1 | Let X(ti) be numerical solution and Y(ti) the exact solution. We have Taylor Expansion |
2 | And by assumption X(ti) = Y(ti) and in Euler's method we use exact derivative at ti. Thus And so X(ti+h) = Y(ti+h) + O(h2) and so local error is O(h2) |
3 | Accumulating this error over 1/h steps gives a global error of order h |