Boyd Indices in Function Spaces

• Boyd indices are numerical invariants that measure the scaling behavior of rearrangement-invariant Banach function spaces via dilation operators.
• They provide a framework to determine boundedness conditions for operators like the Hardy–Littlewood maximal operator and the Hilbert transform.
• Their duality relations and geometric criteria make them essential tools for interpolation, spectral classification, and operator theory in harmonic analysis.

A Boyd index is a numerical invariant associated with a rearrangement-invariant (r.i.) Banach function space, quantifying how the space responds to dilation operations and providing a precise measure of its scaling behavior. Lower and upper Boyd indices, standardly denoted $\alpha_X$ and $\beta_X$ for a space $X$, play a fundamental role in harmonic analysis, particularly in the interpolation, spectral theory, and boundedness properties of operators such as the Hilbert and Hardy–Littlewood transforms. Boyd indices encapsulate both geometric and functional-analytic aspects of function spaces and are equivalently characterized by explicit operator norm growth rates, geometric weight conditions, and duality relations.

1. Formal Definitions of Boyd Indices

Let $X$ be a rearrangement-invariant Banach function space on a measurable domain, equipped with norm $\|\cdot\|_X$. Let $E_t$ (or $D_t$ in some sources) denote the dilation operator defined by

$$(E_t f)(s) = f(ts), \,\quad \text{with suitable truncation to the underlying domain.}$$

For $t > 0$, the operator norm is $\|E_t\|_{X\to X}$.

The lower and upper Boyd indices of $\beta_X$0 are defined as follows (Curbera et al., 2019, Agora et al., 2024): $\beta_X$1

$\beta_X$2

More generally, using the alternative notation $\beta_X$3 and considering real line or abstract settings, the indices are given by

$\beta_X$4

provided submultiplicativity of $\beta_X$5 ensures well-defined limits.

In the particular case of $\beta_X$6 spaces on $\beta_X$7, direct calculation yields

$\beta_X$8

2. Basic Properties, Bounds, and Duality

The Boyd indices satisfy fundamental inequalities and encode duality behavior:

• Always $\beta_X$9.
• Boyd indices are non-trivial if $X$0.
• Duality properties: For the associate (Köthe dual) space $X$1,

$X$2

This reflects the intrinsic relationship between dilation invariance in $X$3 and $X$4 (Curbera et al., 2019).

Additionally, “fundamental indices” $X$5 and $X$6 are defined by

$X$7

and satisfy

$X$8

with the duality relations $X$9, $X$0, where $X$1 is the Hölder conjugate exponent.

3. Functional Significance and Boundedness Criteria

The Boyd indices are central to characterizations of boundedness for classical operators in harmonic analysis:

• Hardy–Littlewood maximal operator $X$2: $X$3 is bounded if and only if $X$4.
• Hilbert transform $X$5: $X$6 is bounded if and only if $X$7 and $X$8. For $X$9, both conditions reduce to the classical $\|\cdot\|_X$0 range.

In weighted Lorentz spaces $\|\cdot\|_X$1, explicit geometric conditions on weights $\|\cdot\|_X$2 governing $\|\cdot\|_X$3 (for $\|\cdot\|_X$4) and $\|\cdot\|_X$5 (for $\|\cdot\|_X$6) enter directly into the computation of $\|\cdot\|_X$7 and $\|\cdot\|_X$8, leading to equivalent operator norm characterizations (Agora et al., 2024).

4. Role in Rearrangement-Invariant and Lorentz Spaces

Boyd indices are robust invariants for r.i. Banach function spaces and crucial in describing their interpolation, maximal function behavior, and operator spectra.

In the setting of weighted Lorentz spaces,

$\|\cdot\|_X$9

the Boyd indices admit explicit formulas in terms of geometric envelopes of the weights,

$E_t$0

These indices govern the strong and weak-type boundedness of Hardy–Littlewood and Hilbert transforms on these spaces (Agora et al., 2024).

Space	$E_t$1	$E_t$2	Strong-type $E_t$3	Bounded $E_t$4
$E_t$5	$E_t$6	$E_t$7	$E_t$8	$E_t$9
r.i. X	varies	varies	$D_t$0	$D_t$1, $D_t$2
$D_t$3	explicit via $D_t$4	explicit via $D_t$5	same as $D_t$6-type	same as $D_t$7-type

5. Spectral Theory and the Finite Hilbert Transform

Boyd indices have deep applications in the spectral analysis of singular integral operators, particularly the finite Hilbert transform $D_t$8 on $D_t$9. In "equal-index" (or "fundamental type") spaces where $$(E_t f)(s) = f(ts), \,\quad \text{with suitable truncation to the underlying domain.}$$0, the entire spectral picture is determined by the value $$(E_t f)(s) = f(ts), \,\quad \text{with suitable truncation to the underlying domain.}$$1: $$(E_t f)(s) = f(ts), \,\quad \text{with suitable truncation to the underlying domain.}$$2 For $$(E_t f)(s) = f(ts), \,\quad \text{with suitable truncation to the underlying domain.}$$3, the point spectrum is the interior of $$(E_t f)(s) = f(ts), \,\quad \text{with suitable truncation to the underlying domain.}$$4 and the residual spectrum is empty. For $$(E_t f)(s) = f(ts), \,\quad \text{with suitable truncation to the underlying domain.}$$5, the point spectrum is empty and the residual spectrum is the interior of $$(E_t f)(s) = f(ts), \,\quad \text{with suitable truncation to the underlying domain.}$$6.

For more general r.i. spaces, the indices $$(E_t f)(s) = f(ts), \,\quad \text{with suitable truncation to the underlying domain.}$$7 determine whether $$(E_t f)(s) = f(ts), \,\quad \text{with suitable truncation to the underlying domain.}$$8 is in the point, continuous, or residual spectrum and hence which spectral decomposition pattern $$(E_t f)(s) = f(ts), \,\quad \text{with suitable truncation to the underlying domain.}$$9 exhibits (Curbera et al., 2019).

6. Geometric and Operator-Theoretic Criteria

Explicit geometric conditions on the weights and “submultiplicativity” properties translate to inequalities on $t > 0$0 and $t > 0$1 as follows:

• $t > 0$2 for $t > 0$3 $t > 0$4 $t > 0$5,
• $t > 0$6 as $t > 0$7 $t > 0$8 $t > 0$9.

These conditions comprehensively govern boundedness of maximal and Hilbert transforms and are used in sharp proof techniques, including reductions to submultiplicativity, duality, and interpolation theorems. For $\|E_t\|_{X\to X}$0, strong and weak-type boundedness of $\|E_t\|_{X\to X}$1 coincide due to these geometric criteria (Agora et al., 2024).

7. Interpolation, Spectral Classification, and Consequences

Boyd’s indices are natural endpoints for interpolation theorems; spaces with non-trivial indices ($\|E_t\|_{X\to X}$2) form the canonical setting for singular integral operators. Through interpolation, one deduces

$\|E_t\|_{X\to X}$3

with exact spectral classification possible when $\|E_t\|_{X\to X}$4 or for Lorentz and certain interpolation spaces (Curbera et al., 2019).

A complete classification of point, continuous, and residual spectra for finite Hilbert transforms is obtained solely in terms of $\|E_t\|_{X\to X}$5 (or $\|E_t\|_{X\to X}$6), unifying classical results and extending Widom’s theory. Thus, the Boyd indices serve as both necessary and sufficient parameters for a wide spectrum of boundedness, interpolation, and spectral properties in abstract harmonic analysis and function space theory (Curbera et al., 2019, Agora et al., 2024).