MA 102 (Mathematics 11) UIT Guwahati Tutorial Sheet No. 3 Linear Algebra April 3, 2023 1. True or False? Give justifications, (2) Let A be an mx n matrix. Then there exist b and b’ such that Ax =b has a unique solution but Ax = b’ has infinitely many solutions (b) Let x and y be nonzero vectors in ” such that x”y. . Then x and y are linearly independent. (©) Let 1,52 and Ss be distinct subsets of R* such that span(S U S2) = span(S; U Ss). Then span(S2) = span(S). (@) The column spaces of A and rref(A) are equal Solution: a) False. Let null(A) = {z|Ax = 0}. If Ax = b is consistent then $= {2|Ax = b} = null(A) +y for some y € S. b) True. Since x,y are nonzero vectors, if ‘they are linearly dependent then x scalar a which leads to contradiction because a = y™x/|ly|? = 0. ¢) False. For example consider n = 2, S; = {{1,0]"}, Se = {[1,1]7} and Ss = {(1, 2)". 4) False. Consider A= [ 1g |: Then area) = lt ‘ay for a nonzero 00 3) [-1 4] [2 2. Check whether the ser $=] _° : alla linearly independent, . whether the set S= 41 3 |.} 9}. 3 ].| 1 | } istinearly independent. 1 2 oJ} lo 3. Let $ be a subspace of R* and x,y € S be linearly independent. (a) Show that if we R*\ 5 then {x,y u} is linearly independent, (©) Hu,v €R*\ 8 are linearly independent then does it imply that (1, ¥,4,V} is linearly indepen- dent? Solution: a) If there exist c1,c2,¢5 not all zeros, such that €1x-+ cay + eyu = otherwise it will contradict that x,y € $ are Ll, But then u contradict ue R*\ S. ©, then ey + 0, (CQ)e + (-B)y which will b) No. For example take x = % 1 1 i 1 | 28d S = span{x, y}, 04, Let A € R"™", Show that row(A™A) = row(A), that is, ATA and A are row equivalent. yTy = 0 which implies Solution: Ax = 0 = ATAx = 0. Also ATAx = 0 = x"ATA3 y= Ax =0. ee Um a ©) am om tod ocban 8 =|] tn A = Od Bx = 0 are equivalent and by the above argument Bx =0 and A” Ax = 0 are equivalent. Hence B and ATA are row equivalent, which will imply, row(A) = row(4"A), Mc da (na) eee Avoca | ATA 0] ‘Then Cx =0 and A?Ax = 0 are equivalent, and Cx = 0 and Ax =0 are equivalent. Hence A and Care row equivalent and row(A) = row(ATA), 5. Let A € B® be such that rref(A) has the 1*, 3" and the 5 column as the only pivot columns. (a) Find two linearly independent solutions of Ax = 0. (b) Show that the columns ay, ag and as ( the 1*, 3° and the 5" column of A) form a basis of the column space of A. (©) Can the sets {ay,a2), {a1,a3,a4} and {a3,a4,a5) be linearly independent? -in ~au 1 ° Solution: ») u=] 0 | and v= | —aa |, where dy i the (i,j)th entry of rref(A). ° 1 ° 0 ») Proved in the class. By inspection one can check that. = ayaa) and ay = duds + dasa. Hence aa = Gyzay and ay = dea; + day and the result follows. ©) The sets {21,29}, {a;,a3,a4} will not be linearly independent follows from part (b). The set {@s,a4,a5) will be linearly independent if dia # 0 6. True or False? Give justifications (©) Hy) and {u,v} are two diffrent linearly independent subsets of R?, then {x, u} and {y,v} are also linearly independent sets, or {(3) [2] esen an ((3] [2 }enon (©) IFS is a subspace of R" then x-+ $ is a subspace if and only if x € S (@) W the diagonal entries of a 4x4 upper triangular matrix A are 1,2,3 and 4 then $= (x € RE ‘Ax = 2x) is subspace of R! but $2 = (x € RY | Ax = 5x) is not c= 5x) isnot. gy 4 ay fa 9 Tt S= 9] 5]./2].1 3 |. Find the valves ofa for which span(s) 4 Rd Page 2Ifa diagonal entry of a 3x3 upper triangular mats dependent, is zer0, then show that the columns are linearly Determine whether b € Col(A) and w € Row(A) where -fio- 3 aa [22 ]ove ff ow weta tl 11-3 i A=]0 2 1],b=|1) ond w=[2 4 -5] 1-1-4 0. Also determine the rank and nullity of A. 10. Determine bases of the column space, row space and null space of A where 11 0 o1-1 o1-1 LL. True or False? Give justifications A (a) If Sis a subspace of R of dimension n, then $ = BR”. (b) For any two matrices A and B for which AB is defined, rank(AB) < min(rank(A),raak(B)) (©) If =[4 Bi], then rank(C) < rank(A) + rank(B) ca) 00= [4 B Jen kr 2a +k) Solution: (a) True. Ify €R" but not in 5, then for any basis B of S, BU {y) is LI (0) True, row(AB) ¢ row(B) and col(AB) ¢ col(A) (¢) True, Let rank(A) = k, rank(B) =r and let the columns ai,,...,a4, of A form a basis of col(A) and the columns bj,,... bj, of B form a basis of col(B). Then 0,5... @iysbjus-+sDje spans col[A B). Hence rank|A B) r +k 1@iy of A forms a basis of col(A) then 12, Lot A € R™", Ifrank(A) = rank(A®) then show that rank(A2) = rank(A®), Is rank( AS) = rank( 4%)? Page 3Solution: Note that col(A®) C col(A). rank(A?) = rank(A) implies col(A?) = col(A).. Again note that col(A%) € col(A2). If y € col(A2), then y = A?z = A(Az) for some z © B" Since col(A2) = col(A), Az = An for some u € R", Hence y = A(A®u) for some u € B®, that is 1 € col(A®). Hence col(A%) = col(?). Dy similar argument one can show that rank(A) = rank(A*1) for all k € N, (One could have argued similarly by considering the row space.) 13, Show that for any two m x n matrices A and B, rank(A + B) < rank(A) + rank(B). Solution: Since A+ B = [AB] [ a } rank(A + B) < rank|A B] < rank(A) + rank(B). see pndMA 102 (Mathematics IT) IIT Guwahati ‘Tutorial Sheet No, 2 Linear Algebra March 27, 2023 WLU = [ur = up] ER™™ and Vim [vy + vq] ER™". Let — 0| gen bbe a diagonal matrix, where o1,...,¢p are nonzero real numbers. Show that A = USVT can be expressed as an outer product expansion of the form As ony} +t aptgy]. 2 tetas [431 42] bea block upper triangular matix where cach block is an nx mate IFA c Ag 0 td An ate invertible, then show that i invetble and A = [ for some matrix C Determine C. Au Aud ae ie An nxn matrix. Show that A can be factored as a product of block matrices a-[5 ie 4] where B= Ag Ajj! and $:= Az — An Az! Aya. The matrix $ is called the Schur complement of Ay, in A, Show that A is invertible <=> $ is invertible, | E RO*+")x(™49) he such that Ay, is an m x m invertible matrix and Az is an 4, Let A be 4 x 3 matrix such that rank(A) = 3. Then show that there exi that BA= Is is a 3x 4 matrix B such Solution: The rref of A is of the form [/s,0]". Heneo there exists a invertible P such that PA=(Is,0)". Take B = [Js,0)P, then BA = Iy, 5. Let A € RS be invertible with row sums 1. Show that the sum of all the elements of A~ is 5, hhas row sums 1 and the result follows, Solution: Let e = (1,1,1,1,1]”. Then Ae = (1,1,1,1,17 =e, which gives A~le =e, ie, A ‘| 6. True oF False? Give justifications, (@) I for all Ae R™", AB = A th(0) WA ond B aro n x matsices such that AB = Jy then A and B are invertible and BA = In implies there exists an invertible P such hat PAB = P, where PA is in rref. (6) 1.4 is am im xm matrix with at least one nonzero row then A is row equivalent to a matrix 37 with all nonzero rows. (@) Hall the columns of an m xn nonzero matrix A are equal then rank(A) = 1 (©) 1A is an m x m matzix with a zero column then the rref of A will again have a zero column. {f) If Pis any invertible matrix such that PA is defined then Ax = b and PAx = Pb are equivalent. Solutio (a) True, take A = Jy. (b) True. Observation: If P is invertible then rank(P) = n. AB = I implies there exists an invertible P such that PAB = P, where PA is in rref, Since P's invertible, PAB cannot have a zero row, hence PA cannot have a zero row. So PA = In or A= Pl and B= P. AB = implies B(AB)B~ BA(BB™) = BA (©) True. If the rref of A has a zero row, say Gj, then replace a; with a + 4s, where &; is some nonzero row of the rref. (@) Tie. (Bach row of A is a multiple of some nonzero row of A.) (¢) Te. (0) Te. 10 7. Let AE RS be such that rref(A) = }0 1 —5 0 —3]. If the first, second and fourth columns oo atmo [3] [5] 1 2] then determine the matrix A. 0 1-3040 5 oo 130 2 reA= lq 9 001-1 0 0 0000 1 “1 3 2 1 8 We Ae) =|] Aes =| 4 | and Aeg = |" then determine A 3 -4 5 9. Give example of three 3x 4 matrices in rref with different number of pivots. Deternii number of different types of 3 x 4 matrices in reef © the total Page 2Solution: Answer of only second part, rst ae exercnen Clon soopnoeagpoo- 3 leading columns: [ 910} foo 1 of Jor of fora | (3) in nuraber Ooo i lo 001) loo 0 1 fo o1 001 oO} forse o Pleading columns: 10.0 0 1] fo 0 0 1] and soon: (2) in number coo loo | 1 leading column: (!) in number. O leading column: (4) in number. Total number (4) + (8) + (1) + (8) = 15, 10. By using Gauss Jordan elimination find the inverse of the matrix 2 4 67 10 12 11, Using LU factorization of the matrix A solve the system of linear system Ax = b, where the aug mented matrix [Alb] as given below: 1 12. Express the intertible matrix A 1 8s : [: i 13. If A € R™" be upper triangular and B is any matrix such that AB = I, then show that each diagonal entry of A is nonzero. ] ot enya : Solution: Suppose J = ref(A) = PA, where P is invertible. Now, if A has at least one vero diagonal entry, consider the least i such that ay ~ 0, then the corenponding colunm af fs nee 8 pivot column. Thus, R has less than m pivot eons, and so has a 2t0 tow, Conscoueret AB = PAB = PI = P has @ zero row, which is not possible because P is invertible, | =n (f A) 15. Show that $= {{ rt ] ER | x = 225 a is a subspace of RS, as Page 3(a) Find {u,v} such that span(u,v} = S. (b) Find av! such that span{u, v'} = span{v, v"} = 5. (©) Find an u’ such that span{u’,v'} is not a subspace of S. Geometrically what will be the picture of $ and span{u',’} ? 2 1 2 Solution: a) Since $ = {a| 0 | +8 1 |a,8 € R}, one choice can be u = [: and i o 1 1 1 0 b)Take any v’ € § but not in span{u} or span{v}. For example take v! = u+v. ©) Take u’ not in S, then span{u’,’} will correspond to a plane in R° and will intersect the plane associated with S in a line given by span{v'} ree End 4 Page 4MA 102 (Mathematics II) IIT Guwahati ‘Tutorial Sheet No.1 Linear Algebra March 20, 2023 1. Let wand v be vectors in R". Prove or disprove the following statements. (a) The equality (u, v) = }(llu-+ vI? = [lu — vl?) holds, (b) The equality [ju + vil = [lu — vl] holds if and only if u and v are orthogonal (6) There exist u and v such that j[ul| = 1, lvl = 2 and (u, v) =3. 2, Let u and v be vectors in R". Show that [lu + vil? + [lu ~ vl/? = 2ul/? + 2liv|?, What does this soy about parallelogram in R2? Further, show that |(u, v)| = [[ullllv if and only if u = av for some scalar a. Furthermore, show that u+v and w—v are orthogonal <=> [lull = [lvl Solution: First partis easy. For the second part, suppose that ||vil #0. Define a == (u, v)/ilvil? and w = u—av, Then v 1 w and u=w-+tav => |jull? = lal]? + [w||?. Since Jal = |[ull/Ilvl. it follows that w = 0. Hainan cm onl ae [th eeL3) ath -atb sotton (Pa [2] = = one = [ ‘tne w= —2u-+4v. Similarly, (b) w = 2u+3v. ] iia ees Check whether the following statements are True or False? Give proper justifications (@) IA is the matrix obtained from A by replacing the ith colamn a of A by 2a; then the systems fx = 0 and Ax = 0 are equivalent. (b) If the rref of a 5 x 5 matrix A has the third column as [1,2,0,0,0]" then [1,~2.1,0,0}" is a solution of Ax = 0. (c) For ann xn matrix A, the systems Ax = 0 and ATx = 0 are equivalent Sotto: 0 Fe tae = [ True. ‘The first two columns are pivot columns and the third of the rref of A are [1,0,1, #, 4), (0,1, 2,#,*],[0,0,0, », Pry fon ‘False. Consider A = [ i | or [ co | | then (1,-1]? isa solution of Ax = 0 but not of ix = 0 snot, Hence the first three rows[ay] is the sum of its diagonal entries and is denoted by tx(A), be. 5. The tee of an mx 1 matrix tr(A) San tot a Prove the following: if A and B are n x n matrices and ot is scalar, then 1, tr(A +B) = tr(A) + (B); 2 tr(ad) = ate( A); 3. tr(AB) = tr(BA). Solution: Easy. 6. Suppose that x and y are two distinct solutions ofthe system Ax = b. Prove that there are infinitely many solutions to this system. Interpret your findings geometrically. AER, + (1— Aly is also a solution. This means that for the Solution: Show that for each hh the points x and y is in the set of case when A is a 3 x 3 matrix the entire line passing throug] solutions o(ta(ra} off tip fes a] oft a] 8 a] Solution: (a) No, first one has rank one and the other two. 10-3 0.1.5 |. So they are row- 0 (b) Both are row equivalent to the row reduced echelon form 00 equivalent, () No, two matrices must have the same size, in order to be row equivalent 8, Find all the solutions ofthe linear system with the augmented matrix [A | b] as given below: 12a 4|2 5 6 7 8I5 9 10 11 12|8 (a) Find 6 such that Ax = doesnot have a solution. (b) By changing exactly one entry ofA, find an A such that Ax = b will be consistent for all b € B®, #).0/ > Solution; Solution set= 9] § | +a) {’] +8] > [laser 0 0 4 a) Since Ry = 2R2 ~ Ry, where Ri is the ith row of A, take bt such that 0 # 2b ~ bf ) Since Ry = 2R2 ~ Ry and rank(A) = 2, change any one entry in the last row of A then rank(A) = 3. Page 29. Determine the reduced row echelon form and the rank of the following matrices 1104 24624 110 2202 12311 of? 33] lr i101] ©l24800 aaa 1101 a 61726 9 12040 ; 001-10 Solution: (ejrref(a)=] 9 8 2 to 00000 pe Find the coefficients a, b, c,d so that the graph of y = ax’ +z? + cr+d passes through (1,2), (—1,6), (2,3) and (0, 1) Solution: Find the values solving a+b-+c = 1, -a+b—e = 5,4a+2b+c= 1 applying Gaussian elimination. 1 31 - O} and] 3 5 1 | are row equivalent. 1 rey 0 aa Show that matrices [ weet End 4