Lecture Notes in Differential Equations - Bruce E. Shapiro
Lecture Notes in Differential Equations - Bruce E. Shapiro Lecture Notes in Differential Equations - Bruce E. Shapiro
222 LESSON 26. GENERAL EXISTENCE THEORY* Proof. Use induction. For n = 1, the formula gives which is true. ‖T y − y‖ ≤ 1 − K ‖T y − y‖ = ‖T y − y‖ (26.24) 1 − K For n > 1 suppose that equation 26.23 holds. Then ‖T n+1 y − y‖ = ‖T n+1 y − T n y + T n y − y‖ (26.25) ≤ ‖T n+1 y − T n y‖ + ‖T n y − y‖ (triangle ineqality) (26.26) ≤ ‖T n+1 y − T n y‖ + 1 − Kn ‖T y − y‖ (by (26.23)) (26.27) 1 − K = ‖T n T y − T n y‖ + 1 − Kn ‖T y − y‖ (26.28) 1 − K ≤ K n ‖T y − y‖ + 1 − Kn ‖T y − y‖ (because T is a contraction) 1 − K (26.29) = (1 − K)Kn + (1 − K n ) ‖T y − y‖ (26.30) 1 − K = 1 − Kn+1 ‖T y − y‖ (26.31) 1 − K which proves the conjecture for n + 1. Definition 26.9. Let V be a vector space let T be an operator on V. Then we say y is a fixed point of T if T y = y. Note that in the vector space of functions, since the vectors are functions, the fixed point is a function. Theorem 26.10. Contraction Mapping Theorem 1 Let T be a contraction on a normed vector space V . Then T has a unique fixed point u ∈ V such that T u = u. Furthermore, any sequence of vectors v 1 , v 2 , . . . defined by v k = T v k−1 converges to the unique fixed point T u = u. We denote this by v k → u. Proof. 2 Let ɛ > 0 be given and let v ∈ V. 1 The contraction mapping theorem is sometimes called the Banach Fixed Point Theorem. 2 The proof follows “Proof of Banach Fixed Point Theorem,” Encyclopedia of Mathematics (Volume 2, 54A20:2034), PlanetMath.org.
223 Since K n /(1 − K) → 0 as n → ∞ (because T is a contraction, K < 1), given any v ∈ V, it is possible to choose an integer N such that K n ‖T v − v‖ 1 − K for all n > N. Pick any such integer N. < ɛ (26.32) Choose any two integers m ≥ n ≥ N, and define the sequence ⎫ v 0 = v v 1 = T v v 2 = T v 1 ⎪⎬ Then since T is a contraction, From Lemma 26.8 we have . v n = T v n−1 . ⎪⎭ (26.33) ‖v m − v n ‖ = ‖T m v − T n v‖ (26.34) = ‖T n T m−n v − T n v‖ (26.35) ≤ K n ‖T m−n v − v‖ (26.36) ‖v m − v n ‖ ≤ K n 1 − Km−n ‖T v − v‖ 1 − K (26.37) = Kn − K m ‖T v − v‖ 1 − K (26.38) ≤ Kn ‖T v − v‖ < ɛ (26.39) 1 − K Therefore v n is a Cauchy sequence, and every Cauchy sequence on a complete normed vector space converges. Hence v n → u for some u ∈ V. Either u is a fixed point of T or it is not a fixed point of T . Suppose that u is not a fixed point of T . Then T u ≠ u and hence there exists some δ > 0 such that ‖T u − u‖ > δ (26.40) On the other hand, because v n → u, there exists an integer N such that for all n > N, ‖v n − u‖ < δ/2 (26.41)
- Page 179 and 180: Lesson 20 The Wronskian We have see
- Page 181 and 182: 173 Definition 20.1. The Wronskian
- Page 183 and 184: 175 Example 20.3. Show that y = sin
- Page 185 and 186: 177 and therefore the system of equ
- Page 187 and 188: Lesson 21 Reduction of Order The me
- Page 189 and 190: 181 The method of reduction of orde
- Page 191 and 192: 183 Plugging these into Bessel’s
- Page 193 and 194: 185 Example 21.5. Find a second sol
- Page 195 and 196: Lesson 22 Non-homogeneous Equations
- Page 197 and 198: 189 where r 1 and r 2 are the roots
- Page 199 and 200: 191 This is a first order linear eq
- Page 201 and 202: 193 Theorem 22.5. Properties of the
- Page 203 and 204: 195 where (∫ ν(t) = exp ) −r 2
- Page 205 and 206: 197 The characteristic equation is
- Page 207 and 208: Lesson 23 Method of Annihilators In
- Page 209 and 210: 201 Theorem 23.5. (D 2 − 2aD + (a
- Page 211 and 212: 203 The method of annihilators is r
- Page 213 and 214: Lesson 24 Variation of Parameters T
- Page 215 and 216: 207 Substituting into equation (24.
- Page 217 and 218: 209 Example 24.3. Solve the initial
- Page 219 and 220: Lesson 25 Harmonic Oscillations If
- Page 221 and 222: 213 It is standard to define a new
- Page 223 and 224: 215 As with the unforced case, we c
- Page 225 and 226: Lesson 26 General Existence Theory*
- Page 227 and 228: 219 In the case just proven, there
- Page 229: 221 Theorem 26.5. Under the same co
- Page 233 and 234: 225 for any φ ∈ V. Let g, h be f
- Page 235 and 236: Lesson 27 Higher Order Linear Equat
- Page 237 and 238: 229 L n+1 (e rt y) = e rt a n (D +
- Page 239 and 240: 231 Example 27.2. Find the general
- Page 241 and 242: 233 Differentiating, u ′ (t) = d
- Page 243 and 244: 235 Integrating, − 2K |t − t 0
- Page 245 and 246: 237 a closed form expression for a
- Page 247 and 248: 239 Example 27.6. Find the general
- Page 249 and 250: 241 The characteristic equation is
- Page 251 and 252: 243 The Wronskian In this section w
- Page 253 and 254: 245 Certainly every φ(t) given by
- Page 255 and 256: 247 the differential equation. Over
- Page 257 and 258: 249 By the lemma, to obtain the der
- Page 259 and 260: 251 Example 27.14. Find the general
- Page 261 and 262: 253 So that f(t) = a n (t)y[ (n) +
- Page 263 and 264: Lesson 28 Series Solutions In many
- Page 265 and 266: 257 Changing the index of the secon
- Page 267 and 268: 259 Since the first two terms (corr
- Page 269 and 270: 261 Hence ∞∑ ∞∑ ∞∑ 0 =
- Page 271 and 272: 263 has an analytic solution at t =
- Page 273 and 274: 265 By the triangle inequality, |(k
- Page 275 and 276: 267 Table 28.1: Table of Special Fu
- Page 277 and 278: 269 Thus y = a 0 ( 1 + 1 6 t3 + 1 +
- Page 279 and 280: 271 into (28.114) and collect terms
223<br />
S<strong>in</strong>ce K n /(1 − K) → 0 as n → ∞ (because T is a contraction, K < 1),<br />
given any v ∈ V, it is possible to choose an <strong>in</strong>teger N such that<br />
K n ‖T v − v‖<br />
1 − K<br />
for all n > N. Pick any such <strong>in</strong>teger N.<br />
< ɛ (26.32)<br />
Choose any two <strong>in</strong>tegers m ≥ n ≥ N, and def<strong>in</strong>e the sequence<br />
⎫<br />
v 0 = v<br />
v 1 = T v<br />
v 2 = T v 1<br />
⎪⎬<br />
Then s<strong>in</strong>ce T is a contraction,<br />
From Lemma 26.8 we have<br />
.<br />
v n = T v n−1<br />
.<br />
⎪⎭<br />
(26.33)<br />
‖v m − v n ‖ = ‖T m v − T n v‖ (26.34)<br />
= ‖T n T m−n v − T n v‖ (26.35)<br />
≤ K n ‖T m−n v − v‖ (26.36)<br />
‖v m − v n ‖ ≤ K n 1 − Km−n<br />
‖T v − v‖<br />
1 − K<br />
(26.37)<br />
= Kn − K m<br />
‖T v − v‖<br />
1 − K<br />
(26.38)<br />
≤<br />
Kn<br />
‖T v − v‖ < ɛ (26.39)<br />
1 − K<br />
Therefore v n is a Cauchy sequence, and every Cauchy sequence on a complete<br />
normed vector space converges. Hence v n → u for some u ∈ V.<br />
Either u is a fixed po<strong>in</strong>t of T or it is not a fixed po<strong>in</strong>t of T .<br />
Suppose that u is not a fixed po<strong>in</strong>t of T . Then T u ≠ u and hence there<br />
exists some δ > 0 such that<br />
‖T u − u‖ > δ (26.40)<br />
On the other hand, because v n → u, there exists an <strong>in</strong>teger N such that<br />
for all n > N,<br />
‖v n − u‖ < δ/2 (26.41)