Weierstrass Approximation Theorem

Join the discussion
Ask a follow-up here, or get your own question answered by working scientists, mathematicians and engineers — people, not an autocomplete.
Real named experts · corrections over time · the nuance an AI answer skips
1 reply · 2K views
adpc
Messages
5
Reaction score
0

Homework Statement


Show that if f is continuously differentiable on [a, b], then there is a sequence of
polynomials pn converging uniformly to f such that p'n converge uniformly to f' as
well.

Homework Equations


The Attempt at a Solution


Let pn(t) = cn t^n
Use uniform convergence and integrate from a to x to get:
[tex]f(x) - f(a) = \lim_{n \rightarrow \infty} \left[ \frac{c_n}{n+1} ( x^{n+1} - a^{n+1}) \right][/tex]

Now what is next?? How do I show that f(a) converges to the right side of the limit? so that then i can conclude that f(x) = lim cn x^(n+1) / (n+1)
 
Last edited:
Physics news on Phys.org
You seem to be confused. The Weierstrass approximation theorem tells you that, given any continuous function [tex]g[/tex] on [tex][a, b][/tex], there exists some sequence [tex](p_n)[/tex] of polynomials which converges uniformly to [tex]g[/tex] on [tex][a, b][/tex]. But it does not tell you anything about the form of the [tex]p_n[/tex]; in particular, you cannot assume that they have the form [tex]p_n(t) = c_n t^n[/tex]. So this approach will not get you anywhere.

Your intuition that you should be approximating [tex]f'[/tex] and integrating is, however, correct. Think about what integration does with accumulated errors in an estimate: if [tex]|p_n(t) - f'(t)| < \varepsilon[/tex], uniformly in [tex]t \in [a, b][/tex], can you use that to construct a polynomial that estimates [tex]f[/tex] with some error related somehow to [tex]\varepsilon[/tex]?