dof_array.flat_norm is CL-specific

[kaushikcfd]

cf.

meshmode/meshmode/dof_array.py

Lines 645 to 646 in f87576d

	import pyopencl.array as cl
	assert isinstance(ary, cl.Array)

[kaushikcfd]

Why does flat_norm flatten a group array before computing the norm?

In [2]: a = np.random.rand(3, 3)

In [3]: np.linalg.norm(a)
Out[3]: 1.899203808354709

In [4]: np.linalg.norm(a.reshape(-1))
Out[4]: 1.899203808354709

[kaushikcfd]

Why does flat_norm flatten a group array before computing the norm?

Oops forgot my 450 here. Feel free to ignore this. Numpy's default norm being Frobenius norm, didn't help :P.

[inducer]

To be fair, I think numpy defaults to Frobenius, which is equivalent to a flattened 2-norm for 2D arrays. For (say 3D) tensor product elements, the shape will likely be (nelements, Np, Np, Np), which would just be an error in numpy, making the flatten necessary after all.