My question is: is Wk2(R)⊗Wk2(R) dense in Wk2(R2), and more generally is this true in Rd?
I found this post:
Tensor products of functions generate dense subspace?
which shows the above type of result for C∞c. So my guess is that the answer should be affirmative, maybe requiring the assumption that k>d/2?
Answer
Yes, you can get this result from C∞c, because C∞c(R2) is dense in f∈Wk2(R2) for any k. So, any Wk2 function can be approximated by smooth functions with compact support, which in turn are approximated by sums of products of univariate smooth functions (even in the stronger sense, C∞c).
But it may be easier to apply the Fourier transform, which transforms Wk2(R2) to a weighted L2 space. Since ^u⊗v=ˆu⊗ˆv, the question reduces to its analog for Lebesgue spaces. Then we observe that the characteristic functions of cubes have a dense linear span, and they belong to the tensor product.
No comments:
Post a Comment