Help with this presentation. |
Go to Start (closes Help) |
Enjoy your time! Press to close. | The source code lies on GitHub. |
A.Ayzenberg | V.Buchstaber | V.Cherepanov | M.Masuda | G.Solomadin | K.Sorokin |
$\Rightarrow$ | \[ \begin{pmatrix} \ast & \ast & \ast & \ast \\ \ast & \ast & & \\ \ast & & \ast & \\ \ast & & & \ast \end{pmatrix} \] | |
Star graph | Arrowhead matrix |
$\Rightarrow$ | \[ \begin{pmatrix} \ast & \ast & \ast & \ast \\ \ast & \ast & \ast & \ast \\ \ast & \ast & \ast & \ast \\ \ast & \ast & \ast & \ast \end{pmatrix} \] | |
Full graph | Full matrix |
$\Rightarrow$ | \[ \begin{pmatrix} \ast & \ast & & \\ \ast & \ast & \ast & \\ & \ast & \ast & \ast \\ & & \ast & \ast \end{pmatrix} \] | |
Path graph | Tridiagonal matrix |
$\Rightarrow$ | \[ \begin{pmatrix} \ast & \ast & & & \ast\\ \ast & \ast & \ast & & \\ & \ast & \ast & \ast & \\ & & \ast & \ast & \ast \\ \ast & & & \ast & \ast \end{pmatrix} \] | |
Cycle graph | Periodic tridiag- onal matrix |
Original matrix: |
|
At each step we represent a matrix $A_i$ as $Q\cdot R$ and take $A_{i+1}=R\cdot Q$. Here $Q$ is orgthogonal, $R$ is upper triangular. |
Fill in the matrix above. Press "u" to upload. Press "q" to make QR step. After 0 iterations: |
|
This matrix is defined earlier. Press "u" to reload. Hold "t" to run Toda flow. Time: 0 seconds: |
|
Definition: $\Gamma$ is said to have diagonalizable type, if there exists a Morse-Smale flow (or cascade) on $M_{\Gamma,\lambda}$, all of whose limit points are diagonal matrices.
There are $n!$ diagonal matrices with spectrum $\lambda$, given by all possible permutations of $\lambda_i$: \[ A_\sigma=\text{diag}(\lambda_{\sigma(1)},\ldots,\lambda_{\sigma(n)}),\text{ where }\sigma\in\Sigma_n. \] We want these points to be the only stationary points of the diagonalizing flow.
In other words, we want to figure out for which graphs the matrix can be diagonalized within its sparseness type.
$\Rightarrow$ | \[ \begin{pmatrix} \ast & \ast & \ast \\ \ast & \ast & \\ \ast & & \ast \end{pmatrix} \] | $\Rightarrow$ | \[ \begin{pmatrix} \ast & \ast & \\ \ast & \ast & \ast \\ & \ast & \ast \end{pmatrix} \] |
Which graphs correspond to staircase matrices?
Definition: Graph $\Gamma$ is called indifference graph if it is the intersection graph of some collection of unit intervals on a line $\mathbb{R}$.
They are also called unit interval graphs or proper interval graphs. The term "indifference" and the notion were introduced by Roberts'69. His idea was similar to tolerance spaces but on a line $\mathbb{R}$.
Theorem (Mertzios'08): $\Gamma$ represents staircase matrices iff $\Gamma$ is an indifference graph.
This is a torus.
Move cursor to pick an element and act on a matrix! |
$\circlearrowright$ |
|
|||||||||
You may go back to reset a matrix (I take first 3x3 block). |
Definition: $X$ is called equivariantly formal if $H^{odd}(X)=0$.
Corollary: If $\Gamma$ has non-diagonalizable induced subgraph, then $\Gamma$ is also non-diagonalizable.
Induced subgraph. | Not an induced subgraph. |
Claw $St_3$ | Holes $Cy_k$, $k\geq 4$ | $Net$ | $Sun$ |
We only need to prove that manifolds $M_{\Gamma,\lambda}$ corresponding to these graphs, are not equivariantly formal.
Theorem (Masuda-Panov'06): Manifold $X$ of complexity $0$ is equivariantly
formal iff its orbit space is a homology polytope.
This means that $X$ should be acyclic, and all its faces should be acyclic.
|
|
Case: $St_3$ |
\[ \begin{pmatrix} \ast & \ast & \ast & \ast \\ \ast & \ast & & \\ \ast & & \ast & \\ \ast & & & \ast \end{pmatrix} \] |
Theorem (Gal-Januszkiewicz, unpublished):
$M_{St_3,\lambda}/T$ is a solid torus with boundary subdivided into hexagons.
Therefore, from Masuda-Panov theorem it follows that $M_{St_3,\lambda}$ is not equivariantly formal. |
|
Case: $Cy_k$, $k\geq 4$ |
\[ \begin{pmatrix} \ast & \ast & & \ast \\ \ast & \ast & \ast & \\ & \ast & \ast & \ast \\ \ast & & \ast & \ast \end{pmatrix} \] |
Case: $Cy_k$, $k\geq 4$ |
\[ \begin{pmatrix} \ast & \ast & & \ast \\ \ast & \ast & \ast & \\ & \ast & \ast & \ast \\ \ast & & \ast & \ast \end{pmatrix} \] |
Case: $Cy_k$, $k\geq 4$ |
\[ \begin{pmatrix} \ast & \ast & & \ast \\ \ast & \ast & \ast & \\ & \ast & \ast & \ast \\ \ast & & \ast & \ast \end{pmatrix} \] |
Cases: |
|
\[ \begin{pmatrix} \ast & \ast & \ast & \ast & & \\ \ast & \ast & \ast & & \ast & \\ \ast & \ast & \ast & & & \ast \\ \ast & & & \ast & & \\ & \ast & & & \ast & \\ & & \ast & & & \ast \\ \end{pmatrix}, \begin{pmatrix} \ast & \ast & \ast & & \ast & \ast \\ \ast & \ast & \ast & \ast & & \ast\\ \ast & \ast & \ast & \ast & \ast & \\ & \ast & \ast & \ast & & \\ \ast& & \ast & & \ast & \\ \ast & \ast & & & & \ast \\ \end{pmatrix} \] |
Cases: |
|
\[ \begin{pmatrix} \ast & \ast & \ast & \ast & & \\ \ast & \ast & \ast & & \ast & \\ \ast & \ast & \ast & & & \ast \\ \ast & & & \ast & & \\ & \ast & & & \ast & \\ & & \ast & & & \ast \\ \end{pmatrix}, \begin{pmatrix} \ast & \ast & \ast & & \ast & \ast \\ \ast & \ast & \ast & \ast & & \ast\\ \ast & \ast & \ast & \ast & \ast & \\ & \ast & \ast & \ast & & \\ \ast& & \ast & & \ast & \\ \ast & \ast & & & & \ast \\ \end{pmatrix} \] |
Let $\Gamma$ be a graph on $n$ vertices. A graphical matroid is a matroid corresponding to the collection \[ \{e_i-e_j\mid \text{ where } \{i,j\}\in E_\Gamma\}. \] Here $e_1,\ldots,e_n$ is a basis of $\Bbbk^n$.
The flat of this matroid corresponds to the subdivision of $[n]$ into subsets $V_1,\ldots,V_r$, for which the induced graphs $\Gamma|_{V_i}$ are connected. We call such subdivisions clusterings.
Let $\mathcal{L}(\Gamma)$ denote the geometric lattice of the graphical matroid.
This corresponds to the "15" game on
The edge graph is the Cayley graph of
$\Sigma_4$ with the standard generators
\[
(1,2), (2,3), (3,4)
\]
|
Cast:
|
Anton Ayzenberg Victor Buchstaber Vlad Cherepanov Mikiya Masuda Grigory Solomadin Kostya Sorokin |
Real torus actions:
|
Vladimir Gorchakov |
Discussion on discrete torus actions:
|
Li Yu |
Cohomology of Hessenberg varieties:
|
Mikiya Masuda Takashi Sato |
Representation theory and PR of staircase matrix manifolds:
|
Hiraku Abe Tatsuya Horiguchi |
Discussion on arrow matrices:
|
Tadeusz Januszkiewicz |
Discussion on symplectic implosion and arrow matrices:
|
Megumi Harada |
Advice on Cauchy interlacing law:
|
Alexey Naumov |
Cyclic Toda flows:
|
Nadya Khoroshavkina |
Skew symmetric matrices
and other Lie types: |
Semyon Abramyan |
Computations sanity check:
|
Anton Medvedev |
GKM computations team:
|
Eugenea Akhmedova Alexander Demin Sveta Gavrilova Eugene Zhukov Anton Medvedev Grigory Taroyan |
Gauss parallelization and C++ team:
|
Oleg Kachan Eduard Tulchinskiy |
Algebraic matroid theory:
|
David Mikheev |
Complexity of star matrices:
|
Anna Kosovskaya |
Thanks for organizing the Data analysis
school in Voronovo, where this project was born:
|
Evgeny Sokolov |
Thanks for an idea to use html in math work:
|
Ilya Schurov |
The slides were made using:
|
Reveal.js $\KaTeX$ Math.js model-viewer |
Used data:
|
Equilateral Goldberg polyhedron by
Gerry in Québec
Wikipedia Running letters by Ganesh Prasad |
Inspired by:
|
"The Matrix" movie |
|
Immanuel Kant: space and time are the a priori forms of sensible intuition.
We are able study topology, because we have intuition about the space.
We can study dynamical systems, algorithms, and do any form of logical reasoning, because
we have intuition about time.
No way to learn abstract algebra, based on kantian concepts.
Unless... well, algebra stems from
groups and symmetries, and symmetries are the manifestation of a priori aesthetic
perception, inherent to humans.
|