

Study with the several resources on Docsity
Earn points by helping other students or get them with a premium plan
Prepare for your exams
Study with the several resources on Docsity
Earn points to download
Earn points by helping other students or get them with a premium plan
Linear Algebra, real polynomials, determinants, Pontrjagin duality,Chinese Remainder Theorem for polynomials, isomorphic, Hermitian matrix,Hilbert matrix,Pontrjagin dual,convolution.
Typology: Exercises
1 / 3
This page cannot be seen from the preview
Don't miss anything!


Math 55a: Honors Abstract Algebra Homework Assignment #10 (10 November 2010): Linear Algebra X: Signatures and real polynomials; determinants and distances; Pontrjagin duality^0
Quotes? We don’t need no stinkin’ quotes! —adapted or misquoted from Blazing Saddles (1974), in turn adapted or mis- quoted from The Treasure of the Sierra Madre^1
Using the signature of a pairing to describe the roots of a real polynomial without repeated factors:
∏m j=1 Pj^ for some polynomials^ Pj^ such that there is no polynomial Q of positive degree that divides more than one of the Pj. Let A be the F -algebra F [X]/(P ), and for each j let Aj be the F -algebra F [X]/(Pj ). [Recall that (f ) is the ideal in A[X] generated by f .] Prove that the algebra homomorphism A →
∏m j=1 Aj^ whose^ j-th coordinate is reduction mod^ Pj^ is an isomorphism. Deduce (using a result from the previous problem set) that if the Pj are irreducible then F [X]/(P ) is a product of fields. [Yes, the usual Chinese Remainder Theorem can be phrased in such terms as well, but linear algebra makes the polynomial version simpler.] ii) Now suppose F = R and the Pj are irreducible. Show that Aj is isomorphic with R or C according as deg(Pj ) = 1 or 2. Let 〈·, ·〉 be the pairing on A defined by 〈f, g〉 = tr(f g), that is, the trace of the multiplication-by-f g map on the finite- dimensional real vector space A. Prove that this pairing is nondegenerate, and that its signature is (r, s) where s is the number of j for which Pj is quadratic and r − s is the number of j for which Pj is linear. iii) Suppose that moreover each Pj is linear and none equals X, so that the roots are all real and nonzero (as happens if P is the characteristic polynomial of an invertible Hermitian matrix, though if we have such a matrix we already know how to count the roots of each sign.) Construct a polynomial for which the signature formula of part (ii) yields the counts of positive and negative roots of P.
The formula 〈f, g〉 = tr(f g) yields a pairing on F [X]/(P ) for any F and P. It is non- degenerate if and only if P has no repeated factors; this will be easier to see once we have developed some more field algebra, but for now you can easily check it if P factors completely in F [X] (i.e. has all roots in F ).
Next some classical product formulas for determinants:
V (x 1 + t, x 2 + t,... , xn + t) = V (x 1 , x 2 ,... , xn) T (t)
for all t and xi. Use this to derive inductively the formula
∏n− 1 i=
∏n j=i+1(xj^ −^ xi) for the Vandermonde determinant ∆(x 1 , x 2 ,... , xn) = det V (x 1 , x 2 ,... , xn). What is the determinant of the n × n matrix whose (j, k) entry is
∑n i=1 x
j+k− 2 i? (^0) The “j” in “Pontrjagin” is pronounced as a “y”. (^1) According to Wikipedia’s “Stinking badges” page. Yes, Wikipedia has a page on “Stinking badges”!
det(A) = ∆(x 1 ,... , xn)∆(y 1 ,... , yn)
/ (^) ∏n
i=
∏^ n
j=
(xi + yj )
where ∆ is the Vandermonde determinant of the previous problem. [It follows via Cramer that each entry of A−^1 is a product of linear polynomials in the xi and yj ; in particular this explains the form of the inverse of the Hilbert matrix, which has xi = i and yj = j − 1.] ii) In particular, if F = R, xi = yi > 0 for each i, and the xi are distinct, deduce that the symmetric matrix A is positive definite (without invoking the calculus interpretation of the associated inner product on Rn). iii) Now let ∫ V be the inner product space of continuous functions on (0, 1) with 〈f, g〉 = 1 0 f^ (t)g(t)^ dt, and^ W^ the subspace spanned by the functions^ t
xi (^) for some distinct nonnegative xi ∈ R. Give a formula for the distance from W to the element tx^ of V for any real x ≥ 0.
This is the key to one of the proofs we’ll give next term of M¨untz’s theorem on sequences {xi} such that the span of {txi^ } is dense in the space of continuous functions on [0, 1].
The remaining problems concern Fourier analysis on finite abelian groups. The Pontrjagin dual Ĝ of a finite abelian group G is the set of homomorphisms from G to the multiplicative group C∗. Pointwise multiplication gives Ĝ the structure of an abelian group (that is, the product of ̂g 1 , ̂g 2 ∈ Ĝ is the homomorphism g 7 → ĝ 1 (g)̂g 2 (g), and likewise for the identity and group inverse). While the definition doesn’t say this, any ̂g must be a root of unity, because gn^ = 1 for some integer n > 0,^2 whence (̂g(g))n^ = ̂g(gn) = 1. It follows that |̂g(g)| = 1 for all g ∈ G and ̂g ∈ Ĝ. Elements of Ĝ are also called “characters” of G. We next explore Pontrjagin duality for finite abelian groups and some applications.
Thus if G is the product of groups Z/nj Z then Ĝ ∼= G. In particular #(Ĝ) = #(G). It turns out that every finite abelian group G is of the form
∏r j=1 Z/nj^ Z, but we won’t need this to prove that #( Ĝ ) = #(G) because we will obtain this fact in the course of proving the next few results.
g∈G ̂g(g) = 0. ii) Let CG^ be the complex inner product space of functions G → C with the usual inner product 〈f 1 , f 2 〉 =
g∈G f^1 (g)^ f^2 (g). Prove that distinct characters of^ G, considered as elements of CG, are orthogonal. Deduce that #(Ĝ ) ≤ #(G).