Theorem · The trace–determinant classificationFor x′ = A·x with trace τ and determinant Δ, the origin is a saddle if Δ < 0. If Δ > 0 it is a node when τ² − 4Δ > 0, a spiral when τ² − 4Δ < 0 and τ ≠ 0, and a centre when τ = 0.
The roots add to τ and multiply to Δ. If Δ < 0, the product is negative, so the roots are real and opposite in sign: a saddle.
If Δ > 0 and τ² − 4Δ > 0 the roots are real and have the same sign, which is the sign of τ: a node. If τ² − 4Δ < 0 the roots are τ/2 ± i·√(4Δ − τ²)/2, a complex pair whose real part is τ/2: a spiral if that is not zero. If τ = 0 and Δ > 0 the roots are the purely imaginary ±i·√Δ, and the real solutions are cosines and sines of the same angle: closed curves around the origin.
□
Δ < 0 saddle
Δ > 0, τ² − 4Δ > 0 node (stable if τ < 0)
Δ > 0, τ² − 4Δ < 0, τ ≠ 0 spiral (stable if τ < 0)
Δ > 0, τ = 0 centre
Δ > 0, τ² − 4Δ = 0 degenerate or star node
The classification is the whole of the story for a linear system in the plane. Two exact numbers, computed from four entries, sort every matrix with non-zero determinant. The dividing line between nodes and spirals is the parabola Δ = τ²/4 in the plane whose axes are τ and Δ, and the lab prints τ² − 4Δ as its own tile so that the side of the parabola is a number and not an impression.
Example · A spiralTake the matrix with rows 1 2 and −2 1. Then τ = 2 and Δ = 1 + 4 = 5, and τ² − 4Δ = 4 − 20 = −16. The discriminant is negative, so the roots are complex: λ = 1 ± 2i.
The real part 1 is positive, so orbits turn around the origin and spiral out: an unstable spiral. The lab prints the roots as 1 ± 2i and calls the origin an unstable spiral.
The other pictures
For rows 1 2 and 3 0, Δ = −6 < 0, so the origin is a saddle with eigenvalues −2 and 3. For rows −1 0 and 0 −2, τ = −3 and Δ = 2, so τ² − 4Δ = 1 > 0 and the roots are real and negative: a stable node, where every orbit comes into the origin. For rows 0 2 and −2 0, τ = 0 and Δ = 4, so τ² − 4Δ = −16 and λ = ±2i: a centre, with closed orbits.
On the parabola itself, τ² − 4Δ = 0 and there is one repeated root. The matrix with rows −1 1 and 0 −1 has τ = −2, Δ = 1 and the repeated root −1. It has only one eigenline, and orbits bend round it as they come in: a degenerate node. A matrix that is a multiple of the identity also has a repeated root, with every direction an eigenvector, and is a star node instead.
The line Δ = 0 is not in the table. There one root is zero and the origin is not an isolated equilibrium, which the next lesson treats along with the stability criterion.