Trace-determinant plane (stability analysis)

Consider an ODE system described by the matrix A=(abcd)A = \begin{pmatrix} a & b \\ c & d \end{pmatrix}. As we have seen, the eigenvalues of this matrix determine the solutions to the system.

det(AIλ)=0λ2(a+d)λ+adbc=0λ2tr(A)λ+det(A)=0tr(A)2±tr2(A)4det(A)=λ. \begin{align*} \det(A - I \lambda) &= 0 \\ \lambda^2 - (a+d)\lambda + ad-bc &= 0 \\ \lambda^2 - \mathrm{tr}(A)\lambda + \det(A) &= 0 \\ \frac{\mathrm{tr}(A)}{2} \pm \sqrt{\frac{\mathrm{tr}^2(A)}{4} - \det(A)} &= \lambda. \end{align*}

The discriminant of the quadratic equation formula will determine whether the eigenvalues are real or complex, and will determine the behavior of the system.

  1. If tr(a)>0\mathrm{tr}(a)>0 then the real part is positive, so the solution is unstable.

  2. When det(A)>tr2(A)/4\det(A) > \mathrm{tr}^2(A)/4 the eigenvalue is imaginary, so the solution is a spiral.

  3. If det(A)<0\det(A)<0 then at least one eigenvalue is positive, so the solution is unstable.

Therefore stable solutions are only found in the top-left quadrant of the Poincaré diagram below.