optimal scaling factors that improve the rate, and reports accuracy comparable or better than sinc at equal term count, with brief extensions to quadrature and rootfinding.

What looks new is the combination plus the claimed full convergence analysis and optimal scalings for Hermite (not just another mapped spectral method). Hermite and these transforms are each standard; packaging them for endpoint singularities with explicit scaling theory and a head-to-head against sinc is a solid incremental contribution if the proofs hold. The abstract is clear about the claims and does not read like circular fitting: scalings are described as derived, and the benchmarks are external (sinc, known singularity classes).

The soft spots are almost entirely information limits, not identified technical holes. We cannot see the function-class hypotheses, the asymptotic analysis of the mapped Hermite coefficients, the error bounds, or the numerical tables. So the root-exponential rates, optimality of the scalings, and the sinc comparison remain unverified. The weakest assumption is exactly the one the reader flagged: that the singularity structures covered by the analysis match the ones people actually care about, and that the asymptotics survive those maps. That is not a manufactured flaw; it is the usual abstract-only gap. No free parameters or invented entities show up in the abstract.

Who this is for: people who already work on spectral methods for singular integrands, mapped bases, or high-order quadrature/rootfinding. A serious referee in approximation theory or numerical analysis should see the full paper. I would not cite it from the abstract alone, and I would not put it in next week’s reading group without the PDF, but I would accept it for peer review rather than desk-reject. If the theorems and tables are clean, it is a useful competitive option next to sinc; if not, the referee will say so. Send it on.

Referee Report

2 major / 2 minor

Summary. The manuscript (available here only as an abstract) proposes Hermite spectral approximation for functions with endpoint singularities, combined with single-exponential (SE), double-exponential (DE), and error-function (EF) transforms. It claims a comprehensive convergence analysis: without scaling the methods converge at a root-exponential rate; with scaling, optimal scaling factors are derived for each transform and are said to improve the rate substantially. Numerical comparisons with the sinc method are asserted to show comparable or superior accuracy at equal term count. Extensions to quadrature and rootfinding are also indicated.

Significance. If the stated rates, the derivation of optimal scaling factors, and the competitive performance versus sinc are rigorously established under clearly delineated function classes, the work would supply a useful Hermite-based alternative for endpoint singularities, with practical carry-over to quadrature and rootfinding. Explicit, derived (rather than purely fitted) scaling factors and a systematic comparison to sinc would be genuine contributions to constructive approximation for singular integrands and related spectral methods.

major comments (2)

Only the abstract is available for review. The central claims—root-exponential rates without scaling, existence and optimality of scaling factors that improve those rates, and comparable/superior accuracy versus sinc at equal term count—rest on theorems, asymptotic analysis of Hermite coefficients after the SE/DE/EF maps, and numerical tables that are not present. Without those materials the load-bearing claims cannot be checked for correctness, scope of hypotheses, or numerical support.
Abstract: the weakest load-bearing assumption is that the function classes and singularity structures under which the optimal scaling factors are derived (and under which the root-exponential rates are proved) actually cover the endpoint singularities of practical interest, and that the asymptotic analysis of the Hermite coefficients after the exponential maps remains valid for those classes. The abstract does not state the precise hypotheses, so coverage and validity cannot be assessed.

minor comments (2)

Abstract wording: “Numerical comparisons with sinc method are present” should read “are presented”; “Hermite method has comparable or superior accuracy performance” is slightly awkward and could be tightened once the full text is available.
Abstract: the three transforms (SE, DE, EF) and the two regimes (with/without scaling) should be tied to explicit theorem or section numbers in the full manuscript so that readers can locate the corresponding error bounds and scaling formulae.

Circularity Check

0 steps flagged

No significant circularity identifiable from the abstract; claims are standard constructive approximation analysis against external benchmarks.

full rationale

Only the abstract is available, so no derivation chain, equations, or self-citations can be inspected for reduction-by-construction. From the abstract alone the work presents Hermite spectral approximation with SE/DE/EF exponential maps, proves root-exponential rates without scaling, derives optimal scaling factors that improve the rate, and compares accuracy to the sinc method at equal term count. These are framed as analytic convergence results and numerical benchmarks, not as fitted parameters renamed as predictions, self-definitional identities, or uniqueness theorems imported from the same authors. External comparison to sinc and the stated existence of derived (not data-fitted) scaling factors are consistent with a self-contained approximation-theory paper. Per the hard rules, circularity may be claimed only when a specific quote exhibits Eq. X = Eq. Y by construction or a fitted input called a prediction; no such quote exists here. Residual risk is the usual abstract-only incompleteness (function-class coverage for the asymptotics), which is a correctness/scope concern, not circularity. Score 0 with empty steps is therefore the warranted finding.

Axiom & Free-Parameter Ledger

0 free parameters · 3 axioms · 0 invented entities

Abstract-only review: free parameters and axioms must be inferred from stated method ingredients. Scaling factors are claimed to be optimally derived (not free fits), so they are not listed as free parameters. Background assumptions are standard approximation-theory hypotheses on singularity structure and transform regularity that any such paper must invoke; invented entities are none.

axioms (3)

domain assumption Target functions belong to classes with endpoint singularities that become sufficiently regular after SE, DE, or EF transformation so that Hermite coefficients decay at the claimed rates.
Required for any root-exponential or improved rate; abstract asserts comprehensive analysis without listing the precise function spaces.
standard math Standard properties of Hermite functions (orthogonality, asymptotic decay of coefficients for analytic/entire functions of suitable growth) hold after the exponential change of variable.
Classical spectral-theory background used by any Hermite spectral convergence proof.
domain assumption The derived scaling factors are asymptotically optimal for the stated transforms and singularity classes.
Abstract claims optimality; correctness of that claim is load-bearing for the 'significantly improved' rate statement.

pith-pipeline@v1.1.0-grok45 · 6004 in / 2156 out tokens · 20214 ms · 2026-07-15T04:26:02.759921+00:00 · methodology

0 comments

read the original abstract

In this paper we introduce Hermite spectral approximation for functions with endpoint singularities using exponential transforms, including single exponential (SE), double exponential (DE) and error function (EF) transforms, and present a comprehensive convergence analysis for these approximations without and with scaling. In the case without scaling, we show that these methods converge at some root-exponential rate. In the case with scaling, we derive optimal scaling factors for each of exponential transforms and show that the convergence rate of Hermite spectral approximation can be significantly improved. Numerical comparisons with sinc method are present and it is shown that Hermite method has comparable or superior accuracy performance when using the same number of terms. Extensions to quadrature and rootfinding algorithm are also discussed.

Hermite spectral approximation for functions with endpoint singularities using exponential transforms

Core claim

What carries the argument

Load-bearing premise

What would settle it

discussion (0)