1

Consider the Navier Stokes equations for the steady, two-dimensional flow of an inviscid, incompressible fluid:

\begin{equation} \frac{\partial u}{\partial x}+\frac{\partial v}{\partial y}=0, \end{equation} \begin{equation} u\frac{\partial u}{\partial x}+v \frac{\partial u}{\partial y}+\frac{1}{\rho} \frac{\partial p}{\partial x}=0, \end{equation} \begin{equation} u\frac{\partial v}{\partial x}+v \frac{\partial v}{\partial y}+\frac{1}{\rho} \frac{\partial p}{\partial y}=0, \end{equation}

I am interested to find the eigenvalues and the characteristic equation for this system. Naturally, the first step is to put the system into the matrix form. Let $w=(u,v,p)^T,$ then we have $$\begin{pmatrix}1&0&0\\u&0&\frac{1}{\rho}\\0&u&0\end{pmatrix}\frac{\partial}{\partial x}w+\begin{pmatrix}0&1&0\\v&0&0\\0&v&\frac{1}{\rho}\end{pmatrix}\frac{\partial}{\partial y}w=0$$ First matrix is invertible, and so we can write the above equation as $$\frac{\partial}{\partial x}w+\begin{pmatrix}0&1&0\\0&\frac{v}{u}&\frac{1}{u\rho}\\v\rho&-\rho u&0\end{pmatrix}\frac{\partial}{\partial y}w=0.$$ We can find the eigenvalues from the last matrix (denote it $A$), by solving $$det(A-\lambda I)=det \begin{pmatrix}-\lambda&1&0\\0&\frac{v}{u}-\lambda&\frac{1}{u\rho}\\v\rho&-\rho u&-\lambda\end{pmatrix}=0$$ $$\implies \lambda^2(\frac{v}{u}-\lambda)-\lambda+\frac{v}{u}=0$$ $$\implies \lambda_1=\frac{v}{u}, \lambda_{2,3}=\pm i.$$

Hence, we have one real eigenvalue of the system $\lambda=\frac{v}{u}$. What I am not sure of is how to find the corresponding characteristic equation, and if this is related to the characteristic polynomial of the matrix. My feeling is that the characteristic equation is $\frac{dy}{dx}=\frac{v}{u},$ but i don't know how to show this using the methods I know for just one PDE (not the system). Thank you!

kirkos73
  • 103
  • 1
    Typo in the first matrix equation: It should be a $v$ rather than a $u$ in the bottom middle entry of the second matrix. – John Barber Sep 18 '24 at 17:06
  • how can a non linear pde have eigenfunctions? – user619894 Sep 18 '24 at 17:30
  • @user619894 I think the above NS equations are technically quasi-linear. – kirkos73 Sep 18 '24 at 19:38
  • The proper way to do what you are trying to do is via Riemann invariants. It is an involved process so I would suggest reviewing a textbook on conservation laws first. Also, I doubt you will find anything useful since some of the eigenvalues are imaginary, indicating that the system is elliptic – whpowell96 Sep 18 '24 at 21:14
  • @whpowell96 I will check it out, thank you. In this similar question however, the implication seems to be that the characteristic direction is very straightforward: https://math.stackexchange.com/questions/799360/method-of-characteristics-for-a-system-of-pdes Of course, the computation for part b) is omitted so it very well may be more involved than it looks. – kirkos73 Sep 18 '24 at 22:26
  • That problem is linear and hyperbolic. Your equation is not – whpowell96 Sep 18 '24 at 22:49

0 Answers0