
Linear Algebra Linear Algebra Math Forum 
 LinkBack  Thread Tools  Display Modes 
July 23rd, 2017, 03:04 AM  #1 
Newbie Joined: May 2017 From: Germany Posts: 5 Thanks: 0  Need help with three (small) problems)
hello, I've been preparing for an upcoming exam for a firstsemester linear algebra lecture this week, and for that I've reworked through all of the old assignments. There are a total of 3 tasks I've been struggling with. It's probably all relatively simple stuff. 1. Two matrices A, B are called similar if there exist invertible matrices S, T with B = SAT. Now I have to proof that a) this is an equivalence relation (that's no problem), and that b) two matrices are similar iff they have the same rank; also there is a distinct representant C for each class, which on the diagonal has rank(C) many 1's and otherwise only 0 entries. I totally get why this is true, but I'm not sure how you're supposed to proof this formally. Also, perhaps more relevantly, it seems to me that you can bring any matrix in such a form using elementary elimination or permutation line operations (add a* one line to another line or swap two lines (but do you have to proof this or can you just state that?)) and that each operation is equivalent to the multiplication with an elimination/permutation matrix. Each of those matrices is invertible therefore their product is invertible ... but then why do you need both S and T? Wouldn't S alone do the job? 2. Let U = <(1, 2, 1)^T, (2,1,2)^T> and let Û = <(1, 0, 1)^T> let p be the projection from V = R^3 to U along Û, so im(p) = U, ker(p) = Û. Proof that for all x ∈ V, p(x) = A(B^1)x where $\text{A} = \begin{pmatrix} 1 & 2 & 0\\ 2 & 1 & 0\\ 1 & 2 & 0 \end{pmatrix} \\ \text{B} = \begin{pmatrix} 1 & 2 & 1\\ 2 & 1 & 0\\ 1 & 2 & 1 \end{pmatrix}$ This one is confusing me a bit. It's possible that the problem isn't stated properly. Wouldn't the matrix representation of p depend on the basis of V? The assignment does not state that the basis of V consists of the three vectors which describe U and Û. But if it did, wouldn't all x ∈ U just have the form (a, b, 0)? Though, even then I can't quite proof that A(B^1)x = x... I know that Ax = Bx would hold and so x = (B^1)Ax but that's not sufficient. I think the proof works without really using the specic numbers. 3. (This one in particular ought to be easy because it's from a very early assignment) Let A be an upper triangular matrix with diagonal entries ≠ 0. Prove by induction that for each vector b there exists exactly one vector x so that Ax = b (so basically a system of equations which is already in a nice form) and that xn = and for k < n, xk = I don't see how induction would work in principle over multiple systems and I can't get it to work over entires of one system. I remember that I already failed on this one when I first had to do it. Okay, those are the three. Any solutions are appreciated! Last edited by skipjack; July 23rd, 2017 at 03:16 PM. 
July 23rd, 2017, 05:30 AM  #2 
Newbie Joined: May 2017 From: Germany Posts: 5 Thanks: 0 
Edit: since they don't show up properly (anoymore), for 2. that's A = B = 
July 23rd, 2017, 10:20 AM  #3 
Senior Member Joined: Dec 2012 From: Hong Kong Posts: 853 Thanks: 311 Math Focus: Stochastic processes, statistical inference, data mining, computational linguistics 
1. Are you familiar with coordinate systems? This would be much, much easier to get intuitively if you are. The first part of b) should be relatively simple to show  separately show that rank B = rank AT and rank AT = rank SAT. The invertibility of S and T should be a big hint. 3. Induce on n. The basis step is trivial. For the inductive step, partition the matrix so that you 'solve' the system A' x = b' where A' and b' are of size n  1. Then you can show that the last component of x is also unique, and you're done. 
July 23rd, 2017, 11:44 AM  #4  
Newbie Joined: May 2017 From: Germany Posts: 5 Thanks: 0  Quote:
I'll look into the other two and see whether I can apply what you said. Edit for OP: in 2. A = ((1, 2, 0), (2,1,0), (1,2,0)) B = ((1, 2, 1), (2,1,0), (1,2,1)) Edit2: yeah, the similar => same rank implication isn't hard. the way back is the problem. Last edited by sty silver; July 23rd, 2017 at 11:58 AM.  
July 24th, 2017, 10:41 AM  #5 
Senior Member Joined: Dec 2012 From: Hong Kong Posts: 853 Thanks: 311 Math Focus: Stochastic processes, statistical inference, data mining, computational linguistics 
2. $p(x) = \begin{bmatrix} 1 & 2 & 1 \\ 2 & 1 & 0 \\ 1 & 2 & 1 \\ \end{bmatrix}\begin{bmatrix} 1 & 0 & 0 \\ 0 & 1 & 0 \\ 0 & 0 & 0 \\ \end{bmatrix}\begin{bmatrix} 1 & 2 & 1 \\ 2 & 1 & 0\\ 1 & 2 & 1 \end{bmatrix}^{1}x$ I think you can fill in the rest (i.e. multiply the first two matrices). Note that there is no 'the basis of V'; every vector space has an infinite number of bases. 

Tags 
problems, small 
Thread Tools  
Display Modes  

Similar Threads  
Thread  Thread Starter  Forum  Replies  Last Post 
A small question  yashrs  Probability and Statistics  1  September 15th, 2015 09:55 PM 
Small change.  jiasyuen  Calculus  5  October 5th, 2014 07:13 PM 
Small problem  thinkfast  Algebra  9  January 28th, 2012 11:07 PM 
Small inequality  utoetu  Real Analysis  4  January 9th, 2011 06:50 AM 