package PCI.ASQ.pca;
import java.util.Iterator;
import java.util.List;

//import OEU.LYG4DQS4D.LYG9DWithDoubleQuickSort4D;
import OEU.LYG4DQS4D.LYG9DWithDoubleTopSort4D;
//import OSI.PCI.ASQ.statistic.LYG4DWithDoubleQuickSort4D;
import PCI.ASQ.basic.Distance;
import PCI.ASQ.demension.AMV_MVS_VSQ_2D;
import PCI.ASQ.demension.AMV_MVS_VSQ_3D;

//作者,著作权人： 罗瑶光，浏阳
public class FindPCAMeanDistance{
  //oberserverPCAScale：确定主要有效最短路径的观测数
  //sortRangeScale：确定坐标距离排序时的相似成份多少来确定堆栈冗余比，避免堆栈溢出。
  //求坐标团的主要有效距离成份集的平均压强算法
  //适用于 观测/预测坐标集中某精度邻接坐标集的 平均距离 来确定 紧凑度，压力，压强，斥力等。
  //思想：罗瑶光 20191225
  //实现：罗瑶光
  public static double findMeanDistanceFromPositions2D(List<AMV_MVS_VSQ_2D> position2Ds
      , double oberserverPCAScale, int sortRangeScale) {
    Iterator<AMV_MVS_VSQ_2D> outPosition2DIterator= position2Ds.iterator();
    double outMean= 0;
    while(outPosition2DIterator.hasNext()) {
      AMV_MVS_VSQ_2D outPosition2D= outPosition2DIterator.next();
      double[] distance= new double[position2Ds.size()];
      int i= 0;
      Iterator<AMV_MVS_VSQ_2D> inPosition2DIterator= position2Ds.iterator();
      //取坐标点所有距离集合先
      while(inPosition2DIterator.hasNext()) {
        AMV_MVS_VSQ_2D inPosition2D= inPosition2DIterator.next();
        distance[i++]= Distance.getDistance2D(outPosition2D, inPosition2D);
      }
      //距离非对称缺陷小高峰过滤极速快排 从小到大
      //distance= new LYG4DWithDoubleQuickSort4D().sort(distance, sortRangeScale);
      distance= new LYG9DWithDoubleTopSort4D().sort(distance, sortRangeScale, 4);//算法应用更新
      //仅仅取精度内坐标点距离求平均值，0 为本身所以从 1 开始
      double inMean= 0;
      oberserverPCAScale= oberserverPCAScale>= position2Ds.size()
          ? position2Ds.size()- 1: oberserverPCAScale;
      oberserverPCAScale= oberserverPCAScale< 0? 0: oberserverPCAScale;  
      for(i= 1; i<= oberserverPCAScale; i++) {
        inMean+= distance[i];
      }
      inMean/= oberserverPCAScale;
      outMean+= inMean;
    }
    return outMean/ position2Ds.size();
  }
  
  public static double findMeanDistanceFromPositions3D(List<AMV_MVS_VSQ_3D> position3Ds
      , double oberserverPCAScale, int sortRangeScale) {
    Iterator<AMV_MVS_VSQ_3D> outPosition3DIterator= position3Ds.iterator();
    double outMean= 0;
    while(outPosition3DIterator.hasNext()) {
      AMV_MVS_VSQ_3D outPosition3D= outPosition3DIterator.next();
      double[] distance= new double[position3Ds.size()];
      int i= 0;
      Iterator<AMV_MVS_VSQ_3D> inPosition3DIterator= position3Ds.iterator();
      //取坐标点所有距离集合先
      while(inPosition3DIterator.hasNext()) {
        AMV_MVS_VSQ_3D inPosition3D= inPosition3DIterator.next();
        distance[i++]= Distance.getDistance3D(outPosition3D, inPosition3D);
      }
      //距离非对称缺陷小高峰过滤极速快排 从小到大
      //distance= new LYG4DWithDoubleQuickSort4D().sort(distance, sortRangeScale);
      distance= new LYG9DWithDoubleTopSort4D().sort(distance, sortRangeScale, 4);//算法应用更新
      //仅仅取精度内坐标点距离求平均值，0 为本身所以从 1 开始
      double inMean= 0;
      oberserverPCAScale= oberserverPCAScale>= position3Ds.size()
          ? position3Ds.size()- 1: oberserverPCAScale;
      oberserverPCAScale= oberserverPCAScale< 0? 0: oberserverPCAScale;  
      for(i= 1; i<= oberserverPCAScale; i++) {
        inMean+= distance[i];
      }
      inMean/= oberserverPCAScale;
      outMean+= inMean;
    }
    return outMean/ position3Ds.size();
  }
  
  //2D坐标团中 每一个坐标的斥力预测算法。
  public static double[] findPascalMeanDistanceByEachPositions2D(List<AMV_MVS_VSQ_2D> position2Ds
      , double oberserverPCAScale, int sortRangeScale) {
    double[] pascal= new double[position2Ds.size()];
    Iterator<AMV_MVS_VSQ_2D> outPosition2DIterator= position2Ds.iterator();
    double outMean= 0;
    int positionPoint=0;
    while(outPosition2DIterator.hasNext()) {
      AMV_MVS_VSQ_2D outPosition2D= outPosition2DIterator.next();
      double[] distance= new double[position2Ds.size()];
      int i= 0;
      Iterator<AMV_MVS_VSQ_2D> inPosition2DIterator= position2Ds.iterator();
      //取坐标点所有距离集合先
      while(inPosition2DIterator.hasNext()) {
        AMV_MVS_VSQ_2D inPosition2D= inPosition2DIterator.next();
        distance[i++]= Distance.getDistance2D(outPosition2D, inPosition2D);
      }
      //距离非对称缺陷小高峰过滤极速快排 从小到大
      //distance= new LYG4DWithDoubleQuickSort4D().sort(distance, sortRangeScale);
      distance= new LYG9DWithDoubleTopSort4D().sort(distance, sortRangeScale, 4);//算法应用更新
      //仅仅取精度内坐标点距离求平均值，0 为本身所以从 1 开始
      double inMean= 0;
      oberserverPCAScale= oberserverPCAScale>= position2Ds.size()
          ? position2Ds.size()- 1: oberserverPCAScale;
      oberserverPCAScale= oberserverPCAScale< 0? 0: oberserverPCAScale;  
      for(i= 1; i<= oberserverPCAScale; i++) {
        inMean+= distance[i];
      }
      inMean/= oberserverPCAScale;
      outMean+= inMean;
      //采集坐标点形成斥力参照的主要最短路径成份。
      pascal[positionPoint++]= inMean;
    }
    outMean/= position2Ds.size();
    for(int i= 0; i< pascal.length; i++) {
      //比值获取 不稳定 观测数据。
      pascal[i]/= outMean;
    }
    return pascal;
  }
  
  //3D坐标团中 每一个坐标的斥力预测算法。
  public static double[] findPascalMeanDistanceByEachPositions3D(List<AMV_MVS_VSQ_3D> position3Ds
      , double oberserverPCAScale, int sortRangeScale) {
    double[] pascal= new double[position3Ds.size()];
    Iterator<AMV_MVS_VSQ_3D> outPosition3DIterator= position3Ds.iterator();
    double outMean= 0;
    int positionPoint=0;
    while(outPosition3DIterator.hasNext()) {
      AMV_MVS_VSQ_3D outPosition3D= outPosition3DIterator.next();
      double[] distance= new double[position3Ds.size()];
      int i= 0;
      Iterator<AMV_MVS_VSQ_3D> inPosition3DIterator= position3Ds.iterator();
      //取坐标点所有距离集合先
      while(inPosition3DIterator.hasNext()) {
        AMV_MVS_VSQ_3D inPosition3D= inPosition3DIterator.next();
        distance[i++]= Distance.getDistance3D(outPosition3D, inPosition3D);
      }
      //距离非对称缺陷小高峰过滤极速快排 从小到大
      //distance= new LYG4DWithDoubleQuickSort4D().sort(distance, sortRangeScale);
      distance= new LYG9DWithDoubleTopSort4D().sort(distance, sortRangeScale, 4);//算法应用更新
      //仅仅取精度内坐标点距离求平均值，0 为本身所以从 1 开始
      double inMean= 0;
      oberserverPCAScale= oberserverPCAScale>= position3Ds.size()
          ? position3Ds.size()- 1: oberserverPCAScale;
      oberserverPCAScale= oberserverPCAScale< 0? 0: oberserverPCAScale;  
      for(i= 1; i<= oberserverPCAScale; i++) {
        inMean+= distance[i];
      }
      inMean/= oberserverPCAScale;
      outMean+= inMean;
      //采集坐标点形成斥力参照的主要最短路径成份。
      pascal[positionPoint++]= inMean;
    }
    outMean/= position3Ds.size();
    for(int i= 0; i< pascal.length; i++) {
      //比值获取 不稳定 观测数据。
      pascal[i]/= outMean;
    }
    return pascal;
  }
}