Polynomial Approximation
1. **Problem statement:**
Given a compact set $K$ in $\mathbb{R}$ disjoint from the interval $I=[0,1]$, and $J$ a minimal closed interval containing both $I$ and $K$. Show for each $n \in \mathbb{N}$ there exists a continuous function $f_n$ such that:
- $f_n(x)=0$ for $x \in I$
- $f_n(x)=n+1$ for $x \in K$
- $0 \leq f_n(x) \leq n+1$ for $x \in J$
Further, prove there is a polynomial $P_n$ with real coefficients satisfying:
- $|P_n(x)| \leq 1$ for $x \in I$
- $P_n(x) \geq 1$ for $x \in K$
- $|P_n(x)| \leq n+2$ for $x \in J$
Part 1.1.b: Define a norm on space $P$ of polynomials with real coefficients by
$$
||P|| = \sup_{x \in [0,1]} |P(x)|
$$
Show this norm defines a normed space structure on $P$.
Part 1.1.c: Using part 1.1.a, prove the linear functional $L_a : P \to \mathbb{R}$ given by:
$$
L_a(P) = P(a)
$$
is continuous if and only if $a \in [0,1]$.
Determine necessary and sufficient conditions on $\alpha, \beta$ so that the linear functional:
$$
L_{\alpha, \beta}(P) = \int_{\alpha}^{\beta} P(x) dx
$$
is continuous.
2. **Problem statement:**
Suppose $P_n \subset P$ is the set of all polynomials of degree $\leq n$. Show the natural evaluation map
$$
f_p : P_n \to \mathbb{R}^{n+1}
$$
given by evaluating polynomial coefficients (or values at $n+1$ points) is an isomorphism from $P_n$ onto $\mathbb{R}^{n+1}$.
---
2. **Step-by-step solution:**
### Part 1.1.a: Constructing continuous functions $f_n$ and polynomials $P_n$
1. Since $K$ and $I$ are disjoint and compact sets (with $I=[0,1]$), by Urysohn's lemma, there exists continuous functions $f_n: J \to [0,n+1]$ with $f_n|_I = 0$ and $f_n|_K = n+1$.
2. By the Stone-Weierstrass theorem, continuous functions on compact sets can be uniformly approximated by polynomials. Thus, for each $n$, there exists $P_n$ polynomial such that:
$$
|P_n(x) - f_n(x)| < 1 \quad \forall x \in J
$$
3. Then, since $f_n(x) = 0$ on $I$, we get
$$
|P_n(x)| \leq |P_n(x)-f_n(x)| + |f_n(x)| < 1 + 0 = 1 \quad \forall x \in I
$$
4. For $x \in K$, $f_n(x) = n+1$, so
$$
P_n(x) \geq f_n(x) - |P_n(x) - f_n(x)| > (n+1) - 1 = n
$$
In particular, $P_n(x) \geq 1$ for $x \in K$.
5. For $x \in J$,
$$
|P_n(x)| \leq |P_n(x) - f_n(x)| + |f_n(x)| < 1 + (n+1) = n+2
$$
Thus, all desired inequalities hold.
### Part 1.1.b: Norm on polynomial space $P$
1. Define norm
$$
||P|| = \sup_{x \in [0,1]} |P(x)|
$$
2. This defines a norm since:
- Positivity: $||P|| \geq 0$, and $||P|| = 0$ iff $P=0$ (polynomial zero everywhere means $P=0$).
- Homogeneity: For any scalar $c$, $||c P|| = |c| \, ||P||$ by absolute value properties.
- Triangle inequality: $||P+Q|| \leq ||P|| + ||Q||$ by supremum and triangle inequality.
Hence, $(P,||\cdot||)$ is a normed vector space.
### Part 1.1.c: Continuity of evaluation linear functionals
1. Consider $L_a(P) = P(a)$ for fixed $a$.
2. If $a \in [0,1]$, then
$$
|L_a(P)| = |P(a)| \leq ||P||
$$
So $L_a$ is bounded, hence continuous.
3. If $a \notin [0,1]$, choose polynomials that vanish on $[0,1]$ but are arbitrarily large at $a$. Then $L_a$ is not bounded, hence not continuous.
4. For integral linear functional
$$
L_{\alpha, \beta}(P) = \int_{\alpha}^{\beta} P(x) dx
$$
5. By definition,
$$
|L_{\alpha, \beta}(P)| \leq \int_{\alpha}^{\beta} |P(x)| dx \leq |\beta - \alpha| \, \sup_{x \in [\alpha, \beta]} |P(x)|
$$
6. For $L_{\alpha, \beta}$ to be continuous on $P$ with norm $||P|| = \sup_{x \in [0,1]} |P(x)|$, the interval $[\alpha, \beta]$ must lie within $[0,1]$. Otherwise, $||P|| = \sup_{[0,1]}$ doesn’t control the integral outside.
### Part 2: Isomorphism $f_p$ from $P_n$ to $\mathbb{R}^{n+1}$
1. $P_n$ is the finite-dimensional vector space of polynomials degree $\leq n$, dimension $n+1$.
2. Define $f_p$ evaluation or coefficient map:
$$
f_p: P_n \to \mathbb{R}^{n+1}
$$
by mapping polynomial to its coefficients or its values at $n+1$ distinct points.
3. By fundamental theorem of algebra and linear algebra, $f_p$ is a bijective linear map (isomorphism).
---
**Final answers:**
- Functions $f_n$ and polynomials $P_n$ exist by Urysohn lemma and Stone-Weierstrass approximation, satisfying the bounds.
- Supremum norm $||P|| = \sup_{[0,1]} |P(x)|$ defines a norm on polynomials $P$.
- Evaluation functionals $L_a$ are continuous iff $a \in [0,1]$.
- Integral functionals $L_{\alpha, \beta}$ are continuous iff $[\alpha, \beta] \subseteq [0,1]$.
- The evaluation map $f_p: P_n \to \mathbb{R}^{n+1}$ is an isomorphism.