For the sake of completeness

Let’s prove the completeness of \ell^p. The argument consists of two steps.

Claim 1. Suppose X is a normed space in which every absolutely convergent series converges; that is, \sum_{n=1}^{\infty} x_n converges whenever x_n\in X are such that \sum_{n=1}^{\infty} \|x_n\| converges. Then the space is complete.

Proof. Take a Cauchy sequence \{y_n\}\subset X. For j=1,2,\dots find an integer n_j such that \|y_n-y_m\|<2^{-j} as long as n,m\ge n_j. (This is possible because the sequence is Cauchy.) Also let n_0=1 and consider the series \sum_{j=1}^{\infty} (y_{n_{j}}-y_{n_{j-1}}). By the hypothesis this series converges. Its partial sums simplify (telescope) to y_{n_j}-y_1. Hence the subsequence \{y_{n_j}\} has a limit. It remains to apply a general theorem about metric spaces: if a Cauchy sequence has a convergent subsequence, then the entire sequence converges. This proves Claim 1.

Claim 2. Every absolutely convergent series in \ell^p converges.

Proof. The elements of \ell^p are functions from \mathbb N to \mathbb C, so let’s write them as such: f_j\colon \mathbb N\to \mathbb C. (This avoids confusion of indices.) Suppose the series \sum_{j=1}^{\infty} \|f_j\| converges. Then for any n the series \sum_{j=1}^{\infty} |f_j(n)| also converges, by Comparison Test. Hence \sum_{j=1}^{\infty} f_j(n) converges (absolutely convergent implies convergent for series of real or complex numbers). Let f(n) = \sum_{j=1}^{\infty} f_j(n). So far the convergence is only pointwise, so we are not done. We still have to show that the series converges in \ell^p, that is, its tails have small \ell^2 norm: \sum_{n=1}^\infty |\sum_{j=k}^{\infty} f_j(n)|^p \to 0 as k\to\infty.

What we need now is a dominating function, so that we can apply the Dominated Convergence Theorem. Namely, we need a function g\colon \mathbb N\to [0,\infty) such that
(1) \sum_{n=1}^{\infty} g(n)<\infty, and
(2) |\sum_{j=k}^{\infty} f_j(n)|^p \le g(n) for all k,n.

Set g=(\sum_{j=1}^{\infty} |f_j|)^p. Then (2) follows from the triangle inequality. Also, g is the increasing limit of functions g_k =(\sum_{j=1}^k |f_j|)^p, for which we have
\sum_n g_k(n) \le (\sum_{j=1}^k \|f_j\|)^p \le (\sum_{j=1}^{\infty} \|f_j\|)^p<\infty
using the triangle inequality in \ell^p. Therefore, \sum_n g(n)<\infty by the Monotone Convergence Theorem.

Almost norming functionals, Part 2

Let E be a real Banach space with the dual E^*. Fix \delta\in (0,1) and call a linear functional e^*\in E^* almost norming for e if |e|=|e^*|=1 and e^*(e)\ge \delta. In Part 1 I showed that in any Banach space there exists a continuous selection of almost norming functionals. Here I will prove that there is no uniformly continuous selection in \ell_1.

Claim. Let S be the unit sphere in \ell_1^n, the n-dimensional \ell_1-space.  Suppose that f\colon S\to \ell_{\infty}^n is a map such that f(e) is almost norming e in the above sense. Then the modulus of continuity \omega_f satisfies \omega_f(2/n)\ge 2\delta.

(If an uniformly continuous selection was available in \ell_1, it would yield selections in \ell_1^n with a modulus of continuity independent of n.)

Proof. Write f=(f_1,\dots,f_n). For any \epsilon\in \{-1,1\}^n we have n^{-1}\epsilon \in S, hence

\sum\limits_{i=1}^n \epsilon_i f_i(n^{-1}\epsilon)\ge n\delta for all \epsilon\in \{-1,1\}^n. Sum over all \epsilon and change the order of summation:

\sum\limits_{i=1}^n \sum\limits_{\epsilon}\epsilon_i f_i(n^{-1}\epsilon)\ge n2^n\delta

There exists i\in\{1,2,\dots,n\} such that

\sum\limits_{\epsilon}\epsilon_i f_i(n^{-1}\epsilon) \ge 2^n \delta

Fix this i from now on. Define \tilde \epsilon to be the same \pm vector as \epsilon, but with the ith component flipped. Rewrite the previous sum as

\sum\limits_{\epsilon} -\epsilon_i f_i(n^{-1}\tilde \epsilon)\ge 2^n\delta

and add them together:

\sum\limits_{\epsilon}\epsilon_i [f_i(n^{-1}\epsilon)-f_i(n^{-1}\tilde \epsilon)]\ge 2^{n+1}\delta

Since \|n^{-1}\epsilon-n^{-1}\tilde \epsilon\|=2/n, it follows that 2^n \omega_f(2/n) \ge 2^{n+1}\delta, as claimed.

A relation between polynomials

This is a brief foray into algebra from a 2006 REU project at Texas A&M.

Given two polynomials P,Q \in \mathbb C[z_1,\dots,z_n], we write Q\preccurlyeq P if there is a differential operator T\in \mathbb C[\frac{\partial}{\partial z_1},\dots, \frac{\partial}{\partial z_n}] such that Q=T P.

The relation \preccurlyeq  is reflexive and transitive, but is not antisymmetric. If both Q\preccurlyeq P and Q\preccurlyeq P hold, we say that P and Q are \partial-equivalent, denoted P\thicksim Q.

A polynomial is \partial -homogeneous if it is \partial -equivalent to a homogeneous polynomial. Obviously, any polynomial in one variable has this property. Polynomials in more than one variable usually do not have it.

The interesting thing about \partial -homogeneous polynomials is that they are refinable, meaning that one has a nontrivial identity of the form P(z)=\sum_{j\in\mathbb Z^n} c_{j} P(\lambda z-j) where c_{j}\in \mathbb C, j\in \mathbb Z^n, and only finitely many of the coefficients c_j are nonzero. The value of \lambda does not matter as long as |\lambda|\ne 0,1. Conversely, every \lambda -refinable polynomial is \partial -homogeneous.

Controlled bilipschitz extension

A map f\colon X\to Y is L-bilipschitz if L^{-1} |a-b| \le |f(a)-f(b)| \le L |a-b| for all a,b\in X. This definition makes sense if X and Y are general metric spaces, but let’s suppose they are subsets on the plane \mathbb R^2.

Definition 1. A set A\subset \mathbb R^2 has the BL extension property if any bilipschitz map f\colon A\to\mathbb R^2 can be extended to a bilipschitz map F\colon \mathbb R^2\to\mathbb R^2. (Extension means that F is required to agree with f on A.)

Lines and circles have the BL extension property. This was proved in early 1980s independently by Tukia, Jerison and Kenig, and Latfullin.

Definition 2. A set A\subset \mathbb R^2 has the controlled BL extension property if there exists a constant C such that any L-bilipschitz map f\colon A\to\mathbb R^2 can be extended to a C L-bilipschitz map F\colon \mathbb R^2\to\mathbb R^2.

Clearly, Definition 2 asks for more than Definition 1. I can prove that a line has the controlled BL extension property, even with a modest constant such as C=2000. (Incidentally, one cannot take C=1.) I still can’t prove the controlled BL extension property for a circle.

Update: extension from line is done in this paper.

WeBWork class roster import

One way to import classroster into WeBWork (at SU):

  1. Download the roster from Blackboard Grade Center and import it into a spreadsheet
  2. The first four columns A,B,C,D will be Last Name, First Name, UserName, Student ID.
  3. Append the column with the function

    (second row shown). Or


    if using OpenOffice.

  4. Using WeBWork file manager, create a file roster.lst and paste this new column into it.
  5. Use the Import Users command under Classlist editor.

Almost norming functionals, Part 1

Let E be a real Banach space with the dual E^*. By the Hahn-Banach theorem, for every unit vector e\in E there exists a functional e^*\in E^* of unit norm such that e^*(e)=1. One says that e^* is a norming functional for e. In general, one cannot choose e^* so that it depends continuously on e. For example, the 2-dimensional space with \ell_1 norm does not allow such a continuous selection.

Fix \delta\in (0,1) and call a linear functional e^* almost norming for e if |e|=|e^*|=1 and e^*(e)\ge \delta. In any Banach space there exists a continuous selection of almost norming functionals.

Continue reading “Almost norming functionals, Part 1”