The Computable Differential Equation Lecture ... - Bruce E. Shapiro

The Computable Differential Equation Lecture ... - Bruce E. Shapiro The Computable Differential Equation Lecture ... - Bruce E. Shapiro

bruce.shapiro.com
from bruce.shapiro.com More from this publisher
21.04.2015 Views

138 CHAPTER 7. DELAY DIFFERENTIAL EQUATIONS Since f is strongly Lipshitz, and therefore |f(t, y 1 , . . . , y m ) − f(, ỹ 1 , . . . , ỹ m )| ≤ K max 1≤j≤m |y j − ỹ j | (7.36) |y(t) − ỹ(t)| ≤ |θ(t 0 ) − ˜θ(t 0 )| + K ≤ v(t 0 ) + K ∫ t ∫ t Applying the definition of v(t) one more time, By Gronwall’s inequality, t 0 max |θ(x − τ j) − ˜θ(x − τ j )|dx (7.37) 1≤j≤m t 0 v(x)dx (7.38) v(t) ≤ v(t 0 ) + K from which 7.30 follows immediately. ∫ t t 0 v(x)dx (7.39) ∫ t v(t) ≤ v(t 0 ) exp v(x)dx t 0 (7.40) 7.2 Method Of Steps Consider the DDE with a single delay, y ′ (t) = f(t − τ, y(t − τ)) (7.41) y(t) = g(t), t 0 − τ ≤ t ≤ t 0 (7.42) where we will assume that g(t 0 − τ) = g(t 0 ). Then on the interval [t 0 , t 0 + τ] we have ∫ t y(t) = g(t 0 ) + f(s − τ, y(s − τ))ds (7.43) t 0 = g(t 0 ) + ∫ t t 0 f(s − τ, g(s − τ))ds (7.44) The right hand side is fully explicit and can be evaluated. This gives the solution on the first “step,” [t 0 , t 0 + τ]. Next, assume we have already solved the DDE on [t 0 , t 0 + kτ] for some integer k > 0. (This statement is immediately true for k = 1 once we have solved the first interval). Then on subsequent steps [t 0 + kτ, t 0 + (k + 1)τ], for k = 1, 2, . . . , y(t) = y(t 0 + kτ) + ∫ t t 0 +kτ f(s − τ, y(s − τ))ds (7.45) Since the integrals are evaluated in order, i.e., the integral is solved on the interval [t 0 + (k − 1)τ, t 0 + kτ] first, everything on the right-hand side of the equation is already known, so the integral can be evaluated. Math 582B, Spring 2007 California State University Northridge c○2007, B.E.Shapiro Last revised: May 23, 2007

CHAPTER 7. DELAY DIFFERENTIAL EQUATIONS 139 Example 7.1. Solve the linear test equation on the interval [0, 3]. Solution. On [0, 1], y ′ = λy(t − 1) (7.46) y(t) = 1, −1 ≤ t ≤ 0 (7.47) y(t) = y(0) + = 1 + λ ∫ t 0 ∫ t 0 λy(s − 1)ds (7.48) ds (7.49) = 1 + λt (7.50) At t = 1 we have y(1) = 1 + λ, so on the interval [1, 2], ⇐= y(t) = y(1) + ∫ t = 1 + λ + λ = 1 + λ + λ = 1 + λ + λ λy(s − 1)ds (7.51) 1 ∫ t 1 ∫ t (1 + λs)| s→s−1 ds (7.52) (1 + λ(s − 1))ds (7.53) 1 ( −1 + t + λ ) 2 − tλ + t2 λ 2 (7.54) = 1 + tλ + λ2 2 − tλ2 + t2 λ 2 2 (7.55) The next iteration gives the solution on [2, 3] as ⇐= ∫ t y(t) = y(2) + (1 + (s − 1)λ + λ2 2 2 − (s − 1)λ2 + (s − 1)2 λ 2 ) ds (7.56) 2 = 1 + λ2 2 − 4λ3 3 + t3 λ 3 ( ) λ 2 6 + t2 2 − λ3 + t(λ − λ 2 + 2λ 3 ) (7.57) This technique is repetitive (much like Picard’s method) and easily implemented in Mathematica. The following code will solve the DDE y ′ = λy(t − τ) (7.58) y(t) = g(t), −τ < t < 0 (7.59) It will print out the solution, one step at a time, for n steps: methodOfSteps[g , t , tau , n ] := Module[{y0, y, s, i}, y0 = g[0]; c○2007, B.E.Shapiro Last revised: May 23, 2007 Math 582B, Spring 2007 California State University Northridge

138 CHAPTER 7. DELAY DIFFERENTIAL EQUATIONS<br />

Since f is strongly Lipshitz,<br />

and therefore<br />

|f(t, y 1 , . . . , y m ) − f(, ỹ 1 , . . . , ỹ m )| ≤ K max<br />

1≤j≤m |y j − ỹ j | (7.36)<br />

|y(t) − ỹ(t)| ≤ |θ(t 0 ) − ˜θ(t 0 )| + K<br />

≤ v(t 0 ) + K<br />

∫ t<br />

∫ t<br />

Applying the definition of v(t) one more time,<br />

By Gronwall’s inequality,<br />

t 0<br />

max |θ(x − τ j) − ˜θ(x − τ j )|dx (7.37)<br />

1≤j≤m<br />

t 0<br />

v(x)dx (7.38)<br />

v(t) ≤ v(t 0 ) + K<br />

from which 7.30 follows immediately.<br />

∫ t<br />

t 0<br />

v(x)dx (7.39)<br />

∫ t<br />

v(t) ≤ v(t 0 ) exp v(x)dx<br />

t 0<br />

(7.40)<br />

7.2 Method Of Steps<br />

Consider the DDE with a single delay,<br />

y ′ (t) = f(t − τ, y(t − τ)) (7.41)<br />

y(t) = g(t), t 0 − τ ≤ t ≤ t 0 (7.42)<br />

where we will assume that g(t 0 − τ) = g(t 0 ). <strong>The</strong>n on the interval [t 0 , t 0 + τ] we<br />

have<br />

∫ t<br />

y(t) = g(t 0 ) + f(s − τ, y(s − τ))ds (7.43)<br />

t 0<br />

= g(t 0 ) +<br />

∫ t<br />

t 0<br />

f(s − τ, g(s − τ))ds (7.44)<br />

<strong>The</strong> right hand side is fully explicit and can be evaluated. This gives the solution<br />

on the first “step,” [t 0 , t 0 + τ]. Next, assume we have already solved the DDE on<br />

[t 0 , t 0 + kτ] for some integer k > 0. (This statement is immediately true for k = 1<br />

once we have solved the first interval). <strong>The</strong>n on subsequent steps [t 0 + kτ, t 0 + (k +<br />

1)τ], for k = 1, 2, . . . ,<br />

y(t) = y(t 0 + kτ) +<br />

∫ t<br />

t 0 +kτ<br />

f(s − τ, y(s − τ))ds (7.45)<br />

Since the integrals are evaluated in order, i.e., the integral is solved on the interval<br />

[t 0 + (k − 1)τ, t 0 + kτ] first, everything on the right-hand side of the equation is<br />

already known, so the integral can be evaluated.<br />

Math 582B, Spring 2007<br />

California State University Northridge<br />

c○2007, B.E.<strong>Shapiro</strong><br />

Last revised: May 23, 2007

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

Saved successfully!

Ooh no, something went wrong!