math572_notes - 1. Thurs 1/6 1 1. IVP for ODEs y = f ( y )...

Info iconThis preview shows pages 1–5. Sign up to view the full content.

View Full Document Right Arrow Icon

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: 1. Thurs 1/6 1 1. IVP for ODEs y = f ( y ) , y (0) = y : find y ( t ) ex y = y , y (0) = y y ( t ) = y e t y = sin y , y (0) = y y ( t ) = ? def An IVP is well-posed if the following conditions are satisfied. 1. A solution exists. 2. The solution is unique. 3. The solution depends continuously on the data (i.e. y , f ). def f satisfies a Lipschitz condition on a domain D if there exists a constant L st | f ( y )- f ( u ) | L | y- u | for all y , u D . note : we typically assume that f ( y ) is smooth, so L = max | f y | thm If f satisfies a Lipschitz condition, then the IVP is well-posed. ex y = ay + b , y (0) = y L = | a | , y ( t ) = y e at + b a ( e at- 1) y = y 1 / 2 , y (0) = 0 L = for y 0 , non-unique solution , y ( t ) = 0 , t 2 4 y = y 2 , y (0) = 1 L = for y 1 , y ( t ) = 1 1- t , lim t 1 y ( t ) = 2 Eulers method h = t = step size , t n = nh y n = y ( t n ) : exact solution , u n : numerical approximation y = f ( y ) : differential equation u n +1- u n h = f ( u n ) : finite-difference scheme u n +1 = u n + hf ( u n ) , u = y ex y = y , y = 1 u = 1 u 1 = u + hf ( u ) = 1 + h u 2 = u 1 + hf ( u 1 ) = (1 + h ) + h (1 + h ) = (1 + h ) 2 u n = (1 + h ) n t n = 1 nh = 1 , y n = y (1) = e = 2 . 7182818 . . . n h u n y n- u n ( y n- u n ) /h 10 0.1 2.5937425 0.1245 1.245 20 0.05 2.6532977 0.0650 1.300 40 0.025 2.6850638 0.0332 1.328 80 0.0125 2.7014849 0.0168 1.344 3 Eulers method y = y , y (0) = 1 u n +1 = u n + hu n , u = 1 0.5 1 1.5 2 1 2 3 4 5 6 7 8 h=1 0.5 1 1.5 2 1 2 3 4 5 6 7 8 h=1/2 0.5 1 1.5 2 1 2 3 4 5 6 7 8 h=1/4 0.5 1 1.5 2 1 2 3 4 5 6 7 8 h=1/8 note 1. The solid line is the exact solution y ( t ). The circles are the numerical solution u n . The dashed line is the piecewise linear interpolant of u n . 2. For a fixed time t , the error decreases as h 0. 3. For a fixed stepsize h , the error increases as t . 4 convergence proof (special case) y = y , y = 1 y ( t ) = e t u n +1 = u n + hu n , u = 1 u n = (1 + h ) n consider t = 1 , h = 1 n lim h u n = lim n 1 + 1 n ! n = lim n exp n ln 1 + 1 n !! = exp lim n n ln 1 + 1 n !! lim n n ln 1 + 1 n ! = 0 = lim n ln (1 + 1 n ) 1 n = lim n 1 1 + 1 n - 1 n 2 1- 1 n 2 = 1 lim h u n = e = y (1) Eulers method converges hw : y = ay + b , y (0) = y convergence proof (general case) y = f ( y ) , y (0) = y u n +1 = u n + hf ( u n ) , u : given fix t > , set h = t/n t = nh = t n , y n = y ( t n ) goal : lim h u n = y n y n +1 = y n + hf ( y n ) + n , n : local truncation error step 1 : estimate n y n +1 = y ( t n +1 ) = y ( t n + h ) = y ( t n ) + hy ( t n ) + h 2 2 y 00 ( e t ) y n +1 = y n + hf ( y n ) + h 2 2 y 00 ( e t ) n = h 2 2 y 00 ( e t ) step 2 : analyze error , e n = y n- u n : error e n +1 = y n +1- u n +1 = y n + hf ( y n ) + n- ( u n + hf ( u n )) = y n- u n + h ( f ( y n )- f ( u n )) + n = e n...
View Full Document

Page1 / 83

math572_notes - 1. Thurs 1/6 1 1. IVP for ODEs y = f ( y )...

This preview shows document pages 1 - 5. Sign up to view the full document.

View Full Document Right Arrow Icon
Ask a homework question - tutors are online