Multi-Toeplitz Kernels: Theory and Structures

• Multi-Toeplitz kernels are multi-index translation-invariant constructs modeled by p‐level Toeplitz matrices and Hardy-space kernels.
• They encompass generalized, matrix-, and block-valued variants, leveraging inner–outer factorizations and tensor-product unitaries to exhibit near-invariance.
• Their study enables precise spectral and multiplier analyses through techniques like Carleson embeddings and Beurling–Malliavin density thresholds.

Multi-Toeplitz kernels arise in several closely related senses across operator theory, Hardy-space analysis, and structured matrix theory. In the finite-dimensional multilevel setting, a $p$-level Toeplitz matrix is the matrix model of a discrete kernel

$$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$

so that dependence is only on coordinatewise differences; this is the multi-index analogue of the classical Toeplitz condition (Cao et al., 2019). In Hardy-space operator theory, “Toeplitz kernel” usually denotes the kernel of a Toeplitz operator, while more recent work considers generalized Toeplitz kernels, projected paired kernels, vector-valued and matrix-valued Toeplitz kernels, and bidisk or multi-shift analogues, all of which retain a common theme: invariance or near-invariance under an appropriate backward shift, together with a representation by inner–outer factorizations and model spaces (Anjali et al., 4 Jul 2025).

1. Definitions and terminological scope

The classical scalar Toeplitz kernel on the disk is

$$\ker T_\phi=\{f\in H^2:T_\phi f=0\},\qquad T_\phi f=P_+(\phi f),$$

with $P_+$ the orthogonal projection onto $H^2$. A fundamental example is the model space

$K_\theta=H^2\ominus \theta H^2=\ker T_{\bar\theta}$

for an inner function $\theta$ (Anjali et al., 4 Jul 2025). In the upper half-plane one similarly defines $T_U f=P_+(Uf)$ on $\mathcal H_+^2$, and $\ker T_U\subset \mathcal H_+^2$ is again a Toeplitz kernel (Anjali et al., 4 Jul 2025).

A broader usage appears in generalized Toeplitz operators. In the disk setting, fixing a closed subspace $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$0 and a simply invariant subspace $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$1, one defines

$$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$2

and the associated generalized Toeplitz kernel is

$$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$3

When $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$4 and $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$5, this collapses to the usual $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$6 (Anjali et al., 4 Jul 2025). In this sense, generalized Toeplitz kernels are kernels of a family of Wiener–Hopf type operators rather than only the canonical $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$7 Toeplitz operators.

A distinct but related finite-dimensional meaning is supplied by multilevel Toeplitz matrices. There a $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$8-level Toeplitz matrix corresponds to a discrete multi-Toeplitz kernel on a finite multi-index set, again depending only on coordinatewise differences (Cao et al., 2019). This identifies “multi-Toeplitz kernel” with a finite section or discretization of a multi-index translation-invariant kernel.

The literature also uses block, vector-valued, and matrix-valued versions. For $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$9, the matrix-valued Toeplitz operator on $$\ker T_\phi=\{f\in H^2:T_\phi f=0\},\qquad T_\phi f=P_+(\phi f),$$0 is

$$\ker T_\phi=\{f\in H^2:T_\phi f=0\},\qquad T_\phi f=P_+(\phi f),$$1

and its kernel is a vector-valued Toeplitz kernel (Nicolas, 2010). For matrix-valued truncated Toeplitz operators, one works on model spaces $$\ker T_\phi=\{f\in H^2:T_\phi f=0\},\qquad T_\phi f=P_+(\phi f),$$2 and studies kernels of compressed multiplication operators $$\ker T_\phi=\{f\in H^2:T_\phi f=0\},\qquad T_\phi f=P_+(\phi f),$$3 (O'Loughlin, 2020). In the upper half-plane block setting, kernels of $$\ker T_\phi=\{f\in H^2:T_\phi f=0\},\qquad T_\phi f=P_+(\phi f),$$4 block Toeplitz operators may be “of scalar type,” meaning

$$\ker T_\phi=\{f\in H^2:T_\phi f=0\},\qquad T_\phi f=P_+(\phi f),$$5

for a scalar space $$\ker T_\phi=\{f\in H^2:T_\phi f=0\},\qquad T_\phi f=P_+(\phi f),$$6 and a fixed vector $$\ker T_\phi=\{f\in H^2:T_\phi f=0\},\qquad T_\phi f=P_+(\phi f),$$7 (Câmara et al., 2018).

A recurrent misconception is that “multi-Toeplitz” always means “many variables.” The surveyed literature shows at least four technically different uses: multilevel difference kernels on $$\ker T_\phi=\{f\in H^2:T_\phi f=0\},\qquad T_\phi f=P_+(\phi f),$$8, generalized kernels indexed by pairs $$\ker T_\phi=\{f\in H^2:T_\phi f=0\},\qquad T_\phi f=P_+(\phi f),$$9, matrix- or block-valued Toeplitz kernels, and multi-shift or bidisk analogues (Cao et al., 2019).

2. Multilevel Toeplitz matrices as discrete multi-index kernels

A classical $P_+$0 Toeplitz matrix $P_+$1 has entries

$P_+$2

so the associated kernel $P_+$3 depends only on the difference $P_+$4. The multilevel generalization is recursive: a $P_+$5-level Toeplitz matrix $P_+$6 is a block Toeplitz matrix whose blocks are themselves $P_+$7-level Toeplitz matrices (Cao et al., 2019). For sizes $P_+$8 and

$P_+$9

the full matrix has size $H^2$0, and the $H^2$1-block is $H^2$2.

Under lexicographic ordering of multi-indices $H^2$3, a $H^2$4-level Toeplitz matrix corresponds to

$H^2$5

which is precisely the discrete multi-Toeplitz kernel on a finite grid (Cao et al., 2019). The $H^2$6-level case is “Toeplitz in blocks, and each block Toeplitz,” while the $H^2$7-level case is “Toeplitz of Toeplitz-blocks of Toeplitz-blocks.”

A central structural result is that every multilevel Toeplitz matrix is unitarily similar to a complex symmetric matrix. If $H^2$8 is a $H^2$9-level Toeplitz matrix, then there exists a unitary matrix $K_\theta=H^2\ominus \theta H^2=\ker T_{\bar\theta}$0 such that

$K_\theta=H^2\ominus \theta H^2=\ker T_{\bar\theta}$1

is symmetric, where $K_\theta=H^2\ominus \theta H^2=\ker T_{\bar\theta}$2 is built from tensor products of one-level unitaries chosen according to the parity of each level size $K_\theta=H^2\ominus \theta H^2=\ker T_{\bar\theta}$3 (Cao et al., 2019). The paper also gives a parity-free variant using

$K_\theta=H^2\ominus \theta H^2=\ker T_{\bar\theta}$4

with $K_\theta=H^2\ominus \theta H^2=\ker T_{\bar\theta}$5 the flip matrix; then

$K_\theta=H^2\ominus \theta H^2=\ker T_{\bar\theta}$6

is symmetric for a tensor-product unitary $K_\theta=H^2\ominus \theta H^2=\ker T_{\bar\theta}$7 of the same form (Cao et al., 2019).

For the special case $K_\theta=H^2\ominus \theta H^2=\ker T_{\bar\theta}$8, the image of $K_\theta=H^2\ominus \theta H^2=\ker T_{\bar\theta}$9-level Toeplitz matrices under the explicit unitary $\theta$0 is characterized exactly. A $\theta$1 complex symmetric matrix $\theta$2 is of the form

$\theta$3

for some $\theta$4-level Toeplitz matrix $\theta$5 if and only if $\theta$6 has constant anti-diagonals at each level (Cao et al., 2019). Here a $\theta$7-level block means a $\theta$8 block in the recursive decomposition, and “constant anti-diagonals at each level” imposes anti-diagonal constancy in every such block scale.

This finite-dimensional theory is constructive rather than purely existential. The same unitary depends only on the shape $\theta$9, not on the Toeplitz entries, so for a fixed grid the unitary symmetrizes every $T_U f=P_+(Uf)$0-level Toeplitz matrix of that shape (Cao et al., 2019). A plausible implication is that finite sections of discrete multi-Toeplitz operators can be analyzed through structured complex symmetric models.

3. Toeplitz kernels, generalized kernels, and multipliers

In the generalized Hardy-space setting, a central organizing device is the minimal generalized Toeplitz kernel generated by a single nonzero vector. For $T_U f=P_+(Uf)$1, the minimal generalized Toeplitz kernel containing $T_U f=P_+(Uf)$2 is

$T_U f=P_+(Uf)$3

If $T_U f=P_+(Uf)$4 is the inner–outer factorization of $T_U f=P_+(Uf)$5, then

$T_U f=P_+(Uf)$6

(Anjali et al., 4 Jul 2025). This gives an explicit symbol for the minimal kernel in terms of the inner part of $T_U f=P_+(Uf)$7 and an outer denominator.

A maximal vector for $T_U f=P_+(Uf)$8 is a function $T_U f=P_+(Uf)$9 such that

$\mathcal H_+^2$0

The characterization is explicit: $\mathcal H_+^2$1 for some outer $\mathcal H_+^2$2 (Anjali et al., 4 Jul 2025). In the upper half-plane analogue, the same role is played by

$\mathcal H_+^2$3

with $\mathcal H_+^2$4 inner and $\mathcal H_+^2$5 outer in $\mathcal H_+^2$6 (Anjali et al., 4 Jul 2025).

The main multiplier theorem identifies when an analytic function $\mathcal H_+^2$7 maps one generalized Toeplitz kernel into another. For nontrivial kernels

$\mathcal H_+^2$8

one has

$\mathcal H_+^2$9

if and only if $\ker T_U\subset \mathcal H_+^2$0 and $\ker T_U\subset \mathcal H_+^2$1; equivalently, $\ker T_U\subset \mathcal H_+^2$2 maps some maximal vector of $\ker T_U\subset \mathcal H_+^2$3 into $\ker T_U\subset \mathcal H_+^2$4 (Anjali et al., 4 Jul 2025). In the upper half-plane, the corresponding statement is formulated with $\ker T_U\subset \mathcal H_+^2$5 and maximal vectors in $\ker T_U\subset \mathcal H_+^2$6 (Anjali et al., 4 Jul 2025).

These results isolate two components of the multiplier problem. One is a Carleson-type embedding condition encoded as $\ker T_U\subset \mathcal H_+^2$7. The other is a symbolic condition, expressed by the Nevanlinna–Smirnov requirement $\ker T_U\subset \mathcal H_+^2$8 in the disk or $\ker T_U\subset \mathcal H_+^2$9 for kernel inclusions in the upper half-plane (Anjali et al., 4 Jul 2025). In particular,

$$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$00

(Anjali et al., 4 Jul 2025).

A different generalization is furnished by paired operators on $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$01,

$$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$02

with paired kernel

$$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$03

and projected paired kernel

$$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$04

If $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$05, then

$$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$06

so projected paired kernels recover classical Toeplitz kernels; if $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$07, they behave as natural closed-space generalizations of kernels of unbounded Toeplitz operators $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$08 (Câmara et al., 2023). This extends the Toeplitz-kernel perspective beyond bounded symbols while preserving near-invariance and minimal-kernel phenomena.

4. Vector-valued, block, and matrix-valued Toeplitz kernels

In vector-valued Hardy spaces, kernels of matrix-valued Toeplitz operators are nearly $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$09-invariant subspaces. For $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$10, a closed subspace $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$11 is nearly $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$12-invariant if every $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$13 with $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$14 satisfies $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$15. The kernels of matrix-valued Toeplitz operators are examples of such subspaces (Nicolas, 2010).

The vector-valued analogue of Hitt’s theorem states that if $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$16 is nontrivial and nearly $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$17-invariant, then

$$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$18

for an outer matrix $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$19 and an inner $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$20, with multiplication by $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$21 acting isometrically from $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$22 onto $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$23 (Nicolas, 2010). This is the basic model-space form for vector-valued Toeplitz kernels.

A matrix-valued Sarason theorem characterizes when $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$24 is an isometry. If $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$25 is the matrix-valued contractive function obtained from $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$26 by the Herglotz construction, then

$$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$27

(Nicolas, 2010). In turn, the matrix-valued Hayashi theorem states that a nearly $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$28-invariant space $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$29 is the kernel of a Toeplitz operator precisely when the divisibility $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$30 holds, the pair $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$31 is special, and $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$32 is rigid; then

$$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$33

(Nicolas, 2010).

For $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$34 block Toeplitz operators on the upper half-plane, a principal phenomenon is “scalar-type” structure: $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$35 for a scalar space $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$36 and a fixed vector function $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$37 (Câmara et al., 2018). Under a Wiener–Hopf factorization

$$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$38

the kernel is

$$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$39

(Câmara et al., 2018). More generally, if $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$40 with left-invertible $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$41 and $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$42, then the kernel admits a complete description in terms of $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$43, $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$44, their left inverses, and an overlap set $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$45; when the relevant intersection is trivial, the scalar-type reduction follows (Câmara et al., 2018).

The $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$46 theory confirms that scalar-type behavior is not automatic. Every $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$47 has a minimal Toeplitz kernel $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$48 given by an explicit matrix-valued symbol (O'Loughlin, 2020). However, not every vector-valued Toeplitz kernel has a maximal function: if

$$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$49

then $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$50 does not have a maximal function (O'Loughlin, 2020). In the Hilbert-space case $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$51, a nonzero matricial Toeplitz kernel has a maximal function if and only if $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$52 and the corresponding scalar-type model is not shift invariant (O'Loughlin, 2020). This sharply separates genuinely multicomponent kernels from those reducible to a single vector times a scalar model space.

5. Near invariance, model spaces, and multi-shift analogues

Near-invariance under backward shifts is the unifying geometric property of Toeplitz kernels. In one variable, a closed subspace $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$53 is nearly $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$54-invariant if

$$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$55

Such subspaces admit the representation

$$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$56

where $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$57 is inner with $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$58, and multiplication by $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$59 is isometric (Zhu et al., 2024). In the same paper, every nearly $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$60-invariant subspace of the vector-valued one-variable Hardy space is shown to be the kernel of a Toeplitz operator constructed from a symbol

$$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$61

for appropriate outer $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$62 (Zhu et al., 2024).

On the bidisk $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$63, the naive extension of near invariance with respect to each coordinate backward shift separately is not compatible with Toeplitz kernels: the paper gives an example of a Toeplitz kernel that fails this naive condition (Zhu et al., 2024). The correct object is the product shift

$$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$64

with

$$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$65

and a subspace $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$66 is nearly $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$67-invariant when

$$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$68

(Zhu et al., 2024).

This reformulation uses the Wold decomposition

$$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$69

so $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$70 becomes a vector-valued Hardy space in the single variable $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$71 (Zhu et al., 2024). The resulting model theorem states that a closed subspace $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$72 is nearly $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$73-invariant if and only if

$$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$74

with $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$75 and $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$76 an isometric multiplier (Zhu et al., 2024). Moreover, for $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$77, every kernel $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$78 on the bidisk is nearly $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$79-invariant (Zhu et al., 2024).

The same conceptual move extends to commuting pure isometric tuples $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$80. If the tuple satisfies

$$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$81

and $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$82, then a bounded operator $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$83 is called a general Toeplitz operator if

$$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$84

and the kernel of every such $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$85 is nearly $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$86-invariant (Zhu et al., 2024). This makes “multi-Toeplitz kernel” precise in an abstract operator-theoretic sense: the kernel of a Toeplitz-type operator associated with a commuting multi-shift is a near-invariant subspace for the composite shift.

Projected paired kernels also fit this pattern. They are nearly $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$87-invariant for every $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$88, hence nearly $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$89-invariant, and their closures are again nearly $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$90-invariant (Câmara et al., 2023). Consequently, their closures can be represented as $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$91 with $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$92 outer and $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$93 or $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$94, placing them inside the Hitt–Hayashi paradigm (Câmara et al., 2023).

6. Structural criteria, spectral connections, and applications

Several structural criteria determine when multi-Toeplitz kernels reduce to simpler models. In the scalar setting, every nontrivial scalar Toeplitz kernel has a maximal function, and if $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$95 is maximal for $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$96, then

$$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$97

(O'Loughlin, 2020). For multiple scalar generators $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$98, the minimal Toeplitz kernel containing them is $$K(i_1,\dots,i_p; j_1,\dots,j_p)=a_{(i_1-j_1,\dots,i_p-j_p)},$$99 whenever $$\ker T_\phi=\{f\in H^2:T_\phi f=0\},\qquad T_\phi f=P_+(\phi f),$$00 and $$\ker T_\phi=\{f\in H^2:T_\phi f=0\},\qquad T_\phi f=P_+(\phi f),$$01 (O'Loughlin, 2020). For two scalar functions $$\ker T_\phi=\{f\in H^2:T_\phi f=0\},\qquad T_\phi f=P_+(\phi f),$$02, the condition

$$\ker T_\phi=\{f\in H^2:T_\phi f=0\},\qquad T_\phi f=P_+(\phi f),$$03

holds if and only if $$\ker T_\phi=\{f\in H^2:T_\phi f=0\},\qquad T_\phi f=P_+(\phi f),$$04 is cyclic for the backward shift on $$\ker T_\phi=\{f\in H^2:T_\phi f=0\},\qquad T_\phi f=P_+(\phi f),$$05 (O'Loughlin, 2020). This identifies backward-shift cyclicity in the Smirnov class as the obstruction to the existence of a proper minimal Toeplitz kernel generated by the pair.

The generalized-kernel multiplier theory connects these questions to Beurling–Malliavin density, Pólya sequences, and the spectral theory of entire functions. For a discrete sequence $$\ker T_\phi=\{f\in H^2:T_\phi f=0\},\qquad T_\phi f=P_+(\phi f),$$06, Makarov–Poltoratski formulas express the interior and exterior Beurling–Malliavin densities through nontriviality thresholds for kernels of symbols $$\ker T_\phi=\{f\in H^2:T_\phi f=0\},\qquad T_\phi f=P_+(\phi f),$$07 and $$\ker T_\phi=\{f\in H^2:T_\phi f=0\},\qquad T_\phi f=P_+(\phi f),$$08 (Anjali et al., 4 Jul 2025). In the example with

$$\ker T_\phi=\{f\in H^2:T_\phi f=0\},\qquad T_\phi f=P_+(\phi f),$$09

the paper states

$$\ker T_\phi=\{f\in H^2:T_\phi f=0\},\qquad T_\phi f=P_+(\phi f),$$10

where $$\ker T_\phi=\{f\in H^2:T_\phi f=0\},\qquad T_\phi f=P_+(\phi f),$$11 for $$\ker T_\phi=\{f\in H^2:T_\phi f=0\},\qquad T_\phi f=P_+(\phi f),$$12 (Anjali et al., 4 Jul 2025). Thus existence of nontrivial $$\ker T_\phi=\{f\in H^2:T_\phi f=0\},\qquad T_\phi f=P_+(\phi f),$$13-multipliers between Toeplitz kernels is controlled by a precise density threshold.

A related measure-free criterion is given for meromorphic symbols in the upper half-plane: $$\ker T_\phi=\{f\in H^2:T_\phi f=0\},\qquad T_\phi f=P_+(\phi f),$$14 (Anjali et al., 4 Jul 2025). This replaces explicit Carleson conditions by a kernel nontriviality and dimension condition for a single Toeplitz symbol.

Matrix-valued truncated Toeplitz operators provide a different application domain. If $$\ker T_\phi=\{f\in H^2:T_\phi f=0\},\qquad T_\phi f=P_+(\phi f),$$15 is a bounded matrix-valued truncated Toeplitz operator with symbol $$\ker T_\phi=\{f\in H^2:T_\phi f=0\},\qquad T_\phi f=P_+(\phi f),$$16, $$\ker T_\phi=\{f\in H^2:T_\phi f=0\},\qquad T_\phi f=P_+(\phi f),$$17, then

$$\ker T_\phi=\{f\in H^2:T_\phi f=0\},\qquad T_\phi f=P_+(\phi f),$$18

is nearly $$\ker T_\phi=\{f\in H^2:T_\phi f=0\},\qquad T_\phi f=P_+(\phi f),$$19-invariant with finite defect $$\ker T_\phi=\{f\in H^2:T_\phi f=0\},\qquad T_\phi f=P_+(\phi f),$$20 (O'Loughlin, 2020). More precisely, the kernel is the isometric image of an $$\ker T_\phi=\{f\in H^2:T_\phi f=0\},\qquad T_\phi f=P_+(\phi f),$$21-invariant subspace of a vector-valued Hardy space: $$\ker T_\phi=\{f\in H^2:T_\phi f=0\},\qquad T_\phi f=P_+(\phi f),$$22 in the nonvanishing-at-zero case, with an isometric norm identity for the $$\ker T_\phi=\{f\in H^2:T_\phi f=0\},\qquad T_\phi f=P_+(\phi f),$$23 coefficients (O'Loughlin, 2020). The same paper shows that the modified MTTO is equivalent after extension to a block Toeplitz operator

$$\ker T_\phi=\{f\in H^2:T_\phi f=0\},\qquad T_\phi f=P_+(\phi f),$$24

so kernel and Fredholm structure can be transferred to a full Toeplitz setting (O'Loughlin, 2020).

The bidisk and multilevel matrix theories suggest a shared principle. In one case, a multilevel Toeplitz kernel on a finite grid is unitarily transformed into a complex symmetric kernel by tensor products of one-dimensional flip-plus-phase unitaries (Cao et al., 2019). In the other, Toeplitz kernels on the bidisk or for commuting isometric tuples are converted into one-variable vector-valued model spaces by passing to the composite shift $$\ker T_\phi=\{f\in H^2:T_\phi f=0\},\qquad T_\phi f=P_+(\phi f),$$25 (Zhu et al., 2024). This suggests that many “multi-Toeplitz” constructions are best understood by separating coordinates through tensor products or composite shifts rather than by treating each variable independently.

Across these settings, the persistent structural features are difference dependence or Toeplitz covariance, near-invariance under a backward shift or composite shift, inner–outer factorization, and reduction to model-space or scalar-type forms under additional hypotheses. What varies is the ambient category: finite multilevel matrices, generalized Hardy-space kernels, block or matrix-valued Toeplitz operators, truncated Toeplitz operators, or multi-shift operator tuples (Cao et al., 2019).