For positive integers m and 1≤p≤∞, let us define the function ∥⋅∥m,p as follows.
∥u∥m,p=0≤∣α∣≤m∑∥Dαu∥ppp1,1≤p<∞
∥u∥m,∞=0≤∣α∣≤mmax∥Dαu∥∞,p=∞
This becomes the norm of the Sobolev space.
Explanation
Simply put, the Sobolev space is a set of functions whose derivatives up to the m-th order all belong to Lp. While the Lp space has useful properties and is an important space, it might be somewhat insufficient for solving differential equations. We need to consider spaces with more properties than the Lp space, and that is precisely the Sobolev space.
Clearly, if m=0 then W0,p=Lp. Also, since C0∞ is dense in Lp, when 1≤p<∞ then W00,p=LP. Additionally, the following embedding exists, making Wm,p treatable as a subspace of Lp.
W0m,p(Ω)→Wm,p(Ω)→Lp(Ω)
The following three function spaces all have the Sobolev norm ∥⋅∥m,p. For positive integers m and 1≤p≤∞,
Wm,p(Ω):={u∈Lp(Ω):Dαu∈Lp(Ω)∀0≤∣α∣≤m}
Hm,p(Ω):=the completion of {u∈Cm(Ω):∥u∥m,p<∞}
W0m,p(Ω):=the closure of C0∞(Ω) in Wm,p(Ω)
Here, (1)~(3) are called Sobolev spaces. Hm,p and Wm,p can actually be shown to be the same.
Symbols used to denote the Sobolev space include Hm,p,Wm,p,Pm,p,Lpm, among others. Also, before the name Sobolev space became established, it was sometimes called a Beppo Levi space.