- Explain eigenvectors and eigenvalues in terms of a matrix transformation.
- Use eigenstructure to interpret Fibonacci updates and long-term ratios.
- Interpret phase-plane behavior in the presented population and spring models.
- Recognize eigenvalue applications in iterated systems and graphs.
Eigenvector and Eigenvalue Foundations
A matrix transformation generally changes a vector's length and direction. An eigenvector is a nonzero vector that the transformation scales without changing its line of direction, and its eigenvalue is the applied scale factor.
An eigenvalue of 1 leaves the length of its associated eigenvectors unchanged. Over the real numbers, a matrix that rotates every nonzero vector by 90 degrees has no nonzero eigenvectors.
What distinguishes an eigenvector from a vector whose direction is changed by a matrix transformation?
An eigenvector is nonzero and is scaled without changing its line of direction; the scale factor is its eigenvalue.
Fibonacci Growth Through Eigenvectors
Each Fibonacci term is obtained by adding the preceding two terms. The Fibonacci update matrix maps (F[n−1], F[n−2]) to (F[n], F[n−1]), and repeated multiplication advances the pair one term at a time.
If both starting numbers are positive and every new term is the sum of the previous two, what happens to the ratio of consecutive terms?
The ratio converges to the golden ratio. For positive starting numbers, this limiting behavior is determined by the recurrence and its matrix rather than by the particular starting terms.
Population Models and Phase Planes
A phase plane displays solution trajectories for a system of differential equations under different initial conditions. A point's derivative components determine its instantaneous direction of motion in that plane.
- In the simplified fox–rabbit model, F′ = F.
- In the simplified fox–rabbit model, R′ = 3R − F.
Which two initial-state lines preserve their population relationship over time in the simplified fox–rabbit model?
A state beginning on the rabbit axis remains on that axis, and a state with twice as many foxes as rabbits preserves that population ratio.
Spring Dynamics and Complex Eigenvalues
- The derivative of position is velocity: x′ = v.
- The spring model obeys mv′ = −kx − bv.
Under the idealized assumptions presented, an undamped mass–spring system oscillates indefinitely around equilibrium. For the undamped linear system presented, purely imaginary eigenvalues produce rotational phase-plane motion without decay toward or divergence from equilibrium.
| Condition | Behavior |
|---|---|
| Negative real component | Decay and eventual convergence to equilibrium in the damped spring system presented |
| More negative real component | Faster decay in the damping comparison presented |
| Positive real component | Growth and motion away from equilibrium in the spring phase-plane model presented |
For the compared undamped trajectories with the same initial state or displacement, what changes when the spring constant increases?
The maximum velocity increases, but the maximum position reached does not increase.
Long-Term Behavior in Matrix Models
In the presented human–zombie scenario, linear hourly population updates form a matrix iteration. An eigenvector identifies the equilibrium population ratio, and an eigenvalue of 1 leaves a state on that equilibrium eigenvector unchanged by the next hourly update.
Transition probabilities for a random walk on a network can be organized into a matrix. For the described network, an eigenvector of that transition matrix gives the long-run percentage of time spent at each node.
After many iterations in the presented human–zombie scenario, what population ratio does the system approach?
The system converges to twice as many zombies as humans.
Graph Spectra and Further Applications
For the unweighted graph representation presented, an adjacency matrix places 1 at entries for connected node pairs and 0 at entries for unconnected pairs.
For the graph setting discussed, the eigenvalues of a bipartite graph's adjacency matrix occur in positive-negative pairs, and a paired adjacency spectrum can identify a graph as bipartite.
- Eigenvalues can help analyze how quickly a disease spreads through a population.
- Eigenvalue methods are used in data compression.
- Eigenvalue methods are used in facial recognition.
In the presented dating-network example, what happens to the positive-negative eigenvalue pairing when a connection is added within one of the two node sets?
The positive-negative eigenvalue pairing is removed.
Common misconceptions
Watch out: The zero vector counts as an eigenvector.
Instead: The zero vector does not count as an eigenvector.
Watch out: In the fox–rabbit differential-equation model, the coefficient matrix outputs new populations.
Instead: It outputs instantaneous population rates of change rather than new populations.
Watch out: For the compared undamped trajectories starting from the same displacement, increasing the spring constant increases the maximum position reached.
Instead: Increasing the spring constant does not increase the maximum position reached in that comparison.
Lesson recap
- An eigenvector is a nonzero vector scaled without a change to its line of direction, and its eigenvalue is the scale factor.
- As the Fibonacci term index increases, coordinate pairs approach an eigenvector line whose associated eigenvalue is the golden ratio.
- A point's derivative components determine its instantaneous direction of motion in a phase plane.
- In the presented damped spring system, a negative real eigenvalue component corresponds to decay and eventual convergence to equilibrium.
- For the described random-walk network, a transition-matrix eigenvector gives the long-run percentage of time spent at each node.
- For the graph setting discussed, a positive-negative paired adjacency spectrum can identify a graph as bipartite.