5.17 Phase portraits

5.17 Phase portraits

IB Maths: Applications and Interpretation HL: Revision notes

Studying this course? Keep your progress in one place.

Key facts

  • A phase portrait shows trajectories (x(t),y(t))(x(t),y(t)) of ddt(xy)=M(xy)\frac{d}{dt}\binom{x}{y}=M\binom{x}{y}; the origin is the only equilibrium.
  • Eigenvectors give straight-line trajectories: solutions eλtve^{\lambda t}\mathbf{v} move out if λ>0\lambda>0 and in if λ<0\lambda<0.
  • Real distinct eigenvalues: (xy)=Aeλ1tv1+Beλ2tv2\binom{x}{y}=Ae^{\lambda_1t}\mathbf{v}_1+Be^{\lambda_2t}\mathbf{v}_2.
  • Both positive: source. Both negative: sink. Opposite signs: saddle point.
  • Complex λ=p±qi\lambda=p\pm qi: p>0p>0 spiral out, p<0p<0 spiral in, p=0p=0 closed ellipses.

The phase plane

A phase portrait shows the paths of two linked quantities in the xyxy-plane, with arrows for the direction of time.

A coupled linear system is dxdt=ax+by\frac{dx}{dt}=ax+by, dydt=cx+dy\frac{dy}{dt}=cx+dy, or in matrix form ddt(xy)=M(xy)\frac{d}{dt}\binom{x}{y}=M\binom{x}{y} with M=(abcd)M=\begin{pmatrix} a & b \\ c & d \end{pmatrix}.

A phase portrait shows trajectories: the paths (x(t),y(t))(x(t),y(t)) as tt increases, with arrows. At each point the direction of travel is M(xy)M\binom{x}{y}. For the systems in this course the origin is the only equilibrium point, where both rates are zero.

−2−1.5−1−0.50.511.52−4−3−2−11234xyEquilibrium (0, 0)y = x²y = 0.5x²y = −x²y = −0.5x²
Phase portrait of dx/dt = −x, dy/dt = −2y: trajectories y = Cx² all move in towards the equilibrium at the origin.
  • Direction at (x,y)(x,y)M(xy)M\binom{x}{y}
  • Equilibriumthe origin, where both rates are zero

At (1,0)(1,0) with M=(2134)M=\begin{pmatrix} 2 & 1 \\ 3 & 4 \end{pmatrix}, the direction is...

Eigenvectors and straight lines

A solution that starts on an eigenvector stays on that line, moving out if λ>0\lambda>0 and in if λ<0\lambda<0.

Find the eigenvalues λ\lambda from det⁡(M−λI)=0\det(M-\lambda I)=0, then an eigenvector v\mathbf{v} for each from (M−λI)v=0(M-\lambda I)\mathbf{v}=\mathbf{0} (a GDC will do this too). You are given only distinct, non-zero eigenvalues.

Starting on an eigenvector, the solution eλtve^{\lambda t}\mathbf{v} stays on that line through the origin: the straight-line trajectories.

Example: M=(4123)M=\begin{pmatrix} 4 & 1 \\ 2 & 3 \end{pmatrix} has λ2−7λ+10=0\lambda^2-7\lambda+10=0, so λ=2\lambda=2 with (1−2)\binom{1}{-2} and λ=5\lambda=5 with (11)\binom{1}{1}.

−3−2−1123−4−3−2−11234xyorigineigenvector (1, −2), λ = 2eigenvector (1, 1), λ = 5
The straight-line trajectories lie along the eigenvectors (1, −2) and (1, 1).

λ=−3\lambda=-3 with eigenvector v\mathbf{v}. A solution on that line...

Exact solutions

For real distinct eigenvalues, combine the two straight-line solutions and fit the initial conditions.

When λ1≠λ2\lambda_1\ne\lambda_2 are real, the general solution combines the two straight-line solutions. Use the initial conditions to find AA and BB.

For M=(4123)M=\begin{pmatrix} 4 & 1 \\ 2 & 3 \end{pmatrix} with x=3x=3, y=0y=0 at t=0t=0: A+B=3A+B=3 and −2A+B=0-2A+B=0, so A=1A=1, B=2B=2 (with λ1=2\lambda_1=2, v1=(1−2)\mathbf{v}_1=\binom{1}{-2}, λ2=5\lambda_2=5, v2=(11)\mathbf{v}_2=\binom{1}{1}):

x=e2t+2e5t,y=−2e2t+2e5t.x=e^{2t}+2e^{5t},\qquad y=-2e^{2t}+2e^{5t}.

For complex or imaginary eigenvalues describe the behaviour qualitatively.

−55101520253035404550−1010203040xyt = 0: (3, 0)Solution
The trajectory x = e^2t + 2e^5t, y = −2e^2t + 2e^5t starts at (3, 0) when t = 0 and moves away from the origin.
  • (xy)\binom{x}{y}Aeλ1tv1+Beλ2tv2Ae^{\lambda_1t}\mathbf{v}_1+Be^{\lambda_2t}\mathbf{v}_2

In the example, A+B=3A+B=3 and −2A+B=0-2A+B=0. What is BB?

Nodes and saddle points

The signs of two real eigenvalues decide whether the origin is a source, a sink or a saddle.

Both positive: every solution moves away from the origin, an unstable node (source). Both negative: every solution moves towards it, a stable node (sink), approaching along the eigenvector with the smaller ∣λ∣|\lambda|.

Opposite signs: a saddle point. Solutions on the line of the negative eigenvalue go in, but every other solution is turned away and leaves along the line of the positive eigenvalue.

To sketch: draw the eigenvector lines with arrows first, then curved trajectories that follow them.

Both positive

Origin:
Source (unstable node)
Motion:
Away from the origin

Both negative

Origin:
Sink (stable node)
Motion:
Towards the origin

Opposite signs

Origin:
Saddle point
Motion:
Unstable
−0.50.511.522.533.544.5512345xystart (1, 1)
A saddle (eigenvalues 1 and −1): the path comes in along the y-direction, then is thrown out along the x-direction.

Eigenvalues −1-1 and 33. The origin is...

Spirals and ellipses

The real part pp decides growth or decay; the imaginary part makes the path rotate.

For λ=p±qi\lambda=p\pm qi with q≠0q\ne0 the solutions rotate around the origin. p>0p>0: spiral away. p<0p<0: spiral towards the origin. p=0p=0 (imaginary): closed ellipses with neither growth nor decay.

To find the sense of rotation test one point: at (1,0)(1,0) the direction is (a,c)(a,c), so c>0c>0 is anticlockwise and c<0c<0 clockwise.

Example: dxdt=−x−4y\frac{dx}{dt}=-x-4y, dydt=x−y\frac{dy}{dt}=x-y has λ=−1±2i\lambda=-1\pm2i, so a spiral towards the origin; at (1,0)(1,0) the direction is (−1,1)(-1,1): anticlockwise.

−1−0.8−0.6−0.4−0.200.20.40.60.811.2−0.4−0.20.20.40.6xy(1, 0)origin
Trajectory spiralling in anticlockwise from (1, 0): λ = −1 ± 2i.

p>0p>0

Trajectory:
Spiral away
Size:
Grows

p<0p<0

Trajectory:
Spiral towards the origin
Size:
Decays

p=0p=0

Trajectory:
Closed ellipses
Size:
Constant size

λ=2±i\lambda=2\pm i. The trajectories...

Reading a phase portrait

The portrait answers what happens to both quantities in the long term; give the eigenvalue as the reason.

A phase portrait answers: what happens to the two quantities in the long term?

  • Stable: a sink or inward spiral means both settle at the equilibrium.
  • Unstable: a source or outward spiral means both grow without bound.
  • Cycles: imaginary eigenvalues give populations that rise and fall repeatedly.
  • Saddle: the outcome depends on the starting point, and only one special line leads to equilibrium.

Always state the conclusion in the context of the question, using the eigenvalues as the reason.

−1−0.8−0.6−0.4−0.200.20.40.60.811.2−1−0.50.51xyStart (1, 0)Inward spiral
Eigenvalues with negative real part give an inward spiral: x = e^(−0.2t) cos t, y = e^(−0.2t) sin t both settle at the origin (illustrative).

Sink or inward spiral

Long term:
Both tend to the equilibrium
Reason:
Negative real parts

Source or outward spiral

Long term:
Both grow without bound
Reason:
Positive real parts

Imaginary

Long term:
Repeated cycles
Reason:
Real part zero

Saddle

Long term:
Depends on the start
Reason:
Opposite signs

Eigenvalues −0.5±2i-0.5\pm2i. Long-term behaviour?

Try an exam question

The populations xx and yy satisfy dxdt=−x−4y\frac{dx}{dt}=-x-4y and dydt=x−y\frac{dy}{dt}=x-y. Find the eigenvalues of the system and describe the long-term behaviour, including the direction of rotation.

[4 marks]

That's the notes covered.

Carry on to the next subtopic.

Try a question

Find the eigenvalues of the matrix (4123)\begin{pmatrix} 4 & 1 \\ 2 & 3 \end{pmatrix} that represents the system.

1 mark, marked against the mark scheme

Revision notes