Scribd Logo
Upload

Welcome to Scribd!

  • 36 views

    Extending An Orthonormal Rational Set of Vectors Into An Orthonormal Rational Basis

    Uploaded by

    Epic Win
    1) The document presents a proof that any set of orthonormal rational vectors in a vector space over the rational numbers can be extended to form a rational orthonormal basis. 2) It introduces a lemma showing that a single rational vector can be transformed into the first basis vector of an orthonormal basis using a rational transformation matrix. 3) An induction proof then shows that any set of orthonormal rational vectors can be extended to a full orthonormal rational basis by applying the lemma and induction hypothesis on subspaces.

    Copyright:

    Attribution Non-Commercial (BY-NC)
    Download as PDF, TXT or read online from Scribd

    Extending An Orthonormal Rational Set of Vectors Into An Orthonormal Rational Basis

    Uploaded by

    Epic Win
    36 views4 pages
    1) The document presents a proof that any set of orthonormal rational vectors in a vector space over the rational numbers can be extended to form a rational orthonormal basis. 2) It introduces a lemma showing that a single rational vector can be transformed into the first basis vector of an orthonormal basis using a rational transformation matrix. 3) An induction proof then shows that any set of orthonormal rational vectors can be extended to a full orthonormal rational basis by applying the lemma and induction hypothesis on subspaces.

    Original Title

    Lin

    Copyright

    © Attribution Non-Commercial (BY-NC)

    Available Formats

    PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    1) The document presents a proof that any set of orthonormal rational vectors in a vector space over the rational numbers can be extended to form a rational orthonormal basis. 2) It introduces a lemma showing that a single rational vector can be transformed into the first basis vector of an orthonormal basis using a rational transformation matrix. 3) An induction proof then shows that any set of orthonormal rational vectors can be extended to a full orthonormal rational basis by applying the lemma and induction hypothesis on subspaces.

    Copyright:

    Attribution Non-Commercial (BY-NC)
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    36 views4 pages

    Extending An Orthonormal Rational Set of Vectors Into An Orthonormal Rational Basis

    Uploaded by

    Epic Win
    1) The document presents a proof that any set of orthonormal rational vectors in a vector space over the rational numbers can be extended to form a rational orthonormal basis. 2) It introduces a lemma showing that a single rational vector can be transformed into the first basis vector of an orthonormal basis using a rational transformation matrix. 3) An induction proof then shows that any set of orthonormal rational vectors can be extended to a full orthonormal rational basis by applying the lemma and induction hypothesis on subspaces.

    Copyright:

    Attribution Non-Commercial (BY-NC)
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    of 4

    Extending an orthonormal rational set of vectors into an orthonormal rational basis

    Calvin Lin Zhiwei August 17, 2006


    Abstract This paper was developed as an answer to a question posed at The University of Chicagos Research Experience for Undergraduates (REU) about the behaviour of rational matrices. The results here have direct application to questions about the volumes of lattice hypercubes in 4k + 2 dimensions. The initial results about orthonormal basis in vector spaces over elds of characteristic zero have independent interest.

    Denition. A rational vector is a vector whose coordinates are all rational. The vector is normal if the sum of the squares of its coordinates is equal to one. Two dierent vectors are orthonormal if their standard dot product is zero. If we were working in Rn , then the Gram-Schmidt procedure would extend any orthogonal set of vectors. However, this procedure requires division by the norm of the vector. It is not immediately obvious how this procedure can be adapted to work in Qn . In this paper, all vectors will be treated as column vectors. In particular, ej refers to the j th standard basis vector, 1 in the j th entry and 0 otherwise. Lemma. Given a rational vector a with norm one, there exists a matrix A GLn (Q) such that the rst column of the matrix is a (Ae1 = a) and the columns of A are orthonormal. Proof. If a = e1 , take A = I. Now suppose a = e1 . Let the ith coordinate of this vector be ai . Consider the linear transformation that leaves the 1

    vector e1 + a xed and reverses the direction of all other vectors orthogonal to e1 + a. The geometric interpretation of this linear transformation is a reection through the vector e1 + a. From the reection, Ae1 = a. Moreover, for all i = 1 Aei = 2 ai ei (e1 + a) (e1 + a) ei . (e1 + a) ei = 2 2 ||e1 + a|| 2 + 2a1

    Since these vectors have rational coordinates, the entries of this matrix are rational. Also, this reection preserves orthogonality and distance, hence the columns of A are orthonormal and thus form a basis for our space. Note that A2 = AAT = I. Comment. It can be veried that the following matrix in explicit form satises the conditions in the lemma. a1 a2 a3 ... an a a2 a1 1 a2 a3 a2 an ... 2 2a1 +1 a1 +1 a1 +1 2 a 1 a3 1 a3 a2 a3 an ... A = a3 a1 +1 a1 +1 a1 +1 . . . . .. . . . . . . . . . 2 a 1 an 1 an a3 an a2 ... an a1 +1 a1 +1 a1 +1 Theorem. In the vector space Qn , any set of orthonormal rational vectors can be extended to an orthonormal rational basis. Proof. We will prove this theorem by induction on n. The base case n = 1 is obvious. For all n, if the set contains exactly one vector, the result follows from the previous lemma. Assume that the statement is true for n1 and that our set contains two or more vectors. Take the k orthonormal vectors in Qn labeled v1 , v2 , . . . , vk . Use v1 as vector a in the lemma to construct the matrix A and consider A1 v1 , A1 v2 , . . . , A1 vk . Since A1 v1 = e1 , the other k 1 vectors lie in the subspace spanned by e2 , e3 , . . . , en By our induction hypothesis, in this subspace of n 1 dimensions, the k 1 rational vectors can be extended to an orthonormal rational basis. If we augment the vectors in this basis with 0 as the rst entry and transform them by A, we get an extension of our n dimensional orthonormal rational vectors into an orthonormal rational basis. 2

    Comment. This proof can be extended to any eld of characteristic zero such as R and Qp by replacing the word rational. We will use this theorem to solve a problem posted by Dr. Paul Sally at the start of the REU. a1,1 a1,2 . . . a1,n a2,1 a2,2 . . . a2,n a3,1 a3,2 . . . a3,n Corollary. Given M GLn (Q), M = , such that . . .. . . . . . . . . an,1 an2 . . . ann M T M = diag(1, 1, . . . , 1, x, x). Then, x can be written as the sum of squares of two rational numbers. Proof. Since x is the length of a vector and the determinant of M T M is not zero, we have x > 0. 1 1 Consider L = diag(1, 1, . . . , 1, , ). Since LT M T M L = I and M L x x is in the orthogonal group On (R), we have LM M T LT = I and the sum of the squares of the entries of each row of M L equals 1. Thus, 1 2 1 2 a2 + a2 + . . . + a2 j,1 j,2 j,n2 + aj,n1 + aj,n = 1 x x Given the rst n 2 orthonormal rational column vectors in the matrix M , they can be extended to include two more orthonormal rational vectors b1 and b2 to form an orthonormal basis. Let bj,i refer to the j th coordinate of the vector bi . Since this is an orthonormal basis,
    2 2 a2 + a2 + . . . + a2 j,1 j,2 j,n2 + bj,1 + bj,2 = 1

    Since the norm of b1 is 1, it is non-zero in some coordinate k. Then,


    2 2 a2 a2 k,n1 + ak,n k,n1 + ak,n = 2 1 a2 a2 . . . a2 bk,1 + b2 k,1 k,2 k,n2 k,2

    x= =

    ak,n1 bk,1 + ak,n bk,2 b2 + b2 k,1 k,2

    ak,n1 bk,2 ak,n bk,1 b2 + b2 k,1 k,2

    can be written as the sum of 2 rational squares. 3

    I would like to thank Robert Young, Jonathon Nieder and Nicholas Longo for helpful comments. Information about the volumes of lattice hypercubes in Rn can be found in [1].

    References
    [1] Judith Sally and Paul J. Sally Jr., Developing Mathematics Vertically, To be published.

    You might also like