Adams-Moulton Three-Step Method Explained
Translated by
ScribdTranslationsAdams-Moulton Three-Step Method Explained
Translated by
ScribdTranslationsAdams-Moulton three-step method
The initial values w are calculated.0= a0, w1= a1, w2= a2(with the Runge-Kutta method), and is applied
the formula:
It is left as an exercise to verify the coefficients of formula (29) by solving the system of
equations given in (26).
It can be shown that the local truncation error |w i– y(tiin the Adams method-
The three-step Moulton is given by the expression:
for some μi∈ti-2, ti+1That is to say, this method is also of the order of h.4Therefore, they are compared
always the results of applying the Adams-Bashforth method of n + 1 steps, against the method
of Adams-Moulton of n steps.
The pseudocode of the algorithm for this method is shown below.
This method requires fewer points and has the same precision as the previous one, but it has the
difficulty of having to solve an equation at each step, which may be nonlinear, in which case
A method of approximation for solutions of nonlinear equations should be applied.
Example
Let's consider the following initial value problem:
y' = y - t2+ 1, 0 ≤ t ≤ 2, y(0) = 0.5
The four-step Adams-Bashforth (A-B) methods and the Adams-Moulton method will be applied.
three steps (A-M), both with step size h = 0.2 for the mesh in the domain [0, 2]. With this
step size, the mesh of points results in:
ti= 0.2.i, for i = 0, ..., 10.
The A-B method applied to this problem, with f(t,y) = y - t2+ 1 and taking it = 0.2 i, have by
difference equation:
Similarly, the A-M method applied to this problem, with the same expression for f(t,y) and
the same values for lost, have as a difference equation:
It is clearly seen here that the A-Mtene method by difference equation is an expression.
implicit for wi+1In this case, the unknown w can be solved.i+1, to obtain the equation:
The results obtained by applying these equations are shown in the following table.
The exact values come from the exact solution of the IVP, y(t) = (t+1) 2- 0.5 etNotes well
show the comparison of these values graphically, due to the great precision of the results
obtained, which causes the errors to be on the order of 10-3.
Table 1
In the example, the implicit Adams-Moulton method gave better results than the method.
Adams-Bashforth explicit of the same order. This generally occurs, but the methods
implicit have the intrinsic weakness that they must first convert algebraically the
method in an explicit representation of wi+1This procedure is not always possible, such as
ocurre por ejemplo en el siguiente problema elemental de valor inicial:
Given that f(t) = ey, the three-step Adams-Moulton method has the difference equation
the next:
and this equation cannot be solved for wi+1To solve the equation, some must be applied
numerical method.
Predictor-Corrector Method
In practice, implicit multistep methods are not used as shown here. They are used
to improve the approximations obtained with explicit methods. The combination of a method
Explicit with an implicit one is called a predictor-corrector method: The explicit method
predict an approximation, and the implicit method corrects it.
Let's consider the following fourth-order method to solve an initial value problem. The
the first step is to calculate the initial values w0, w1,w2y w3for the Adams method
Fourth-order Bashforth. To do this, the Runge-Kutta method can be used. The next step
(0)
it consists of calculating a first approximation w4at point t4from the mesh using as a predictor
the Adams-Bashforth method:
(38)
Then, this approach is improved using the three-step Adams-Moulton method as
corrector, entering the value of w4(0) on the right side:
(39)
In this procedure, the only new evaluation of the function that needs to be calculated is f(t4,
w4(0)
) in the equation of the corrector. The rest of the evaluations of f had already been calculated for
the previous approach. Then, the value w is used.4(1)as an approximation of y(t4), and it repeats the
Technique that involves using the Adams-Bashforth method as a predictor and as a corrector the
(1)
from Adams-Moulton to obtain w 5(0) y w5, the initial and improved approximations of y(t5), and so
successively.
Below is the pseudocode for the four-step Adams predictor-corrector method.
steps.
Example
Given the initial value problem from the previous example:
y ' = y - t2+1, 0 ≤ t ≤ 2, y(0) = 0.5 40
We now apply the Adams predictor-corrector method given in formulas (38) and (39),
having previously applied Runge-Kutta to determine the starting values for the
predictor-corrector, and the values shown in the following table were obtained. In it
The corresponding values of the exact solution and the truncation error are also listed.
local.
Table 2
It can be seen, by comparing the results shown in tables 1 and 2, that the predictor method-
the corrector improves the results obtained with the Adams-Bashforth method.