Vectors
[12/100%]Is the limit of a sequence of quadratic forms a quadratic form?
If is a positive, symmetric, sesquilinear form, is it necessarily true that
for all and ?
Find a coordinatized proof of the Riesz representation theorem.
The Riesz representation theorem says that to each bounded linear functional on a Hilbert space there corresponds a vector in such that for all . The statement is "invariant" or "coordinate-free," and therefore, according to current mathematical ethics, it is mandatory that the proof be such. The trouble is that most coordinate-free proofs are so elegant that they conceal what is really going on.
The unit ball of every Hilbert space is strictly convex.
In a real vector space (and hence, in particular, in a complex vector space) the segment joining two distinct vectors and is, by definition, the set of all vectors of the form , where . A subset of a real vector space is convex if, for each pair of vectors that it contains, it contains all the vectors of the segment joining them. The unit ball is the set of all vectors with .
If is a point of the segment joining two distinct vectors and , and if , then is called an interior point of that segment. If a point of a convex set does not belong to the interior of any segment in the set, then it is called an extreme point of the set. A closed convex set in a Hilbert space is called strictly convex if all its boundary points are extreme points.
Construct, for every infinite-dimensional Hilbert space, a simple continuous curve with the property that every two non-overlapping chords of it are orthogonal.
A continuous curve in a Hilbert space is a continuous function from the closed unit interval into ; the curve is simple if the function is one-to-one. The chord of the curve determined by the parameter interval is the vector . Two chords, determined by the intervals and , are non-overlapping if the intervals and have at most an end-point in common.
Any two normalized crinkled arcs are unitarily equivalent to reparametrizations of one another.
A crinkled arc is a simple continuous curve with every two non-overlapping chords orthogonal (see Problem 5). There are three trivial senses in which crinkled arcs are not unique. (1) Translation: fix a vector and replace the arc by . Remedy: normalize so that . (2) Scale: fix a positive number and replace the arc by . Remedy: normalize so that . (3) Span: fix a Hilbert space and replace (the range space of ) by . Remedy: normalize so that the span of the range of is . A crinkled arc is called normalized in case all three of these normalizations have been applied to it.
There are two other useful ways in which one crinkled arc can be changed into another. One is reparametrization: fix an increasing homeomorphism of onto itself and replace by . The other is unitary equivalence: fix a unitary operator on and replace by .
Does there exist a Hilbert space whose linear dimension is ?
There exists a total set in a Hilbert space that continues to be total when any one element is omitted but ceases to be total when any two elements are omitted.
A subset of a Hilbert space is total if its span is the entire space.
If a set remains total after the omission of each finite subset, then it has at least one infinite subset whose omission leaves it total also.
A subset of a Hilbert space is total if its span is the entire space.
If , and if
determine the span of the set of all 's in . Generalize (to other collections of vectors), and specialize (to finite-dimensional spaces).
The Hilbert space consists, by definition, of all infinite sequences of complex numbers such that . The vector operations are coordinatewise and the inner product is defined by
Does there exist an infinite total set such that every infinite subset of it is total?
A subset of a Hilbert space is total if its span is the entire space.
If is an orthonormal basis for a Hilbert space , and if is an orthonormal set in such that
then the vectors span (and hence form an orthonormal basis for ).