Skip to main content

Estimates of probabilistic widths of the diagonal operator of finite-dimensional sets with the Gaussian measure

Abstract

In this paper, we estimate the asymptotic orders of probabilistic and average widths of the compact embedding operators from the Sobolev space W 2 r (T) into L q (T) (1q) with the Gaussian measure.

MSC:41A10, 41A46, 42A61, 46C99.

1 Introduction and main results

Problems of n-widths in the approximation theory have by now been studied in depth. A great deal of classical problems have been solved, and interesting new developments have appeared. For example, the problems of probabilistic, average and stochastic widths, which can reflect the behavior of function on the whole class and give information about the measure of the elements in the class that can be approximated to this or that degree, are the problems of this kind. For the results related to the probabilistic, average and stochastic widths, the reader may be referred to Sul’din [1, 2], Traub et al. [3], Maiorov [47], Mathé [812], Sun [13, 14], and Ritter [15]. The new developments in this direction can be found in Fang’s papers [1621]. Moreover, Carl and Pajor [22] proved an inequality with respect to the Gelfand numbers of an operator u from 1 N into a Hilbert space, from which one can immediately derive the inequality related to the Kolmogorov numbers by the known duality. In this article we continue the previous works and prove the estimates of probabilistic widths of the diagonal operators from R m onto  q m .

First, we recall some useful concepts. Let W be a bounded subset of a normed linear space X with the norm X , and F N be an N-dimensional subspace of X. The quantity

e(W, F N ,X)= sup x W e(x, F N ,X),

where

e(x, F N ,X)= inf y F N x y X

is called the deviation of W from F N . It shows how well the ‘worst’ elements of W can be approximated by F N . The number

d N (W,X)= inf F N e(W, F N ,X)= inf F N sup x W inf y F N x y X ,

where F N runs through all possible linear subspaces of X of dimension at most N, is called the Kolmogorov’s N-width of W in X. Assume that W contains a Borel field consisting of open subsets of W and equipped with a probabilistic measure μ defined on . That is, μ is a σ-additive nonnegative function on , and μ(W)=1. Let δ[0,1] be an arbitrary number. The corresponding probabilistic Kolmogorov’s (N,δ)-width of a set W with a measure μ in the space X is defined by

d N , δ (W,μ,X)= inf G δ d N (WG,X),
(1)

where G δ runs through all possible subsets in with measure μ( G δ )δ. The p-average Kolmogorov’s N-width is defined by

d N ( a ) ( W , μ , X ) p = inf F N ( W e ( x , F N , X ) p d μ ( x ) ) 1 / p ,0<p<,
(2)

where F N in (2) runs over all linear subspaces of X of dimension at most N. Let p m be an m-dimensional normed space of vectors x=( x 1 ,, x m ) R m , with a norm

x p m ={ ( i = 1 m | x i | p ) 1 / p , 1 p < , max 1 i m | x i | , p = .

Consider in R m the standard Gaussian measure v= v m , which is defined as

v(G)= ( 2 π ) m / 2 G exp ( 1 2 x 2 2 ) dx,

where G is any Borel subset in R m . Obviously, v( R m )=1.

Denote by B p m (ρ)={x p m : x p ρ} the ball of radius ρ in p m . Let B p m = B p m (1).

Let N=0,1, , δ[0,1) be arbitrary and T m be a linear invertible operator from R m onto q m . We define the probabilistic (N,δ)-width of the operator acting in space R m equipped with the Gaussian measure v in q m -norm:

d N , δ ( T m : R m q m , v ) = inf G inf L N e ( T m ( R m G ) , L N , q m ) ,

where v(G)<δ, dim L N N.

Maiorov in [6] proved the following result.

Theorem A [6]

For m>2N, δ(0,1/2], 1q2, then

d N , δ ( R m , q m , v ) m 1 q 1 2 m + ln ( 1 / δ ) .

In [22], Carl and Pajor proved the following result with respect to Gelfand numbers of an operator with values in a Hilbert space.

Theorem B [22]

Let T be an operator from 1 m into a Hilbert space H. Then

d N (T)C ( log ( m N + 1 ) N ) 1 / 2 T

for 1Nm, m=1,2, , where C>0 is a universal constant.

Detailed facts about the usual widths, such as the Kolmogorov’s N-widths and N th Gelfand numbers (or Gelfand N-widths) of T were given in the books [2326].

Remark 1

  1. (a)

    Theorem A shows the asymptotic expression of the probabilistic widths of the identity embedding from R m into q m , 1q2.

  2. (b)

    Theorem B gives the upper estimate of Gelfand numbers of operators from 1 m into a Hilbert space, and some of its striking applications in the geometry of Banach spaces and Rademacher processes can be found in [22]. By the dual relation, it is easy to obtain the similar upper estimate of the Kolmogorov’s N-widths d N (T) of operators from 2 m into m , i.e.,

    d N (T)C ( log ( m N + 1 ) N ) 1 / 2 T.
  3. (c)

    Motivated by Theorems A and B, in general cases, here we investigate the asymptotic estimate of probabilistic widths for diagonal operators from R m onto q m , 1q.

Now we are in a position to formulate our main results.

Theorem 1 For m>N, δ(0,1/2], then

d N , δ ( T m : R m m , v ) C T m ( 1 + ( 1 / N ) ln ( 1 / δ ) ) ln ( e m / N ) .

Theorem 2 For m>2N, δ(0,1/2], then

d N , δ ( T m : R m 1 m , v ) C T m m + ln ( 1 / δ ) .

2 Proof of main results

In order to prove Theorems 1 and 2, we also need some auxiliary assertions.

Lemma 1 Let δ(0, 2 / e π ], and let T m be a bounded linear invertible operator from R m onto m . Then, for any vector z R m ,

v ( x : | ( T m x , z ) | 2 T m ln ( 1 / δ ) z 2 ) δ.

Proof First, assume that T m is a diagonal operator of R m , i.e., T m x= ( λ i x i ) i = 1 m , for any x R m . Without loss of generality, assume that the sequence of the absolute of eigenvalues  λ i , i=1,,m, is arranged non-increasingly, i.e., | λ 1 || λ 2 || λ m |>0. It is known that

T m = max 1 i m | λ i |=| λ 1 |.

Since v is invariant with respect to orthogonal transformation of R m , it suffices to prove the lemma for the vector z =( z 2 ,0,,0). Let ε(0, e 1 ] be arbitrary. We have

v ( x : | ( T m x , z ) | T m ln ( 1 / ε ) z 2 ) = v ( x : | λ 1 x 1 | z 2 | λ 1 | ln ( 1 / ε ) z 2 ) = v ( x : | x 1 | ln ( 1 / ε ) ) = 2 π ( 1 / 2 ) ln ( 1 / ε ) exp ( t 2 ) d t ( 2 ε π ) 1 / 2 .
(3)

Here we use the inequality

u exp ( t 2 ) dt< 1 2 u exp ( u 2 ) ,u 1 2 .

From (3) we obtain the assertion of the lemma by δ= 2 ε / π .

Next, assume that T m is a symmetric transformation of R m , then there is an orthogonal matrix U of order m such that the matrix U T m U T is a diagonal matrix. Since the Gaussian measure is invariant for orthogonal transformation, the result holds for symmetric transformation T m .

Finally, assume that T m is a general invertible linear transformation from R m onto R m , then there are two matrices U and S such that T m =US, where U is an orthogonal matrix and S is a positive definite symmetric matrix. As the same reason above, the result holds for the transformation T m .

Thus Lemma 1 is proved. □

The following inequality will be used (see [27]). For any integers N and m with m>N1, there exists a subspace H of R m of dimension dimHmN such that for any xH,

x 2 c 0 ( ln ( e m / N ) N ) 1 / 2 x 1 ,
(4)

where c 0 is an absolute constant.

Let G R m be a set. We introduce in R m another norm for the operator T m : R m m :

x G = sup y R m G | ( T m y , x ) | .

Lemma 2 For any δ(0,1/2] and an arbitrary operator T m from R m onto m , there exists a subset G= G δ of R m with measure v(G)δ such that

sup z B 1 m H z G c 1 T m ( 1 N ln exp ( a N ) δ ln e m N ) 1 / 2 ,

where a and c 1 are absolute constants.

Proof Let k=[8(N+1)/ln(em/N)] and consider the k 1 -net

S= { ( s 1 / k , , s m / k ) : s 1 , , s m Z , | s 1 | + + | s m | k }

for the B 1 m in m -norm. Using the inequality ( m ) ( e m ) , we estimate the cardinality of S:

cardS = 1 k 2 ( m + ) = 1 k 2 ( 4 e m ) ( 8 e m k ) k e a N ,
(5)

where a is some absolute constant.

Consider the polyhedron Q= B 1 m k 1 B m . Let Q be the set of extremal points of Q. The set Q consists of vectors with k coordinates equal to ± k 1 and the remaining coordinates zero. This implies that Q S, and hence card Q exp(aN).

Let ε=δ/exp(aN). In R m we consider the set G= s S G s , where

G s = { y R m : | ( T m y , s ) | 2 T m ln ( 1 / ε ) s 2 } .

Let z B 1 m H be any point, and sS be a point closest to z in m -norm. Then z=s+t for some tQ. From Lemma 1,

z G s G + t G 2 T m ln ( 1 / ε ) s 2 + t G .
(6)

Using the definition of Q, we have t 2 2 t 1 t k 1 . From this and the definition of Q and G,

t G max t Q t G max t S Q t G 2 T m ln ( 1 / ε ) max t S Q t 2 2 T m k 1 ln ( 1 / ε ) .

Therefore from (6),

z G 2 T m ln ( 1 / ε ) ( z 2 + t 2 ) + t G 2 T m ln ( 1 / ε ) ( z 2 + 2 k 1 / 2 ) .
(7)

Since zH, it follows from the inequalities (4) and (7) that

z G c 1 T m k 1 ln ( 1 / ε ) c 2 T m ( 1 / N ) ln ( 1 / ε ) ln ( e m / N ) .

Using Lemma 1 and the inequality (5), we can estimate the measure of G:

v(G) s S v( G s )εcardSεexp(aN)=δ.

Thus, Lemma 2 is proved. □

Proof of Theorem 1 Using the duality in R m and Lemma 2, we have

sup x R m G inf y H T m x y = sup x R m G sup z H B 1 m | ( T m x , z ) | = sup z H B 1 m z G c T m ( 1 / N ) ln ( 1 / ε ) ln ( e m / N ) = c T m ( 1 / N ) ln ( exp ( a N ) / δ ) ln ( e m / N ) ,

where H is the orthogonal complement of H and dim H N. The proof of Theorem 1 is completed. □

Let us proceed to the proof of Theorem 2. For this, we first prove four lemmas. We introduce a definition. For arbitrary ε>0, the ε-cardinality of a subset K of 1 m is defined to be

N ε (K)=min { N : z 1 , , z N R m , e ( K , { z 1 , , z N } ) ε } ,

where

e ( K , { z 1 , , z N } ) = sup x K min i = 1 , , N x z i 1

is the deviation of K from the set { z 1 ,, z N } in 1 m .

Let λ be a Lebesgue measure in , normalized by the condition λ(B)=1, where B= B 2 m . We consider Kolmogorov’s (N,δ)-width of the ball B with measure λ in the 1 m -norm:

d N , δ d N , δ ( T m : B 1 m , λ ) = inf G inf L e ( T m ( B G ) , L , 1 m ) ,

where the T m is as above, the infima are over all possible subsets GB of measure λ(G)δ and all subspaces L R m with dimLN.

Lemma 3 Suppose that DB is an arbitrary subset with measure λ(D)δ. Then, for any ε> d N , δ ,

N ε ( T m ( B D ) ) ( 1 + 4 T m ε d N , δ ) m .

Proof Let h>0 be any number, and let H be any subspace of with dimHN such that

e ( T m ( B D ) , H , 1 m ) h d N , δ .
(8)

Let ε =ε d N , δ . We consider the set Q=2( T m (BD))H. Let Q ε ={ z 1 ,, z N } be the maximal subset of Q such that z i z j 1 ε for all ij. Clearly, by maximality Q ε is a ε -net of Q for 1 . The balls z i +( ε /2) B 1 m are disjoint and all contained in Q+( ε /2) B 1 m . Therefore, taking volumes we can obtain

i = 1 N vol ( z i + ( ε / 2 ) B 1 m ) vol ( Q + ( ε / 2 ) B 1 m ) .

Hence, we have

N ( ε / 2 ) vol ( B 1 m ) vol ( Q + ( ε / 2 ) B 1 m ) .
(9)

By Q=2( T m (BD))H2 T m (B)2 T m ( B 1 m )2 T m ( B 1 m ) and (9), we have

N ( ε / 2 ) m vol ( B 1 m ) ( 2 T m + ( ε / 2 ) ) m vol ( B 1 m ) ,

that is,

N ( 1 + 4 T m ε ) m .
(10)

Now, we need to establish e( T m (BD),{ z 1 ,, z N })ε. Since tQH, we have from (8)

sup x T m ( B D ) min t Q x t 1 sup x T m ( B D ) min t Q min h 0 H ( x h 0 1 + h 0 t 1 ) sup x T m ( B D ) min t Q x h 0 1 + sup x T m ( B D ) min t Q min h 0 H h 0 t 1 d N , δ + h .
(11)

From the inequality (11), the definition of Q ε and (8), it follows that

e ( T m ( B D ) , { z 1 , , z N } ) = sup x T m ( B D ) min i = 1 , , N x z i 1 sup x T m ( B D ) min i = 1 , , N min t Q ( x t 1 + t z i 1 ) sup x T m ( B D ) min t Q ( x t 1 + min i = 1 , , N t z i 1 ) = sup x T m ( B D ) min t Q x t 1 + ε d N , δ + h + ε = ε + h .

Consequently, letting h0, we get that e( T m (BD),{ z 1 ,, z N })ε, which together with (10) completes the proof of Lemma 3. □

From the relation (see [28])

vol ( B p m ) = [ 2 Γ ( 1 / p + 1 ) ] m /Γ(m/p+1),1p,

the balls B p m satisfy the inequalities

( c p m ) m / p <vol ( B p m ) < ( c p m ) m / p ,
(12)

where Γ is the Euler Γ-function, and c p , c p depend only on p.

To estimate N ε ( T m (BD)) from below, we now need another auxiliary result.

Lemma 4 If T m is a diagonal operator from R m onto 1 m , then

| det ( T m ) | =| i = 1 m λ i ( T m )| ( T m m ) m ,

where λ i ( T m ), i=1,,m, are non-zero eigenvalues of the operator T m rearranged as usual so that | λ i ( T m )| is non-increasing and each eigenvalue is repeated according to its multiplicity.

Proof It is known that

T m = ( i = 1 m ( λ i ( T m ) ) 2 ) 1 / 2 .

Accordingly,

m | λ m ( T m ) | T m m | λ 1 ( T m ) | .

Obviously,

| λ m ( T m ) | m | i = 1 m λ i ( T m )| | λ m ( T 1 ) | m ,

from which the result of Lemma 4 follows immediately. □

Lemma 5 If δ[0,1] and λ(D)δ, then

N ε ( T m ( B D ) ) 1 3 (1δ) ( c 0 T m / ε ) m .

Proof We first establish the inequality

N ε ( T m ( B D ) ) 1 3 λ ( T m ( B D ) ) λ ( B 1 m ( 2 ε ) ) .
(13)

Indeed, suppose that (13) does not hold. Then, for N= N ε ( T m (BD)) and some set of points z 1 ,, z N , by

ε sup x T m ( B D ) min i = 1 , , N x z i 1 1 λ ( T m ( B D ) ) T m ( B D ) min i x z i 1 λ ( d x ) 1 λ ( T m ( B D ) ) T m ( B D ) i = 1 N ( z i + B 1 m ( 2 ε ) ) min i x z i 1 λ ( d x ) 1 λ ( T m ( B D ) ) ( 2 ε ) λ ( T m ( B D ) i = 1 N ( z i + B 1 m ( 2 ε ) ) ) = 2 ε λ ( T m ( B D ) ) N λ ( B 1 m ( 2 ε ) ) λ ( T m ( B D ) ) = 2 ε ( 1 N λ ( B 1 m ( 2 ε ) ) λ ( T m ( B D ) ) ) 2 ε 2 3 = 4 3 ε ,

we have obtained a contradiction.

In the sequel, we may as well assume that T m is a diagonal operator from R m onto 1 m . Using the inequality (13), (12) and Lemma 4, we have

N ε ( T m ( B D ) ) 1 3 λ ( T m ( B D ) ) λ ( B 1 m ( 2 ε ) ) = 1 3 | det ( T m ) | λ ( B D ) λ ( B 1 m ( 2 ε ) ) 1 3 c ( T m m ) m ( 1 δ ) vol ( B ) vol ( B 1 m ( 2 ε ) ) 1 3 ( c 1 T m m ) m ( 1 δ ) ( c 2 m 2 ε ) m = 1 3 ( 1 δ ) ( c 0 T m ε ) m .

Next, assume that T m is a symmetric transformation of R m , then there is an orthogonal matrix U of order m such that the matrix U T m U T is a diagonal matrix. Since the Lebesgue measure is invariant for orthogonal transformation, the result holds for symmetric transformation T m .

Finally, in the general case, T m is a general invertible linear transformation from R m onto 1 m , then there are two matrices U and S such that T m =US, where U is an orthogonal matrix and S is a positive definite symmetric matrix. As the same reason above, the result holds for the transformation T m .

Thus, we complete the proof of Lemma 5. □

Lemma 6 If δ[0,1/2], then

d N , δ ( T m : B 1 m , λ ) c 1 T m .

Proof From Lemma 3 and Lemma 5, we get

c ( 1 + 4 T m ε d N , δ ) m N ε ( T m ( B D ) ) 1 3 (1δ) ( c 0 T m ε ) m 1 6 ( c 0 T m ε ) m .
(14)

Let ε=5 d N , δ . Taking the logarithm of the inequality (14), we get the inequality

d N , δ ( T m : B 1 m , λ ) c 1 T m

for some constants c 1 and c 2 and N with m c 2 N. Lemma 6 is proved. □

Proof of Theorem 2 According to Lemma 6, for any δ[0,1/2] and any subspace L R m with dimLN, there is a set KB with Lebesgue measure λ(K)>δ such that

e ( T m x , L , 1 m ) c 1 T m
(15)

for any element xK. On the unit sphere S m 1 ={x R m : x 2 =1}, we consider the subset K ={x/ x 2 :xK}.

Let λ S m 1 be a Lebesgue measure on the sphere S m 1 . We prove that

λ S m 1 ( K ) >δ λ S m 1 ( S m 1 ) .
(16)

Indeed, assume that

λ S m 1 ( K ) δ λ S m 1 ( S m 1 ) .

We introduce in R m a polar system of coordinates (r,s), where r0 and s S m 1 , and consider in B the cone

C= { ( r , s ) : 0 r 1 , s K } .

Then

λ ( K ) λ ( C ) = 1 vol ( B ) 0 1 r m 1 d r K λ S m 1 ( d s ) δ vol ( B ) 0 1 r m 1 d r S m 1 λ S m 1 ( d s ) = δ .

We have obtained a contradiction.

Consider the set K t ={(r,s):rt,s K }, t0. Using the inequality (16), we estimate the Gaussian measure of K t :

v ( K t ) = ( 2 π ) m / 2 K t exp ( 1 2 u 2 2 ) ( d u ) = ( 2 π ) m / 2 t r m 1 exp ( r 2 2 ) d r K λ S m 1 ( d s ) ( 2 π ) m / 2 δ t r m 1 exp ( r 2 2 ) d r S m 1 λ S m 1 ( d s ) = δ v ( x : x 2 t ) .
(17)

A direct computation shows that for t m ,

v ( x 2 t ) exp ( t 2 ) andv ( x 2 m ) c 0 >0,

where c 0 is some absolute constant. It follows from this and (17) that for t 0 =max{ m , ln ( 1 / δ ) } and for any δ(0, c 0 ],

v( K t 0 )δv ( x 2 > max { m , ln ( 1 / δ ) } ) > δ 2 .
(18)

For any element y=rs K t 0 , we have from (15)

e ( T m y , L , 1 m ) = r e ( T m s , L , 1 m ) c 1 r T m c 2 T m max { m , ln ( 1 / δ ) } c 2 2 T m m + ln ( 1 / δ ) .
(19)

Since is an arbitrary subspace with dimLN, it follows from (18) and (19) that

d N , δ 2 ( T m : R m 1 m , v ) c 2 2 T m m + ln ( 1 / δ ) .

Theorem 2 is a direct consequence of this. □

References

  1. Sul’din P: Wiener measure and its applications to approximation theory I. Izv. Vysš. Učebn. Zaved., Mat. 1959, 6: 145–158. (in Russian)

    MathSciNet  Google Scholar 

  2. Sul’din P: Wiener measure and its applications to approximation theory II. Izv. Vysš. Učebn. Zaved., Mat. 1960, 5: 165–179. (in Russian)

    MathSciNet  Google Scholar 

  3. Traub JF, Wasilkowski GW, Woźniakowski H: Information-Based Complexity. Academic Press, New York; 1988.

    MATH  Google Scholar 

  4. Maiorov VE: Discretization of the problem of diameters. Usp. Mat. Nauk 1975, 30: 179–180.

    MathSciNet  MATH  Google Scholar 

  5. Maiorov VE: On linear widths of Sobolev classes and chains of extremal subspaces. Mat. Sb. 1980, 113(115):437–463. English transl. in Math. USSR Sb. 41 (1982)

    MathSciNet  Google Scholar 

  6. Maiorov VE:Kolmogorov’s (N,δ)-widths of the spaces of the smooth functions. Russ. Acad. Sci. Sb. Math. 1994, 79: 265–279.

    MathSciNet  Google Scholar 

  7. Maiorov VE: Linear widths of function spaces equipped with the Gaussian measure. J. Approx. Theory 1994, 77: 74–88. 10.1006/jath.1994.1035

    Article  MathSciNet  MATH  Google Scholar 

  8. Mathé P: S -Numbers in information-based complexity. J. Complex. 1990, 6: 41–66. 10.1016/0885-064X(90)90011-2

    Article  MATH  Google Scholar 

  9. Mathé P: Random approximation of Sobolev embeddings. J. Complex. 1991, 7: 261–281. 10.1016/0885-064X(91)90036-W

    Article  MATH  Google Scholar 

  10. Mathé P: A minimax principle for the optimal error of Monte Carlo methods. Constr. Approx. 1993, 9: 23–29. 10.1007/BF01229334

    Article  MathSciNet  MATH  Google Scholar 

  11. Mathé P: On optimal random nets. J. Complex. 1993, 9: 171–180. 10.1006/jcom.1993.1012

    Article  MATH  Google Scholar 

  12. Mathé, P: Approximation theory of stochastic numerical methods. Habilitationsschrift, Fachbereich Mathematik, Freie Universität Berlin, Berlin (1994)

  13. Sun Y, Wang C: μ -Average n -widths on the Wiener space. J. Complex. 1994, 10: 428–436. 10.1006/jcom.1994.1022

    Article  MATH  Google Scholar 

  14. Sun Y, Wang C: Average error bounds of best approximation of continuous functions on the Wiener space. J. Complex. 1995, 11: 74–104. 10.1006/jcom.1995.1003

    Article  MATH  Google Scholar 

  15. Ritter K Lecture Notes in Math. 1733. In Average-Case Analysis of Numerical Problems. Springer, New York; 2000.

    Chapter  Google Scholar 

  16. Fang G, Qian L: Approximation characteristics for diagonal operators in different computational settings. J. Approx. Theory 2006, 140: 178–190. 10.1016/j.jat.2005.12.005

    Article  MathSciNet  MATH  Google Scholar 

  17. Fang G, Qian L: Optimization on class of operator equations in the probabilistic case setting. Sci. China Ser. A 2007, 50: 100–104. 10.1007/s11425-007-2041-9

    Article  MathSciNet  MATH  Google Scholar 

  18. Fang G, Ye P: Probabilistic and average linear widths of Sobolev space with Gaussian measure. J. Complex. 2003, 19: 73–84. 10.1016/S0885-064X(02)00005-5

    Article  MathSciNet  MATH  Google Scholar 

  19. Fang G, Ye P:Probabilistic and average linear widths of Sobolev space with Gaussian measure in L norm. Constr. Approx. 2004, 20: 159–172.

    MathSciNet  MATH  Google Scholar 

  20. Chen G, Fang G: Probabilistic and average widths of multivariate Sobolev spaces with mixed derivative equipped with the Gaussian measure. J. Complex. 2004, 20: 858–875. 10.1016/j.jco.2004.04.001

    Article  MathSciNet  MATH  Google Scholar 

  21. Chen G, Fang G: Linear widths of multivariate function spaces equipped with the Gaussian measure. J. Approx. Theory 2005, 132: 77–96. 10.1016/j.jat.2004.10.012

    Article  MathSciNet  MATH  Google Scholar 

  22. Carl B, Pajor A: Gelfand numbers of operators with values in a Hilbert space. Invent. Math. 1988, 94: 479–504. 10.1007/BF01394273

    Article  MathSciNet  MATH  Google Scholar 

  23. Pinkus A: n-Widths in Approximation Theory. Springer, New York; 1985.

    Book  MATH  Google Scholar 

  24. Carl B, Stephani I: Entropy, Compactness and the Approximation of Operators. Cambridge University Press, Cambridge; 1990.

    Book  MATH  Google Scholar 

  25. Pietsch A: Operator Ideals. North-Holland, Amsterdam; 1980.

    MATH  Google Scholar 

  26. Pietsch A: Eigenvalues and s-Numbers. Geest und Portig, Leipzig; 1987.

    MATH  Google Scholar 

  27. Gluskin ED: Norm of random matrices and widths of finite-dimensional sets. Mat. Sb. 1983, 120(162):180–189. English transl. in Math. USSR Sb. 48 (1984)

    MathSciNet  Google Scholar 

  28. Pisier G: The Volume of Convex Bodies and Banach Space Geometry. Cambridge University Press, Cambridge; 1989.

    Book  MATH  Google Scholar 

Download references

Acknowledgements

The authors thank the editor and the referees for their valuable suggestions to improve the quality of this paper. The present investigations was supported by the Natural Science Foundation of Inner Mongolia Province of China under Grant 2011MS0103.

Author information

Authors and Affiliations

Authors

Corresponding author

Correspondence to Yuewu Li.

Additional information

Competing interests

The authors declare that they have no competing interests.

Authors’ contributions

Both authors completed the paper together. They also read and approved the final manuscript.

Rights and permissions

Open Access This article is distributed under the terms of the Creative Commons Attribution 2.0 International License (https://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Reprints and permissions

About this article

Cite this article

Zhou, J., Li, Y. Estimates of probabilistic widths of the diagonal operator of finite-dimensional sets with the Gaussian measure. J Inequal Appl 2013, 277 (2013). https://doi.org/10.1186/1029-242X-2013-277

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1186/1029-242X-2013-277

Keywords