Another matrix rank proof

Tags: matrix, proof, rank
 HW Helper P: 3,225 Let A and B me matrices such that the product AB is defined. One has to proove that r(AB) <= r(A) and r(AB) <= r(B). My first thoughts are: let A be 'mxn' and B be 'nxp', so AB is 'mxp'. Further on, we know that r(A) <= min{m, n}, r(B) <= min{n, p} and r(AB) <= min{m, p}. I'm stuck here, although I tried to work something out with the inequalities, but without success. Any hints would be appreciated.
 P: 3,174 you should look on the vector space of the homogenous solutions of Bx=0. and use the fact that dim of this space equals n-rankB, also you should look on ABx=0. from this you can easily prove that rank(AB)<=rank(B). now i forgot how to prove the second inequality.
HW Helper
P: 3,225
 Quote by loop quantum gravity you should look on the vector space of the homogenous solutions of Bx=0. and use the fact that dim of this space equals n-rankB, also you should look on ABx=0. from this you can easily prove that rank(AB)<=rank(B). now i forgot how to prove the second inequality.
Hm, I can't exactly say I follow.

The only way that r(AB) <= r(A) would follow from this is that d2 >= d1, where d2 is the dimension of the vector space of solutions to ABx=0, and d1 the dimension of the vector space of solutions to Ax=0. But why would I conclude that d2>=d1?

P: 3,174

Another matrix rank proof

ok, you need to show that U={x|Bx=0} is a subset of V={x|ABx=0}, then
dim U<=dim V
and dim U=n-rankB and dimV=n-rankAB, n is the number of columns of the matrix, the number of columns of AB is the same as of B.
the same way you can prove that rankAB<=rankA, with the use of rankAB=rank(AB)^t=rank(B^tA^t).
and rankA=rankA^t.
 P: 3,174 if you ask why dimU<=dimV, then it follow from a thoerem that if U is subspace of V then (and V is finite) the inequality holds.
 HW Helper P: 3,225 Ok, I got a grip on it now. Thanks!
 Sci Advisor HW Helper P: 9,398 Woah, boys, you're making this way way to complicated. What is the rank? Just the dimension of the image. This is cleary a submultiplicative function (take some linearly independent vectors, applying a linear map can only make it more dependent).
HW Helper
P: 3,225
 Quote by matt grime Woah, boys, you're making this way way to complicated. What is the rank? Just the dimension of the image. This is cleary a submultiplicative function (take some linearly independent vectors, applying a linear map can only make it more dependent).
Ok, the definition of the rank of a mxn matrix A is r(A) = dim [{C1, ..., Cn}], where C1, ..., Cn are the columns of the matrix. What exactly does it mean 'to be a submultiplicative function' and how exactly do I apply this to the problem?
 Sci Advisor HW Helper P: 2,020 The image of AB is a subet of the image of A. So r(AB) <= r(A). The second inequality follows from the first, as r(AB) = r((AB)^t) = r(B^t A^t) <= r(B^t) = r(B)
HW Helper
P: 3,225
 Quote by morphism The image of AB is a subet of the image of A.
How is the image of a matrix defined?
 Sci Advisor HW Helper P: 2,020 The set { Av in R^m : v is in R^n }, if A is mxn, which is the same as the image of the linear transformation T : R^n -> R^m given by Tv = Av. So a matrix can simply be thought of as a linear transformation.