Papers
Topics
Authors
Recent
Search
2000 character limit reached

Totik-Ditzian Modulus of Continuity

Updated 2 February 2026
  • Totik-Ditzian modulus of continuity is a quantitative measure that adapts step-sizes to match domain geometry and weight structures for optimal error bounds.
  • It underpins direct and inverse approximation theorems by linking the modulus with weighted Peetre K-functionals to precisely characterize function smoothness.
  • It is applied in estimating errors for positive linear operators, refining classical estimates while effectively handling endpoint singularities and unbounded domains.

The Totik-Ditzian-type modulus of continuity (commonly referred to as the Ditzian–Totik modulus of smoothness or continuity) is a central tool in quantitative approximation theory for positive linear operators, especially when working on finite or infinite intervals and when capturing fine local regularity in the presence of endpoint singularities or weights. Invented by Z. Ditzian and V. Totik, it generalizes the classical modulus by introducing a variable step-size function that adapts its scale to the geometry or weight structure of the domain, producing estimates that are optimal in both local and global senses for the order of convergence, characterization of function spaces, and direct/inverse approximation theorems.

1. Formal Definitions and Step-Weight Schemes

For r1r\geq1 and an admissible step-weight ϕ ⁣:I[0,)\phi\colon I\to[0,\infty) on a domain IRI\subset\mathbb R, the rrth-order Ditzian–Totik modulus of smoothness for fC(I)f\in C(I) is defined as

ωϕr(f,t)=sup0<htsupx,khϕ(x)IΔhϕrf(x),\omega_\phi^r(f,t) = \sup_{0<h\le t} \sup_{x,k\,h\,\phi(x)\in I} \left| \Delta_{h\phi}^r f(x) \right|,

where

Δhϕrf(x)=k=0r(1)rk(rk)f(x+khϕ(x)).\Delta_{h\phi}^r f(x) = \sum_{k=0}^r (-1)^{r-k} \binom{r}{k}\, f(x + k\,h\,\phi(x)).

For the classical case on [0,1][0,1] one sets ϕ(x)=x(1x)\phi(x)=\sqrt{x(1-x)}; for unbounded domains like [0,)[0,\infty) one employs ϕ(x)=x\phi(x)=\sqrt{x} or ϕ(x)=x(1+x)\phi(x)=\sqrt{x(1+x)} (Rao et al., 2015, Deo et al., 4 Mar 2025, Pratap et al., 2018, Yadav et al., 2019, Wafi et al., 2015).

The first-order modulus is particularly common,

ωϕ(f,t)=sup0<ht{f(x+hϕ(x)2)f(xhϕ(x)2):x±hϕ(x)2I},\omega_{\phi}(f,t) = \sup_{0<h\le t} \left\{\big|f(x+\tfrac{h\phi(x)}{2}) - f(x-\tfrac{h\phi(x)}{2})\big| : x \pm \tfrac{h\phi(x)}{2} \in I \right\},

with analogous forms on [0,1][0,1], [0,)[0,\infty), or weighted spaces.

2. Equivalence with Weighted Peetre K-Functionals

Central to Totik–Ditzian theory is the tight equivalence between their modulus of smoothness and an appropriately weighted Peetre KK-functional,

Kr,ϕ(f,t)=infgWr(I){fg+trϕrg(r)},K_{r,\phi}(f, t) = \inf_{g\in W^r(I)} \big\{ \|f-g\| + t^r \|\phi^r g^{(r)}\| \big\},

where Wr(I)W^r(I) comprises functions having sufficient smoothness and whose weighted derivatives are bounded. The equivalence,

C1Kr,ϕ(f,t)ωϕr(f,t)C2Kr,ϕ(f,t),C_1\, K_{r,\phi}(f,t)\leq \omega_\phi^r(f,t)\leq C_2\,K_{r,\phi}(f,t),

with C1,C2>0C_1, C_2 > 0 independent of ff and tt, allows the modulus and KK-functional to be used interchangeably for direct/inverse theorems and regularity classification (Kajla et al., 2018, Pratap et al., 2018, Yadav et al., 2019, Rao et al., 2015). For weighted approximation, the KK-functional is often adapted further with a weight w(x)w(x), as in

Kr,ϕ(f,t)w=infg{w(fg)+trwϕrg(r)},K_{r,\phi}(f, t)_w = \inf_{g} \left\{ \|w(f-g)\| + t^r \|w\phi^r g^{(r)}\| \right\},

which is fundamental for endpoint singularities (Lu et al., 2010).

3. Application to Operator Error Estimates: Direct and Inverse Theorems

Most refined error estimates for positive approximation operators (Bernstein, Baskakov, Szász–Mirakjan, Durrmeyer variants, Bézier variants, etc.) are expressible in Totik–Ditzian terms: Ln(f;x)f(x)Cωϕr(f;γn(x)),|L_n(f;x) - f(x)| \leq C\, \omega_\phi^r \left( f; \gamma_n(x)\right), where γn(x)\gamma_n(x) reflects a local scale determined by moments or the kernel structure, often γn(x)n1/rϕ(x)α\gamma_n(x)\sim n^{-1/r}\phi(x)^\alpha for some α\alpha; this sharpens the error estimate compared to classical moduli, particularly near endpoints (Kajla et al., 2018, Deo et al., 4 Mar 2025, Rao et al., 2015, Singh et al., 2020). For example, for Bézier-variants of Bernstein–Durrmeyer operators on [0,1][0,1],

D~n,M(2),p(f;x)f(x)Cωφ2(f,1n+2),|\widetilde D_{n,M}^{(2),p}(f;x)-f(x)| \leq C\, \omega_\varphi^2 \left(f, \frac{1}{\sqrt{n+2}}\right),

with φ(x)=x(1x)\varphi(x)=\sqrt{x(1-x)} (Singh et al., 2020).

The inverse theorem typically has the contrapositive form: if the error Ln(f;x)f(x)|L_n(f;x)-f(x)| decays sufficiently fast (e.g., O(nλ)O(n^{-\lambda})), then ff necessarily lies in the Lipschitz–Zygmund or weighted space determined by ωϕr(f,t)=O(tλ)\omega_\phi^r(f,t) = O(t^\lambda) (Lu et al., 2010).

4. Modulus of Continuity for Endpoint Singularities and Weighted Spaces

Totik–Ditzian moduli natively accommodate singularities at endpoints through the step-weight ϕ\phi and auxiliary weight w(x)w(x),

ωϕr(f,t)w:=supwΔhϕrf(x)I,\omega_\phi^r(f,t)_w := \sup_{\cdots} \|w \Delta_{h\phi}^r f(x)\|_{I^\prime},

where II^\prime may be an interior or boundary strip and Δhϕr\Delta_{h\phi}^r transitions between weighted and ordinary differences as xx approaches the endpoint. This mechanism allows functions with f(x)xαf(x)\sim x^{-\alpha} or f(x)(1x)βf(x)\sim (1-x)^{-\beta}, provided w(x)w(x) absorbs the singularity (Lu et al., 2010).

5. Specialized Moduli and Adaptations to Unbounded Domains

On [0,)[0,\infty), the choice ϕ(x)=x(1+x)\phi(x)=\sqrt{x(1+x)} or ϕ(x)=x\phi(x)=\sqrt{x} matches the scaling of kernel variances for Szász, Baskakov, or Kantorovich–Charlier operators, yielding

ωϕ2(f;t)=sup0<htsupx±hϕ(x)[0,)f(x+hϕ(x))2f(x)+f(xhϕ(x)).\omega_\phi^2(f; t) = \sup_{0<h\le t} \sup_{x\pm h\phi(x)\in [0,\infty)} |f(x+h\phi(x))-2f(x)+f(x-h\phi(x))|.

For the Bézier–Gupta–Srivastava family one obtains

Lnm,c,a(f;x)f(x)Cωϕβ(f,Mn,2(x)),ϕ(x)=x(1+cx),|L_n^{m,c,a}(f;x)-f(x)| \leq C \omega_\phi^{\beta}\left(f, \sqrt{M_{n,2}(x)}\right), \qquad \phi(x)=\sqrt{x(1+cx)},

with the second central moment Mn,2(x)x(1+cx)/nM_{n,2}(x)\sim x(1+cx)/n providing spatially adaptive rates (Pratap et al., 2018, Deo et al., 4 Mar 2025, Yadav et al., 2019).

6. Quantitative Voronovskaja-Type Theorems

Totik–Ditzian moduli underpin generalized Voronovskaja results, which characterize the main term in the asymptotic expansion and provide error bounds in the same scale. For generalized Bernstein–Durrmeyer operators,

limnn[Gn,ρ(α)(f;x)f(x)]=12xρf(x)+(1+ρ)x(1x)2ρf(x),\lim_{n\to\infty} n\left[\mathcal G_{n,\rho}^{(\alpha)}(f;x)-f(x)\right] = \frac{1-2x}{\rho}f'(x) + \frac{(1+\rho)x(1-x)}{2\rho}f''(x),

with remainder decay governed by ωϕ2(f;n1/2)\omega_\phi^2(f; n^{-1/2}) (Kajla et al., 2018). For modified Szász–Mirakjan–Kantorovich or (p,q)-Bernstein–Stancu operators, analogous expansions and quantitative remainders in DT-modulus scale are derived (Yadav et al., 2019, Khan et al., 2016).

7. Numerical Implications and Extensions

Maple or other computational illustrations confirm that operators designed to fit Totik–Ditzian estimates converge locallly at rates predicted by the modulus. Pointwise error curves are sharper and more uniform when measured in the DT-modulus rather than the classical modulus, especially for functions of bounded variation or with endpoint behavior (Kajla et al., 2018, Singh et al., 2020). The flexibility in step-weight and order allows the theory to be extended to weighted Lipschitz-type spaces, almost-Lipschitz classes, and function spaces characterized by smoothness in the DT sense (Deo et al., 4 Mar 2025, Lu et al., 2010).

Table: Canonical Step-Weights in Totik–Ditzian Modulus

Domain Step-Weight ϕ(x)\phi(x) Operator Class
[0,1][0,1] x(1x)\sqrt{x(1-x)} Bernstein, Bernstein–Durrmeyer
[0,)[0,\infty) x\sqrt{x}, x(1+x)\sqrt{x(1+x)} Szász–Mirakjan, Baskakov, Charlier

These choices ensure that the modulus adapts to the intrinsic scaling of the kernels, leading to optimal direct and inverse approximation theorems for the relevant operator families (Rao et al., 2015, Wafi et al., 2015, Yadav et al., 2019, Deo et al., 4 Mar 2025).

References

Papers referenced in the above exposition include:

  • "Generalized Bernstein-Durrmeyer Operators of Blending Type" (Kajla et al., 2018)
  • "Durrmeyer type operators linked with Boas-Buck type polynomials" (Deo et al., 4 Mar 2025)
  • "Stancu-variant of generalized Baskakov operators" (Rao et al., 2015)
  • "Statistical approximation by (p,q)(p,q)-analogue of Bernstein-Stancu Operators" (Khan et al., 2016)
  • "Optimal bounds on the modulus of continuity of the uncentered Hardy-Littlewood maximal function" (Aldaz et al., 2010)
  • "Approximation properties by some modified Szasz-Mirakjan-Kantorovich operators" (Yadav et al., 2019)
  • "Kantorovich form of generalized Szasz-type operators with certain parameters using Charlier polynomials" (Wafi et al., 2015)
  • "Direct and Inverse Estimates for Combinations of Bernstein Polynomials with Endpoint Singularities" (Lu et al., 2010)
  • "Bézier Variant of generalized Bernstein-Durrmeyer type operators" (Singh et al., 2020)
  • "Rate of convergence of Gupta-Srivastava operators based on certain parameters" (Pratap et al., 2018)

The Totik-Ditzian modulus remains indispensable as the canonical smoothness measure for modern approximation theory, providing the exact framework for optimal convergence rates, regularity characterizations, and analysis of positive linear operators on diverse domains and function spaces.

Topic to Video (Beta)

No one has generated a video about this topic yet.

Whiteboard

No one has generated a whiteboard explanation for this topic yet.

Follow Topic

Get notified by email when new papers are published related to Totik-Ditzian-Type Modulus of Continuity.