Let T be a real number with T>0 and let d be a natural number. Adopt the notation B[0,T] and λ[0,T] of the restricted Lebesgue measure space on a compact interval, so that ([0,T],B[0,T],λ[0,T]) is a measure space. Write x⋅y for the dot product and ∣x∣ for the Euclidean norm of elements x,y of Euclidean space Rd.
1. (Square-integrable maps.) A map u:[0,T]→Rd with components u1,…,ud is square-integrable if each component ui is measurable from ([0,T],B[0,T]) to R equipped with its Borel σ-algebra, and the Lebesgue integral of the nonnegative function t↦∣u(t)∣2 is finite:
∫[0,T]∣u∣2dλ[0,T]<∞.
Write L2([0,T];Rd) for the set of all square-integrable maps [0,T]→Rd, equipped with the pointwise operations (u+v)(t)=u(t)+v(t) and (cu)(t)=cu(t) for real c.
2. (Almost-everywhere equality.) For u,v∈L2([0,T];Rd) write u∼v if there is a set N∈B[0,T] with λ[0,T](N)=0 such that u(t)=v(t) for every t∈[0,T]∖N.
3. (The space L2.) L2([0,T];Rd) is the set of equivalence classes of the relation ∼ on L2([0,T];Rd). The class of u is written [u], and the operations on L2([0,T];Rd) are [u]+[v]=[u+v] and c[u]=[cu] for real c.
4. (Pairing, norm, and metric.) For u,v∈L2([0,T];Rd) set
⟨[u],[v]⟩L2=∫[0,T]u⋅vdλ[0,T],∥[u]∥L2=(∫[0,T]∣u∣2dλ[0,T])1/2,
the second being formed with the nonnegative square root, and
dL2([u],[v])=∥[u]−[v]∥L2.
5. (Notational convention.) An element of L2([0,T];Rd) is denoted by the same symbol as a representative of it: one writes u for [u], ⟨u,v⟩L2 for ⟨[u],[v]⟩L2, and ∥u∥L2 for ∥[u]∥L2.