06.09.2021 Views

A First Course in Linear Algebra, 2017a

A First Course in Linear Algebra, 2017a

A First Course in Linear Algebra, 2017a

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

4.11. Orthogonality and the Gram Schmidt Process 241<br />

Theorem 4.139: Approximation Theorem<br />

Let W be a subspace of R n and Y any po<strong>in</strong>t <strong>in</strong> R n .LetZ be the po<strong>in</strong>t whose position vector is the<br />

orthogonal projection of Y onto W .<br />

Then, Z is the po<strong>in</strong>t <strong>in</strong> W closest to Y .<br />

Proof. <strong>First</strong> Z is certa<strong>in</strong>ly a po<strong>in</strong>t <strong>in</strong> W s<strong>in</strong>ce it is <strong>in</strong> the span of a basis of W.<br />

To show that Z is the po<strong>in</strong>t <strong>in</strong> W closest to Y , we wish to show that |⃗y −⃗z 1 | > |⃗y−⃗z| for all⃗z 1 ≠⃗z ∈ W.<br />

We beg<strong>in</strong> by writ<strong>in</strong>g ⃗y −⃗z 1 =(⃗y −⃗z)+(⃗z −⃗z 1 ). Now, the vector ⃗y −⃗z is orthogonal to W, and⃗z −⃗z 1 is<br />

conta<strong>in</strong>ed <strong>in</strong> W. Therefore these vectors are orthogonal to each other. By the Pythagorean Theorem, we<br />

have that<br />

‖⃗y −⃗z 1 ‖ 2 = ‖⃗y −⃗z‖ 2 + ‖⃗z −⃗z 1 ‖ 2 > ‖⃗y −⃗z‖ 2<br />

This follows because⃗z ≠⃗z 1 so ‖⃗z −⃗z 1 ‖ 2 > 0.<br />

Hence, ‖⃗y −⃗z 1 ‖ 2 > ‖⃗y −⃗z‖ 2 . Tak<strong>in</strong>g the square root of each side, we obta<strong>in</strong> the desired result.<br />

Consider the follow<strong>in</strong>g example.<br />

Example 4.140: Orthogonal Projection<br />

Let W be the plane through the orig<strong>in</strong> given by the equation x − 2y + z = 0.<br />

F<strong>in</strong>d the po<strong>in</strong>t <strong>in</strong> W closest to the po<strong>in</strong>t Y =(1,0,3).<br />

♠<br />

Solution. We must first f<strong>in</strong>d an orthogonal basis for W. Notice that W is characterized by all po<strong>in</strong>ts (a,b,c)<br />

where c = 2b − a. Inotherwords,<br />

⎡<br />

a<br />

⎤ ⎡<br />

1<br />

⎤ ⎡ ⎤<br />

0<br />

W = ⎣ b<br />

2b − a<br />

⎦ = a⎣<br />

0<br />

−1<br />

⎦ + b⎣<br />

1 ⎦, a,b ∈ R<br />

2<br />

We can thus write W as<br />

W = span{⃗u 1 ,⃗u 2 }<br />

⎧⎡<br />

⎤ ⎡<br />

⎨ 1<br />

= span ⎣ 0 ⎦, ⎣<br />

⎩<br />

−1<br />

Notice that this span is a basis of W as it is l<strong>in</strong>early <strong>in</strong>dependent. We will use the Gram-Schmidt<br />

Process to convert this to an orthogonal basis, {⃗w 1 ,⃗w 2 }. In this case, as we remarked it is only necessary<br />

to f<strong>in</strong>d an orthogonal basis, and it is not required that it be orthonormal.<br />

⃗w 2<br />

⎡<br />

⃗w 1 = ⃗u 1 = ⎣<br />

1<br />

0<br />

−1<br />

⎤<br />

⎦<br />

( ) ⃗u2 •⃗w 1<br />

= ⃗u 2 −<br />

‖⃗w 1 ‖ 2 ⃗w 1<br />

0<br />

1<br />

2<br />

⎤⎫<br />

⎬<br />

⎦<br />

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!