Linear Algebra Exercises-n-Answers.pdf
Linear Algebra Exercises-n-Answers.pdf
Linear Algebra Exercises-n-Answers.pdf
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
<strong>Answers</strong> to <strong>Exercises</strong> 137<br />
(iii) each ⃗κ j is orthogonal to all of the ⃗κ m ’s prior to it (that is, with m < j). With those inductive<br />
hypotheses, consider ⃗κ i+1 .<br />
⃗κ i+1 = ⃗ β i+1 − proj [⃗κ1 ](β i+1 ) − proj [⃗κ2 ](β i+1 ) − · · · − proj [⃗κi ](β i+1 )<br />
= ⃗ β i+1 − β i+1 ⃗κ 1<br />
⃗κ 1 ⃗κ 1<br />
· ⃗κ 1 − β i+1 ⃗κ 2<br />
⃗κ 2 ⃗κ 2<br />
· ⃗κ 2 − · · · − β i+1 ⃗κ i<br />
⃗κ i ⃗κ i<br />
· ⃗κ i<br />
By the inductive assumption (ii) we can expand each ⃗κ j into a linear combination of ⃗ β 1 , . . . , ⃗ β j<br />
= β ⃗ β<br />
i+1 − ⃗ i+1 ⃗κ 1<br />
·<br />
⃗κ 1 ⃗κ ⃗β 1<br />
1<br />
β<br />
− ⃗ i+1 ⃗κ<br />
(<br />
2<br />
· linear combination of β<br />
⃗κ 2 ⃗κ ⃗ 1 , ⃗ )<br />
β<br />
β 2 − · · · − ⃗ i+1 ⃗κ<br />
(<br />
i<br />
· linear combination of β<br />
2 ⃗κ i ⃗κ ⃗ 1 , . . . , ⃗ )<br />
β i<br />
i<br />
The fractions are scalars so this is a linear combination of linear combinations of β ⃗ 1 , . . . , β ⃗ i+1 . It is<br />
therefore just a linear combination of β ⃗ 1 , . . . , β ⃗ i+1 . Now, (i) it cannot sum to the zero vector because<br />
the equation would then describe a nontrivial linear relationship among the β’s ⃗ that are given as<br />
members of a basis (the relationship is nontrivial because the coefficient of β ⃗ i+1 is 1). Also, (ii) the<br />
equation gives ⃗κ i+1 as a combination of β ⃗ 1 , . . . , β ⃗ i+1 . Finally, for (iii), consider ⃗κ j ⃗κ i+1 ; as in the i = 3<br />
case, the dot product of ⃗κ j with ⃗κ i+1 = β ⃗ i+1 − proj [⃗κ1 ]( β ⃗ i+1 ) − · · · − proj [⃗κi ]( β ⃗ i+1 ) can be rewritten to<br />
(<br />
give two kinds of terms, ⃗κ ⃗βi+1 j − proj [⃗κj]( β ⃗ )<br />
i+1 ) (which is zero because the projection is orthogonal)<br />
and ⃗κ j proj [⃗κm ]( β ⃗ i+1 ) with m ≠ j and m < i + 1 (which is zero because by the hypothesis (iii) the<br />
vectors ⃗κ j and ⃗κ m are orthogonal).<br />
Subsection Three.VI.3: Projection Into a Subspace<br />
Three.VI.3.10<br />
(a) When bases for the subspaces<br />
( (<br />
1 2<br />
B M = 〈 〉 B<br />
−1)<br />
N = 〈 〉<br />
−1)<br />
are concatenated<br />
( ) (<br />
⌢ 1 2<br />
B = B M BN = 〈 , 〉<br />
−1 −1)<br />
and the given vector is represented (<br />
3<br />
−2<br />
)<br />
( ) (<br />
1 2<br />
= 1 · + 1 ·<br />
−1 −1)<br />
then the answer comes from retaining the M part and dropping the N part.<br />
( (<br />
3 1<br />
proj M,N ( ) =<br />
−2)<br />
−1)<br />
(b) When the bases<br />
( (<br />
1 1<br />
B M = 〈 〉 B<br />
1)<br />
N 〈 〉<br />
−2)<br />
are concatenated, and the vector is represented,<br />
( ( (<br />
1 1 1<br />
= (4/3) · − (1/3) ·<br />
2)<br />
1)<br />
−2)<br />
then retaining only the M part gives this answer.<br />
(<br />
1<br />
proj M,N ( ) =<br />
2)<br />
(c) With these bases<br />
⎛<br />
⎞<br />
⎛<br />
⎞<br />
( )<br />
4/3<br />
4/3<br />
⎛<br />
⎞<br />
B M = 〈 ⎝ 1 −1⎠ , ⎝ 0 0⎠〉 B N = 〈 ⎝ 1 0⎠〉<br />
0 1<br />
1