SOR bug

[lukeolson]

From the example on wikipedia:
https://en.wikipedia.org/wiki/Successive_over-relaxation#Example

A = np.array([[ 4, -1, -6,  0],
              [-5, -4, 10,  8],
              [ 0,  9,  4, -2],
              [ 1,  0, -7,  5.]])
b = np.array([2, 21, -12, -6.])
x = np.array([0, 0, 0, 0.])

pyamg.relaxation.relaxation.sor(A, x, b, 0.5, iterations=1)
print(x)

gives

[ 0.25     -2.9375    5.109375  6.503125]

whereas the first iteration should be

[0.25  −2.78125  1.6289062  0.5152344]

[Sushmey]

I think the discrepancy is because the Gauss-Seidel iteration being used internally in SOR doesn't consider the relaxation factor while calculating the next x_i. It just processes the whole matrix without the relaxation factor.

I've tried to resolve this with a commit and I'll make the PR referencing this issue.