# What's the mathematical meaning of characteristic decomposition of Euler equations

 Register Blogs Members List Search Today's Posts Mark Forums Read September 2, 2018, 23:46 What's the mathematical meaning of characteristic decomposition of Euler equations #1 Senior Member   Join Date: Oct 2017 Location: United States Posts: 229 Blog Entries: 1 Rep Power: 8 Greetings, When people apply nonreflecting BCs for compressible Euler equations, typically what we do is studying the characteristic decomposition of the compressible Euler equations, break the coefficient matrix into a diagonal matrix. But what baffles me is that, what's the mathematical meaning of this decomposition? What do those variables (namely the eigenvalues and eigenvectors from the decomposition) mean mathematically? Why in this way can we impose BCs? Thanks.   September 3, 2018, 04:16 #2
Senior Member

Filippo Maria Denaro
Join Date: Jul 2010
Posts: 6,266
Rep Power: 67   Quote:
 Originally Posted by TurbJet Greetings, When people apply nonreflecting BCs for compressible Euler equations, typically what we do is studying the characteristic decomposition of the compressible Euler equations, break the coefficient matrix into a diagonal matrix. But what baffles me is that, what's the mathematical meaning of this decomposition? What do those variables (namely the eigenvalues and eigenvectors from the decomposition) mean mathematically? Why in this way can we impose BCs? Thanks.

From a mathematical point of view, by using the characteristic method you can reduce the PDE system to an ODE system with initial value problem.
In terms of linear algebra, the original matrix is pre and post multiplicated my matrices of eigenvectors to get a diagonal matrix where only eigenvalue appear. Of course, the resolved variables are not the original one but the transformed one. That is also at the origin of the Riemann invariant for omo-entropic flows. Such variables are called invariant because they remain constant along the characteristic curves, allowing for an exact mathematical solution. However, this can be done exactly only for particular 1D assumption.   September 3, 2018, 11:11 #3
Senior Member

Join Date: Oct 2017
Location: United States
Posts: 229
Blog Entries: 1
Rep Power: 8 Quote:
 Originally Posted by FMDenaro From a mathematical point of view, by using the characteristic method you can reduce the PDE system to an ODE system with initial value problem. In terms of linear algebra, the original matrix is pre and post multiplicated my matrices of eigenvectors to get a diagonal matrix where only eigenvalue appear. Of course, the resolved variables are not the original one but the transformed one. That is also at the origin of the Riemann invariant for omo-entropic flows. Such variables are called invariant because they remain constant along the characteristic curves, allowing for an exact mathematical solution. However, this can be done exactly only for particular 1D assumption.
Thanks for the reply! And do you have any idea about the mathematical or physical meaning of left & right eigenvectors? This is what most people have used when they do characteristic decomposition.   September 3, 2018, 12:40 #4
Senior Member

Filippo Maria Denaro
Join Date: Jul 2010
Posts: 6,266
Rep Power: 67   Quote:
 Originally Posted by TurbJet Thanks for the reply! And do you have any idea about the mathematical or physical meaning of left & right eigenvectors? This is what most people have used when they do characteristic decomposition.

It is a linear algebra topic. Consider the matrix A and the following eigenvalues problems where the right and left eigenvectors appear

(A- lambda(k) I).rk = 0
lk^T .(A- lambda(k) I) = 0

so that you get a basis in R^N since lj^T.rk = 0.

After determining the eigenvectors, you can now build the matrix R =[r1,r2,...rN] and L =R^-1 such that

D = L.A.R

is the diagonal matrix having the eigenvalues as entries.
Therefore, the eigenvectors are used to build rotation matrices in such a way to express the diagonal form. Practically, you are rewriting the original matrix in terms of the principal axes   September 3, 2018, 22:07 #5
Senior Member

Join Date: Oct 2017
Location: United States
Posts: 229
Blog Entries: 1
Rep Power: 8 Quote:
 Originally Posted by FMDenaro It is a linear algebra topic. Consider the matrix A and the following eigenvalues problems where the right and left eigenvectors appear (A- lambda(k) I).rk = 0 lk^T .(A- lambda(k) I) = 0 so that you get a basis in R^N since lj^T.rk = 0. After determining the eigenvectors, you can now build the matrix R =[r1,r2,...rN] and L =R^-1 such that D = L.A.R is the diagonal matrix having the eigenvalues as entries. Therefore, the eigenvectors are used to build rotation matrices in such a way to express the diagonal form. Practically, you are rewriting the original matrix in terms of the principal axes
So the sole purpose of using left/right eigenvectors is just to find out the principle axis of the matrix? Do the vectors themselves have some physical meaning?   September 4, 2018, 04:34 #6 Senior Member   Filippo Maria Denaro Join Date: Jul 2010 Posts: 6,266 Rep Power: 67   the eigenvectors are tangent to the integral curves in the the variable space, have a look to the book of Leveque   September 4, 2018, 15:13 #7 Senior Member   Martin Hegedus Join Date: Feb 2011 Posts: 500 Rep Power: 18 The eigenvectors can be thought of as filters. The state variables are dotted with the eigenvector to get the variable of interest. The physical variables of interest are entropy, vorticity, and the +/- acoustic waves. At this time, I'm going to probably do some fudging. We always seem to use the term "eigenvector", however, I view the first three eigenvectors (the ones for entropy and vorticity which are multiplied by velocity) as an eigentensor. Not sure if eigentensor is an actual mathmatical term, but that's what it seems, at least to me, to be.  Tags boundary condition, characteristics, compressible flow problem, euler equations Thread Tools Search this Thread Show Printable Version Email this Page Search this Thread: Advanced Search Display Modes Linear Mode Switch to Hybrid Mode Switch to Threaded Mode Posting Rules You may not post new threads You may not post replies You may not post attachments You may not edit your posts BB code is On Smilies are On [IMG] code is On HTML code is OffTrackbacks are Off Pingbacks are On Refbacks are On Forum Rules Similar Threads Thread Thread Starter Forum Replies Last Post francescofaggiano SU2 0 May 4, 2016 05:05 bubble45 Main CFD Forum 9 March 16, 2015 18:09 Christian87 Main CFD Forum 0 April 5, 2013 14:51 RameshK Main CFD Forum 3 May 30, 2012 11:41 Jan Ramboer Main CFD Forum 2 August 19, 1999 02:58

All times are GMT -4. The time now is 09:28.