package bartMachine;

import java.io.Serializable;

/**

* This portion of the code initializes the Gibbs sampler

*

* @author Adam Kapelner and Justin Bleich

*/

@SuppressWarnings("serial")

public abstract class bartMachine_d_init extends bartMachine_c_debug implements Serializable{

/** during debugging, we may want to fix sigsq */

protected transient double fixed_sigsq;

/** the number of the current Gibbs sample */

protected int gibbs_sample_num;

/** cached current sum of residuals vector */

protected transient double[] sum_resids_vec;

/** Initializes the Gibbs sampler setting all zero entries and moves the counter to the first sample */

protected void SetupGibbsSampling(){

InitGibbsSamplingData();

InitizializeSigsq();

InitializeTrees();

InitializeMus();

if (tree_illust){

InitTreeIllustrations();

}

//the zeroth gibbs sample is the initialization we just did; now we're onto the first in the chain

gibbs_sample_num = 1;

sum_resids_vec = new double[n];

}

/** Initializes the vectors that hold information across the Gibbs sampler */

protected void InitGibbsSamplingData(){

//now initialize the gibbs sampler array for trees and error variances

gibbs_samples_of_bart_trees = new bartMachineTreeNode[num_gibbs_total_iterations + 1][num_trees];

gibbs_samples_of_bart_trees_after_burn_in = new bartMachineTreeNode[num_gibbs_total_iterations - num_gibbs_burn_in + 1][num_trees];

gibbs_samples_of_sigsq = new double[num_gibbs_total_iterations + 1];

gibbs_samples_of_sigsq_after_burn_in = new double[num_gibbs_total_iterations - num_gibbs_burn_in];

accept_reject_mh = new boolean[num_gibbs_total_iterations + 1][num_trees];

accept_reject_mh_steps = new char[num_gibbs_total_iterations + 1][num_trees];

}

/** Initializes the tree structures in the zeroth Gibbs sample to be merely stumps */

protected void InitializeTrees() {

//create the array of trees for the zeroth gibbs sample

bartMachineTreeNode[] bart_trees = new bartMachineTreeNode[num_trees];

for (int i = 0; i < num_trees; i++){

bartMachineTreeNode stump = new bartMachineTreeNode(this);

stump.setStumpData(X_y, y_trans, p);

bart_trees[i] = stump;

}

gibbs_samples_of_bart_trees[0] = bart_trees;

}

/** Initializes the leaf structure (the mean predictions) by setting them to zero (in the transformed scale, this is the center of the range) */

protected void InitializeMus() {

for (bartMachineTreeNode stump : gibbs_samples_of_bart_trees[0]){

stump.y_pred = 0;

}

}

/** Initializes the first variance value by drawing from the prior */

protected void InitizializeSigsq() {

gibbs_samples_of_sigsq[0] = StatToolbox.sample_from_inv_gamma(hyper_nu / 2, 2 / (hyper_nu * hyper_lambda));

}

/** this is the number of posterior Gibbs samples afte burn-in (thinning was never implemented) */

public int numSamplesAfterBurningAndThinning(){

return num_gibbs_total_iterations - num_gibbs_burn_in;

}

public void setNumGibbsBurnIn(int num_gibbs_burn_in){

this.num_gibbs_burn_in = num_gibbs_burn_in;

}

public void setNumGibbsTotalIterations(int num_gibbs_total_iterations){

this.num_gibbs_total_iterations = num_gibbs_total_iterations;

}

public void setSigsq(double fixed_sigsq){

this.fixed_sigsq = fixed_sigsq;

}

public boolean[][] getAcceptRejectMH(){

return accept_reject_mh;

}

}