Dot diagrams and Jordan canonical forms

psie
Messages
315
Reaction score
40
Homework Statement
Why is the number of dots ##r_i## in row ##i## of a dot diagram given by ##r_i=\max\{j:p_j\geq i\}##, where ##p_j## are the number of dots in column ##j##?
Relevant Equations
Some familiarity with Jordan canonical forms is required I think. I'll try to explain the rest below.
We know that a Jordan canonical form is simply the matrix representation of an operator (whose characteristic polynomial splits) with respect to a special basis called a Jordan canonical basis. This basis consists of a disjoint union of cycles/chains of generalized eigenvectors. Take all the ##n## cycles that correspond to a certain eigenvalue ##\lambda## and take their union, which we denote ##\gamma=\gamma_1\cup\gamma_2\cup\cdots\cup\gamma_n##. Note that ##\gamma_i## may have different lengths ##p_1,p_2,\ldots,p_n##. To be concrete, suppose ##n=3## and ##p_1=3, p_2=2## and ##p_3=1##. Then \begin{align*}\gamma_1&=\{(T-\lambda I)^{2}(v_1),(T-\lambda I)(v_1),v_1\};\\ \gamma_2&=\{(T-\lambda I)(v_2),v_2\};\\ \gamma_n&=\{v_3\}.\end{align*} Then the matrix representation of ##T## restricted to ##\operatorname{span}(\gamma)## is a so-called Jordan block. We can visualize a Jordan block with the help of a dot diagram as follows: $$\begin{array}{ccc}\bullet&\bullet&\bullet \\ \bullet&\bullet&\\ \bullet \end{array}$$Here the first dots in the first row are the initial vectors in the cycle; thus the bottom dots in each column are ##v_1,v_2,v_3##, from left to right respectively. Dot diagrams are always ordered in decreasing lengths of cycle going from left to right.

Suppose now a dot diagram has ##k## columns and ##m## rows, with ##p_j## dots in column ##j## and ##r_i## dots in row ##i##. I need to show by induction on ##m=p_1## that ##p_j=\max\{i:r_i\geq j\}## for ##1\leq j\leq k## and ##r_i=\max\{j:p_j\geq i\}## for ##1\leq i\leq m##. The induction step is causing me great trouble (the base case I think I manage by myself). Any help would be very appreciated.
 
Physics news on Phys.org
I get it now I think. We have a dot at ##(i,j)## if and only if ##p_j\geq i## and ##r_i\geq j##.
 
For the sake of math trivia, a matrix with a non-trivial Jordan block is an (counter) example of a matrix that can't be diagonalized; not even over ##\mathbb C ##
 
psie said:
Homework Statement: Why is the number of dots ##r_i## in row ##i## of a dot diagram given by ##r_i=\max\{j:p_j\geq i\}##, where ##p_j## are the number of dots in column ##j##?
Relevant Equations: Some familiarity with Jordan canonical forms is required I think. I'll try to explain the rest below.

We know that a Jordan canonical form is simply the matrix representation of an operator (whose characteristic polynomial splits) with respect to a special basis called a Jordan canonical basis. This basis consists of a disjoint union of cycles/chains of generalized eigenvectors. Take all the ##n## cycles that correspond to a certain eigenvalue ##\lambda## and take their union, which we denote ##\gamma=\gamma_1\cup\gamma_2\cup\cdots\cup\gamma_n##. Note that ##\gamma_i## may have different lengths ##p_1,p_2,\ldots,p_n##. To be concrete, suppose ##n=3## and ##p_1=3, p_2=2## and ##p_3=1##. Then \begin{align*}\gamma_1&=\{(T-\lambda I)^{2}(v_1),(T-\lambda I)(v_1),v_1\};\\ \gamma_2&=\{(T-\lambda I)(v_2),v_2\};\\ \gamma_n&=\{v_3\}.\end{align*} Then the matrix representation of ##T## restricted to ##\operatorname{span}(\gamma)## is a so-called Jordan block. We can visualize a Jordan block with the help of a dot diagram as follows: $$\begin{array}{ccc}\bullet&\bullet&\bullet \\ \bullet&\bullet&\\ \bullet \end{array}$$Here the first dots in the first row are the initial vectors in the cycle; thus the bottom dots in each column are ##v_1,v_2,v_3##, from left to right respectively. Dot diagrams are always ordered in decreasing lengths of cycle going from left to right.

Suppose now a dot diagram has ##k## columns and ##m## rows, with ##p_j## dots in column ##j## and ##r_i## dots in row ##i##. I need to show by induction on ##m=p_1## that ##p_j=\max\{i:r_i\geq j\}## for ##1\leq j\leq k## and ##r_i=\max\{j:p_j\geq i\}## for ##1\leq i\leq m##. The induction step is causing me great trouble (the base case I think I manage by myself). Any help would be very appreciated.

"The highest-numbered column tall enough to reach row i is max{j : p_j ≥ i}"

If column heights are [4, 3, 3, 1]:

Row 3 gets dots from columns with height ≥ 3
That's columns 1, 2, and 3 (but not 4, since p₄ = 1 < 3)
So r₃ = 3 = max{j : p_j ≥ 3}
The botton line is - r_i counts how many columns are tall enough to reach row i, which is exactly max{j : p_j ≥ i}.
 
There are two things I don't understand about this problem. First, when finding the nth root of a number, there should in theory be n solutions. However, the formula produces n+1 roots. Here is how. The first root is simply ##\left(r\right)^{\left(\frac{1}{n}\right)}##. Then you multiply this first root by n additional expressions given by the formula, as you go through k=0,1,...n-1. So you end up with n+1 roots, which cannot be correct. Let me illustrate what I mean. For this...
Back
Top