Foundations — Eigenvalues of A — system modes, stability

This page assumes you know nothing yet. Every letter, arrow, and squiggle in the parent note Eigenvalues of A — system modes, stability is built here from the ground up, in the order you need them.

0. What is a "state" and why an arrow?

Plain words. The state of a system is the shortest list of numbers that fully describes "where it is right now". For a spinning spacecraft it might be [tilt angle, tilt speed]. We stack that list into a column called the state vector $\mathbf{x}$.

The picture. Think of a dot living in a room. Its position along each wall is one number. Two numbers = a dot on a flat page; three = a dot in a box. The dot is the state.

Why the topic needs it. Eigenvalues describe how a whole state moves through time. No state vector, nothing to move. Built more carefully in State-Space Representation.

1. The dot symbol: $\dot{\mathbf{x}}$ (a rate of change)

Plain words. A dot over a letter means "how fast this thing is changing, right now, per second". $\dot{x}$ reads "x-dot" = the speed of $x$.

The picture. Stand on the moving dot from before. The little arrow showing which way it drifts next second — that arrow is $\dot{\mathbf{x}}$. Long arrow = changing fast; no arrow = frozen.

Why the topic needs it. The whole subject is one equation, $\dot{\mathbf{x}}=A\mathbf{x}$: "the drift arrow at any point is decided by the matrix $A$ acting on where you are."

2. The matrix $A$: a machine that bends arrows

Plain words. A matrix is a grid of numbers that eats a vector and spits out a new vector. $A\mathbf{x}$ means "run the state through the grid".

The picture. Feed the room a wind pattern: at every point the wind blows a certain direction and strength. $A$ is that wind field. Where you are ($\mathbf{x}$) decides the wind you feel ($\dot{\mathbf{x}}=A\mathbf{x}$).

Why the topic needs it. $A$ carries every rate, every coupling, all the physics. Everything else is decoding $A$.

3. Eigen-direction: where the wind blows straight

Plain words. Most arrows get turned by $A$. But a few special arrows come out pointing the same way — only longer or shorter. Those are eigenvectors $\mathbf{v}$; the stretch factor is the eigenvalue $\lambda$.

The picture. Walk the wind field. On most paths the wind shoves you sideways. On the eigen-line the wind pushes exactly along the line you're standing on — straight out or straight back in. No turning, pure scaling.

Why the "eigen"? German for "own / characteristic". These are $A$'s own directions. Along them the swirling matrix becomes plain multiplication — a scalar. That single fact is what collapses the coupled mess into $n$ easy pieces. Deepened in Diagonalization and Modal Decomposition.

4. The letters $I$, $\det$, and the hunt for $\lambda$

Plain words. $I$ is the identity matrix — the "do-nothing" grid ($I\mathbf{v}=\mathbf{v}$), 1's on the diagonal, 0's elsewhere. We need it so $\lambda \mathbf{v}$ can be written as $\lambda I\mathbf{v}$ (a matrix times $\mathbf{v}$), letting us factor.

$A\mathbf{v}=\lambda \mathbf{v}\Rightarrow A\mathbf{v}-\lambda I\mathbf{v}=0\Rightarrow (A-\lambda I)\mathbf{v}=0.$

Plain words for $\det$. The determinant $\det (M)$ of a matrix is one number measuring how much $M$ scales area (in 2-D) or volume (in 3-D). If $\det (M)=0$, $M$ squashes the whole space flat onto a line or point.

The picture. Take the unit square. Apply $M$ to its corners. The area of the resulting parallelogram = $|\det M|$. Zero area means everything collapsed — the map is not reversible.

Why this tool and not trial-and-error? Setting the determinant to zero turns "search all $\lambda$" into a single polynomial equation whose roots ARE the eigenvalues. See Characteristic Polynomial & Determinants.

5. The exponential $e^{\lambda t}$: the shape of "grow or shrink"

Plain words. $e$ is a fixed number, about $2.718$. The function $e^{\lambda t}$ is the unique curve that grows in proportion to its own size — the natural language of anything whose speed depends on its amount (radioactivity, interest, decaying wobbles).

The picture. Three curves versus time:

• $\lambda <0$: swoops down to zero (decay). ← stable.
• $\lambda =0$: flat line (holds).
• $\lambda >0$: rockets up (blow-up). ← unstable.

Why the topic needs it. Once $A\mathbf{v}=\lambda \mathbf{v}$ shrinks the vector ODE to $\dot{s}=\lambda s$, the answer is $s(t)=e^{\lambda t}$. Every "mode" of the system is one of these curves riding along its eigenvector.

6. Complex numbers and $j$: where oscillation hides

Plain words. Some eigenvalues are complex: $\lambda =\sigma +j\omega$, where $j=\sqrt{-1}$ (engineers write $j$, not $i$). $\sigma$ (real part) and $\omega$ (imaginary part) are just two ordinary numbers glued together.

The picture — the complex plane. A flat map: go right by $\sigma$, up by $\omega$. Each eigenvalue is a dot on this map. The parent's whole stability rule is "which side of the vertical line is the dot on?"

More on how $\sigma ,\omega$ map to physical damping in Damping Ratio and Natural Frequency and Transfer Functions and Poles.

7. Putting the alphabet together

$\underset{\text{drift arrow}}{\underbrace{\dot{\mathbf{x}}}}=\underset{\text{wind machine}}{\underbrace{A}}\underset{\text{state dot}}{\underbrace{\mathbf{x}}}\underset{\text{eigen-directions}}{\overset{A\mathbf{v}=\lambda \mathbf{v}}{\to }}\underset{\text{one scalar}}{\underbrace{\dot{s}=\lambda s}}\Rightarrow \underset{\text{grow/shrink}}{\underbrace{s=e^{\lambda t}}}$

Every symbol in the parent note is now one of these earned pieces.

Equipment checklist

Read each question, answer aloud, then reveal. If any stumps you, reread that section before the parent note.