SearcharxivSearch

arXiv subjects

Nigel J. Kalton

Publications and source records attributed to Nigel J. Kalton.

At least 19 recordsLinked to original sources

Euclidean structures and operator theory in Banach spaces

We present a general method to extend results on Hilbert space operators to the Banach space setting by representing certain sets of Banach space operators $Γ$ on a Hilbert space. Our assumption on $Γ$ is expressed in terms of $α$-boundedness for a Euclidean structure $α$ on the underlying Banach space $X$. This notion is originally motivated by $\mathcal{R}$- or $γ$-boundedness of sets of operators, but, for example, any operator ideal from the Euclidean space $\ell^2_n$ to $X$ defines such a structure. Therefore, our method is quite flexible. Conversely we show that $Γ$ has to be $α$-bounded for some Euclidean structure $α$ to be representable on a Hilbert space. By choosing the Euclidean structure $α$ accordingly, we get a unified and more general approach to classical factorization and extension theorems. Furthermore we use these Euclidean structures to build vector-valued function spaces and define an interpolation method based on these spaces, which has formulations modelled after both the real and the complex interpolation method. Using our representation theorem we prove a transference principle for sectorial operators on a Banach space, enabling us to extend Hilbert space results for sectorial operators to the Banach space setting. We define generalizations of the classical square function estimates in $L^p$-spaces and establish, via the $H^\infty$-calculus, a version of Littlewood-Paley theory and associated spaces of fractional smoothness for a rather large class of sectorial operators. Our results for sectorial operators lead to some sophisticated counterexamples.

math.FA

Complex interpolation and twisted twisted Hilbert spaces

We show that Rochberg's generalizared interpolation spaces $\mathscr Z^{(n)}$ arising from analytic families of Banach spaces form exact sequences $0\to \mathscr Z^{(n)} \to \mathscr Z^{(n+k)} \to \mathscr Z^{(k)} \to 0$. We study some structural properties of those sequences; in particular, we show that nontriviality, having strictly singular quotient map, or having strictly cosingular embedding depend only on the basic case $n=k=1$. If we focus on the case of Hilbert spaces obtained from the interpolation scale of $\ell_p$ spaces, then $\mathscr Z^{(2)}$ becomes the well-known Kalton-Peck $Z_2$ space; we then show that $\mathscr Z^{(n)}$ is (or embeds in, or is a quotient of) a twisted Hilbert space only if $n=1,2$, which solves a problem posed by David Yost; and that it does not contain $\ell_2$ complemented unless $n=1$. We construct another nontrivial twisted sum of $Z_2$ with itself that contains $\ell_2$ complemented.

math.FA

Set-functions and factorization

If $ϕ$ is a submeasure satisfying an appropriate lower estimate we give a quantitative result on the total mass of a measure $μ$ satisfying $0\leμ\leϕ.$ We give a dual result for supermeasures and then use these results to investigate convexity on non-locally convex quasi-Banach lattices. We then show how to use these results to extend some factorization theorems due to Pisier to the setting of quasi-Banach spaces. We conclude by showing that if $X$ is a quasi-Banach space of cotype two then any operator $T:C(Ω)\to X$ is 2-absolutely summing and factors through a Hilbert space and discussing general factorization theorems for cotype two spaces.

math.FA

Frames of translates

We give necessary and sufficient conditions for a subfamily of regularly spaced translates of a function to form a frame (resp. a Riesz basis) for its span. One consequence is that ifthetranslates are taken only from a subset of the natural numbers, then this family is a frame if and only if it is a Riesz basis. We also consider arbitrary sequences of translates and show that for sparse sets, having an upper frame bound is equivalent to the family being a frame sequence. Finally, we use the fractional Hausdorff dimension to identify classes of exact frame sequences.

math.FA

Uniqueness of unconditional bases in c_0-products

We give counterexamples to a conjecture of Bourgain, Casazza, Lindenstrauss and Tzafriri that if X has a unique unconditional basis (up to permutation) then so does c_0(X). In particular, we show that for Tsirelson's space T, every unconditional basis of c_0(T) must be equivalent to a subsequence of the canonical basis but c_0(T) still fails to have a unique unconditional basis. We also give some positive results including a simpler proof that c_0(l_1)has a unique unconditional basis.

math.FA

Some Applications of Operator-Valued Herglotz Functions

We consider operator-valued Herglotz functions and their applications to self-adjoint perturbations of self-adjoint operators and self-adjoint extensions of densely defined closed symmetric operators. Our applications include model operators for both situations, linear fractional transformations for Herglotz operators, results on Friedrichs and Krein extensions, and realization theorems for classes of Herglotz operators. Moreover, we study the concrete case of Schrödinger operators on a half-line and provide two illustrations of Livsic's result [44] on quasi-hermitian extensions in the special case of densely defined symmetric operators with deficiency indices (1,1).

math.FA

Spectral characterization of sums of commutators II

For countably generated ideals, $\Jc$, of $B(\Hil)$, geometric stability is necessary for the canonical spectral characterization of sums of $(\Jc,B(\Hil))$--commutators to hold. This answers a question raised by Dykema, Figiel, Weiss and Wodzicki. There are some ideals, $\Jc$, having quasi--nilpotent elements that are not sums of $(\Jc,B(\Hil))$--commutators. Also, every trace on every geometrically stable ideal is a spectral trace.

math.FA

Spectral characterization of sums of commutators I

Suppose $\Cal J$ is a two-sided quasi-Banach ideal of compact operators on a separable infinite-dimensional Hilbert space $\Cal H$. We show that an operator $T\in\Cal J$ can be expressed as finite linear combination of commutators $[A,B]$ where $A\in\Cal J$ and $B\in\Cal B(\Cal H)$ if and only its eigenvalues $(λ_n)$ (arranged in decreasing order of absolute value, repeated according to algebraic multiplicity and augmented by zeros if necessary) satisfy the condition that the diagonal operator $\diag\{\frac1n(λ_1+\cdots +λ_n)\}$ is a member of $\Cal J.$ This answers (for quasi-Banach ideals) a question raised by Dykema, Figiel, Weiss and Wodzicki.

math.FA

Distances between Banach spaces

The main object of the paper is to study the distance between Banach spaces introduced by Kadets. For Banach spaces $X$ and $Y$, the Kadets distance is defined to be the infimum of the Hausdorff distance $d(B_X,B_Y)$ between the respective closed unit balls over all isometric linear embeddings of $X$ and $Y$ into a common Banach space $Z.$ This is compared with the Gromov-Hausdorff distance which is defined to be the infimum of $d(B_X,B_Y)$ over all isometric embeddings into a common metric space $Z$. We prove continuity type results for the Kadets distance including a result that shows that this notion of distance has applications to the theory of complex interpolation.

math.FA

Nonlinear equations and weighted norm inequalities

We study connections between the problem of the existence of positive solutions for certain nonlinear equations and weighted norm inequalities. In particular, we obtain explicit criteria for the solvability of the Dirichlet problem $$\aligned -& Δu = v \, u^q + w, \quad u \ge 0 \quad \text {on} \quad Ω, \\ &u = 0 \quad \text {on} \quad \partial Ω, \endaligned $$ on a regular domain $Ω$ in $\bold R^n$ in the ``superlinear case'' $q > 1$. The coefficients $v, w$ are arbitrary positive measurable functions (or measures) on $Ω$. We also consider more general nonlinear differential and integral equations, and study the spaces of coefficients and solutions naturally associated with these problems, as well as the corresponding capacities. Our characterizations of the existence of positive solutions take into account the interplay between $v$, $w$, and the corresponding Green's kernel. They are not only sufficient, but also necessary, and are established without any a priori regularity assumptions on $v$ and $w$; we also obtain sharp two-sided estimates of solutions up to the boundary. Some of our results are new even if $v \equiv 1$ and $Ω$ is a ball or half-space. The corresponding weighted norm inequalities are proved for integral operators with kernels satisfying a refined version of the so-called $3 G$-inequality by an elementary ``integration by parts'' argument. This also gives a new unified proof for some classical inequalities including the Carleson measure theorem for Poisson integrals and trace inequalities for Riesz potentials and Green potentials.

math.FA

Uniqueness of unconditional bases in Banach spaces

We prove a general result on complemented unconditional basic sequences in Banach lattices and apply it to give some new examples of spaces with unique unconditional basis. We show that Tsirelson space and certain Nakano spaces have the unique unconditional bases. We also construct an example of a space with a unique unconditional basis with a complemented subspace failing to have a unique unconditional basis.

math.FA

Kernels of surjections from ${\cal L}_1$-spaces with an application to Sidon sets

If $Q$ is a surjection from $L^1(μ)$, $μ$ $σ$-finite, onto a Banach space containing $c_0$ then (*) $\ker Q$ is uncomplemented in its second dual. If $Q$ is a surjection from an ${\cal L}_1$-space onto a Banach space containing uniformly $\ell_n^\infty$ ($n=1,2,\dots$) then (**) there exists a bounded linear operator from $\ker Q$ into a Hilbert space which is not 2-absolutely summing. Let $S$ be an infinite Sidon set in the dual group $Γ$ of a compact abelian group $G$. Then $L^1_{\tilde{S}}(G)=\{f\in L^1(G): \hat{f}(γ)=0$ for $γ\in S\}$ satisfies (*) and (**) hence $L^1_{\tilde{S}}(G)$ is not an ${\cal L}_1$-space and is not isomorphic to a Banach lattice.

math.FA

Twisted sums, Fenchel-Orlicz spaces and property (M)

We study certain twisted sums of Orlicz spaces with non-trivial type which can be viewed as Fenchel-Orlicz spaces on ${\rm {\bf R}}^2$. We then show that a large class of Fenchel-Orlicz spaces on ${\rm {\bf R}}^n$ can be renormed to have property (M). In particular this gives a new construction of the twisted Hilbert space $Z_2$ and shows it has property (M), after an appropriate renorming.

math.FA

Subspaces of rearrangement-invariant spaces

We prove a number of results concerning the embedding of a Banach lattice $X$ into an r.i. space $Y$. For example we show that if $Y$ is an r.i. space on $[0,\infty)$ which is $p$-convex for some $p>2$ and has nontrivial concavity then any Banach lattice $X$ which is $r$-convex for some $r>2$ and embeds into $Y$ must embed as a sublattice. Similar conclusions can be drawn under a variety of hypotheses on $Y$; if $X$ is an r.i. space on $[0,1]$ one can replace the hypotheses of $r$-convexity for some $r>2$ by $X\neq L_2.$ We also show that if $Y$ is an order-continuous Banach lattice which contains no complemented sublattice lattice-isomorphic to $\ell_2,$ $X$ is an order-continuous Banach lattice so that $\ell_2$ is not complementably lattice finitely representable in $X$ and $X$ is isomorphic to a complemented subpace of $Y$ then $X$ is isomorphic to a complemented sublattice of $Y^N$ for some integer $N.$

math.FA