Abstract
The main problem investigated in this paper is that of restricted invertibility of linear operators acting on finite dimensionall p -spaces. Our initial motivation to study such questions lies in their applications. The results obtained below enable us to complete earlier work on the structure of complemented subspaces ofL p -spaces which have extremal euclidean distance.
LetA be a realn ×n matrix considered as a linear operator onl n p ; l ≦p ≦ ∞. By restricted invertibility ofA, we mean the existence of a subset σ of {1, 2, …,n} such that |σ| ∼n andA acts as an isomorphism when restricted to the linear span of the unit vectorse i n i=1 There are various conditions under which this property holds. For instance, if the norm ‖A‖ p ofA is bounded by a constant independent ofn and the diagonal ofA is the identity matrix, then there exists an index set σ, |σ| ∼n, for which (Rσ) has a bounded inverse σ stands for the restriction map). This is achieved by simply constructing the set σ so that ••R σ(A-I)R σ••p< 12 .
The casep=2 is of particular interest. Although the problem is purely Hilbertian, the proofs involve besides the spacel 2 also the spacel 1. The methods are probabilistic and combinatorial. Crucial use is made of Grothendieck’s theorem.
The paper also contains a nice application to the behavior of the trigonometric system on sets of positive measure, generalizing results on harmonic density. Given a subsetB of the circleT of positive Lebesgue measure, there exists a subset Λ of the integersZ of positive density dens Λ > 0 such that {fx137-1} whenever the support of the Fourier transform\(\overset{\lower0.5em\hbox{$\smash{\scriptscriptstyle\frown}$}}{f} \) off lies in Λ. The matrices involved here are Laurent matrices.
The problem of restricted invertibility is meaningful beyond the class ofl p -spaces, as is shown in a separate section. However, most of the paper uses specificl p -techniques and complete results are obtained only in the context ofl p -spaces.
Similar content being viewed by others
References
K. I. Babenko,On conjugate functions, Dokl. Akad. Nauk SSSR62 (1948), 157–160 (Russian).
G. Bennett,Probability inequalities for sums of independent random variables, J. Am. Stat. Assoc.57 (1962), 33–45.
G. Bennett, L. E. Dor, V. Goodman, W. B. Johnson and C. M. Newman,On uncomplemented subspaces of L p, 1 <p < 2, Isr. J. Math.26 (1977), 178–187.
J. Bourgain,New Classes of ℒ p -spaces, Lecture Notes in Math.889, Springer-Verlag, Berlin, 1981.
J. Bourgain,A remark on finite dimensional P λ-spaces, Studia Math.72 (1981), 87–91.
J. Bourgain, P. G. Casazza, J. Lindenstrauss and L. Tzafriri,Banach spaces with a unique unconditional basis, up to permutation, Memoirs Am. Math. Soc.322 (1985).
S. Chevet,Series de variables aleatoires Gaussians a valeurs dans E \(\hat \otimes \) E F. Applications aux espaces de Wiener abstraits, Seminaire Maurey-Schwartz, 1977–78, Expose 19, Ecole Polytech., Paris.
J. Elton,Sign-embeddings of l 1n , Trans. Am. Math. Soc.279 (1983), 113–124.
T. Figiel and W. B. Johnson,Large subspaces of l n∞ and estimates of the Gordon-Lewis constant, Isr. J. Math.37 (1980), 92–112.
T. Figiel, J. Lindenstrauss and V. Milman,The dimension of almost spherical sections of convex bodies, Acta Math.139 (1977), 53–94.
T. Figiel, W. B. Johnson and G. Schechtman,Random sign-embeddings from l rn ; 2 <r < ∞, Proc. Am. Math. Soc., to appear.
Y. Gordon,Some inequalities for Gaussian processes and applications, Isr. J. Math.50 (1985), 265–289.
W. B. Johnson and G. Schechtman,On subspaces of L 1 with maximal distances to Euclidean space, Proc. Research Workshop on Banach Space Theory, University of Iowa (Bor-Luh-Lin, ed.), 1981, pp. 83–96.
W. B. Johnson and L. Jones,Every L p operator is an L 2 operator, Proc. Am. Math. Soc.72 (1978), 309–312.
B. S. Kashin,Some properties of matrices bounded operators from space l n2 to l m2 , Izvestiya Akademii Nauk Armyanoskoi SSR, Matematika,15, No. 5 (1980), 379–394.
Y. Katznelson and L. Tzafriri,On power bounded operators, J. Funct. Anal.68 (1986), 313–328.
J. L. Krivine,Theoremes de factorisation dans les espaces reticules, Seminaire Maurey-Schwartz, 1973–74, Exposes 22–23, Ecole Polytech., Paris.
S. Kwapien,Isomorphic characterisations of inner product spaces by orthogonal series with vector valued coefficients, Studia Math.44 (1972), 583–595.
D. R. Lewis,Finite dimensional subspaces of L p , Studia Math.63 (1978), 207–212.
J. Lindenstrauss and L. Tzafriri,Classical Banach Spaces. I. Sequence Spaces, Springer-Verlag, Berlin, 1979.
J. Lindenstrauss and L. Tzafriri,Classical Banach Spaces. II, Function Spaces, Springer-Verlag, Berlin, 1979.
B. Maurey,Theoremes de factorisations pour les operateures a valeurs dans un espace L p , Asterisque11, Soc. Math. France, 1974.
A. Pajor,Sous espace l nl des espaces de Banach, Travaux En Cours, Hermann, Paris, 1985.
A. Pietsch,Absolute p-summierende Abbildugen in normierten Raumen, Studia Math.28 (1967), 333–353.
W. Rudin,Trigonometric series with gaps, J. Math. Mech.9 (1960), 203–227.
I. Z. Ruzsa,On difference sets, Studia Sci. Math. Hungar.13 (1978), 319–326.
N. Sauer,On the density of families of sets, J. Comb. Theory Ser. A13 (1972), 145–147.
G. Schechtman,Fine embeddings of finite dimensional subspaces of L p , 1 ≦p < 2,into l 1 m , Proc. Am. Math. Soc.94 (1985), 617–623.
S. Shelah,A combinatorial problem: stability and order for models and theories in infinitary languages, Pacific J. Math.41 (1972), 247–261.
N. Tomczak-Jaegermann,Computing 2-summing norm with few vectors, Ark. Mat.17 (1979), 173–177.
V. N. Vapnik and A. Ya. Cervonenkis,On uniform convergence of the frequencies of events to their probabilities, SIAM Theory of Prob. and Its Appl.16 (1971), 264–280.
Author information
Authors and Affiliations
Rights and permissions
About this article
Cite this article
Bourgain, J., Tzafriri, L. Invertibility of ‘large’ submatrices with applications to the geometry of Banach spaces and harmonic analysis. Israel J. Math. 57, 137–224 (1987). https://doi.org/10.1007/BF02772174
Received:
Issue Date:
DOI: https://doi.org/10.1007/BF02772174