Dr. Marques Sophie Office 519 Linear algebra Spring Semester 2016 marques@cims.nyu.edu Problem Set # 8 The following equations are considered over the reals numbers. All the answer should be justified unless mentioned differently. Problem 1 : Find a basis for the set of vector in R3 on the plane x ´ 3y ` 2z “ 0. Solution : Note that the plane¨in the ˛ question that we denote P is defined by x “ x 3y ´ 2z. So that any point x “ ˝ y ‚ on this plane is of the form z ˛ ¨ ˛ ¨ ˛ ¨ ˛ ¨ ´2 3 3y ´ 2z x ‚ “ y ˝ 1 ‚` z ˝ 0 ‚ y x “ ˝ y ‚“ ˝ 1 0 z z So that ˛ ˛ ¨ ´2 3 P “ Spant˝ 1 ‚, ˝ 0 ‚u 1 0 ¨ ˛ ˛ ¨ ´2 3 t˝ 1 ‚, ˝ 0 ‚u is a linearly independent set since the vectors are not multiples. So 1 0 ˛ ¨ ˛ ¨ ´2 3 that t˝ 1 ‚, ˝ 0 ‚u is not a basis. 1 0 ¨ Problem 2 : Find bases for NulpAq and ColpAq where ¨ ˛ ´2 4 ´2 ´4 A “ ˝ 2 ´6 ´3 1 ‚ ´3 8 2 ´3 Solution : We row reduce A ¨ ˛ 1 0 6 5 A „ Row reduce „ ˝ 0 2 5 3 ‚ 0 0 0 0 We know that a basis for column space is formed by the set of the pivot columns of A. We see via the echelon form of A, B that the pivot columns of A are the column 1 1 ˛ ˛ ¨ 4 ´2 and 2. Thus, t˝ 2 ‚, ˝ ´6 ‚u is a basis for ColpAq. NulpAq “ tx P R4 |Ax “ 0u. ´8 ´3 Since¨A and ˛ B are row equivalent, we know that NulpAq “ NulpBq. From Bx “ 0 where x1 ˚ x2 ‹ ‹ x“˚ ˝ x3 ‚ we get the system x4 " x1 ` 6x3 ` 5x4 “ 0 2x2 ` 5x3 ` 3x4 “ 0 ¨ which is equivalent to " ¨ x1 “ ´6x3 ´ 5x4 x2 “ ´5{2x3 ´ 3{2x4 ˛ x1 ˚ x2 ‹ ‹ So that any element x “ ˚ ˝ x3 ‚ P NulpAq is of the x4 ˛ ¨ ¨ ´6 ´6x3 ´ 5x4 ˚ ´5{2 ˚ ´5{2x3 ´ 3{2x4 ‹ ‹ “ x3 ˚ ˚ ‚ ˝ 1 ˝ x3 0 x4 where x3 and x4 can be any scalar. form ˛ ´5 ‹ ‹ ˚ ‹ ` x4 ˚ ´3{2 ‹ ‚ ˝ 0 ‚ 1 ˛ ¨ ˛ ¨ ´5 ´6 ˚ ´5{2 ‹ ˚ ´3{2 ‹ ˚ So that we know that NulpAq “ Spant˚ ˝ 1 ‚, ˝ 0 1 0 dimpNulpAqq ¨ “ 2 number ˛ ¨ of free ˛ variable. ´6 ´5 ˚ ´5{2 ‹ ˚ ´3{2 ‹ ‹ ˚ ‹ So that t˚ ˝ 1 ‚, ˝ 0 ‚u is a basis for NulpAq. 0 1 ¨ ˛ ‹ ‹u and we also know that ‚ Problem 3 : In the vector space of all real valued functions, find a basis for the subspace spanned by tsinptq, sinp2tq, sinptqcosptqu. Solution : Note that sinp2tq “ 2sinptqcosptq, by the spannin set theorem we know that Spantsinptq, sinp2tq, sinptqcosptqu “ Spantsinptq, sinp2tqu Note also that sinptq and sinp2tq are not multiples. So tsinptq, sinp2tqu is also a linearly independent set. As a consequence, by the definition of a basis, tsinptq, sinp2tqu is a 2 basis for Spantsinptq, sinp2tq, sinptqcosptqu. Problem 4 : Is the set of polynomial tp1 , p2 u where p1 ptq “ 1 ` t2 and p2 ptq “ 1 ´ t2 linearly independent or not ? Solution : There is no non zero c scalar such that cp1 ptq “ p2 ptq or p2 ptq “ cp1 ptq if so then one would have that 1 “ ´1 which is impossible. So they are not multiples. So, the set tp1 , p2 u is linearly independent. Problem 5 : Given vectors u1 , ¨ ¨ ¨ , up and w in V, show that w is a linear combination of u1 , ¨ ¨ ¨ , up if and only if rwsB is a linear combination of the coordinate vectors ru1 sB , ¨ ¨ ¨ , rup sB . Solution : Let u1 , ¨ ¨ ¨ , up and w in V. Suppose that w is a linear combination of u2 , ¨ ¨ ¨ , up . That is there is scalar c1 , ¨ ¨ ¨ , cp , such that w “ c1 u1 ` ¨ ¨ ¨ ` cp up . Thus rwsB “ rc1 u1 ` ¨ ¨ ¨ ` cp up sB , since the coordinate map is linear this implies that rwsB “ c1 ru1 sB ` ¨ ¨ ¨ ` cp rup sB Thus, rwsB is a linear combination of the coordinate vectors ru1 sB , ¨ ¨ ¨ , rup sB . For the converse suppose rwsB is a linear combination of the coordinate vectors ru1 sB , ¨ ¨ ¨ , rup sB . That is there is scalar c1 , ¨ ¨ ¨ , cp , such that rwsB “ c1 ru1 sB ` ¨ ¨ ¨ ` cp rup sB since the coordinate map is linear this implies that rwsB “ rc1 u1 ` ¨ ¨ ¨ ` cp up sB . The coordinate map being one to one implies that then. w “ c1 u1 ` ¨ ¨ ¨ ` cp up Problem 6 : Let p1 ptq “ 1 ` t2 , p2 ptq “ t ´ 3t2 and p3 ptq “ 1 ` t ´ 3t2 . 1. Use coordinate vectors to show that these polynomials form a basis for P2 . 2. Consider the basis B “ tp1 , p2 , p3 u for P2 . Find q in P2 , given that ¨ ˛ ´1 rqsB “ ˝ 1 ‚ 2 Solution : 1. We consider the standard basis for P2 which is S “ t1, t, t2 u. The coordinates vectors of these polynomial for the standard basis are then : ¨ ˛ ¨ ˛ ¨ ˛ 1 0 1 rp1 sS “ ˝ 0 ‚, rp2 sS “ ˝ 1 ‚, rp3 sS “ ˝ 1 ‚ 1 ´3 ´3 3 In order to prove that that tp1 , p2 , p3 u form a basis one can prove that the matrix A “ rrp1 sS , rp2 sS , rp3 sS s whose column are the coordinate vectors of the polynomials is invertible. For this, one can row reduce the matrix A and see if there is a pivot in each row. ˛ ¨ 1 0 1 A „ Row reduce „ ˝ 0 1 1 ‚ 0 0 ´1 Since we have a pivot in each row by the invertible matrix theorem we know that A is invertible, , thus the column of A form a basis of R3 . Since the coordinate map is an isomorphism form R3 to P2 , the corresponding polynomials tp1 , p2 , p3 u form a basis for P2 . 2. By definition if ¨ ˛ ´1 rqsB “ ˝ 1 ‚ 2 Then one has q “ ´p1 ` p2 ` 2p3 . Thus qptq “ 1 ` 3t ´ 10t2 . Problem 7 : The first four Laguerre polynomials are 1, 1 ´ t, 2 ´ 4t ` t2 and 6 ´ 18t ` 9t2 ´ t3 . Show that these polynomials form a basis of P3 . Solution : Let p1 ptq “ 1, p2 ptq “ 1 ´ t, p3 ptq “ 2 ´ 4t ` t2 , p4 ptq “ 6 ´ 18t ` 9t2 ´ t3 . We consider the standard basis for P3 which is S “ t1, t, t2 , t3 u. The coordinates vectors of these polynomial for the standard basis are then : ˛ ˛ ¨ ˛ ¨ ¨ ¨ ˛ 6 2 1 1 ‹ ‹ ˚ ‹ ˚ ˚ ˚ 0 ‹ ‹ , rp2 sS “ ˚ ´1 ‹ , rp3 sS “ ˚ ´4 ‹ , rp4 sS “ ˚ ´18 ‹ , rp1 sS “ ˚ ˝ 9 ‚ ˝ 1 ‚ ˝ 0 ‚ ˝ 0 ‚ ´1 0 0 0 In order to prove that that tp1 , p2 , p3 , p4 u form a basis one can prove that the matrix A “ rrp1 sS , rp2 sS , rp3 sS , rp4 sS s whose column are the coordinate vectors of the polynomial is invertible. But ¨ ˛ 1 1 2 6 ˚ 0 ´1 ´4 ´18 ‹ ‹ A“˚ ˝ 0 0 1 9 ‚ 0 0 0 ´1 The matrix has a pivots position in each row thus by the invertible matrix theorem we know that A is invertible, thus the column of A form a basis of R4 and since the coordinate map is an isomorphism from P3 onto R4 , thus tp1 , ¨ ¨ ¨ , p4 u form a basis for P3 . Problem 8 : Let H be an n-dimensional subspace of a n-dimensional vector space V. Show that H “ V. Solution : Let H be an n-dimensional subspace of a n-dimensional vector space V. 4 If dimpVq “ dimpHq “ 0 then H “ V “ t0u and the statement is true. Now suppose dimpVq “ dimpHq “ n ą 0. By definition of the dimension, we know that H contains a basis S consisting of n vectors, but this vectors being also vector of V since V is a subspace of H, we have then n linearly independent vectors on a space V of dimension n, thus the basis theorem applied to V tells us that S is also a basis for V. Thus H “ V “ SpanpSq. Problem 9 : Let V and W be vector spaces and T : V Ñ W be a linear transformation. Let H be a subspace of V and let TpHq be the set off images of vectors in H. Then TpHq is a subspace of W. Prove that dimpTpHqq ď dimpHq. Solution : Since H is a nonzero subspace of a finite dimensional vector space V, H is finite-dimensional and has a basis. Let tu1 , ¨ ¨ ¨ , up u be a basis for H. We show that the set tTpu1 q, ¨ ¨ ¨ , Tpup qu span TpHq. Let y be in TpHq. Then there is a vector x in H with Tpxq “ y. Since x is in H and tu1 , ¨ ¨ ¨ , up u is a basis for H, one can write x “ c1 u1 ` ¨ ¨ ¨ ` cp up for some scalars c1 , ¨ ¨ ¨ , cp . Since the transformation T is linear, one has y “ Tpxq “ Tpc1 u1 ` ¨ ¨ ¨ ` cp up q “ c1 Tpu1 q ` ¨ ¨ ¨ ` cp Tpup q Thus y is a linear combination of Tpu1 q, ¨ ¨ ¨ , Tpup q and tTpu1 q, ¨ ¨ ¨ , Tpup qu spans TpHq. By the Spanning set theorem, this set contains a basis for TpHq. This basis then has not more than p vectors and dimpTpHqq ď p “ dimpHq. Problem 10 : Let V and W be vector spaces and T : V Ñ W be a linear transformation. Let H be a nonzero subspace of V, and suppose T is a one-to-one (linear) mapping of V into W. Prove that dimpTpHqq “ dimpHq. If T happens to be one-to-one mapping of V onto W, then dimpVq “ dimpWq. Isomorphic finite-dimensional vector spaces have the same dimension. Solution Since H is a nonzero subspace of a finite dimensional vector space V, H is finite-dimensional and has a basis. Let tu1 , ¨ ¨ ¨ , up u be a basis for H, we have shown in the previous exercise that tTpu1 q, ¨ ¨ ¨ , Tpup qu span TpHq. So we just need to show that tTpu1 q, ¨ ¨ ¨ , Tpup qu are linearly independent. For this let c1 , ¨ ¨ ¨ cp be scalar such that c1 Tpu1 q ` ¨ ¨ ¨ ` cp Tpup q “ 0 Since T is linear, then the previous equality is equivalent to Tpc1 u1 ` ¨ ¨ ¨ ` cp up q “ 0 But since T is supposed to be one-to-one this implies that c1 u1 ` ¨ ¨ ¨ ` cp up “ 0 But since tu1 , ¨ ¨ ¨ , up u form a basis for H by assumption there are in particular linearly independent and the previous equality implies c1 “ ¨ ¨ ¨ “ cp “ 0 5 Thus tTpu1 q, ¨ ¨ ¨ , Tpup qu is linearly independent and we knew it span TpHq thus form a basis of TpHq. As a consequence, dimpTpHqq “ dimpHq “ p 6