Scribd
Upload
71 views3 pages

Formal Integration of First Order Differen-Tial Equations

This document discusses methods for solving first order differential equations. It covers solving homogeneous and inhomogeneous linear equations, as well as nonlinear equations. For linear equations, it describes using an integrating factor to find a closed form solution. For nonlinear equations, it proposes using Newton's method to linearize the equation and solve the linearized system iteratively to converge on a solution. The overall cost of this Newton's method for finding a truncation of order N is O(M(N)) operations. Exercises are provided to implement these methods in Sage and analyze their computational costs.

Uploaded by

Olivier Ruatta
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
71 views3 pages

Formal Integration of First Order Differen-Tial Equations

This document discusses methods for solving first order differential equations. It covers solving homogeneous and inhomogeneous linear equations, as well as nonlinear equations. For linear equations, it describes using an integrating factor to find a closed form solution. For nonlinear equations, it proposes using Newton's method to linearize the equation and solve the linearized system iteratively to converge on a solution. The overall cost of this Newton's method for finding a truncation of order N is O(M(N)) operations. Exercises are provided to implement these methods in Sage and analyze their computational costs.

Uploaded by

Olivier Ruatta
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 3

Formal integration of rst order dierential equations

by Olivier Ruatta Limoges University - Master ACSYON

1 Homogeneous linear dierential equation of the rst order


Consider an equation of type: y (X) = a(X) y(X) where a(X) K[[X]]. The equation 1 admits as solution: y(X) = y0 e
a(X)

(1)

(2)

This way, a truncation of the solution of equation 1 at order N can be computed using O(M (N ) log (N )) operations in K (the log (N ) factor is in red since in reality, one can show that inversion can be done using O(M (N )) using the algorithm we already give).

2 Inhomogeneous rst order linear dierential equation


Consider an equation of type: y (X) = a(X) y(X) + b(X) (3)

where a(X) and b(X) K[[X]]. We consider the solution of the homogeneous equation (removing b(X)) and make the multiplicative constant of the solution 2 as a serie, i.e. consider: y(X) = f (X)e and plug it in 3, we obtain: f (X) = b(X)e
a(X) a(X)

(4) (5)

Section 3

So one has to solve the equation 5 then compute the product in equation 4. Everything can be done using O(M (N)) operations in K to compute the truncation at order N of the solution of equation 3. The two preceding cases are used as black boxes in the following section.

3 Non-linear rst order dierential equations


First, we give a result, it will be use a lot in text below. Proposition 1. Let F (X , Y ) K[[X , Y ]] a bivariate formal power serie such that F for all serie S(X) we can compute the truncation of F (X , S(X)) and Y (X , S(X)) at order N using O(L(N )) operations in K; then the dierential system: y (X) = F (X , y(X)) y(0) = a (6)

has a FPS solution and a truncation at order N of this solution can be computed using O(L(N ) + M (N )) operation in K. Proof. The idea of the proof is to apply the Newton method to: : y y F (X , y).

The principle is to linearize around an approximation and to annihilate the linear part to rene the approximation. First, considering (y + h), using Taylor expansion we nd: F (X , y)h + O(h2) (7) (y + h) = (y) + h Y this way, we have to solve a rst order dierential equation to compute h annihilating the linear part : F (X , y) h (y). (8) h= Y The serie h(X) can be computed by the method deduced from equations 4 and 5. We introduce the following sequence: y0 = a, yk+1 = yk + hk , hk = e
F (X ,yk) Y

(F (X , yk) yk )e

Y (X ,yk)

+O X 2

k+1

The quadratic convergence of the sequence yk can be shown by induction proving k k simultaneously that both (yk) = O X 2 and hk = O X 2 . At the end of this period, we will be able to show a complexity theorem implying that the overall cost of this algorithm to have a truncation of y(X) at order N is in O(M (N)).

Exercises

4 Exercises
Exercise 1. Let f (X) = sin (X), make a SAGE program taking an integer N as input and computing a truncation of the solution of: y (X) = f (X)y(X) at order N . What is the cost of this approach.

Exercise 2. Let f (X) = 1 X and g(X) = eX , make a SAGE program taking an integer N as input and computing a truncation of the solution of: y (X) = f (X)y(X) + g(X) at order N . What is the cost of this approach.

Exercise 3. Let F (X , Y ) = XY 2, make a SAGE program taking an integer N as input and computing a truncation of the solution of: y (X) = F (X , Y (X)) = XY (X)2 at order N . What is the cost of this approach.

You might also like