Suppose v1,,vmv_1,\dots,v_m is a basis of VV and WW is finite-dimensional. Suppose TL(V,W)T \in \mathcal L(V,W). Prove that there exists a basis w1,,wnw_1,\dots,w_n of WW such that all the entries in the first column of M(T)\mathcal M(T) (with respect to the bases v1,,vmv_1,\dots,v_m and w1,,wnw_1,\dots,w_n) are 00 except for possibly a 11 in the first row, first column.

(In this exercise, unlike 3c/3, you are given the basis of VV instead of being able to choose a basis of VV.)


If T(v1)0T(v_1) \ne 0 define w1=T(v1)w_1 = T(v_1) then extend to a basis w1,,wmw_1,\dots,w_m of WW. We have

Tv1=w1=j=1mAj,1wjTv_1 = w_1 = \sum_{j=1}^m A_{j,1}w_j

Clearly w1=1w1+0w2++0wmw_1 = 1w_1 + 0w_2 + \dots + 0w_m, and since w1,,wmw_1,\dots,w_m are independent this is the only linear combination that gives w1w_1 meaning Aj,1=1A_{j,1} = 1 if j=1j=1 and Aj,1=0A_{j,1} = 0 otherwise.

If T(v1)=0T(v_1) = 0 then the first column of M(T)\mathcal M(T) must be zero for any basis w1,,wmw_1,\dots,w_m since

Tv1=0=M(T)v1=j=1mAj,1wjTv_1 = 0 = \mathcal M(T)v_1 = \sum_{j=1}^m A_{j,1}w_j

Implies Aj,1=0A_{j,1} = 0 for all jj by the independence of w1,,wmw_1,\dots,w_m.


Interestingly we "hardly" had to choose the basis of WW as we only had to pick w1w_1. If we handpicked all of WW one thing I think we could do is get 3c/3 except without the ordering, ie. the diagonal is littered with ones and zeros as opposed to "all ones, then all zeros".