Papers
Topics
Authors
Recent
2000 character limit reached

Miyachi–Peral Fixed-Time Estimates

Updated 21 December 2025
  • Miyachi–Peral estimates are quantitative bounds that specify the minimal regularity loss needed for the wave propagator on L^p spaces across varied geometric contexts.
  • They leverage Fourier analysis, complex interpolation, and microlocal techniques to rigorously establish sharp spectral multiplier theorems.
  • These estimates underpin applications in well-posedness, endpoint Strichartz analysis, and extend wave multiplier theory to nonelliptic and rough-coefficient operators.

Miyachi-Peral-type fixed-time estimates refer to quantitative regularity bounds for the wave propagator (such as cos(tL)\cos(t\sqrt{L})) acting on LpL^p spaces, quantifying the precise loss of regularity ("smoothing" exponent) necessary for the operator to remain bounded at fixed time. The archetypal context is the study of the wave equation on Rd\mathbb{R}^d or on more general manifolds and differential operators, especially those lacking uniform ellipticity, such as sub-Laplacians on sub-Riemannian spaces or operators with nonsmooth coefficients. These estimates are tightly connected to sharp spectral multiplier theorems, microlocal analysis, and the geometry of the underlying space.

1. Classical Miyachi-Peral Theorem and Sharp Smoothing Indices

The original results of Peral and Miyachi established sharp conditions for the LpL^p boundedness, with loss of derivatives, of the fixed-time wave propagator on Rd\mathbb{R}^d. For A=ΔA=\sqrt{-\Delta}, the estimate

(1+A)aeitAfLpCd,p,atafLp\|(1 + A)^{-a} e^{itA} f\|_{L^p} \leq C_{d,p,a}\langle t\rangle^{a} \|f\|_{L^p}

holds if and only if asp:=(d1)1/p1/2a \geq s_p := (d-1)|1/p - 1/2|, for 1<p<1 < p < \infty (Frey et al., 2020, Kriegler, 2012). This threshold is sharp; no better smoothing is admissible. Equivalently, the operator (IΔ)sp/2eitΔ(I-\Delta)^{-s_p/2} e^{it \sqrt{-\Delta}} is bounded on LpL^p at fixed times if and only if spαs_p \leq \alpha.

The necessity is demonstrated by explicit evaluation on plane wave data. Sufficiency hinges on complex interpolation, Fourier integral operator theory, and deep properties of the Fourier transform.

2. Geometric Generalizations: Sub-Laplacians and Sub-Riemannian Settings

For sub-Laplacians L\mathscr L in divergence form on manifolds MnM^n with a bracket-generating Hamiltonian HH, Miyachi-Peral-type bounds hold in a manner exactly analogous to the Euclidean case (Martini et al., 2018). The central result is

(1+L)α/2cos(tL)LpLp(1+t)α    α(n1)1/p1/2\|(1 + \mathscr L)^{-\alpha/2} \cos(t\sqrt{\mathscr L})\|_{L^p \rightarrow L^p} \lesssim (1 + |t|)^\alpha \implies \alpha \geq (n-1)|1/p - 1/2|

uniformly for all 1<p<1 < p < \infty. This remains valid regardless of the step or the Hausdorff dimension QQ, provided the bracket-generating (Hörmander) condition holds.

The proof relies on a Fourier integral operator parametrix for the wave propagator, properties of the sub-Riemannian geodesic flow, and the stationary phase method to extract the optimal lower bound on α\alpha. Nondegeneracy of the sub-Riemannian exponential map is critical, ensuring microlocal control on oscillatory integral kernels. Notably, the threshold for α\alpha does not worsen even when the underlying geometry is highly degenerate or QnQ \gg n—the necessity remains "Euclidean-type".

3. Extensions to Variable Coefficient and Nonsmooth Operators

Frey and Portal (Frey et al., 2020) showed that for operators L=j=1daj+d(xj)j(aj(xj)j)\mathcal{L} = -\sum_{j=1}^d a_{j+d}(x_j) \partial_j (a_j(x_j) \partial_j) with merely C0,1C^{0,1} (Lipschitz) coefficients, Miyachi-Peral-type fixed-time LpL^p estimates still hold, under the same threshold sps_p. Specifically,

(I+L)sp/2eitLfLpCp,d,aj,tfLp\| (I+\mathcal{L})^{-s_p/2} e^{it \sqrt{\mathcal{L}}} f\|_{L^p} \leq C_{p,d,a_j,t} \|f\|_{L^p}

for all 1<p<1<p<\infty, sp=(d1)1/p1/2s_p = (d-1)|1/p-1/2|.

The sufficiency is achieved not by classical parametrix constructions but via adaptations of tent spaces Tp,2(Rd)T^{p,2}(\mathbb{R}^d), Hardy-Sobolev spaces HLp,sH^{p,s}_{\mathcal{L}}, and a wave-packet transform tailored to the Lipschitz structure. Central to the analysis are algebraic factorization properties and boundedness criteria for the associated (now, non-smooth) Fourier integral operators.

This demonstrates that the critical regularity required for boundedness of the wave propagator does not increase even for rough coefficients, provided certain structural and ellipticity conditions are preserved.

4. Function Space Refinements: Hardy, Tent, and DSβ_\beta Spaces

Classical results are formulated in LpL^p and Hardy spaces H1\mathcal{H}^1. Recent developments have extended Miyachi-Peral-type estimates to ββ-dimensional stable measure spaces DSβ(Rn)DS_\beta(\mathbb{R}^n) (Basak et al., 14 Dec 2025). For TbtT_b^t the wave multiplier,

TbtμLpCtn/pμDSβ\| T_b^t \mu \|_{L^p} \leq C t^{-n/p'} \|\mu\|_{DS_\beta}

holds in several regimes, with the threshold bp=(n+1)/21/pb_p = (n+1)/2 - 1/p and suitable conditions on β\beta and pp. For p2p \geq 2, β(0,n]\beta \in (0,n], and for 1p<21 \leq p < 2, β>n1\beta > n-1 (or >2/p>2/p in n=2n=2), the bound is achieved.

The definition of DSβDS_\beta measures involves atomic decompositions with cancellation and heat extension conditions, interpolating between Hardy space (β=n\beta = n) and the full measure space (β0\beta \to 0). Proofs for p<2p<2 exploit refined two-scale decompositions and Sobolev-type L2L^2 control.

This framework allows endpoint estimates for data beyond classical function spaces, refining and extending known Hardy space theory for wave multipliers.

5. Analytic Tools: Paley-Littlewood, Transference, and Functional Calculus

Spectral multiplier results for the wave equation are often established via Paley–Littlewood decompositions, boundedness of analytic families, and transference principles for regularized groups (Kriegler, 2012). For general self-adjoint or $0$-sectorial operators LL that admit a suitable Paley-Littlewood decomposition and satisfy polynomial growth bounds

(1+L)BeitLXXCta\|(1+L)^{-B} e^{itL}\|_{X \to X} \leq C \langle t \rangle^a

the corresponding fixed-time wave multiplier

(1+L)aeitLXXCta\|(1+L)^{-a} e^{itL} \|_{X \to X} \leq C \langle t \rangle^{a}

follows, mirroring the Euclidean case. This framework allows deduction of Miyachi-Peral-type fixed-time estimates in settings as general as Lie groups (e.g., Heisenberg), under off-diagonal heat or Poisson kernel bounds.

Functional calculus and complex interpolation play key roles in transferring L2L2L^2\to L^2 and end-point L1L^1\to BMO or L1LL^1\to L^\infty estimates to the entire LpL^p scale.

6. Comparative Summary and Sharpness of Bounds

A crucial theme across all settings (Euclidean, sub-Riemannian, nonsmooth) is the universality of the lower bound α(n1)1/p1/2\alpha \geq (n-1)|1/p - 1/2| for fixed-time LpL^p wave propagator estimates. This threshold is sharp and dictated by underlying dispersive geometry and the nature of oscillations in the wave kernel. In no generalization considered—whether increasing the step in Carnot groups, reducing regularity of coefficients, or weakening cancellation assumptions on data—is it possible to extend pp or reduce α\alpha beyond the Euclidean threshold (Martini et al., 2018, Frey et al., 2020, Basak et al., 14 Dec 2025).

The table below summarizes the sharp regularity thresholds in several model contexts:

Context Regularity Threshold (α\alpha or bb) Reference
Euclidean Rn\mathbb{R}^n (n1)1/p1/2(n-1)\left|1/p - 1/2\right| (Martini et al., 2018)
Sub-Laplacians on bracket-generating MM same as Euclidean (Martini et al., 2018)
Lipschitz-structured variable coeff. same as above (Frey et al., 2020)
β\beta-dim. stable measures bp=(n+1)/21/pb_p = (n+1)/2 - 1/p (Basak et al., 14 Dec 2025)

This universality indicates that the critical loss of derivatives in LpL^p estimates for wave equations is a geometric and algebraic invariant, resistant to increased degeneracy or roughness in the operator or the data.

7. Applications, Corollaries, and Ongoing Directions

Miyachi-Peral-type fixed-time bounds have immediate implications for well-posedness, smoothing, and regularity properties of wave equations, including those with measure data (Basak et al., 14 Dec 2025). They underpin endpoint Strichartz bounds, guide spectral multiplier theory for sub-Laplacians, and inform the design of microlocal analysis in nonelliptic or subelliptic geometries.

Recent work demonstrates that these bounds refine classical inequalities by extending optimal wave estimates to broader classes of functions and measures, revealing new phenomena when tracking cancellation, atomicity, or fractal-like properties of measures. The continued focus is on weakening regularity requirements on both operators and data while preserving sharp bounds dictated by the geometry and oscillatory analysis. Further generalizations include noncompact settings, semigroups on abstract Banach spaces, and analysis on manifolds with nontrivial microgeometry or singularities.

These developments further the central theme that the fixed-time regularity loss for the wave propagator is fundamentally constrained by the underlying geometric dispersion, as initially encapsulated by the Miyachi-Peral theorem.

Whiteboard

Follow Topic

Get notified by email when new papers are published related to Miyachi-Peral-Type Fixed-Time Estimates.