Meyers–Serrin theorem explained

Wk,p(\Omega)

for arbitrary domains

\Omega\subseteq\Rn

.

Historical relevance

Originally there were two spaces:

Wk,p(\Omega)

defined as the set of all functions which have weak derivatives of order up to k all of which are in

Lp

and

Hk,p(\Omega)

defined as the closure of the smooth functions with respect to the corresponding Sobolev norm (obtained by summing over the

Lp

norms of the functions and all derivatives). The theorem establishes the equivalence

Wk,p(\Omega)=Hk,p(\Omega)

of both definitions. It is quite surprising that, in contradistinction to many other density theorems, this result does not require any smoothness of the domain

\Omega

. According to the standard reference on Sobolev spaces by Adams and Fournier (p 60): "This result, published in 1964 by Meyers and Serrin ended much confusion about the relationship of these spaces that existed in the literature before that time. It is surprising that this elementary result remained undiscovered for so long."

References