Professional Documents
Culture Documents
Textbook Algebra Complex Analysis and Pluripotential Theory 2 Usuzcamp Urgench Uzbekistan August 8 12 2017 Zair Ibragimov Ebook All Chapter PDF
Textbook Algebra Complex Analysis and Pluripotential Theory 2 Usuzcamp Urgench Uzbekistan August 8 12 2017 Zair Ibragimov Ebook All Chapter PDF
https://textbookfull.com/product/differential-equations-and-
dynamical-systems-2-usuzcamp-urgench-uzbekistan-
august-8-12-2017-abdulla-azamov/
https://textbookfull.com/product/algebra-2-linear-algebra-galois-
theory-representation-theory-group-extensions-and-schur-
multiplier-1st-edition-ramji-lal/
https://textbookfull.com/product/algebra-2-linear-algebra-galois-
theory-representation-theory-group-extensions-and-schur-
multiplier-1st-edition-ramji-lal-2/
https://textbookfull.com/product/a-course-in-analysis-volume-iii-
measure-and-integration-theory-complex-valued-functions-of-a-
complex-variable-1st-edition-niels-jacob/
A Course in Analysis Volume III Measure and Integration
Theory Complex valued Functions of a Complex Variable
1st Edition Niels Jacob
https://textbookfull.com/product/a-course-in-analysis-volume-iii-
measure-and-integration-theory-complex-valued-functions-of-a-
complex-variable-1st-edition-niels-jacob-2/
Zair Ibragimov
Norman Levenberg
Utkir Rozikov
Azimbay Sadullaev Editors
Algebra, Complex
Analysis, and
Pluripotential
Theory
2 USUZCAMP, Urgench, Uzbekistan,
August 8–12, 2017
Springer Proceedings in Mathematics & Statistics
Volume 264
Springer Proceedings in Mathematics & Statistics
This book series features volumes composed of selected contributions from
workshops and conferences in all areas of current research in mathematics and
statistics, including operation research and optimization. In addition to an overall
evaluation of the interest, scientific quality, and timeliness of each proposal at the
hands of the publisher, individual contributions are all refereed to the high quality
standards of leading journals in the field. Thus, this series provides the research
community with well-edited, authoritative reports on developments in the most
exciting areas of mathematical and statistical research today.
Editors
123
Editors
Zair Ibragimov Utkir Rozikov
Department of Mathematics Institute of Mathematics
California State University Uzbekistan Academy of Sciences
Fullerton, CA, USA Tashkent, Uzbekistan
Mathematics Subject Classification (2010): 17A32, 31B15, 32U30, 37K15, 41A17, 46L57, 46S10,
60J10
This Springer imprint is published by the registered company Springer Nature Switzerland AG
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
Preface
v
Acknowledgements
We thank all contributing authors and referees for their efforts. We also thank
Springer for the opportunity to publish this volume.
vii
Contents
ix
x Contents
1 Introduction
In this work, we study removable singular sets for the certain classes of subharmonic
and m−subharmonic functions in the domain D of complex space Cn .
for any ω ∈ F (n−m,n−m) , ω ≥ 0. Here the space F (n−m,n−m) is finite and C ∞ - smooth
on D differential forms of bidegree (n − m, n − m).
The class of such functions are considered in works of Z. Khusanov [13, 14], R.
Harvey and B. Lausson [10], M. Verbitsky [27], D. Joyce [12] and others. In works
[1, 2] first author has proved a series of potential properties of m− subharmonic
functions.
The class of m − sh functions is wider than the class of plurisubharmonic func-
tions, but it strictly contains in the class of subharmonic functions. In addition, the
class of 1 − sh functions coincides with the class of subharmonic functions and the
class of n − sh functions coincides with the class of subharmonic functions.
Hn−2+α = 0, 0 < α ≤ 2,
then any subharmonic function in D\E from the class Li pα (D) subharmonically
extends into the domain D.
In the works of A. Sadullaev and B. Abdullaev (see [20, 22, 23]) a number of
theorems have been proved about removable singular sets for the bounded from above
Removable Singular Sets of m-Subharmonic Functions 3
class m − sh (D\E) and for the class Li pα (D) ∩ m − sh (D\E) of functions. Since
the methods of proving these results are directly related to the problems studied here,
we provide these results.
Theorem 2 (A. Sadullaev, B. Abdullaev [20]) If a closed in D set E is polar, i.e. its
newtonian capacity
C(E) = 0,
H2n−2+α (E) = 0,
In the proof of this theorem, authors used the following theorem of W. Schiffman
[26] (see also [15, 24]): if E ⊂ Rn (x) × Rk (y), n ≥ 1, k ≥ 1, Hn+α (E) = 0, 0 ≤
α ≤ k, then for almost all x 0 ∈ Rn (x) the intersection E ∩ Πx 0 , where Πx 0 = {x =
x 0 }, has a zero α−Hausdorff measure, Hα (E ∩ Πx 0 ) = 0 in Rk (y).
Applying this theorem to E ⊂ Cn = C(n−m)+m ≈ R2(n−m)+2m , we obtain that
for almost all z = (z 1 , z 2 , . . . , z n−m ) ∈ Cn−m the intersection E ∩ Πz has the zero
(2m − 2 + α) Hausdorff measure. (The case m = n is trivial. In this case Πz ≈ Cn
and E ∩ Πz = E.)
From the definition of m − sh functions, the restriction u|Π is subharmonic in
(D\E) ∩ Πz and belongs to the class Li pα (D ∩ Πz ). Hence, for the planes Πz
for which H2m−2+α (E ∩ Πz ) = 0 the restriction u|Π is subharmonic in D ∩ Πz ,
because of the Theorem 1 stated above, since dimC Πz = m.
4 B. I. Abdullaev et al.
Since, the set of such planes has full measure then, they are dense everywhere.
From the fact that the function u belongs to the class Li pα and is continuous in
the domain D, it follows that the restriction u|Π is subharmonic in D ∩ Πz for all
planes Πz . Considering unitary transformations of the space Cn , we get that the
restriction u|Π ∈ sh (D ∩ Π ) for any Π ⊂ Cn , where dim Π = m. This means that
u ∈ m − sh (D) .
3 (n − s, q)-Capacity
where the upper bound is taken all over positive Borel measures concentrated on the
set E and satisfying the condition
1p
μ μ 1 1
Un−s (x) = Un−s (x) p d x ≤ 1, + = 1. (2)
p p q
For p > n
n−s
, that is qs < n the integral
μ
Un−s (x) p d x
|x|≥1
where, now the upper bound is taken over all positive Borel measures concentrated
on the set E and satisfying condition
⎡ ⎤ 1p
μ μ
Un−s (x) = ⎢
⎣ Un−s (x) p d x ⎥
⎦ ≤ 1, p=
n
.
p n−s
B(0,1)
Here ⎛ ⎞ q1
s q
ϕq,s = ⎝ ∇ ϕ(x) d x ⎠ =
x
Rn
⎧ ⎫1
⎨ 2 q2 ⎬q
s! ∂sϕ
= dx .
⎩
α1 +α2 +···+αn =s
α1 !α2 ! . . . αn ! ∂ x1 ∂ x2α2 . . . ∂ xnαn
α1
⎭
Rn
We note that there are constants 0 < A1 < A2 depending only on n, s and q, the
following estimation holds:
n
A1 Cn−s,q (E) ≤ γn−s,q (E) ≤ A2 Cn−s,q (E), 1 < q ≤ . (3)
s
Cn−s,q -capacity has the following metric properties (see [16]):
(a) if qs < n, 0 < α < n − qs and Hα (E) = 0, then Cn−s,q (E) = 0;
(b) if qs < n, n − qs < α and Hα (E) > 0, then Cn−s,q (E) > 0;
(c) if qs = n, ϕ(r ) = |ln r |1−q , q > 1 and Hϕ (E) < ∞, then Cn−s,q (E) = 0;
(d) if qs = n, α > 0 and Hα (E) > 0, then Cn−s,q (E) > 0.
It follows that the dimension the set of zero Cn−s,q -capacity is not greater than
n − qs.
In the classical case n > 2, 0 < s ≤ m, P. Matthila [15] (see also [24]) obtained
the following result: let E be a compact set in Rn+m = Rnx × Rmy . If the newtonian
capacity Cn+m−s (E) = 0, 0 < s ≤ m, defined by the Riesz kernel K (x) = |x|s−m−n
is zero, then for almost all a ∈ Rnx intersection E ∩ {x = a} ⊂ Rmy has a newtonian
6 B. I. Abdullaev et al.
Then, for almost all a ∈ Rnx , the intersection E ∩ {x = a} ⊂ Rmy has zero (m − s, q)-
capacity, i.e
Cm−s,q (E ∩ {x = a}) = 0.
According to the Whitney’s theorem (see [28]), for any plane Πa = {x = a},
such that E ∩ Πa = ∅, the set of restrictions ϕ|Πa = {ψ(y) = ϕ(a, y), ϕ ∈ Φ(E)}
of functions ϕ(x, y) ∈ Φ(E) coincides with the set of functions
Φ(E ∩ Πa ) = ψ(y) ∈ C0∞ (Rm ) : ψ(y) > 1, y ∈ E ∩ Πa .
Obviously,
⎛ ⎛ ⎞ ⎞ q1 ⎛ ⎞ q1
s q s q
⎝ ⎝ ∇ ϕ(x, y) dy ⎠ d x ⎠ ≤ ⎝ ∇ ϕ(x, y) d xd y ⎠
y x,y
Rn Rm Rn Rm
and
⎛ ⎛ ⎞ ⎞ q1 ⎛ ⎛ ⎞ ⎞ q1
s q s q
⎝ inf ⎝ ∇ ϕ(x, y) dy ⎠ d x ⎠ ≤ inf ⎝ ⎝ ∇ ϕ(x, y) dy ⎠ d x ⎠ ≤
ϕ y ϕ y
Rn Rm Rn Rm
⎛ ⎞ q1
s q
≤ inf ⎝ ∇ ϕ(x, y) d xd y ⎠ .
ϕ x,y
Rn Rm
⎛ ⎞ q1
" #q
⎝ γm−s,q (E ∩ Πx ) d x ⎠ ≤ γn+m−s,q (E).
Rn
According to the conditions of the theorem and the estimation (3) we have
" #q
γm−s,q (E ∩ Πx ) d x = 0.
Rn
Cm−s,q (E ∩ Πx ) = 0.
4 Main Results
The main results of this paper are Theorems 5 and 6 which we provide below.
where
p 2m
q= and m > 1, ≤ p < +∞,
p−1 2m − 2
where
p 2m
q= and m ≥ 1, ≤ p < +∞,
p−1 2m − 1
In the proofs of these theorems we will use the Theorems 4, 7 and 8 which will be
provided below on the properties of removable singular sets of subharmonic (n − sh)
functions in the class of L p and L 1p . Here L kp (D)(k is a fixed entire number) denotes
the class of functions having all derivatives up to k, furthermore k-order derivatives
belonging to L p (D).
8 B. I. Abdullaev et al.
p
Cn−2,q (E) = 0, q = .
p−1
Sketch of proof. (1) First we assume that there exists a compact set E, which is
n
removable for all subharmonic functions in D\E of the class L p , n−2 ≤ p < +∞
p
and Cn−2,q (E) > 0, q = p−1 . Then, by the definition of the capacity Cn−2,q , there
exists a positive Borel measure, such that suppμ ⊂ E, μ(E) > 0 and the potential
dμ(y)
u(x) =
|x − y|n−2
belongs to the class L p . The function u(x) is harmonic, and consequently subhar-
monic outside of E . By the assumption, it extends subharmonically into E. However,
it is not subharmonic in D, since −u(x) is a subharmonic function. We came to the
contradiction, i.e., Cn−2,q (E) = 0.
(2) Now suppose that Cn−2,q (E) = 0 and u(x) is subharmonic function in D\E,
and belong to L p (D), 1p + q1 = 1.
Further we use the following statement (see [19]): let E be a compact subset of
the space Rn , and D ⊃ E is a neighborhood of the set E. Then for a fixed integer k
the following statements are equivalent:
(1) Ck,q (E) = 0;
(2) The set of test-functions C0∞ (D\E) is dense in the set of test-functions C0∞ (D)
with respect to the norm L kp .
If now ψ(x) is a positive test-function supp ψ D, then according to this state-
ment there is a sequence of positive basic functions ϕ j (x), supp ϕ j D\E and
converges to ψ(x) with respect to the norm L kp . Then
u(x)Δ(ψ − ϕ j )d x ≤ C u L ψ − ϕ j 2 ,
p Lq
for any positive basic test-function ψ(x) ∈ C0∞ (D). Therefore, the function u(x)
subharmonically extends in whole D. The Theorem 7 is proved.
p
Cn−1,q (E) = 0 , q = .
p−1
Hence, according to Theorem 7, we obtain that for almost all Πz planes restriction
u(z)|Πz is subharmonic in D ∩ Πz .
This implies that the restriction u(z)|Πz is subharmonic in D ∩ Πz for all planes
Πz . Using a unitary change of variables Cn , we get that u(z)|Π is subharmonic in
5 Appendix
For the completeness of the materials of subject, we present one theorem on remov-
able singular sets of sub-solutions of elliptic operators from the class L kp (see [4]):
let G ⊂ Rn be a domain in space Rn and F (G) be the space of distributions in the
domain G and
P (D) = aα D α ,
|α|≤m
∂ |α|
Dα = , α = (α1 , α2 , . . . , αn ) , |α| = |α1 | + |α2 | + · · · + |αn | ,
∂ x1α1 ∂ x2α2 . . . ∂ xnαn
p
Cm−k,q (E) = 0, q = .
p−1
References
1. Abdullayev, B.I.: P -measure in the class of m − wsh functions. J. Sib. Fed. Univ. Math. Phys.
N.7(1), 3–9 (2014)
2. Abdullayev, B.I.: Subharmonic functions on complex Hyperplanes of Cn . J. Sib. Fed. Univ.
Math. Phys.-Krasn. N6(4), 409–416 (2013)
3. Abdullaev, B.I., Imomkulov, S.A.: Removable singularities of subharmonic functions in the
level L p and L 1p . Uzb. Math. J. N.4, 10–14 (1997)
4. Abdullaev, B.I., Yarmetov, Zh.R.: On singular sets of subsolutions of elliptic operators. Bull.
KrasGAU N9, 74–80 (2006)
5. Carleson, L.: Selected Problems on Exceptional Sets. Van Nostrand, Toronto (1967)
6. Dauzhanov, A.Sh.: On the smoothness of generalized solutions of elliptic equations (Russian).
Uzb. Math. J. N2, 9–19 (2006)
7. Dolzhenko, E.P.: On the singularities of continuous harmonic functions. Izv. Akad. Nauk. SSSR
Ser. Math. 28(6), 1251–1270 (1964)
8. Dolzhenko, E.P.: On the representation of continuous harmonic functions in the form of poten-
tials. Izv. Akad. Nauk SSSR Ser. Math. 28(5), 1113–1130 (1964)
9. Harve, R., Polking, J.C.: A notion of capacity which characterizes removable singularites.
Trans. Am. Math. Soc. 169, 183–195 (1968)
10. Harvey, F.R., Lawson Jr., H.B.: p− convexity, p− plurisubharmonicity and the Levi problem.
Indiana Univ. Math. J. 62(N1), 149–169 (2013)
11. Hayman, W.K., Kennedy, P.B.: Subharmonic Functions. Vol 1. London Mathematical Society
Monographs, vol. 9. Academic Press, London (1976)
12. Joyce, D.: Riemannian Holonomy Groups and Calibrated Geometry. Oxford Graduate Texts
in Mathematics, vol. 12. OUP, Oxford (2007)
13. Khusanov, Z.Kh.: Capacity properties of q-subharmonic functions. I, (Russian). Izv. Akad.
Nauk UzSSR Ser. Fiz.-Mat. Nauk 97(1), 41–45 (1990)
14. Khusanov, Z.Kh.: Capacity properties of q-subharmonic functions. II, (Russian). Izv. Akad.
Nauk UzSSR Ser. Fiz.-Mat. Nauk 94(5), 28–33 (1990)
15. Mattila, P.: Integral geometric properties of capacities. Trans. Am. Math. Soc. 266, 539–554
(1981)
16. Maz’ya, V.G., Havin, V.P.: Non-linear potential theory. Russ. Math. Surv. 27(6), 71–148 (1972)
Removable Singular Sets of m-Subharmonic Functions 11
17. Maz’ya, V.G.: On ( p, l)-capacity, inbedding theorems, and the spectrum of a selfadjoint elliptic
operator. Math. USSR-Izv. 7(2), 357–387 (1973)
18. Mel’nikov, M.S., Sinanyan, S.O.: Questions of the theory of approximation of functions of a
complex variable. J. Sov. Math. 5(5), 688–752 (1976)
19. Maz’ya V.G.: Classes of sets and measures connected with imbedding theorems (Russian). In:
Embedding Theorems and Their Applications. Proceedings of the symposium on imbedding
theorems, pp. 142–159. M. Nauka (1970)
20. Sadullaev, A.S., Yarmetov, Zh.R.: Removable singularities of subharmonic functions of class
Li pα . Sb. Math. 186(1), 133–150 (1995)
21. Sadullaev, A.S., Abdullaev, B.: A removable singularity of the bounded above m − wsh func-
tions, (Russian). Dok. Akad Nauk N5, 12–14 (2015)
22. Sadullaev, A.S., Abdullaev, B.: Removable singularity m − wsh functions of the class Li pα ,
(Russian). Bull. Natl. Univ. Uzb. N1, 4–6 (2015)
23. Sadullaev, A.S., Abdullaev, B.I., Sharipov, R.A.: A removable singularity of the bounded above
m − sh functions, (Russian). Uzbek Math. J. N3, 118–124 (2016)
24. Sadullaev, A.S.: Rational approximation and pluripolar sets. Math. USSR-Sb. 47(1), 91–113
(1984)
25. Shapiro, V.L.: Subharmonic functions and Hausdorff measure. J. Differ. Equ. 27(1), 28–45
(1978)
26. Shiffman, B.: On the removal of singularities of analytic sets. Mich. Math. J. 15, 111–120
(1968)
27. Verbitsky, M.: Plurisubharmonic functions in calibrated geometry and q-convexity. Math. Z.
264(4), 939–957 (2010)
28. Whitney, H.: Analytic extensions of differentiable functions defined in closed set. Trans. Am.
Math. Soc. 36, 63–89 (1934)
Extensions of Bernstein’s Lethargy
Theorem
1 Introduction
The formal beginnings of approximation theory date back to 1885, with Weierstrass’
celebrated approximation theorem [31]. The discovery that every continuous function
defined on a closed interval [a, b] can be uniformly approximated as closely as desired
by a polynomial function immediately prompted many new questions. One such
question concerned approximating functions with polynomials of limited degree.
That is, if we limit ourselves to polynomials of degree at most n, what can be said of
the best approximation? As it turns out, there is no unified answer to this question.
In fact, S. N. Bernstein [11] in 1938 showed that there exist functions whose best
approximation converges arbitrarily slowly as the degree of the polynomial rises.
In this paper, we take up this aptly-named “Lethargy Theorem” of Bernstein and
present two extensions. For f ∈ C([0, 1]), the sequence of the best approximation
errors is defined as:
A. G. Aksoy (B)
Department of Mathematical Sciences, Claremont McKenna College,
Claremont, CA 91711, USA
e-mail: aaksoy@cmc.edu
ρ( f, P1 ) ≥ ρ( f, P2 ) ≥ · · ·
ρ( f, Pn ) = dn , for all n ≥ 1.
This remarkable result is called Bernstein’s Lethargy Theorem (BLT) and is used in
the constructive theory of functions [37], and it has been applied to the theory of quasi
analytic functions in several complex variables [33, 34]. Also see [18] and references
therein for an application of BLT to the study Gonchar quasianalytic functions of
several variables.
Bernstein’s proof is based on a compactness argument and only works when the
subspaces are finite dimensional. Note that the density of polynomials in C[0, 1] (the
Weierstrass Approximation Theorem) implies that
lim ρ( f, Pn ) = 0.
n→∞
ρ(x, Yn ) = O(dn ), as n → ∞.
Note that Shapiro’s result is not restricted to finite dimensional subspaces Yn . This
result was later strengthened by Tyuriemskih [39] . He showed that the sequence of
errors of the best approximation from x to Yn , {ρ(x, Yn )} may converge to zero at an
arbitrary slow rate. More precisely, for any expanding sequence {Yn } of subspaces
of X and for any null sequence {dn } of positive numbers, he constructed an element
x ∈ X such that
However, it is also possible that the errors of the best approximation {ρ(x, Yn )} may
converge to zero arbitrarily fast; for results of this type see [9].
We refer the reader to [14] for an application of Tyuriemskih’s Theorem to con-
vergence of sequences of bounded linear operators.
We also refer to [3, 5, 9, 10, 23, 24] for other versions of Bernstein’s Lethargy
Theorem and to [4, 6, 26, 32, 41] for Bernstein’s Lethargy Theorem for Fréchet
spaces.
Given an arbitrary Banach space X , a strictly increasing sequence {Yn } of sub-
spaces of X and a non-increasing null sequence {dn } of non-negative numbers, one
can ask the question whether there exists x ∈ X such that ρ(x, Yn ) = dn for each
n? For a long time, no sequence {dn } of this type was known for which such an
element x exists for all possible Banach spaces X . The only known spaces X in
which the answer is always “yes” are the Hilbert spaces (see [7, 8, 40]). For a gen-
eral (separable) Banach space X , a solution x is known to exist whenever all Yn are
finite-dimensional (see [38]). Moreover, it is known that if X has the above property,
then it is reflexive (see [40]).
Borodin [12] proved the above Lemma 1 by taking (X, · ) to be a Banach space.
Returning to the question posed before, namely given an arbitrary Banach space
X , a strictly increasing sequence {Yn } of subspaces of X and a non-increasing null
sequence {dn } of non-negative numbers, one can ask the question whether there exists
x ∈ X such that ρ(x, Yn ) = dn for each n? For a long time no sequence {dn } of this
type was known for which such an element x exists for all possible Banach spaces
X . Borodin in [12] uses the above lemma for Banach space to establish the existence
of such an element in case of rapidly decreasing sequences; more precisely, in 2006
he proves the following theorem:
Theorem 1 (Borodin [12]) Let X be an arbitrary Banach space (with finite or infi-
nite dimension), Y1 ⊂ Y2 ⊂ · · · be an arbitrary countable system of strictly nested
subspaces in X , and fix a numerical sequence {dn }n≥1 satisfying: there exists a natural
number n 0 ≥ 1 such that
16 A. G. Aksoy
∞
dn > dk for all n ≥ n 0 at which dn > 0. (2)
k=n+1
The condition (2) on the sequence {dn } is the key to the derivation of (3) in
Theorem 1. Based on this result, Konyagin [22] in 2013 takes a further step to
show that, for a general non-increasing null sequence {dn }, the deviation of x ∈ X
from each subspace Yn can range in some interval depending on dn .
Theorem 2 (Konyagin [22]) Let X be a real Banach space, Y1 ⊂ Y2 ⊂ · · · be a
sequence of strictly nested closed linear subspaces of X , and d1 ≥ d2 ≥ · · · be a
non-increasing sequence converging to zero, then there exists an element x ∈ X
such that the distance ρ(x, Yn ) satisfies the inequalities
Note that the condition (2) is satisfied when dn = (2 + ε)−n for ε > 0 arbitrarily
small, however it is not satisfied when dn = 2−n . Of course there are two natural
questions to ask:
Question 1 Is the condition (2) necessary for the results in Theorem 1 to hold, or
does Theorem 1 still hold for the sequence dn = 2−n , n ≥ 1?
Question 2 Under the same conditions given in Theorem 2, can the lower and
upper bounds of ρ(x, Yn ) in (4) be improved?
Both of these questions are answered and two improvements on a theorem of S. N.
Bernstein for Banach spaces are presented in [2]. A positive answer to Question 1 is
obtained in [2] by showing that Theorem 1 can be extended by weakening the strict
inequality in (2) to a non-strict one:
∞
dn ≥ dk , for every n ≥ n 0 . (5)
k=n+1
Clearly, the condition (5) is weaker than (2), but unlike the condition (2), (5) is
∞
satisfied by the sequences {dn }n≥1 verifying dn = dk for all n ≥ n 0 . For a
k=n+1
typical example of such a sequence one can take {dn } = {2−n }. We have also shown
that if X is an arbitrary infinite-dimensional Banach space, {Yn } is a sequence of
strictly nested subspaces of X , and if {dn } is a non-increasing sequence of non-
negative numbers tending to 0, then for any c ∈ (0, 1] we can find xc ∈ X , such that
the distance ρ(xc , Yn ) from xc to Yn satisfies
We prove the above inequality by first improving [12] result for Banach spaces
by weakening his condition on the sequence {dn }. The weakened condition on dn
requires a refinement of Borodin’s construction to extract an element in X , whose
distances from the nested subspaces are precisely the given values dn .
Now, we are ready to state the following theorem [2] which improves the theorem
of Borodin [12].
Theorem 3 (Aksoy, Peng [2]) Let X be an arbitrary infinite-dimensional Banach
space, {Yn }n≥1 be an arbitrary system of strictly nested subspaces with the prop-
erty Y n ⊂ Yn+1 for all n ≥ 1, and let the non-negative numbers {dn }n≥1 satisfy the
following property: there is an integer n 0 ≥ 1 such that
∞
dn ≥ dk , for every n ≥ n 0 .
k=n+1
Then, for any c ∈ (0, 1], there exists an element xc ∈ X (depending on c) such that
Since the sequence {d̃n }n=1,2,...,ni0 −1 ∪ {d̃ni }i≥i0 satisfies the condition (5) and Ỹ n ⊂
Ỹn+1 for all n ≥ 1, then we can apply Theorem 3 to get x ∈ X so that
ρ(x, Ỹn ) = d̃n , for n = 1, . . . , n i0 − 1, and ρ(x, Ỹni ) = d̃ni , for all i ≥ i 0 . (7)
Case 2 if n i < n < n i+1 for some i ≥ i 0 , then the fact that {d̃n } is non-increasing
and Ỹni ⊂ Ỹn ⊂ Ỹni+1 leads to
ρ(x, Ỹn ) ∈ ρ(x, Ỹni+1 ), ρ(x, Ỹni ) = K 2−(i+1) , K 2−i
and
d̃n ∈ K 2−i , K 2−i+1 .
It follows that
ρ(x, Ỹn ) K 2−i−1 K 2−i 1
∈ , = ,1 .
d̃n K 2−i+1 K 2−i 4
1
d̃n ≤ ρ(x, Ỹn ) ≤ d̃n for all n ≥ 1.
4
For c ∈ (0, 1], taking xc = 4cx in the above inequalities, we obtain
Remembering that {(dn , Yn )}n≥1 ⊆ {(d̃n , Ỹn )}n≥1 , we then necessarily have
3 1 ρ(x, Yn )
The interval length makes , 1 the “narrowest” estimating interval of
4 4 dn
that Theorem 4 could provide.
The subspace condition given in Theorem 4 states that the nested sequence {Yn }
has “enough gaps” so that the sequence
satisfies dn ≥ dn+1 → 0 and Yn ⊂ Yn+1 for all n ≥ 1. For another subspace condition
see the results in [1].
Extensions of Bernstein’s Lethargy Theorem 19
Observe that in Konyagin’s paper [22] it is assumed that {Yn } are closed and
strictly increasing. In Borodin’s paper [12], this is not specified, but from the proof
of Theorem 1 it is clear that his proof works only under the assumption that Y n is
strictly included in Yn+1 . The subspace condition Yn ⊂ Yn+1 does not come at the
expense of our assumption to weaken the condition on the sequence dn . This is a
natural condition. To clarify the reason why almost all Lethargy theorems have this
condition on the subspaces, we give the following example.
Example 1 Let X = L ∞ [0, 1] and consider C[0, 1] ⊂ L ∞ [0, 1] and define the sub-
spaces of X as follows:
1. Y1 = space of all polynomials;
2. Yn+1 =span[Yn ∪ { f n }] where f n ∈ C[0, 1] \ Yn , for n ≥ 1.
Observe that by the Weierstrass Theorem we have Y n = C[0, 1] for all n ≥ 1. Take
any f ∈ L ∞ [0, 1] and consider the following cases:
(a) If f ∈ C[0, 1] , then
Note that in the above, we have used the fact that ρ( f, Yn ) = ρ( f, Y n ). Hence
in this case BLT does not hold.
Fréchet spaces are locally convex spaces that are complete with respect to a translation
invariant metric, and they are generalization of Banach spaces which are normed
linear spaces, complete with respect to the metric induced by the norm. However,
there are metric spaces which are not normed spaces. This can be easily seen by
considering the space s the set of all sequence x = (xn ) and defining d(x, y) =
∞
|xi − yi |
2−i as a metric on s. If we let λ ∈ R then
i=1
(1 + |xi − yi |)
∞
1 |λ||xi − yi |
d(λx, λy) = i (1 + |λ||x − y |)
= |λ|d(x, y) (no homogeneity).
i=1
2 1 i
Many Fréchet spaces X can also be constructed using a countable family of semi-
norms ||x||k where X is a complete space with respect to this family of semi-norms.
For example a translation invariant complete metric inducing the topology on X can
be defined as
∞
||x − y||k
d(x, y) = 2−k for x, y ∈ X.
k=0
1 + ||x − y||k
Clearly, every Banach space is a Fréchet space, and the other well known example of
a Fréchet space is the vector space C ∞ [0, 1] of all infinitely differentiable functions
f : [0, 1] → R where the semi-norms are
For more information about Fréchet spaces the reader is referred to [20, 35].
Recently in [4] a version of Bernstein Lethargy Theorem (BLT) was given for
Fréchet spaces. More precisely, let X be an infinite-dimensional Fréchet space and
let V = {Vn } be a nested sequence of subspaces of X such that Vn ⊆ Vn+1 for any
n ∈ N. Let dn be a decreasing sequence of positive numbers tending to 0. Under an
additional natural condition on sup{dist(x, Vn )}, we proved that there exists x ∈ X
and n o ∈ N such that
dn
≤ ρ(x, Vn ) ≤ 3dn ,
3
for any n ≥ n o . By using the above theorem, it is also possible to obtain an extension
of both Shapiro’s [36] and Tyuremskikh’s [39] theorems for Fréchet spaces as well.
Notation. Let (X, . F ) be a Fréchet space and assume that V = {Vn } is a nested
sequence of linear subspaces of X satisfying Vn ⊂ Vn+1 . Let dn,V denote the devia-
tion of Vn from Vn+1 defined as:
Vn = {x ∈ X : xk = 0 for k ≥ n + 1}.
2 1 2
Let dn = and observe that for any x ∈ X, ρ(x, Vn ) ≤ n < . Also observe that
n 2 n
1
dn,V = n+1 which implies that dV = 0.
2
Thus in the case when dV = 0, we cannot even hope to prove Shapiro’s theorem.
Above example also shows that it is impossible to prove the Tyuriemskih Theorem
or Konyagin’s type result in Fréchet spaces without additional assumptions, because
they are stronger statements than Shapiro’s theorem. Note that if X is a Banach space,
then the condition dV = inf{dn,V : n ∈ N} > 0 is satisfied automatically. It can be
seen easily that dn,V = +∞ for Banach spaces. because
ρ(t x, Vn ) = tρ(x, Vn )
and the supremum taking over all v ∈ Vn+1 and Vn is strictly included in Vn+1 . The
next, example illustrates that there is a natural way to build Fréchet spaces where
dn,V = 1 .
Example 3 Let (X, . ) be a Banach space. Define in X an F-norm F by: x F =
x /(1 + x ). Then dn,V = 1 for any n ∈ N independently of V . Because the
mapping
t
t →
1+t
ρ(t x, Vn )
ρ F (t x, Vn ) = →1
1 + ρ(t x, Vn )
as t → ∞.
Theorem 5 (Aksoy, Lewicki, [4]) Let X be a Fréchet space and and assume that
V = {Vn } is a nested sequence of linear subspaces of X satisfying Vn ⊆ Vn+1 , where
the closure is taken with respect to · F . Let dn,V be defined as above and {en } be
a decreasing sequence of positive numbers satisfying
∞
2 j−n (δ j + e j ) < min{dn,V , en−1 }
j=n
with a fixed sequence of positive numbers δ j . Then, there exists x ∈ X such that
Another random document with
no related content on Scribd:
PLEASE READ THIS BEFORE YOU DISTRIBUTE OR USE THIS WORK
1.D. The copyright laws of the place where you are located also
govern what you can do with this work. Copyright laws in most
countries are in a constant state of change. If you are outside
the United States, check the laws of your country in addition to
the terms of this agreement before downloading, copying,
displaying, performing, distributing or creating derivative works
based on this work or any other Project Gutenberg™ work. The
Foundation makes no representations concerning the copyright
status of any work in any country other than the United States.
1.E.6. You may convert to and distribute this work in any binary,
compressed, marked up, nonproprietary or proprietary form,
including any word processing or hypertext form. However, if
you provide access to or distribute copies of a Project
Gutenberg™ work in a format other than “Plain Vanilla ASCII” or
other format used in the official version posted on the official
Project Gutenberg™ website (www.gutenberg.org), you must, at
no additional cost, fee or expense to the user, provide a copy, a
means of exporting a copy, or a means of obtaining a copy upon
request, of the work in its original “Plain Vanilla ASCII” or other
form. Any alternate format must include the full Project
Gutenberg™ License as specified in paragraph 1.E.1.
• You pay a royalty fee of 20% of the gross profits you derive from
the use of Project Gutenberg™ works calculated using the
method you already use to calculate your applicable taxes. The
fee is owed to the owner of the Project Gutenberg™ trademark,
but he has agreed to donate royalties under this paragraph to
the Project Gutenberg Literary Archive Foundation. Royalty
payments must be paid within 60 days following each date on
which you prepare (or are legally required to prepare) your
periodic tax returns. Royalty payments should be clearly marked
as such and sent to the Project Gutenberg Literary Archive
Foundation at the address specified in Section 4, “Information
about donations to the Project Gutenberg Literary Archive
Foundation.”
• You comply with all other terms of this agreement for free
distribution of Project Gutenberg™ works.
1.F.
Most people start at our website which has the main PG search
facility: www.gutenberg.org.