VW  CVM 1.1 
We have referred often to the completeness of the eigenfunctions of the Laplacian in a lattice cell. This means that (in an appropriate sense) you can build any function periodic with respect to a lattice out of lattice waves.
As a small verification of this, we took the complexvalued identity
function
If the approximation were exact, we would expect to see perfect copies of the color wheel on a completely black background, like a quilt sewn together out of swatches just like the original figure. This figure is imperfect in two ways: there are subtle wiggles, as if the original wheel had acquired waves; that imperfection comes from the fact that we included only 30 terms. The second problem is more fundamental.
The periodic extension of the identity function is discontinuous. For example, the value of the extension at the point p should be both p and p (the latter to match what happens in the interval from p to 3p, which is supposed to be a repeat of the piece from p to p). Since our approximation is a continuous function, no matter how many terms we include, it cannot exactly match a discontinuous one. But from the picture, you can see that it's trying very hard to match both. The position corresponding to the number p is colored white, representing 0, halfway between p and p. This is similar to a wellknown property of Fourier series on an interval: at a jump discontinuity they converge to the average of the two onesided limits.

