Misplaced Pages

Helly's selection theorem

Article snapshot taken from Wikipedia with creative commons attribution-sharealike license. Give it a read and then ask your questions in the chat. We can research this topic together.
On convergent subsequences of functions that are locally of bounded total variation

In mathematics, Helly's selection theorem (also called the Helly selection principle) states that a uniformly bounded sequence of monotone real functions admits a convergent subsequence. In other words, it is a sequential compactness theorem for the space of uniformly bounded monotone functions. It is named for the Austrian mathematician Eduard Helly. A more general version of the theorem asserts compactness of the space BVloc of functions locally of bounded total variation that are uniformly bounded at a point.

The theorem has applications throughout mathematical analysis. In probability theory, the result implies compactness of a tight family of measures.

Statement of the theorem

Let (fn)n ∈ N be a sequence of increasing functions mapping a real interval I into the real line R, and suppose that it is uniformly bounded: there are a,b ∈ R such that a ≤ fn ≤ b for every n  ∈  N. Then the sequence (fn)n ∈ N admits a pointwise convergent subsequence.

Proof

Step 1. An increasing function f on an interval I has at most countably many points of discontinuity.

Let A = { x I : f ( y ) f ( x )  as  y x } {\displaystyle A=\{x\in I:f(y)\not \rightarrow f(x){\text{ as }}y\rightarrow x\}} , i.e. the set of discontinuities, then since f is increasing, any x in A satisfies f ( x ) f ( x ) f ( x + ) {\displaystyle f(x^{-})\leq f(x)\leq f(x^{+})} , where f ( x ) = lim y x f ( y ) {\displaystyle f(x^{-})=\lim \limits _{y\uparrow x}f(y)} , f ( x + ) = lim y x f ( y ) {\displaystyle f(x^{+})=\lim \limits _{y\downarrow x}f(y)} , hence by discontinuity, f ( x ) < f ( x + ) {\displaystyle f(x^{-})<f(x^{+})} . Since the set of rational numbers is dense in R, x A [ ( f ( x ) , f ( x + ) ) Q ] {\displaystyle \prod _{x\in A}} is non-empty. Thus the axiom of choice indicates that there is a mapping s from A to Q.

It is sufficient to show that s is injective, which implies that A has a non-larger cardinity than Q, which is countable. Suppose x1,x2A, x1<x2, then f ( x 1 ) < f ( x 1 + ) f ( x 2 ) < f ( x 2 + ) {\displaystyle f(x_{1}^{-})<f(x_{1}^{+})\leq f(x_{2}^{-})<f(x_{2}^{+})} , by the construction of s, we have s(x1)<s(x2). Thus s is injective.

Step 2. Inductive Construction of a subsequence converging at discontinuities and rationals.

Let A n = { x I ; f n ( y ) f n ( x )  as  y x } {\displaystyle A_{n}=\{x\in I;f_{n}(y)\not \rightarrow f_{n}(x){\text{ as }}y\to x\}} , i.e. the discontinuities of fn, A = ( n N A n ) ( I Q ) {\displaystyle A=(\cup _{n\in \mathrm {N} }A_{n})\cup (\mathrm {I} \cap \mathrm {Q} )} , then A is countable, and it can be denoted as {an: nN}.

By the uniform boundedness of (fn)n ∈ N and B-W theorem, there is a subsequence (fn)n ∈ N such that (fn(a1))n ∈ N converges. Suppose (fn)n ∈ N has been chosen such that (fn(ai))n ∈ N converges for i=1,...,k, then by uniform boundedness, there is a subsequence (fn)n ∈ N of (fn)n ∈ N, such that (fn(ak+1))n ∈ N converges, thus (fn)n ∈ N converges for i=1,...,k+1.

Let g k = f k ( k ) {\displaystyle g_{k}=f_{k}^{(k)}} , then gk is a subsequence of fn that converges pointwise in A.

Step 3. gk converges in I except possibly in an at most countable set.

Let h k ( x ) = sup a x , a A g k ( a ) {\displaystyle h_{k}(x)=\sup _{a\leq x,a\in A}g_{k}(a)} , then , hk(a)=gk(a) for aA, hk is increasing, let h ( x ) = lim sup k h k ( x ) {\displaystyle h(x)=\limsup \limits _{k\rightarrow \infty }h_{k}(x)} , then h is increasing, since supremes and limits of increasing functions are increasing, and h ( a ) = lim k g k ( a ) {\displaystyle h(a)=\lim \limits _{k\rightarrow \infty }g_{k}(a)} for aA by Step 2. By Step 1, h has at most countably many discontinuities.

We will show that gk converges at all continuities of h. Let x be a continuity of h, q,r∈ A, q<x<r, then g k ( q ) h ( r ) g k ( x ) h ( x ) g k ( r ) h ( q ) {\displaystyle g_{k}(q)-h(r)\leq g_{k}(x)-h(x)\leq g_{k}(r)-h(q)} ,hence

lim sup k ( g k ( x ) h ( x ) ) lim sup k ( g k ( r ) h ( q ) ) = h ( r ) h ( q ) {\displaystyle \limsup \limits _{k\rightarrow \infty }{\bigl (}g_{k}(x)-h(x){\bigr )}\leq \limsup \limits _{k\rightarrow \infty }{\bigl (}g_{k}(r)-h(q){\bigr )}=h(r)-h(q)}

h ( q ) h ( r ) = lim inf k ( g k ( q ) h ( r ) ) lim inf k ( g k ( x ) h ( x ) ) {\displaystyle h(q)-h(r)=\liminf \limits _{k\rightarrow \infty }{\bigl (}g_{k}(q)-h(r){\bigr )}\leq \liminf \limits _{k\rightarrow \infty }{\bigl (}g_{k}(x)-h(x){\bigr )}}

Thus,

h ( q ) h ( r ) lim inf k ( g k ( x ) h ( x ) ) lim sup k ( g k ( x ) h ( x ) ) h ( r ) h ( q ) {\displaystyle h(q)-h(r)\leq \liminf \limits _{k\rightarrow \infty }{\bigl (}g_{k}(x)-h(x){\bigr )}\leq \limsup \limits _{k\rightarrow \infty }{\bigl (}g_{k}(x)-h(x){\bigr )}\leq h(r)-h(q)}

Since h is continuous at x, by taking the limits q x , r x {\displaystyle q\uparrow x,r\downarrow x} , we have h ( q ) , h ( r ) h ( x ) {\displaystyle h(q),h(r)\rightarrow h(x)} , thus lim k g k ( x ) = h ( x ) {\displaystyle \lim \limits _{k\rightarrow \infty }g_{k}(x)=h(x)}

Step 4. Choosing a subsequence of gk that converges pointwise in I

This can be done with a diagonal process similar to Step 2.


With the above steps we have constructed a subsequence of (fn)n ∈ N that converges pointwise in I.

Generalisation to BVloc

Let U be an open subset of the real line and let fn : U → R, n ∈ N, be a sequence of functions. Suppose that (fn) has uniformly bounded total variation on any W that is compactly embedded in U. That is, for all sets W ⊆ U with compact closure  ⊆ U,

sup n N ( f n L 1 ( W ) + d f n d t L 1 ( W ) ) < + , {\displaystyle \sup _{n\in \mathbf {N} }\left(\|f_{n}\|_{L^{1}(W)}+\|{\frac {\mathrm {d} f_{n}}{\mathrm {d} t}}\|_{L^{1}(W)}\right)<+\infty ,}
where the derivative is taken in the sense of tempered distributions.

Then, there exists a subsequence fnk, k ∈ N, of fn and a function f : U → R, locally of bounded variation, such that

lim k W | f n k ( x ) f ( x ) | d x = 0 ; {\displaystyle \lim _{k\to \infty }\int _{W}{\big |}f_{n_{k}}(x)-f(x){\big |}\,\mathrm {d} x=0;}
  • and, for W compactly embedded in U,
d f d t L 1 ( W ) lim inf k d f n k d t L 1 ( W ) . {\displaystyle \left\|{\frac {\mathrm {d} f}{\mathrm {d} t}}\right\|_{L^{1}(W)}\leq \liminf _{k\to \infty }\left\|{\frac {\mathrm {d} f_{n_{k}}}{\mathrm {d} t}}\right\|_{L^{1}(W)}.}

Further generalizations

There are many generalizations and refinements of Helly's theorem. The following theorem, for BV functions taking values in Banach spaces, is due to Barbu and Precupanu:

Let X be a reflexive, separable Hilbert space and let E be a closed, convex subset of X. Let Δ : X → ; X) with zn(t) ∈ E for all n ∈ N and t ∈ . Then there exists a subsequence znk and functions δz ∈ BV(; X) such that

  • for all t ∈ ,
[ 0 , t ) Δ ( d z n k ) δ ( t ) ; {\displaystyle \int _{[0,t)}\Delta (\mathrm {d} z_{n_{k}})\to \delta (t);}
  • and, for all t ∈ ,
z n k ( t ) z ( t ) E ; {\displaystyle z_{n_{k}}(t)\rightharpoonup z(t)\in E;}
  • and, for all 0 ≤ s < t ≤ T,
[ s , t ) Δ ( d z ) δ ( t ) δ ( s ) . {\displaystyle \int _{[s,t)}\Delta (\mathrm {d} z)\leq \delta (t)-\delta (s).}

See also

References

  1. ^ Ambrosio, Luigi; Fusco, Nicola; Pallara, Diego (2000). Functions of Bounded Variation and Free Discontinuity Problems. Oxford University Press. doi:10.1093/oso/9780198502456.001.0001. ISBN 9780198502456.
  • Rudin, W. (1976). Principles of Mathematical Analysis. International Series in Pure and Applied Mathematics (Third ed.). New York: McGraw-Hill. 167. ISBN 978-0070542358.
  • Barbu, V.; Precupanu, Th. (1986). Convexity and optimization in Banach spaces. Mathematics and its Applications (East European Series). Vol. 10 (Second Romanian ed.). Dordrecht: D. Reidel Publishing Co. xviii+397. ISBN 90-277-1761-3. MR860772
Categories: