Multivariable Fixed Point Method Analysis
Uploaded by
ScribdTranslationsMultivariable Fixed Point Method Analysis
Uploaded by
ScribdTranslationsSan Simón Mayor University
Faculty of Sciences and Technology
NUMERICAL ANALYSIS
(RESEARCH WORK)
Topics: -Multivariable Fixed Point Method
Method of sudden descent
Estudiante:Sarah Daniela Chila Ovando
Career: Chemical Engineering
Fecha:Viernes, 6 de Diciembre de 2017
Teacher: Dr. José A. Soruco Maita
Cochabamba-Bolivia
Multivariable fixed point
The fixed-point method can also be easily extended to
apply it to a system of nonlinear equations. The system of equations by
iterating is obtained by regrouping each of the equations to obtain a
system of equations that separates each of the variables on the left side;
On the right side, an approximation of each variable is inserted, and thus they are calculated.
the new values. These are used to give, in turn, new values, and the process
it is repeated in an iterative form [Nieves et al., 2002], [Burden et al., 2002]. If the
the method is suitable, the values are getting closer and closer to the solution
true. If we start from the following system of equations
the general formula of the method is obtained by rearranging the system of equations
as:
In an iterative manner, the system simply looks like this:
The true solution x* satisfies the equation,
Subtracting equation (2.11) from equation (2.12) leads to,
Using the mean value theorem on the right side of the equation, we have
where ζ is a vector of values between x* and xr.
If ε is definedr= x * -xrthe previous equation will be:
And if all the values of |G′(ζ)|<1, then the error in each variable will decrease.
step by step. For |G′(ζ )|>1 the error will be increasing.
Example:
For the purposes of comparison, the multivariable fixed point method is applied to
system of equations from exercise 2.6, with the initial conditions x0=1, y0=1
y z0=1, that is, the following system of equations:
Solution.
By isolating the first variable from the first equation and so on, the
the iterative system has the following structure,
Numerically, the first iteration is obtained by substituting in the form
simultaneously all the initial conditions in the previous system of equations.
This is how it is obtained:
The previous substitution method is known as the Gauss method.
A known variant called the 'Gauss-Seidel method' uses the
solutions that are being calculated in the following equations; that is, once
when the value of a variable is calculated, it is immediately used in the
successive equations. With this scheme, the iterative system remains of the
next form:
With this method, the following results are obtained in the first iteration.
Table 2.8 shows the results using both fixed point methods.
multivariable.
Numerical analysis. José Alberto Gutiérrez Robles, Miguel Ángel Olmos Gómez, Juan
Martín Casillas González. University of Guadalajara. Chap. 2 Solution of
nonlinear equations. Multivariable fixed point method; Gauss and Gauss-Seidel.
page 55
Method of sudden descent
A Linear Programming (LP) model is one where the decision variables
they are expressed as nonlinear functions either in the objective function and/or
restrictions of an optimization model. This particular characteristic of the
nonlinear models allow addressing problems where there are economies or
diseconomies of scale or in general where the assumptions associated with the
proportionality is not fulfilled.
In this sense, the gradient method (also known as the method of
Cauchy or the steepest descent consists of a specific algorithm for
the resolution of unconstrained NLP models, belonging to the category of
general descent algorithms, where the search for a minimum is
associated with the sequential resolution of a series of one-dimensional problems.
The steps associated with the use of the gradient method or descent more
pronounced consists of:
Example of the Gradient Method
Consider the following nonlinear programming model without constraints. Apply
2 iterations of thegradient methodstarting from the initial point X0=(1,1).
After the second iteration, it is verified that the conditions are met.
necessary first order (d1=(0,0)). Additionally, it can be verified that
the objective function turns out to be convex and consequently the conditions of
first order are sufficient to state that the coordinate (X1, X2) = (-
2,1) is the global optimum or minimum of the problem.
Bibliography.
Numerical analysis. José Alberto Gutiérrez Robles, Miguel Ángel Olmos
Gómez, Juan Martín Casillas González. University of Guadalajara. Chapter 2
Solution of nonlinear equations. Multivariable fixed point method;
Gauss and Gauss-Seidel. p.55
Operations Research. Gradient Method (Cauchy) or of
More Pronounced Descent (Non-Linear Programming).
[Link]