128 CHAPTER 5. FUNCTIONS ON NORMED LINEAR SPACES

that∣∣ fk( vm

m

)∣∣ ≥ 1. Since f is continuous, and vm/m→ 0, it follows that f( vm

m

)→ 0 in V.

However, by Lemma 4.4.7, fk( vm

m

)→ 0, a contradiction.

⇐= If each coordinate function is continuous, then

∥ f (v)− f (v̂)∥W =

∥∥∥∥∥ n

∑k=1

fk (v)wk−n

∑k=1

fk (v̂)wk

∥∥∥∥∥≤ n

∑k=1| fk (v)− fk (v̂)|∥wk∥W

Since each fk is continuous, this shows that f is also. ■

5.3 ComparisonsHere are some useful lemmas about comparisons. Here |·| will be the usual norm but onecould generalize.

Lemma 5.3.1 Suppose S,T are linear, defined on a finite dimensional normed linearspace, S−1 exists, and let δ ∈ (0,1). Then whenever ∥S−T∥ is small enough, it followsthat

|Tv||Sv|

∈ (1−δ ,1+δ ) (5.1)

for all v ̸= 0. Similarly if T−1 exists and ∥S−T∥ is small enough,

|Tv||Sv|

∈ (1−δ ,1+δ )

Proof: Say S−1 exists. Then v → |Sv| is a norm. Then by equivalence of norms,Theorem 4.4.9, there exists η > 0 such that for all v, |Sv| ≥ η |v| . Say ∥T −S∥< r < δη

|Tv||Sv|

=|Sv− (S−T )v|

|Sv|≥ |Sv|−∥T −S∥|v|

|Sv|≥ |Sv|−δη |v|

|Sv|≥ |Sv|−δ |Sv|

|Sv|= 1−δ

|Tv||Sv|

=|Sv+(T −S)v|

|Sv|≤ |Sv|+∥T −S∥|v|

|Sv|≤ |Sv|+δη |v|

|Sv|≤ |Sv|+δ |Sv|

|Sv|= 1+δ

Next suppose that T−1 exists. Then, letting δ̂ be small enough,(

1− δ̂ ,1+ δ̂

)⊆(

11+δ

, 11−δ

). From what was just shown, if ∥S−T∥ is small enough,

|Sv||Tv|

∈(

1− δ̂ ,1+ δ̂

)⊆(

11+δ

,1

1−δ

)so|Tv||Sv|

∈ (1−δ ,1+δ ) . ■

In short, the above lemma says that if one of S,T is invertible and the other is close to it,then it is also invertible and the quotient of |Sv| and |Tv| is close to 1. Then the followinglemma is fairly obvious.

Lemma 5.3.2 Let S,T be n×n matrices which are invertible. Then

o(Tv) = o(Sv) = o(v)

and if L is a continuous linear transformation such that for a < b,

supv ̸=0

|Lv||Sv|

< b, infv ̸=0

|Lv||Sv|

> a

If ∥S−T∥ is small enough, it follows that the same inequalities hold with S replaced withT . Here ∥·∥ denotes the operator norm.

128 CHAPTER 5. FUNCTIONS ON NORMED LINEAR SPACESthat | f; ()| > 1. Since f is continuous, and v,,/m —> 0, it follows that f (%) — 0 in V.However, by Lemma 4.4.7, fy (%) —+ 0, a contradiction.<= If each coordinate function is continuous, thennds Ome ¥ HO (9) wIF) —F Ol =nSD Mie ()| [wellSince each f; is continuous, this shows that f is also.5.3 ComparisonsHere are some useful lemmas about comparisons. Here |-| will be the usual norm but onecould generalize.Lemma 5.3.1 Suppose S,T are linear, defined on a finite dimensional normed linearspace, S| exists, and let & € (0,1). Then whenever ||S—T'|| is small enough, it followsthat|To|— €(1-6,1+6 5.1Sv] ( +6) (5.1)for all v £0. Similarly if T~! exists and \\S — T || is small enough,|To|— €(1-—6,1+6Sv] €(1—6,1+6)Proof: Say S~! exists. Then v — |Sv| is a norm. Then by equivalence of norms,Theorem 4.4.9, there exists 7 > 0 such that for all v, |Sv| > 7 |v|. Say ||T —S|| <<r<6n|Tv| _ |Sv—(S—T)| , |Sv| —||7 Sill], |Sv|—dn |e], [Sv] —6|Se|= =1-6|Sv| |Sv| ~ |Sv| — |Sv| ~ |Sv||Tv _ |Su+(T—S)v| < |Sv| + ||T — S|] |v| < |Sv| +67 |v| < |Sv| + 4 |Sv| 1468|Sv| |Sv| |Sv| |Sv| |Sv|Next suppose that T~! exists. Then, letting 5 be small enough, (1 —6s, 1 +3) C(4: rs): From what was just shown, if || — 7|| is small enough,|Sv| ( e, 8 I I 9 EUPere (1-61 é)c , €(1—6,1+5).™\To| +9) 2 \ 75:75) °° isu € +9)In short, the above lemma says that if one of S,T is invertible and the other is close to it,then it is also invertible and the quotient of |Sv| and |Tv| is close to 1. Then the followinglemma is fairly obvious.Lemma 5.3.2 Let S,T be n x n matrices which are invertible. Theno(Tv) =o(Sv) = o(v)and if L is a continuous linear transformation such that for a < b,Lu Lvsup —— Lo} | <b, inf [Lol | >av0 |Sv| v£0 |Sv|If ||S—T|| is small enough, it follows that the same inequalities hold with S replaced withT. Here ||-|| denotes the operator norm.