Difference between revisions of "Aufgaben:Problem 5"

From Ferienserie MMP2
Jump to: navigation, search
m (nitpicks)
 
(39 intermediate revisions by 8 users not shown)
Line 1: Line 1:
=Problem=
+
Follow the link! (Original Solution Sheldon)
Show, by using the Fourier series, that
+
[https://polybox.ethz.ch/public.php?service=files&t=3cd27159ac0fca5b53c53fe4cff98dcc]
$$\sum_{k=1}^\infty \frac{1}{k^n} = - \frac{(2 \pi i)^n}{2(n!)} B_n,\ n \in 2 \mathbb{Z}_{>0}$$
+
where \(B_n\) are the Bernoulli numbers defined as
+
$$\frac{z}{e^z - 1} = \sum_{n=0}^\infty B_n \frac{z^n}{n!}$$
+
  
''Hint:'' consider \( B_n(x) := \sum_{k \in \mathbb{Z} \setminus \{0\}} k^{-n} e^{ikx},\ n \in \mathbb{Z}_{\geq 2},\ x \in \mathbb{R} \)
+
An alternative solution can be found here (3.). (MuLö Linalg)
 +
[http://www.math.ethz.ch/education/bachelor/lectures/hs2013/math/linalg1/l12]
  
=Solution=
+
== Task ==
As suggested in the problem, we will have a look at the convergent fourier series
+
Determine the center of the algebra of complex d × d-matrices.
$$B_n(x) := \sum_{k \in \mathbb{Z} \setminus \{0\}} k^{-n} e^{ikx},\ n \in \mathbb{Z}_{\geq 2},\ x \in \mathbb{R}$$
+
Hint: the center of an algebra \(A\) is defined as \(Z(A) = \{\phi \in A | \phi \psi = \psi \phi \; \forall \psi \in A \} \).
with Fourier coefficients
+
$$c_k = \begin{cases}
+
0 & \text{for } k = 0\\
+
\frac{1}{k^n} & \text{for } k \neq 0
+
\end{cases}$$
+
  
 +
== Problem 5 (Sheldon) ==
 +
Let big bold letters denote elements of \(\mathbb{C}^{d \times d}\) and small Greek letters complex numbers.
  
It is obvious that \( B_n(x) \) is \(2\pi\)-periodic and, by differentiating every term separately, that
+
Let \(M \in Z\)
$$B'_n = iB_{n-1}$$
+
  
<hr style="border:0;border-top:thin dashed #ccc;background:none;"/>
+
We make the following observations and conclude that \( Z = \{\lambda\mathbb{I}\}\)
  
We will first show that \( B_2(x) = f(x) := \pi^2/3 - \pi x + x^2/2 \)  for  \( x \in [0, 2\pi]\).
 
  
Recall the definition of the Fourier coefficients for \(2\pi\)-periodic functions in \(C^\infty(\mathbb{R})\):
+
Obviously, the elementes of the center are closed under scalar multiplication (they even form a vector space over \(\mathbb{C}\)). \(\mathbb{I} \in Z \Rightarrow \{\lambda\mathbb{I}\} \subseteq Z\)
$$c_k = \frac{1}{2\pi} \int_0^{2\pi} f(x) \: e^{-ikx} \: dx$$
+
  
We can easily verify the following identities:
 
$$\begin{align}
 
\frac{1}{2\pi} \int_0^{2\pi} e^{-ikx} \: dx &\ =\
 
\begin{cases}
 
1 & \text{for } k = 0\\
 
0 & \text{for } k \neq 0
 
\end{cases} \\
 
\frac{1}{2\pi} \int_0^{2\pi} x \: e^{-ikx} \: dx &\ =\
 
\begin{cases}
 
\pi & \text{for } k = 0\\
 
\frac{i}{k} & \text{for } k \neq 0
 
\end{cases} \\
 
\frac{1}{2\pi} \int_0^{2\pi} x^2 \: e^{-ikx} \: dx &\ =\
 
\begin{cases}
 
\frac{4}{3} \pi^2 & \text{for } k = 0\\
 
\frac{2(1+ik\pi)}{k^2} & \text{for } k \neq 0
 
\end{cases}
 
\end{align}$$
 
  
Calculating the Fourier coefficients for \(f(x)\) then yields
+
The theorem about Jordan decomposition states that \(M\) is conjugate to a matrix in Jordan normal form \(J\) and \(J = CMC^{-1}=MCC^{-1}=M\).
$$\begin{align}
+
\hat f(k) &= \frac{1}{2\pi} \int_0^{2\pi} f(x) \: e^{-ikx} \: dx\\
+
&= \frac{1}{2\pi} \left(
+
\frac{\pi^2}{3} \int_0^{2\pi} e^{-ikx} \: dx
+
- \pi \int_0^{2\pi} x \: e^{-ikx} \: dx
+
+ \frac{1}{2} \int_0^{2\pi} x^2 \: e^{-ikx}
+
\right) \\
+
&= \begin{cases}
+
0 & \text{for } k = 0\\
+
\frac{1}{k^2} & \text{for } k \neq 0
+
\end{cases}
+
\end{align}$$
+
which proves that \(B_2(x) = f(x)\) for \(x \in [0, 2\pi]\).
+
  
Thus, we have shown that the \(B_n|_{[0, 2\pi]}\) are polynomials, which are recursively defined by \(B'_n = iB_{n-1}\) and \(\int_0^{2\pi} B_n(x) dx = -i(B_{n+1}(2\pi) - B_{n+1}(0)) = 0\). We also add
+
\(M\) is a diagonal matrix.
$$B_1(x) = -iB'_2(x) = i\pi - ix \qquad B_0(x) = -iB'_1(x) = -1$$
+
  
<hr style="border:0;border-top:thin dashed #ccc;background:none;"/>
+
Suppose to the contrary that \(M\) has a '1' in the superdiagonal, then \(2M\) is supposed to be in the center as well, but it can't, because it is not in Jordan normal form.
  
By induction over \(n\), we will now prove that
 
$$B_n(x) = \sum_{k=0}^n B_{n-k}(0) \frac{(ix)^k}{k!}$$
 
Base case for \(n=0\):
 
$$B_0(0) \frac{(ix)^0}{0!} = B_0(0) =  -1 = B_0(x)$$
 
Inductive step (\(n\!-\!1 \rightarrow n\)):
 
$$\begin{align}
 
B_n(x) &= \int_0^x iB_{n-1}(\tilde x)\ d\tilde x + B_n(0) \\
 
&= \int_0^x i\left(\sum_{k=0}^{n-1} B_{(n-1)-k}(0) \frac{(i\tilde x)^k}{k!} \right) d\tilde x + B_n(0) \\
 
&= \left( \sum_{k=0}^{n-1} B_{(n-1)-k}(0) \frac{i^{k+1}}{k!} \int_0^x \tilde x^k\ d\tilde x \right) + B_n(0) \\
 
&= \left( \sum_{k=0}^{n-1} B_{(n-1)-k}(0) \frac{i^{k+1}}{k!} \frac{x^{k+1}}{k+1} \right) + B_n(0) \\
 
&= \left( \sum_{k=0}^{n-1} B_{n-(k+1)}(0) \frac{(ix)^{k+1}}{(k+1)!} \right) + B_n(0) \\
 
&= \left( \sum_{k=1}^n B_{n-k}(0) \frac{(ix)^k}{k!} \right) + B_n(0) \\
 
&= \sum_{k=0}^n B_{n-k}(0) \frac{(ix)^k}{k!}
 
\end{align}$$
 
  
<hr style="border:0;border-top:thin dashed #ccc;background:none;"/>
+
All diagonal entries of \(M\) are equal.
  
''What's this all leading up to'', you may ask. Find the answer here in next weeks sequel.
+
Let \( M =\begin{pmatrix} \lambda_1 & 0 & \ldots & 0 \\ 0 & \lambda_2 & \ldots & 0 \\ \vdots & &\ddots& \\ 0 & \ldots & 0 & \lambda_d \end{pmatrix} \) and \( E_{ij} \) be the matrix with zero entries except for a '1' in the i-th row and j-th column.
[[User:Nik|Nik]] ([[User talk:Nik|talk]]) 18:28, 25 December 2014 (CET)
+
 
 +
\(\lambda_i E_{ij} = ME_{ij} = E_{ij}M = \lambda_j E_{ij} \Rightarrow \lambda_i = \lambda_j \)
 +
 
 +
Because \( 1 \leq i \leq j \leq d \) can be chosen arbitrary, this proves the other inclusion \(M \in \{\lambda\mathbb{I}\}\).
 +
 
 +
== Problem 5 (Musterloesung Illmanen) ==
 +
Let \(E_{ij} \in \mathbb{K}^{n\times n} \) be the matrix which has the entry 1 at \((i,j)\) and 0 else. From proposition \(AE_{ij} = E_{ij}A ~~ \forall 1 \leq i, j \leq n\). Let \(1 \leq i, j \leq n \) be fixed: \(AE_{ij} \) consists solely out of zeros except for the j-th column which is equal to \((a_{1i}, \ldots , a_{ni})^T\). So we get \( a_{ii} = a_{jj} ~~ \wedge  ~~ a_{ki} = 0 ~~\forall k \neq j\)
 +
 
 +
We can do this for all indices and this leads to \(a_{ij} = 0 \forall i \neq j ~~\wedge ~~ a_{11} = a_{22} = \ldots = a_{nn} \). Therefor \(A\) has to have the form \(\lambda E_n, \lambda \in \mathbb{K}\)
 +
 
 +
===Ilmanen's solution componentwise===
 +
 
 +
Let \(A \in Z(\mathrm{Mat}_d(\mathbb{C}))\).
 +
 
 +
Let \(E_{ij}\) be the \(d \times d\)-matrix with \((E_{ij})_{kl} = \delta_{ik} \delta_{jl} \).
 +
 
 +
Now let \(1 \leq i, j \leq d\) with \(i \neq j\) be fixed and consider
 +
$$(E_{ij} A)_{kl} = \sum \limits_{m}(E_{ij})_{km} A_{ml} = \sum \limits_{m}\delta_{ki} \delta_{jm} A_{ml} = \delta_{ki} A_{jl}.$$
 +
Since \(A\) commutes with all compex \(d \times d\)-matrices, this is the same as
 +
$$(A E_{ij})_{kl} = \sum \limits_{m}A_{km} (E_{ij})_{ml} = \sum \limits_{m} A_{km} \delta_{mi} \delta_{jl} = \delta_{jl} A_{ki}.$$
 +
 
 +
Thus, we have that
 +
$$\delta_{ki} A_{jl} = \delta_{jl} A_{ki}$$
 +
Now set \(k=i, l=j\). We then find
 +
$$A_{ii} = A_{jj}$$
 +
Next, set \(k=l=i\). Still, \(j \neq i\). It follows that
 +
$$A_{ji} = 0.$$
 +
As this holds for an arbitrary choice of \(i \neq j\), the required form for \(A\) is
 +
$$A = \lambda \mathbb{I}_d$$
 +
for some \(\lambda \in \mathbb{C}\). And obviously \(\forall \lambda \in \mathbb{C}: \lambda \mathbb{I} \in Z(\mathrm{Mat}_d(\mathbb{C}))\).

Latest revision as of 14:01, 30 July 2015

Follow the link! (Original Solution Sheldon) [1]

An alternative solution can be found here (3.). (MuLö Linalg) [2]

Task

Determine the center of the algebra of complex d × d-matrices. Hint: the center of an algebra \(A\) is defined as \(Z(A) = \{\phi \in A | \phi \psi = \psi \phi \; \forall \psi \in A \} \).

Problem 5 (Sheldon)

Let big bold letters denote elements of \(\mathbb{C}^{d \times d}\) and small Greek letters complex numbers.

Let \(M \in Z\)

We make the following observations and conclude that \( Z = \{\lambda\mathbb{I}\}\)


Obviously, the elementes of the center are closed under scalar multiplication (they even form a vector space over \(\mathbb{C}\)). \(\mathbb{I} \in Z \Rightarrow \{\lambda\mathbb{I}\} \subseteq Z\)


The theorem about Jordan decomposition states that \(M\) is conjugate to a matrix in Jordan normal form \(J\) and \(J = CMC^{-1}=MCC^{-1}=M\).

\(M\) is a diagonal matrix.

Suppose to the contrary that \(M\) has a '1' in the superdiagonal, then \(2M\) is supposed to be in the center as well, but it can't, because it is not in Jordan normal form.


All diagonal entries of \(M\) are equal.

Let \( M =\begin{pmatrix} \lambda_1 & 0 & \ldots & 0 \\ 0 & \lambda_2 & \ldots & 0 \\ \vdots & &\ddots& \\ 0 & \ldots & 0 & \lambda_d \end{pmatrix} \) and \( E_{ij} \) be the matrix with zero entries except for a '1' in the i-th row and j-th column.

\(\lambda_i E_{ij} = ME_{ij} = E_{ij}M = \lambda_j E_{ij} \Rightarrow \lambda_i = \lambda_j \)

Because \( 1 \leq i \leq j \leq d \) can be chosen arbitrary, this proves the other inclusion \(M \in \{\lambda\mathbb{I}\}\).

Problem 5 (Musterloesung Illmanen)

Let \(E_{ij} \in \mathbb{K}^{n\times n} \) be the matrix which has the entry 1 at \((i,j)\) and 0 else. From proposition \(AE_{ij} = E_{ij}A ~~ \forall 1 \leq i, j \leq n\). Let \(1 \leq i, j \leq n \) be fixed: \(AE_{ij} \) consists solely out of zeros except for the j-th column which is equal to \((a_{1i}, \ldots , a_{ni})^T\). So we get \( a_{ii} = a_{jj} ~~ \wedge ~~ a_{ki} = 0 ~~\forall k \neq j\)

We can do this for all indices and this leads to \(a_{ij} = 0 \forall i \neq j ~~\wedge ~~ a_{11} = a_{22} = \ldots = a_{nn} \). Therefor \(A\) has to have the form \(\lambda E_n, \lambda \in \mathbb{K}\)

Ilmanen's solution componentwise

Let \(A \in Z(\mathrm{Mat}_d(\mathbb{C}))\).

Let \(E_{ij}\) be the \(d \times d\)-matrix with \((E_{ij})_{kl} = \delta_{ik} \delta_{jl} \).

Now let \(1 \leq i, j \leq d\) with \(i \neq j\) be fixed and consider $$(E_{ij} A)_{kl} = \sum \limits_{m}(E_{ij})_{km} A_{ml} = \sum \limits_{m}\delta_{ki} \delta_{jm} A_{ml} = \delta_{ki} A_{jl}.$$ Since \(A\) commutes with all compex \(d \times d\)-matrices, this is the same as $$(A E_{ij})_{kl} = \sum \limits_{m}A_{km} (E_{ij})_{ml} = \sum \limits_{m} A_{km} \delta_{mi} \delta_{jl} = \delta_{jl} A_{ki}.$$

Thus, we have that $$\delta_{ki} A_{jl} = \delta_{jl} A_{ki}$$ Now set \(k=i, l=j\). We then find $$A_{ii} = A_{jj}$$ Next, set \(k=l=i\). Still, \(j \neq i\). It follows that $$A_{ji} = 0.$$ As this holds for an arbitrary choice of \(i \neq j\), the required form for \(A\) is $$A = \lambda \mathbb{I}_d$$ for some \(\lambda \in \mathbb{C}\). And obviously \(\forall \lambda \in \mathbb{C}: \lambda \mathbb{I} \in Z(\mathrm{Mat}_d(\mathbb{C}))\).