Computational Science Stack Exchange is a question and answer site for scientists using computers to solve scientific problems. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

In the FEM we usually use a trick to find the coefficients for shape functions:

Finding $M^{-1}$ in the system $MC=I$, where for example in the linear case: $$ \begin{aligned} M&=[1 ,x_1,y_1;1 ,x_2,y_2;1 ,x_3,y_3],\\ C&=[a_1, b_1,c_1;a_2, b_2,c_2;a_3, b_3,c_3],\\ I&=[1,0,0;0,1,0;0,0,1]. \end{aligned} $$ and $\phi_1=a1+b_1x_1+c_1y_1$ and also $\phi_2,\phi_3$ are defined in the same way. Since $MC=I$ so $M$ is inverse of $C$. But how do we ensure that $M$ is invertible for a higher degree basis?

share|improve this question
    
Oops. Then go ahead and change it back. – David Ketcheson Jan 17 at 10:51
    
@David I tried to be a bit more explicit -- how's that? (Rosa, feel free to revert if you're not happy with it.) – Christian Clason Jan 17 at 10:57
    
@ChristianClason, thanks a lot and no problem. In fact I am so new in fem and enjoy from your comments. – Rosa Jan 17 at 11:15
up vote 4 down vote accepted

This is ensured by choosing the degrees of freedom appropriately. (That's why you need, e.g., additional degrees of freedom on the edges of a triangle for quadratic elements.)

Basically, constructing a (local) finite element space consists of choosing a reference element $K$ (e.g., a triangle) and a polynomial space $P$ (e.g., bivariate polynomials on $K$ of total order up to $2$). You then need to find a set of conditions (e.g., point evaluation in some $x_i\in K$) to make the interpolation problem "for given $f\in C(K)$, find $p\in P$ such that $p(x_i) = f(x_i)$ for all $i$" have a unique solution -- these are then the degrees of freedom you need on each element.

It is then a theorem of linear algebra that the matrix $M$ (which is called Vandermonde matrix) is always invertible. (Note that your equation $MC=I$ is just a compact way of writing the nodal interpolation conditions $\phi_i(x_j,y_j) = \delta_{ij}$.)

share|improve this answer

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.