The material for this section will mainly be taken from Rudin's [1,2]. Additional examples will be handed out as problem sheets.
Hilbert spaces. Definition, Orthonormal bases, Riesz representation theorem H* @ H. l2(\mathbbN) and L2(W).
Banach spaces. Definition. A list of examples of
function spaces which are Banach spaces: lp(\mathbbN),
Lp(W), C0(W), C0,a(W), Wm,p(W), W0m, p(W), W-m, p(W), BV(W).
A function space which is not a Banach space:
S(\mathbbRn).
Banach spaces which are (not necessarily) function spaces: the dual X*, the space of bounded operators L(X, Y) with operator norm.
Bounded operators.
Definition, some examples: differential operators P(D):CmÆ C0,
integral operators of various types (Young's inequality, infinite
matrices on lp, Hilbert-Schmidt operators).
The (Banach-) contraction mapping principle, ``Neumann-series" (a.k.a. the geometric series for (I+K)-1).
General facts. Hahn-Banach, Open Mapping and Closed
Graph theorems. Closed subspaces in general do not have closed
complements.
Adjoint operators, kerT* = (rangeT)^, kerT = ^(rangeT).
Weak convergence. Definition, Banach-Alaoglu theorem.
Riemann-Lebesgue lemma.
Compact operators and Fredholm operators.
Fourier series on \mathbbTn. Uniform convergence of partial sums sN(f) if f is Hölder continuous, convergence in L2(\mathbbTn) for f ‘ L2.
Fourier integral.
Inversion formula for f ‘ C1c(\mathbbRn), pointwise for f with f, [^f] ‘ L1(\mathbbRn) and in the sense of distributions for f ‘ S¢(\mathbbRn).
The Plancherel formula and boundedness of the Fourier transform on L2(\mathbbRn).
Convolution and Fourier multipliers.
Boundedness on L2, Hölder spaces (example of dyadic
decomposition - notes will be provided) and without proof Lp.
Elliptic estimates for constant coefficient operators.
Definition of the spaces Wm, p(W) with m ‘ \mathbbZ and of W0m, p(W) for any open subset W Ã \mathbbRn. Invariance under diffeomorphisms (coordinate changes) of W.
Trace theorems.
Restriction of f ‘ Wm, p(W) to a smooth submanifold makes
sense for suitable values of m and p. In particular one can
restrict any f ‘ W1, p(W) to W if
W is smooth, and if p > n then one can ``restrict to a
point,'' i.e. any f ‘ f ‘ W1, p(W) is defined
everywhere. In fact W1, p(W) Ã C0, 1-n/p(W).
Sobolev inequalities.
||f||Ln/(n-1) £ C ||‹f||L1, and consequences for
other values of p than p = 1. Relation between the Sobolev inequality
and the isoperimetric inequality.
Compactness issues.
The Rellich-Kondrachov compactness theorem (and Ascoli-Arzela of
course).
Sobolev-like spaces. (if time permits)
BV(W) and sets with finite perimeter. Quick solution
to the problem of minimal surfaces.
Solving PDEs by Hilbert-Sobolev space methods. Solution of the Dirichlet problem by minimizing the Dirichlet integral. Solution of Poisson's equation by using the Riesz representation theorem. A proof of the Riemann Mapping Theorem from complex analysis, using real variable methods (if time permits).