pith. sign in

arxiv: 2605.04765 · v2 · pith:FN4ON4ACnew · submitted 2026-05-06 · 🧮 math.NA · cs.NA

A Generalized FC-Gram Approximation Framework with Analysis and Applications

Prakash Nainwal , Akash Anand This is my paper

Pith reviewed 2026-05-08 15:47 UTC · model grok-4.3

classification 🧮 math.NA cs.NA
keywords FC-Gram algorithmGram polynomialsperiodic extensiontrigonometric interpolationconvergence ratesnon-periodic functionsFourier continuationboundary value problems
0
0 comments X

The pith

A generalized FC-Gram framework adds flexibility to Gram polynomial blending and proves convergence rates of O(n to the minus min of r plus beta and d) for non-periodic functions.

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

The paper develops a generalized FC-Gram framework, GenFC, that increases flexibility when constructing the blending continuation of Gram polynomials for periodic extensions of non-periodic functions. This flexibility improves control over the shape of the extension and raises the accuracy of the trigonometric interpolant. The authors prove a convergence theorem establishing that the supremum-norm error on the original interval decays as n to the power of minus the minimum of the smoothness parameter r plus a Fourier-decay parameter beta and the number of Gram polynomials d. The analysis is constructed so that an earlier modified FC-Gram method appears as one special case. Numerical tests confirm the rates and show that the extra flexibility produces smaller errors, with the gains appearing also when the method is embedded in Fourier solvers for two-point boundary value problems.

Core claim

The GenFC framework generalizes the construction of blending continuations for Gram polynomials, recovering the modified FC-Gram method as a special case. It establishes that the supremum-norm error of the trigonometric interpolant on the original interval is bounded by a constant times n to the power of minus the minimum of r plus beta and d, where r measures the smoothness of the target function, d is the number of Gram polynomials, and beta is a parameter between 0 and 1 controlling Fourier decay.

What carries the argument

The flexible blending continuation of Gram polynomials in the GenFC framework, which enables adjustable periodic extensions while preserving controlled boundary behavior needed for the convergence proof.

Load-bearing premise

The flexible blending continuation of Gram polynomials can be constructed with the stated flexibility while preserving controlled boundary behavior and without introducing uncontrolled errors that invalidate the convergence analysis.

What would settle it

Compute the observed supremum-norm error decay rate for increasing numbers of points n on a test function with known smoothness r, using chosen values of d and beta; rates falling below the predicted min(r plus beta, d) would falsify the theorem.

Figures

Figures reproduced from arXiv: 2605.04765 by Akash Anand, Prakash Nainwal.

Figure 1
Figure 1. Figure 1: Shape function η R ℓ defined in (4.4) on [1, b] with b = 2 and d = 5. 15 view at source ↗
Figure 2
Figure 2. Figure 2: Log–log convergence plots for the generalized FC–Gram approximation computed with Gram polynomials of degree d = 3, 4, 5 and periodic extension length b = 2. the decay condition (3.18) with r = 0 and β = (α1 − α2)/α2 = 0.7, which yields a predicted theoretical convergence rate of min(β, d) = 0.7 for all d ≥ 1. Despite the challenging oscillatory behavior near the origin, the generalized FC–Gram approximati… view at source ↗
Figure 4
Figure 4. Figure 4: Log–log convergence plots for the generalized FC–Gram approximation computed with Gram polynomials of degree d = 4 and periodic extension length b = 2, 1.5, 1.25. We also verify that the asymptotic rate is robust to the choice of b. Fixing d = 4 and varying b ∈ {2, 3/2, 5/4}, the predicted rates are confirmed for the smooth function and f(x) = |x − 1/2| 7/2 ; see Figures 4a and 4b. For f(x) = x 1.7 sin(1/x… view at source ↗
Figure 3
Figure 3. Figure 3: Generalized FC–Gram approximation of f(x) = x 1.7 sin(1/x) with b = 2. Left: The function f (solid blue) and its approximation τ b nf (dashed black) on [0, 1] for n = 212; the inset magnifies the oscillatory region near x = 0. Right: Log–log convergence plots for d = 3, 4, 5, asymptotically attaining the predicted rate β = 0.7. 5.2 Effect of β In this section we present through an example how improved regu… view at source ↗
Figure 5
Figure 5. Figure 5: Log–log convergence plots for the generalized FC–Gram approximation of f(x) = (1−x) 3+β using Gram polynomials of degree d = 5, periodic extension length b = 2, and varying β. 18 view at source ↗
Figure 6
Figure 6. Figure 6: Left: shape functions η R 4 for ModFC and GenFC with d = 5 and b = 2. Right: the corresponding blending-to-zero continuations p R,e 4 on [1, b], demonstrating the substantial reduction in sup-norm achievable with GenFC. This is illustrated in view at source ↗
Figure 7
Figure 7. Figure 7: Interpolation errors for ModFC and GenFC applied to f(x) = exp(− cos(kx)), with d = 5 and b = 2, for k = 50, 100, 200. Example 2. Consider f(x) = 1/(x+ε) for ε = 0.1, 0.01, 0.001. As ε → 0 +, the function develops a near￾singular behavior at the left boundary, causing the projection coefficients to grow rapidly in magnitude. This puts particular pressure on the blending continuation: ModFC must smoothly ex… view at source ↗
Figure 8
Figure 8. Figure 8: Interpolation errors for ModFC and GenFC applied to f(x) = 1/(x + ε), with d = 5 and b = 2, for ε = 1/10, 1/100, 1/1000. 5.3.2 Two-Point Boundary Value Problems The approximation advantage of GenFC observed in the preceding examples carries over directly to differ￾ential equation solvers. We take b = 2 and d = 5 throughout in this section, and abbreviate τn(f) := τ 2 n (f). We consider the two-point bounda… view at source ↗
read the original abstract

The FC-Gram algorithm constructs high-order trigonometric approximations of nonperiodic functions by periodically extending them to a larger interval, with the quality of the blending continuation of Gram polynomials over the extension interval directly governing the approximation accuracy. We introduce GenFC, a generalized FC-Gram framework in which the continuation of each Gram polynomial is shaped by a cutoff function satisfying prescribed boundary flatness conditions. We establish a convergence theorem showing that for any such family the GenFC approximation error satisfies $O(n^{-\min(r+\beta,\,d)})$ in the supremum norm on the original interval, where $f \in C^r([0,1])$ has an integrable $(r+1)$th derivative, $d$ is the number of Gram polynomials, and $\beta \in [0,1]$ is the Fourier decay exponent of $f^{(r+1)}$. The modified FC-Gram algorithm, recently introduced by the authors, is recovered as a special case, and several explicit families satisfying these conditions are constructed in the paper. Numerical experiments across smooth, limited-regularity, and rapidly oscillating test cases confirm the theoretical predictions. The framework is further applied to high-order solvers for linear ODEs and parabolic PDEs via backward differentiation formulae (BDF) time-stepping, demonstrating high-order accuracy throughout.

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.

Referee Report

2 major / 2 minor

Summary. The paper introduces the GenFC framework as a generalization of the FC-Gram algorithm, providing greater flexibility in the blending continuation of Gram polynomials to construct periodic extensions of non-periodic functions. It establishes a convergence theorem stating that the trigonometric interpolant converges at the rate O(n^{-min(r+β,d)}) in the supremum norm on the original interval, recovers the modified FC-Gram method of [J. Sci. Comput., 105(1):8, 2025] as a special case, and reports numerical experiments confirming the predicted rates along with improved accuracy that extends to a Fourier continuation solver for two-point boundary value problems.

Significance. If the convergence theorem holds with the generalized blending, the work strengthens the FC-Gram approach by adding tunable flexibility that demonstrably improves practical accuracy while recovering prior results as a special case. The numerical validation of rates and the BVP application are positive contributions to high-order approximation techniques for non-periodic data. The framework's design for controlled boundary behavior is a strength, but significance depends on rigorous bounding of any additional errors from the new blending rules.

major comments (2)
  1. [Convergence theorem] Convergence theorem (abstract and analysis section): the claimed rate O(n^{-min(r+β,d)}) in the sup norm requires that the generalized blending continuation preserves the Fourier-decay bound β without introducing remainder terms that degrade the effective decay or boundary control. The manuscript must supply an explicit error estimate for the generalized blending step (distinct from the special case of modified FC-Gram) to confirm the analysis is not invalidated by perturbations to matching conditions or high-frequency content.
  2. [Numerical experiments] Numerical experiments section: the reported confirmation of rates and accuracy gains lacks error-bar information, explicit data-exclusion rules, and details on how β is chosen per function or blending rule. This leaves the practical performance claims plausible but incompletely verified, directly affecting the strength of the cross-period and application claims.
minor comments (2)
  1. [Framework definition] Clarify the precise definition and selection procedure for the free parameters β and d in the GenFC framework to avoid any appearance of circularity in the rate statement.
  2. [Figures] Ensure all figures include axis labels, legends, and captions that explicitly reference the GenFC variants tested.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the careful reading of our manuscript and the constructive comments on the GenFC framework. We address each major comment below and outline the revisions that will be incorporated to strengthen the presentation and verification of our results.

read point-by-point responses

  1. Referee: [Convergence theorem] Convergence theorem (abstract and analysis section): the claimed rate O(n^{-min(r+β,d)}) in the sup norm requires that the generalized blending continuation preserves the Fourier-decay bound β without introducing remainder terms that degrade the effective decay or boundary control. The manuscript must supply an explicit error estimate for the generalized blending step (distinct from the special case of modified FC-Gram) to confirm the analysis is not invalidated by perturbations to matching conditions or high-frequency content.

    Authors: We appreciate the referee's emphasis on rigor for the generalized case. The convergence analysis in Section 3 is formulated to hold for arbitrary blending parameters that satisfy the boundary-matching conditions of the GenFC construction, with the modified FC-Gram recovered when the blending coefficients reduce to the specific choice in the cited reference. The proof proceeds by bounding the Fourier coefficients of the periodic extension directly from the smoothness and the controlled boundary behavior, without additional remainder terms that would alter the decay exponent β. To make this explicit and address the concern about potential perturbations, we will insert a new lemma (Lemma 3.2) that isolates the error contribution of the generalized blending step and shows that it is absorbed into the existing O(n^{-min(r+β,d)}) bound without degrading the rate or boundary control. revision: yes

  2. Referee: [Numerical experiments] Numerical experiments section: the reported confirmation of rates and accuracy gains lacks error-bar information, explicit data-exclusion rules, and details on how β is chosen per function or blending rule. This leaves the practical performance claims plausible but incompletely verified, directly affecting the strength of the cross-period and application claims.

    Authors: We agree that greater transparency in the numerical section will improve reproducibility and strengthen the validation. In the revised manuscript we will augment the experiments section with: (i) error bars on all convergence plots, computed from repeated runs with random perturbations to the blending parameters within the admissible range; (ii) an explicit statement of the data-exclusion criteria (e.g., exclusion of points within 10^{-12} of machine epsilon); and (iii) a table or paragraph detailing the selection rule for β for each test function and each blending family, together with the concrete values used. These additions will directly support the reported accuracy gains and their extension to the BVP solver. revision: yes

Circularity Check

0 steps flagged

No significant circularity; convergence theorem stated as independently established

full rationale

The paper introduces GenFC with added flexibility in blending Gram polynomials and states that a convergence theorem is established for the rate O(n^{-min(r+β,d)}) in the sup norm. It notes that the prior modified FC-Gram is recovered as a special case, but provides no equations or steps showing that the general theorem reduces by construction to the special case or to fitted parameters. The Fourier-decay parameter β is introduced as part of the general statement rather than derived from data fitting within the same derivation. No self-definitional loops, renamed predictions, or load-bearing self-citations that collapse the central claim are exhibited. The analysis is presented as self-contained against the stated assumptions on boundary behavior and extension errors.

Axiom & Free-Parameter Ledger

2 free parameters · 1 axioms · 1 invented entities

The central claims rest on standard properties of Gram polynomials and trigonometric interpolation together with the new definition of flexible blending; β and d function as tunable parameters whose values affect the stated rate.

free parameters (2)
  • β
    Fourier-decay parameter in [0,1] that enters the convergence exponent and is selected according to the target function or blending choice.
  • d
    Number of Gram polynomials used in the blending continuation; controls both the rate and the computational cost.
axioms (1)
  • standard math Gram polynomials and trigonometric interpolation satisfy the standard approximation and periodicity properties used in the original FC-Gram construction.
    Invoked throughout the definition of the periodic extension and the proof of the convergence theorem.
invented entities (1)
  • GenFC framework no independent evidence
    purpose: Generalized construction of the blending continuation of Gram polynomials with tunable flexibility.
    New object introduced by the paper; no independent external evidence supplied beyond the numerical tests reported here.

pith-pipeline@v0.9.0 · 5508 in / 1489 out tokens · 51991 ms · 2026-05-08T15:47:49.606749+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.