gametheory
Class GameMinimization

java.lang.Object
  extended by gametheory.GameMinimization
All Implemented Interfaces:
mcmcFunction, Uncmin_methods

public class GameMinimization
extends java.lang.Object
implements mcmcFunction, Uncmin_methods

Class to perform the optimization of the GMM function on real auction entry data.


Constructor Summary
GameMinimization(java.util.ArrayList<Game> gameList)
          Creates a new instance of GameMinimization
 
Method Summary
 double[] execute(double[] guess)
          Use Newton-based solver to optimize GMM function.
 Jama.Matrix executeLTE(double[] guess)
          Use Chernozhukov and Hong's (2003) LTE method to optimize the GMM function.
 double f_to_minimize(double[] x)
          Computes GMM objective function.
 void gradient(double[] x, double[] g)
          User-supplied gradient for Uncmin.
 void hessian(double[] x, double[][] h)
          User-supplied Hessian for Uncmin.
 double objectiveFunction(double[] x)
          MCMC objective function for use in the LTE.
 double pi(double[] x)
          LTE probability of proposed parameters.
 
Methods inherited from class java.lang.Object
clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait
 

Constructor Detail

GameMinimization

public GameMinimization(java.util.ArrayList<Game> gameList)
Creates a new instance of GameMinimization

Parameters:
gameList - List of games (observations).
Method Detail

executeLTE

public Jama.Matrix executeLTE(double[] guess)
Use Chernozhukov and Hong's (2003) LTE method to optimize the GMM function.

Parameters:
guess - Initial guess of parameters.
Returns:
MCMC chain containing results of LTE procedure.

execute

public double[] execute(double[] guess)
Use Newton-based solver to optimize GMM function. Newton methods all use the translations of FORTRAN Uncmin methods done by Steve Verrill. See his webpage at

objectiveFunction

public double objectiveFunction(double[] x)
MCMC objective function for use in the LTE.

Specified by:
objectiveFunction in interface mcmcFunction
Parameters:
x - Vector of unknown parameters.
Returns:
Objective function in LTE form.

pi

public double pi(double[] x)
LTE probability of proposed parameters.

Specified by:
pi in interface mcmcFunction
Parameters:
x - Proposed parameter vector.
Returns:
One everywhere, implying no a priori restrictions on permissable parameters

f_to_minimize

public double f_to_minimize(double[] x)
Computes GMM objective function.

Specified by:
f_to_minimize in interface Uncmin_methods
Parameters:
x - Vector of unknown parameters.
Returns:
Value of GMM function at a guess of the parameters.

gradient

public void gradient(double[] x,
                     double[] g)
User-supplied gradient for Uncmin. Not used.

Specified by:
gradient in interface Uncmin_methods
Parameters:
x - Parameters.
g - Gradient.

hessian

public void hessian(double[] x,
                    double[][] h)
User-supplied Hessian for Uncmin. Not used.

Specified by:
hessian in interface Uncmin_methods
Parameters:
x - Parameters.
h - Hessian.