]> git.uio.no Git - u/mrichter/AliRoot.git/blobdiff - PMD/AliPMDClusteringV2.cxx
Fix for DA's
[u/mrichter/AliRoot.git] / PMD / AliPMDClusteringV2.cxx
index 8fd20829e15e7d072caa2b0702a70966eab90204..6d0e87a8a24776926afb2dcf12968b894c41a2d4 100644 (file)
@@ -13,6 +13,8 @@
  * provided "as is" without express or implied warranty.                  *
  **************************************************************************/
 
+/* $Id$ */
+
 //-----------------------------------------------------//
 //                                                     //
 //  Source File : PMDClusteringV2.cxx                  //
    Bhubaneswar 751 005 ( phatak@iopb.res.in ) Given the energy deposited
    ( or ADC value ) in each cell of supermodule ( pmd or cpv ), the code
    builds up superclusters and breaks them into clusters. The input is
-   in array fEdepCell[kNDIMX][kNDIMY] and cluster information is in array
-   fClusters[5][5000]. integer fClno gives total number of clusters in the
-   supermodule.
-
-   fEdepCell, fClno  and fClusters are the only global ( public ) variables.
+   in TObjarray  and cluster information is in TObjArray.
+   integer clno gives total number of clusters in the  supermodule.
+   fClusters is the  global ( public ) variables.
    Others are local ( private ) to the code.
    At the moment, the data is read for whole detector ( all supermodules
    and pmd as well as cpv. This will have to be modify later )
    LAST UPDATE  :  October 23, 2002
 -----------------------------------------------------------------------*/
 
-#include "Riostream.h"
+#include <Riostream.h>
+#include <TMath.h>
 #include <TObjArray.h>
-#include <stdio.h>
+#include <TArrayI.h>
 
+#include "AliPMDcludata.h"
 #include "AliPMDcluster.h"
 #include "AliPMDClustering.h"
 #include "AliPMDClusteringV2.h"
@@ -51,8 +53,9 @@ ClassImp(AliPMDClusteringV2)
 const Double_t AliPMDClusteringV2::fgkSqroot3by2=0.8660254;  // sqrt(3.)/2.
 
 AliPMDClusteringV2::AliPMDClusteringV2():
-  fClno(0),
-  fCutoff(0.0)
+  fPMDclucont(new TObjArray()),
+  fCutoff(0.0),
+  fClusParam(0)
 {
   for(int i = 0; i < kNDIMX; i++)
     {
@@ -60,31 +63,59 @@ AliPMDClusteringV2::AliPMDClusteringV2():
        {
          fCoord[0][i][j] = i+j/2.;
          fCoord[1][i][j] = fgkSqroot3by2*j;
-         fEdepCell[i][j] = 0;
        }
     }
 }
 // ------------------------------------------------------------------------ //
+
+
+AliPMDClusteringV2::AliPMDClusteringV2(const AliPMDClusteringV2& pmdclv2):
+  AliPMDClustering(pmdclv2),
+  fPMDclucont(0),
+  fCutoff(0),
+  fClusParam(0)
+{
+  // copy constructor
+  AliError("Copy constructor not allowed ");
+  
+}
+// ------------------------------------------------------------------------ //
+AliPMDClusteringV2 &AliPMDClusteringV2::operator=(const AliPMDClusteringV2& /*pmdclv2*/)
+{
+  // copy constructor
+  AliError("Assignment operator not allowed ");
+  return *this;
+}
+// ------------------------------------------------------------------------ //
 AliPMDClusteringV2::~AliPMDClusteringV2()
 {
-
+  delete fPMDclucont;
 }
 // ------------------------------------------------------------------------ //
-void AliPMDClusteringV2::DoClust(Int_t idet, Int_t ismn, Double_t celladc[48][96], TObjArray *pmdcont)
+
+void AliPMDClusteringV2::DoClust(Int_t idet, Int_t ismn, 
+                                Int_t celltrack[48][96],
+                                Int_t cellpid[48][96],
+                                Double_t celladc[48][96],
+                                TObjArray *pmdcont)
 {
   // main function to call other necessary functions to do clustering
   //
   AliPMDcluster *pmdcl = 0;
 
-  Int_t    i, i1, i2, j, nmx1, incr, id, jd;
-  Int_t    celldataX[15], celldataY[15];
-  Float_t  clusdata[6];
-  Double_t cutoff, ave;
+  const Float_t ktwobysqrt3 = 1.1547; // 2./sqrt(3.)
+  const Int_t   kNmaxCell   = 19;     // # of cells surrounding a cluster center
+  Int_t    i = 0, j = 0, nmx1 = 0;
+  Int_t    incr = 0, id = 0, jd = 0;
+  Int_t    ndimXr = 0;
+  Int_t    ndimYr = 0;
+  Int_t    celldataX[kNmaxCell], celldataY[kNmaxCell];
+  Int_t    celldataTr[kNmaxCell], celldataPid[kNmaxCell];
+  Float_t  celldataAdc[kNmaxCell];
+  Float_t  clusdata[6] = {0.,0.,0.,0.,0.,0.};  
+  Double_t cutoff = 0., ave = 0.;
+  Double_t edepcell[kNMX];
 
-  const float ktwobysqrt3 = 1.1547; // 2./sqrt(3.)
-
-  Int_t ndimXr =0;
-  Int_t ndimYr =0;
 
   if (ismn < 12)
     {
@@ -96,72 +127,81 @@ void AliPMDClusteringV2::DoClust(Int_t idet, Int_t ismn, Double_t celladc[48][96
       ndimXr = 48;
       ndimYr = 96;
     }
-
-  for (Int_t i =0; i < kNDIMX; i++)
+  
+  for (i =0; i < kNMX; i++)
     {
-      for (Int_t j =0; j < kNDIMY; j++)
-       {
-         fEdepCell[i][j] = 0;
-       }
+     edepcell[i] = 0.;
     }
-
-
+    
   for (id = 0; id < ndimXr; id++)
     {
       for (jd = 0; jd < ndimYr; jd++)
        {
-         j=jd;
-         i=id+(ndimYr/2-1)-(jd/2);
-         
+         j = jd;
+         i = id + (ndimYr/2-1) - (jd/2);
+         Int_t ij = i + j*kNDIMX;
          if (ismn < 12)
            {
-             fEdepCell[i][j] = celladc[jd][id];
+             edepcell[ij]    = celladc[jd][id];
            }
          else if (ismn >= 12 && ismn <= 23)
            {
-             fEdepCell[i][j] = celladc[id][jd];
+            edepcell[ij]    = celladc[id][jd];
            }
 
        }
     }
 
-  Order();          // order the data
+  // the dimension of iord1 is increased twice
+  Int_t iord1[2*kNMX];
+  TMath::Sort((Int_t)kNMX,edepcell,iord1);// order the data
   cutoff = fCutoff; // cutoff used to discard cells having ener. dep.
-  ave=0.;
-  nmx1=-1;
+  ave  = 0.;
+  nmx1 = -1;
 
-  for(j=0;j<kNMX; j++)
+  for(i = 0;i < kNMX; i++)
     {
-      i1 = fIord[0][j];
-      i2 = fIord[1][j];
-      if (fEdepCell[i1][i2] > 0.) {ave = ave + fEdepCell[i1][i2];}
-      if (fEdepCell[i1][i2] > cutoff ) nmx1 = nmx1 + 1;
+      if(edepcell[i] > 0.) 
+       {
+         ave += edepcell[i];
+       }
+      if(edepcell[i] > cutoff )
+       {
+         nmx1++;
+       }
     }
-  // nmx1 --- number of cells having ener dep >= cutoff
-
+  
   AliDebug(1,Form("Number of cells having energy >= %f are %d",cutoff,nmx1));
-
-  if (nmx1 == 0) nmx1 = 1;
-  ave=ave/nmx1;
-
+  
+  if (nmx1 == 0) 
+    {
+      nmx1 = 1;
+    }
+  ave = ave/nmx1;
+  
   AliDebug(1,Form("Number of cells in a SuperM = %d and Average = %f",
                  kNMX,ave));
-          
-  incr = CrClust(ave, cutoff, nmx1);
-  RefClust(incr);
-
-  AliDebug(1,Form("Detector Plane = %d  Serial Module No = %d Number of clusters = %d",idet, ismn, fClno));
   
-  for(i1=0; i1<=fClno; i1++)
+  incr = CrClust(ave, cutoff, nmx1,iord1, edepcell);
+  RefClust(incr,edepcell );
+  
+  Int_t nentries1 = fPMDclucont->GetEntries();
+  AliDebug(1,Form("Detector Plane = %d  Serial Module No = %d Number of clusters = %d",idet, ismn, nentries1));
+  AliDebug(1,Form("Total number of clusters/module = %d",nentries1));
+  for (Int_t ient1 = 0; ient1 < nentries1; ient1++)
     {
-      Float_t cluXC    = (Float_t) fClusters[0][i1];
-      Float_t cluYC    = (Float_t) fClusters[1][i1];
-      Float_t cluADC   = (Float_t) fClusters[2][i1];
-      Float_t cluCELLS = (Float_t) fClusters[3][i1];
-      Float_t sigmaX   = (Float_t) fClusters[4][i1];
-      Float_t sigmaY   = (Float_t) fClusters[5][i1];
+      AliPMDcludata *pmdcludata = 
+       (AliPMDcludata*)fPMDclucont->UncheckedAt(ient1);
+      Float_t cluXC    = pmdcludata->GetClusX();
+      Float_t cluYC    = pmdcludata->GetClusY();
+      Float_t cluADC   = pmdcludata->GetClusADC();
+      Float_t cluCELLS = pmdcludata->GetClusCells();
+      Float_t cluSIGX  = pmdcludata->GetClusSigmaX();
+      Float_t cluSIGY  = pmdcludata->GetClusSigmaY();
+      
       Float_t cluY0    = ktwobysqrt3*cluYC;
       Float_t cluX0    = cluXC - cluY0/2.;
+      
       // 
       // Cluster X centroid is back transformed
       //
@@ -169,664 +209,864 @@ void AliPMDClusteringV2::DoClust(Int_t idet, Int_t ismn, Double_t celladc[48][96
        {
          clusdata[0] = cluX0 - (24-1) + cluY0/2.;
        }
-      else if (ismn >= 12 && ismn <= 23)
+      else if (ismn  >= 12 && ismn <= 23)
        {
          clusdata[0] = cluX0 - (48-1) + cluY0/2.;
        }         
 
-      clusdata[1]      = cluY0;
-      clusdata[2]      = cluADC;
-      clusdata[3]      = cluCELLS;
-      clusdata[4]      = sigmaX;
-      clusdata[5]      = sigmaY;
-
+      clusdata[1]     = cluY0;
+      clusdata[2]     = cluADC;
+      clusdata[3]     = cluCELLS;
+      clusdata[4]     = cluSIGX;
+      clusdata[5]     = cluSIGY;
       //
       // Cells associated with a cluster
       //
-      for (Int_t ihit = 0; ihit < 15; ihit++)
+      for (Int_t ihit = 0; ihit < kNmaxCell; ihit++)
        {
-         celldataX[ihit] = 1;  // dummy nos. -- will be changed
-         celldataY[ihit] = 1;  // dummy nos. -- will be changed
-       }
+         Int_t dummyXY = pmdcludata->GetCellXY(ihit);
+        
+         Int_t celldumY   = dummyXY%10000;
+         Int_t celldumX   = dummyXY/10000;
+          Float_t cellY    = (Float_t) celldumY/10;
+         Float_t cellX    = (Float_t) celldumX/10;
+
+         // 
+         // Cell X centroid is back transformed
+         //
+         if (ismn < 12)
+           {
+             celldataX[ihit] = (Int_t) ((cellX - (24-1) + cellY/2.) + 0.5);
+           }
+         else if (ismn  >= 12 && ismn <= 23)
+           {
+             celldataX[ihit] = (Int_t) ((cellX - (48-1) + cellY/2.) + 0.5 );
+           }     
+         celldataY[ihit]   = (Int_t) (cellY + 0.5);
 
-      pmdcl = new AliPMDcluster(idet, ismn, clusdata, celldataX, celldataY);
-      pmdcont->Add(pmdcl);
-    }
-}
-// ------------------------------------------------------------------------ //
-void AliPMDClusteringV2::Order()
-{
-  // Sorting algorithm
-  // sorts the ADC values from higher to lower
-  //
-  double dd[kNMX];
-  // matrix fEdepCell converted into
-  // one dimensional array dd. adum a place holder for double
-  int i, j, i1, i2, iord1[kNMX];
-  // information of
-  // ordering is stored in iord1, original array not ordered
-  //
-  // define arrays dd and iord1
-  for(i1=0; i1 < kNDIMX; i1++)
-    {
-      for(i2=0; i2 < kNDIMY; i2++)
-       {
-         i        = i1 + i2*kNDIMX;
-         iord1[i] = i;
-         dd[i]    = fEdepCell[i1][i2];
-       }
-    }
-  // sort and store sorting information in iord1
+         Int_t irow = celldataX[ihit];
+         Int_t icol = celldataY[ihit];
 
-  TMath::Sort(kNMX,dd,iord1);
+         if ((irow >= 0 && irow < 48) && (icol >= 0 && icol < 96))
+           {
+             celldataTr[ihit]  = celltrack[irow][icol];
+             celldataPid[ihit] = cellpid[irow][icol];
+             celldataAdc[ihit] = (Float_t) celladc[irow][icol];
+           }
+         else
+           {
+             celldataTr[ihit]  = -1;
+             celldataPid[ihit] = -1;
+             celldataAdc[ihit] = -1;
+           }
 
-  // store the sorted information in fIord for later use
-  for(i=0; i<kNMX; i++)
-    {
-      j  = iord1[i];
-      i2 = j/kNDIMX;
-      i1 = j-i2*kNDIMX;
-      fIord[0][i]=i1;
-      fIord[1][i]=i2;
+       }
+
+      pmdcl = new AliPMDcluster(idet, ismn, clusdata, celldataX, celldataY,
+                               celldataTr, celldataPid, celldataAdc);
+      pmdcont->Add(pmdcl);
     }
+  fPMDclucont->Delete();
 }
 // ------------------------------------------------------------------------ //
-Int_t AliPMDClusteringV2::CrClust(Double_t ave, Double_t cutoff, Int_t nmx1)
+Int_t AliPMDClusteringV2::CrClust(Double_t ave, Double_t cutoff, Int_t nmx1,
+                                 Int_t iord1[], Double_t edepcell[])
 {
   // Does crude clustering
   // Finds out only the big patch by just searching the
   // connected cells
   //
 
-  int i,j,k,id1,id2,icl, numcell;
-  int jd1,jd2, icell, cellcount;
-  int clust[2][5000];
-  static int neibx[6]={1,0,-1,-1,0,1}, neiby[6]={0,1,1,0,-1,-1};
+  Int_t i = 0, j = 0, k = 0, id1 =0, id2 = 0, icl = 0, numcell = 0;
+  Int_t jd1 = 0, jd2 = 0, icell = 0, cellcount = 0;
+  Int_t clust[2][5000];
+  static Int_t neibx[6] = {1,0,-1,-1,0,1}, neiby[6] = {0,1,1,0,-1,-1};
 
   // neibx and neiby define ( incremental ) (i,j) for the neighbours of a
   // cell. There are six neighbours.
   // cellcount --- total number of cells having nonzero ener dep
   // numcell --- number of cells in a given supercluster
-  // ofstream ofl0("cells_loc",ios::out);
-  // initialize fInfocl[2][kNDIMX][kNDIMY]
-
+  
   AliDebug(1,Form("kNMX = %d nmx1 = %d kNDIMX = %d kNDIMY = %d ave = %f cutoff = %f",kNMX,nmx1,kNDIMX,kNDIMY,ave,cutoff));
   
-  for (j=0; j < kNDIMX; j++){
-    for(k=0; k < kNDIMY; k++){
-      fInfocl[0][j][k] = 0;
-      fInfocl[1][j][k] = 0;
+  for (j=0; j < kNDIMX; j++)
+    {
+      for(k=0; k < kNDIMY; k++)
+       {
+         fInfocl[0][j][k] = 0;
+         fInfocl[1][j][k] = 0;
+       }
+    }
+  for(i=0; i < kNMX; i++)
+    {
+      fInfcl[0][i] = -1;
+      
+      j  = iord1[i];
+      id2 = j/kNDIMX;
+      id1 = j-id2*kNDIMX;
+      
+      if(edepcell[j] <= cutoff)
+       {
+         fInfocl[0][id1][id2] = -1;
+       }
     }
-  }
-  for(i=0; i < kNMX; i++){
-    fInfcl[0][i] = -1;
-    id1=fIord[0][i];
-    id2=fIord[1][i];
-    if(fEdepCell[id1][id2] <= cutoff){fInfocl[0][id1][id2]=-1;}
-  }
   // ---------------------------------------------------------------
   // crude clustering begins. Start with cell having largest adc
   // count and loop over the cells in descending order of adc count
   // ---------------------------------------------------------------
-  icl=-1;
-  cellcount=-1;
-  for(icell=0; icell <= nmx1; icell++){
-    id1=fIord[0][icell];
-    id2=fIord[1][icell];
-    if(fInfocl[0][id1][id2] == 0 ){
-      // ---------------------------------------------------------------
-      // icl -- cluster #, numcell -- # of cells in it, clust -- stores
-      // coordinates of the cells in a cluster, fInfocl[0][i1][i2] is 1 for
-      // primary and 2 for secondary cells,
-      // fInfocl[1][i1][i2] stores cluster #
-      // ---------------------------------------------------------------
-      icl=icl+1;
-      numcell=0;
-      cellcount = cellcount + 1;
-      fInfocl[0][id1][id2]=1;
-      fInfocl[1][id1][id2]=icl;
-      fInfcl[0][cellcount]=icl;
-      fInfcl[1][cellcount]=id1;
-      fInfcl[2][cellcount]=id2;
-
-      clust[0][numcell]=id1;
-      clust[1][numcell]=id2;
-      for(i=1; i<5000; i++)clust[0][i] = -1;
-      // ---------------------------------------------------------------
-      // check for adc count in neib. cells. If ne 0 put it in this clust
-      // ---------------------------------------------------------------
-      for(i=0; i<6; i++){
-       jd1=id1+neibx[i];
-       jd2=id2+neiby[i];
-       if( (jd1 >= 0 && jd1 < kNDIMX) && (jd2 >= 0 && jd2 < kNDIMY) &&
-           fInfocl[0][jd1][jd2] == 0){
-         numcell=numcell+1;
-         fInfocl[0][jd1][jd2]=2;
-         fInfocl[1][jd1][jd2]=icl;
-         clust[0][numcell]=jd1;
-         clust[1][numcell]=jd2;
-         cellcount=cellcount+1;
-         fInfcl[0][cellcount]=icl;
-         fInfcl[1][cellcount]=jd1;
-         fInfcl[2][cellcount]=jd2;
-       }
-      }
-      // ---------------------------------------------------------------
-      // check adc count for neighbour's neighbours recursively and
-      // if nonzero, add these to the cluster.
-      // ---------------------------------------------------------------
-      for(i=1;i < 5000;i++){
-       if(clust[0][i] != -1){
-         id1=clust[0][i];
-         id2=clust[1][i];
-         for(j=0; j<6 ; j++){
-           jd1=id1+neibx[j];
-           jd2=id2+neiby[j];
+  icl       = -1;
+  cellcount = -1;
+  for(icell=0; icell <= nmx1; icell++)
+    {
+      j  = iord1[icell];
+      id2 = j/kNDIMX;
+      id1 = j-id2*kNDIMX;
+      if(fInfocl[0][id1][id2] == 0 )
+       {
+         // ---------------------------------------------------------------
+         // icl -- cluster #, numcell -- # of cells in it, clust -- stores
+         // coordinates of the cells in a cluster, fInfocl[0][i1][i2] is 1 for
+         // primary and 2 for secondary cells,
+         // fInfocl[1][i1][i2] stores cluster #
+         // ---------------------------------------------------------------
+         icl++;
+         numcell = 0;
+         cellcount++;
+         fInfocl[0][id1][id2]  = 1;
+         fInfocl[1][id1][id2]  = icl;
+         fInfcl[0][cellcount]  = icl;
+         fInfcl[1][cellcount]  = id1;
+         fInfcl[2][cellcount]  = id2;
+
+         clust[0][numcell]     = id1;
+         clust[1][numcell]     = id2;
+         for(i = 1; i < 5000; i++)
+           {
+             clust[0][i] = -1;
+           }
+         // ---------------------------------------------------------------
+         // check for adc count in neib. cells. If ne 0 put it in this clust
+         // ---------------------------------------------------------------
+         for(i = 0; i < 6; i++)
+           {
+           jd1 = id1 + neibx[i];
+           jd2 = id2 + neiby[i];
            if( (jd1 >= 0 && jd1 < kNDIMX) && (jd2 >= 0 && jd2 < kNDIMY) &&
-               fInfocl[0][jd1][jd2] == 0 ){
-             fInfocl[0][jd1][jd2] = 2;
-             fInfocl[1][jd1][jd2] = icl;
-             numcell              = numcell + 1;
-             clust[0][numcell]    = jd1;
-             clust[1][numcell]    = jd2;
-             cellcount            = cellcount+1;
-             fInfcl[0][cellcount] = icl;
-             fInfcl[1][cellcount] = jd1;
-             fInfcl[2][cellcount] = jd2;
+               fInfocl[0][jd1][jd2] == 0)
+             {
+               numcell++;
+               fInfocl[0][jd1][jd2] = 2;
+               fInfocl[1][jd1][jd2] = icl;
+               clust[0][numcell]    = jd1;
+               clust[1][numcell]    = jd2;
+               cellcount++;
+               fInfcl[0][cellcount] = icl;
+               fInfcl[1][cellcount] = jd1;
+               fInfcl[2][cellcount] = jd2;
+             }
+           }
+         // ---------------------------------------------------------------
+         // check adc count for neighbour's neighbours recursively and
+         // if nonzero, add these to the cluster.
+         // ---------------------------------------------------------------
+         for(i = 1;i < 5000; i++)
+           { 
+             if(clust[0][i] != -1)
+               {
+                 id1 = clust[0][i];
+                 id2 = clust[1][i];
+                 for(j = 0; j < 6 ; j++)
+                   {
+                     jd1 = id1 + neibx[j];
+                     jd2 = id2 + neiby[j];
+                     if( (jd1 >= 0 && jd1 < kNDIMX) && 
+                         (jd2 >= 0 && jd2 < kNDIMY) 
+                         && fInfocl[0][jd1][jd2] == 0 )
+                       {
+                         fInfocl[0][jd1][jd2] = 2;
+                         fInfocl[1][jd1][jd2] = icl;
+                         numcell++;
+                         clust[0][numcell]    = jd1;
+                         clust[1][numcell]    = jd2;
+                         cellcount++;
+                         fInfcl[0][cellcount] = icl;
+                         fInfcl[1][cellcount] = jd1;
+                         fInfcl[2][cellcount] = jd2;
+                       }
+                   }
+               }
            }
-         }
        }
-      }
     }
-  }
-  //  for(icell=0; icell<=cellcount; icell++){
-  //    ofl0 << fInfcl[0][icell] << " " << fInfcl[1][icell] << " " <<
-  //      fInfcl[2][icell] << endl;
-  //  }
   return cellcount;
 }
 // ------------------------------------------------------------------------ //
-void AliPMDClusteringV2::RefClust(Int_t incr)
+  void AliPMDClusteringV2::RefClust(Int_t incr, Double_t edepcell[])
 {
   // Does the refining of clusters
   // Takes the big patch and does gaussian fitting and
   // finds out the more refined clusters
-  //
 
-  const Int_t kndim = 4500;
+  const Float_t ktwobysqrt3 = 1.1547;
+  const Int_t   kNmaxCell   = 19;
+
+  AliPMDcludata *pmdcludata = 0;
 
-  int i, j, k, i1, i2, id, icl, itest;
-  int ihld;
-  int ig, nsupcl;
-  int ncl[kndim], iord[kndim];
+  Int_t i12 = 0;
+  Int_t i = 0, j = 0, k = 0;
+  Int_t i1 = 0, i2 = 0, id = 0, icl = 0, itest = 0, ihld = 0;
+  Int_t ig = 0, nsupcl = 0, clno = 0, clX = 0, clY = 0;
+  Int_t clxy[kNmaxCell];
 
-  double x1, y1, z1, x2, y2, z2;
-  double rr;
+  Float_t  clusdata[6] = {0.,0.,0.,0.,0.,0.};
+  Double_t x1 = 0., y1 = 0., z1 = 0., x2 = 0., y2 = 0., z2 = 0., rr = 0.;
 
-  double x[kndim], y[kndim], z[kndim];
-  double xc[kndim], yc[kndim], zc[kndim], cells[kndim];
-  double rcl[kndim], rcs[kndim];
+  Int_t kndim = incr + 1;
 
-  // fClno counts the final clusters
+  TArrayI testncl;
+  TArrayI testindex;
+
+  Int_t    *ncl, *iord;
+
+  Double_t *x, *y, *z, *xc, *yc, *zc, *cells, *rcl, *rcs;
+
+  ncl   = new Int_t [kndim];
+  iord  = new Int_t [kndim];
+  x     = new Double_t [kndim];
+  y     = new Double_t [kndim];
+  z     = new Double_t [kndim];
+  xc    = new Double_t [kndim];
+  yc    = new Double_t [kndim];
+  zc    = new Double_t [kndim];
+  cells = new Double_t [kndim];
+  rcl   = new Double_t [kndim];
+  rcs   = new Double_t [kndim];
+  
+  for(Int_t kk = 0; kk < 15; kk++)
+    {
+      if( kk < 6 )clusdata[kk] = 0.;
+    }
+   
   // nsupcl =  # of superclusters; ncl[i]= # of cells in supercluster i
   // x, y and z store (x,y) coordinates of and energy deposited in a cell
   // xc, yc store (x,y) coordinates of the cluster center
-  // zc stores the energy deposited in a cluster
-  // rc is cluster radius
-  // finally the cluster information is put in 2-dimensional array clusters
-  // ofstream ofl1("checking.5",ios::app);
+  // zc stores the energy deposited in a cluster, rc is cluster radius
 
-  fClno  = -1;
+  clno   = -1;
   nsupcl = -1;
-  for(i=0; i<4500; i++){ncl[i]=-1;}
-  for(i=0; i<incr; i++){
-    if(fInfcl[0][i] != nsupcl){ nsupcl=nsupcl+1; }
-    if (nsupcl > 4500) {
-      AliWarning("RefClust: Too many superclusters!");
-      nsupcl = 4500;
-      break;
-    }
-    ncl[nsupcl]=ncl[nsupcl]+1;
-  }
 
+  for(i = 0; i < kndim; i++)
+    {
+      ncl[i] = -1;
+    }
+  for(i = 0; i <= incr; i++)
+    {
+      if(fInfcl[0][i] != nsupcl)
+       {
+         nsupcl++;
+       }
+      if (nsupcl > 4500) 
+       {
+         AliWarning("RefClust: Too many superclusters!");
+         nsupcl = 4500;
+         break;
+       }
+      ncl[nsupcl]++;
+    }
+  
   AliDebug(1,Form("Number of cells = %d Number of Superclusters = %d",
                  incr+1,nsupcl+1));
-
-  id=-1;
-  icl=-1;
-  for(i=0; i<nsupcl; i++){
-    if(ncl[i] == 0){
-      id++;
-      icl++;
-      // one  cell super-clusters --> single cluster
-      // cluster center at the centyer of the cell
-      // cluster radius = half cell dimension
-      if (fClno >= 5000) {
-       AliWarning("RefClust: Too many clusters! more than 5000");
-       return;
-      }
-      fClno++;
-      i1 = fInfcl[1][id];
-      i2 = fInfcl[2][id];
-      fClusters[0][fClno] = fCoord[0][i1][i2];
-      fClusters[1][fClno] = fCoord[1][i1][i2];
-      fClusters[2][fClno] = fEdepCell[i1][i2];
-      fClusters[3][fClno] = 1.;
-      fClusters[4][fClno] = 0.0;
-      fClusters[5][fClno] = 0.0;
-      //ofl1 << icl << " " << fCoord[0][i1][i2] << " " << fCoord[1][i1][i2] <<
-      //" " << fEdepCell[i1][i2] << " " << fClusters[3][fClno] <<endl;
-    }else if(ncl[i] == 1){
-      // two cell super-cluster --> single cluster
-      // cluster center is at ener. dep.-weighted mean of two cells
-      // cluster radius == half cell dimension
-      id++;
-      icl++;
-      if (fClno >= 5000) {
-       AliWarning("RefClust: Too many clusters! more than 5000");
-       return;
-      }
-      fClno++;
-      i1   = fInfcl[1][id];
-      i2   = fInfcl[2][id];
-      x1   = fCoord[0][i1][i2];
-      y1   = fCoord[1][i1][i2];
-      z1   = fEdepCell[i1][i2];
-
-      id++;
-      i1   = fInfcl[1][id];
-      i2   = fInfcl[2][id];
-      x2   = fCoord[0][i1][i2];
-      y2   = fCoord[1][i1][i2];
-      z2   = fEdepCell[i1][i2];
-
-      fClusters[0][fClno] = (x1*z1+x2*z2)/(z1+z2);
-      fClusters[1][fClno] = (y1*z1+y2*z2)/(z1+z2);
-      fClusters[2][fClno] = z1+z2;
-      fClusters[3][fClno] = 2.;
-      fClusters[4][fClno] = sqrt(z1*z2)/(z1+z2);
-      fClusters[5][fClno] = 0;  // sigma large nonzero, sigma small zero
-
-      //ofl1 << icl << " " << fClusters[0][fClno] << " " << fClusters[1][fClno]
-      //   << " " << fClusters[2][fClno] << " " <<fClusters[3][fClno] <<endl;
-    }
-    else{
-      id      = id + 1;
-      iord[0] = 0;
-      // super-cluster of more than two cells - broken up into smaller
-      // clusters gaussian centers computed. (peaks separated by > 1 cell)
-      // Begin from cell having largest energy deposited This is first
-      // cluster center
-      // *****************************************************************
-      // NOTE --- POSSIBLE MODIFICATION: ONE MAY NOT BREAKING SUPERCLUSTERS
-      // IF NO. OF CELLS IS NOT TOO LARGE ( SAY 5 OR 6 )
-      // SINCE WE EXPECT THE SUPERCLUSTER 
-      // TO BE A SINGLE CLUSTER
-      //*******************************************************************
-
-      i1      = fInfcl[1][id];
-      i2      = fInfcl[2][id];
-      x[0]    = fCoord[0][i1][i2];
-      y[0]    = fCoord[1][i1][i2];
-      z[0]    = fEdepCell[i1][i2];
-      iord[0] = 0;
-      for(j=1;j<=ncl[i];j++){
-
-       id      = id + 1;
-       i1      = fInfcl[1][id];
-       i2      = fInfcl[2][id];
-       iord[j] = j;
-       x[j]    = fCoord[0][i1][i2];
-       y[j]    = fCoord[1][i1][i2];
-       z[j]    = fEdepCell[i1][i2];
-      }
-      // arranging cells within supercluster in decreasing order
-      for(j=1;j<=ncl[i];j++)
+  
+  id  = -1;
+  icl = -1;
+  for(i = 0; i <= nsupcl; i++)
+    {
+      if(ncl[i] == 0)
        {
-         itest = 0;
-         ihld  = iord[j];
-         for(i1=0; i1<j; i1++)
+         id++;
+         icl++;
+         // one  cell super-clusters --> single cluster
+         // cluster center at the centyer of the cell
+         // cluster radius = half cell dimension
+         if (clno >= 5000) 
            {
-             if(itest == 0 && z[iord[i1]] < z[ihld])
-               {
-                 itest = 1;
-                 for(i2=j-1;i2>=i1;i2--)
-                   {
-                     iord[i2+1] = iord[i2];
-                   }
-                 iord[i1] = ihld;
-               }
+             AliWarning("RefClust: Too many clusters! more than 5000");
+             return;
            }
-       }
+         clno++;
+         i1          = fInfcl[1][id];
+         i2          = fInfcl[2][id];
+         i12         = i1 + i2*kNDIMX;
+         clusdata[0] = fCoord[0][i1][i2];
+         clusdata[1] = fCoord[1][i1][i2];
+         clusdata[2] = edepcell[i12];
+         clusdata[3] = 1.;
+         clusdata[4] = 0.0;
+         clusdata[5] = 0.0;
+         
+         //cell information
+         
+         clY = (Int_t)((ktwobysqrt3*fCoord[1][i1][i2])*10);
+         clX = (Int_t)((fCoord[0][i1][i2] - clY/20.)*10);
+         clxy[0] = clX*10000 + clY ;
 
-      // compute the number of clusters and their centers ( first
-      // guess )
-      // centers must be separated by cells having smaller ener. dep.
-      // neighbouring centers should be either strong or well-separated
-      ig     = 0;
-      xc[ig] = x[iord[0]];
-      yc[ig] = y[iord[0]];
-      zc[ig] = z[iord[0]];
-      for(j=1;j<=ncl[i];j++){
-       itest = -1;
-       x1    = x[iord[j]];
-       y1    = y[iord[j]];
-       for(k=0;k<=ig;k++){
-         x2 = xc[k];
-         y2 = yc[k];
-         rr = Distance(x1,y1,x2,y2);
-         //***************************************************************
-         // finetuning cluster splitting
-         // the numbers zc/4 and zc/10 may need to be changed. 
-         // Also one may need to add one more layer because our 
-         // cells are smaller in absolute scale
-         //****************************************************************
-
-
-         if( rr >= 1.1 && rr < 1.8 && z[iord[j]] > zc[k]/4.)
-           itest++;
-         if( rr >= 1.8 && rr < 2.1 && z[iord[j]] > zc[k]/10.)
-           itest++;
-         if( rr >= 2.1)itest++;
-       }
-       if(itest == ig){
-         ig++;
-         xc[ig] = x1;
-         yc[ig] = y1;
-         zc[ig] = z[iord[j]];
+         for(Int_t icltr = 1; icltr < kNmaxCell; icltr++)
+           {
+             clxy[icltr] = -1;
+           }
+         pmdcludata  = new AliPMDcludata(clusdata,clxy);
+         fPMDclucont->Add(pmdcludata);
+         
+         
        }
-      }
-
-      ClustDetails(ncl[i], ig, x[0], y[0] ,z[0], xc[0], yc[0], zc[0],
-                  rcl[0], rcs[0], cells[0]);
-
-      icl = icl + ig + 1;
-
-      for(j=0; j<=ig; j++)
+      else if(ncl[i] == 1)
        {
-         if (fClno >= 5000)
+         // two cell super-cluster --> single cluster
+         // cluster center is at ener. dep.-weighted mean of two cells
+         // cluster radius == half cell dimension
+         id++;
+         icl++;
+         if (clno >= 5000) 
            {
              AliWarning("RefClust: Too many clusters! more than 5000");
              return;
            }
-         fClno++;
-         fClusters[0][fClno] = xc[j];
-         fClusters[1][fClno] = yc[j];
-         fClusters[2][fClno] = zc[j];
-         fClusters[4][fClno] = rcl[j];
-         fClusters[5][fClno] = rcs[j];
-         if(ig == 0)
-           {
-             fClusters[3][fClno] = ncl[i];
-           }
-         else
+         clno++;
+         i1   = fInfcl[1][id];
+         i2   = fInfcl[2][id];
+         i12  = i1 + i2*kNDIMX;
+         
+         x1   = fCoord[0][i1][i2];
+         y1   = fCoord[1][i1][i2];
+         z1   = edepcell[i12];
+         
+         id++;
+         i1   = fInfcl[1][id];
+         i2   = fInfcl[2][id];
+         i12  = i1 + i2*kNDIMX;
+         
+         x2   = fCoord[0][i1][i2];
+         y2   = fCoord[1][i1][i2];
+         z2   = edepcell[i12];
+         
+         clusdata[0] = (x1*z1+x2*z2)/(z1+z2);
+         clusdata[1] = (y1*z1+y2*z2)/(z1+z2);
+         clusdata[2] = z1+z2;
+         clusdata[3] = 2.;
+         clusdata[4] = (TMath::Sqrt(z1*z2))/(z1+z2);
+         clusdata[5] = 0.0;
+
+          clY = (Int_t)((ktwobysqrt3*y1)*10);
+         clX = (Int_t)((x1 - clY/20.)*10);
+         clxy[0] = clX*10000 + clY ;
+
+         clY = (Int_t)((ktwobysqrt3*y2)*10);
+         clX = (Int_t)((x2 - clY/20.)*10);
+         clxy[1] = clX*10000 + clY ;
+
+         for(Int_t icltr = 2; icltr < kNmaxCell; icltr++)
            {
-             fClusters[3][fClno] = cells[j];
+             clxy[icltr] = -1;
            }
+         pmdcludata  = new AliPMDcludata(clusdata, clxy);
+         fPMDclucont->Add(pmdcludata);
        }
+      else{
+       id++;
+       iord[0] = 0;
+       // super-cluster of more than two cells - broken up into smaller
+       // clusters gaussian centers computed. (peaks separated by > 1 cell)
+       // Begin from cell having largest energy deposited This is first
+       // cluster center
+       // *****************************************************************
+       // NOTE --- POSSIBLE MODIFICATION: ONE MAY NOT BREAKING SUPERCLUSTERS
+       // IF NO. OF CELLS IS NOT TOO LARGE ( SAY 5 OR 6 )
+       // SINCE WE EXPECT THE SUPERCLUSTER 
+       // TO BE A SINGLE CLUSTER
+       //*******************************************************************
+       
+       i1      = fInfcl[1][id];
+       i2      = fInfcl[2][id];
+       i12     = i1 + i2*kNDIMX;
+       
+       x[0]    = fCoord[0][i1][i2];
+       y[0]    = fCoord[1][i1][i2];
+       z[0]    = edepcell[i12];
+       
+       iord[0] = 0;
+       for(j = 1; j <= ncl[i]; j++)
+         {
+           
+           id++;
+           i1      = fInfcl[1][id];
+           i2      = fInfcl[2][id];
+           i12     = i1 + i2*kNDIMX;
+           iord[j] = j;
+           x[j]    = fCoord[0][i1][i2];
+           y[j]    = fCoord[1][i1][i2];
+           z[j]    = edepcell[i12];
+         }
+       
+       // arranging cells within supercluster in decreasing order
+       for(j = 1; j <= ncl[i];j++)
+         {
+           itest = 0;
+           ihld  = iord[j];
+           for(i1 = 0; i1 < j; i1++)
+             {
+               if(itest == 0 && z[iord[i1]] < z[ihld])
+                 {
+                   itest = 1;
+                   for(i2 = j-1;i2 >= i1;i2--)
+                     {
+                       iord[i2+1] = iord[i2];
+                     }
+                   iord[i1] = ihld;
+                 }
+             }
+         }
+       
+       
+       // compute the number of clusters and their centers ( first
+       // guess )
+       // centers must be separated by cells having smaller ener. dep.
+       // neighbouring centers should be either strong or well-separated
+       ig     = 0;
+       xc[ig] = x[iord[0]];
+       yc[ig] = y[iord[0]];
+       zc[ig] = z[iord[0]];
+       for(j = 1; j <= ncl[i]; j++)
+         {
+           itest = -1;
+           x1    = x[iord[j]];
+           y1    = y[iord[j]];
+           for(k = 0; k <= ig; k++)
+             {
+               x2 = xc[k];
+               y2 = yc[k];
+               rr = Distance(x1,y1,x2,y2);
+               //************************************************************
+               // finetuning cluster splitting
+               // the numbers zc/4 and zc/10 may need to be changed. 
+               // Also one may need to add one more layer because our 
+               // cells are smaller in absolute scale
+               //************************************************************
+               
+               
+               if( rr >= 1.1 && rr < 1.8 && z[iord[j]] > zc[k]/4.) itest++;
+               if( rr >= 1.8 && rr < 2.1 && z[iord[j]] > zc[k]/10.) itest++;
+               if( rr >= 2.1)itest++;
+             }
+           
+           if(itest == ig)
+             {
+               ig++;
+               xc[ig] = x1;
+               yc[ig] = y1;
+               zc[ig] = z[iord[j]];
+             }
+         }
+       ClustDetails(ncl[i], ig, x, y ,z, xc, yc, zc, rcl, rcs, cells, 
+                    testncl, testindex);
+       
+       Int_t pp = 0;
+       for(j = 0; j <= ig; j++)
+         { 
+           clno++;
+           if (clno >= 5000)
+             {
+               AliWarning("RefClust: Too many clusters! more than 5000");
+               return;
+             }
+           clusdata[0] = xc[j];
+           clusdata[1] = yc[j];
+           clusdata[2] = zc[j];
+           clusdata[4] = rcl[j];
+           clusdata[5] = rcs[j];
+           if(ig == 0)
+             {
+               clusdata[3] = ncl[i] + 1;
+             }
+           else
+             {
+               clusdata[3] = cells[j];
+             }
+           // cell information
+           Int_t ncellcls =  testncl[j];
+           if( ncellcls < kNmaxCell )
+             {
+               for(Int_t kk = 1; kk <= ncellcls; kk++)
+                 {
+                   Int_t ll =  testindex[pp];
+                    clY = (Int_t)((ktwobysqrt3*y[ll])*10);
+                   clX = (Int_t)((x[ll] - clY/20.)*10);
+                   clxy[kk-1] = clX*10000 + clY ;
 
-
+                   pp++;
+                 }
+               for(Int_t icltr = ncellcls ; icltr < kNmaxCell; icltr++)
+                 {
+                   clxy[icltr] = -1;
+                 }
+             }
+           pmdcludata = new AliPMDcludata(clusdata, clxy);
+           fPMDclucont->Add(pmdcludata);
+         }
+       testncl.Set(0);
+       testindex.Set(0);
+      }
     }
-  }
+  delete [] ncl;
+  delete [] iord;
+  delete [] x;
+  delete [] y;
+  delete [] z;
+  delete [] xc;
+  delete [] yc;
+  delete [] zc;
+  delete [] cells;
+  delete [] rcl;
+  delete [] rcs;
 }
-
-
 // ------------------------------------------------------------------------ //
-
-void AliPMDClusteringV2::ClustDetails(Int_t ncell, Int_t nclust,
-                                     Double_t &x, Double_t &y, Double_t &z,
-                                     Double_t &xc, Double_t &yc, Double_t &zc,
-                                     Double_t &rcl, Double_t &rcs,
-                                     Double_t &cells)
+void AliPMDClusteringV2::ClustDetails(Int_t ncell, Int_t nclust, Double_t x[], 
+                                     Double_t y[], Double_t z[],Double_t xc[],
+                                     Double_t yc[], Double_t zc[],
+                                     Double_t rcl[], Double_t rcs[], 
+                                     Double_t cells[], TArrayI &testncl,
+                                     TArrayI &testindex)
 {
   // function begins
   //
-  
-  const Int_t kndim1 = 4500;
-  const Int_t kndim2 = 10;
-  const Int_t kndim3 = 100;
 
-  int i, j, k, i1, i2;
-  int cluster[kndim1][kndim2];
-  
-  double x1, y1, x2, y2, rr;
-  double sumx, sumy, sumxy, sumxx;
-  double sum, sum1, sumyy;
-  double b, c, r1, r2;
-
-  double xx[kndim1], yy[kndim1], zz[kndim1];
-  double xxc[kndim1], yyc[kndim1];
-
-  double str[kndim1];
-
-  double str1[kndim1];
-  double xcl[kndim1], ycl[kndim1], cln[kndim1];
-  double clustcell[kndim1][kndim3];
-
-  for(i=0; i<=nclust; i++){
-   xxc[i]=*(&xc+i); 
-   yyc[i]=*(&yc+i); 
-   str[i]=0.; 
-   str1[i]=0.;
-  }
-  for(i=0; i<=ncell; i++){
-    xx[i]=*(&x+i); 
-    yy[i]=*(&y+i); 
-    zz[i]=*(&z+i);
-  }
-  // INITIALIZE 
-  for(i=0; i<4500; i++){
-    for(j=0; j<100; j++){
-      clustcell[i][j]=0.;
-    }
-  }
-
-  // INITIALIZE
-  for(i=0;i<4500;i++){
-    for(j=0;j<10;j++){
-      cluster[i][j]=0;
+  Int_t kndim1 = ncell + 1;//ncell
+  Int_t kndim2 = 20;
+  Int_t kndim3 = nclust + 1;//nclust
+
+  Int_t    i = 0, j = 0, k = 0, i1 = 0, i2 = 0;
+  Double_t x1 = 0., y1 = 0., x2 = 0., y2 = 0.;
+  Double_t rr = 0., b = 0., c = 0., r1 = 0., r2 = 0.;
+  Double_t sumx = 0., sumy = 0., sumxy = 0.;
+  Double_t sumxx = 0., sum = 0., sum1 = 0., sumyy = 0.;
+
+  Double_t  *str, *str1, *xcl, *ycl, *cln; 
+  Int_t    **cell;
+  Int_t    ** cluster;
+  Double_t **clustcell;
+  str  = new Double_t [kndim3];
+  str1 = new Double_t [kndim3];
+  xcl  = new Double_t [kndim3];
+  ycl  = new Double_t [kndim3];
+  cln  = new Double_t [kndim3];
+
+  clustcell = new Double_t *[kndim3];
+  cell      = new Int_t    *[kndim3];
+  cluster   = new Int_t    *[kndim1];
+  for(i = 0; i < kndim1; i++)
+    {
+      cluster[i] = new Int_t [kndim2];
     }
-  }
-
-
-  if(nclust > 0){
-    // more than one cluster
-    // checking cells shared between several  clusters.
-    // First check if the cell is within
-    // one cell unit ( nearest neighbour). Else, 
-    // if it is within 1.74 cell units ( next nearest )
-    // Else if it is upto 2 cell units etc.
-
-    for (i=0; i<=ncell; i++){
-      x1            = xx[i];
-      y1            = yy[i];
-      cluster[i][0] = 0;
-      // distance <= 1 cell unit
-      for(j=0; j<=nclust; j++)
+  
+  for(i = 0; i < kndim3; i++)
+    {
+      str[i]  = 0;
+      str1[i] = 0;
+      xcl[i]  = 0;
+      ycl[i]  = 0;
+      cln[i]  = 0;
+      
+      cell[i]    = new Int_t [kndim2];
+      clustcell[i] = new Double_t [kndim1];      
+      for(j = 0; j < kndim1; j++)
        {
-         x2 = xxc[j];
-         y2 = yyc[j];
-         rr = Distance(x1, y1, x2, y2);
-         if(rr <= 1.)
-           {
-             cluster[i][0]++;
-             i1             = cluster[i][0];
-             cluster[i][i1] = j;
-           }
+         clustcell[i][j] = 0;
        }
-      // next nearest neighbour
-      if(cluster[i][0] == 0)
+      for(j = 0; j < kndim2; j++)
+       {
+         cluster[i][j] = 0;
+         cell[i][j] = 0;
+       }
+    }
+  
+  if(nclust > 0)
+    {
+      // more than one cluster
+      // checking cells shared between several  clusters.
+      // First check if the cell is within
+      // one cell unit ( nearest neighbour). Else, 
+      // if it is within 1.74 cell units ( next nearest )
+      // Else if it is upto 2 cell units etc.
+      
+      for (i = 0; i <= ncell; i++)
        {
-         for(j=0; j<=nclust; j++)
+         x1            = x[i];
+         y1            = y[i];
+         cluster[i][0] = 0;
+
+         // distance <= 1 cell unit
+
+         for(j = 0; j <= nclust; j++)
            {
-             x2 = xxc[j];
-             y2 = yyc[j];
+             x2 = xc[j];
+             y2 = yc[j];
              rr = Distance(x1, y1, x2, y2);
-             if(rr <= sqrt(3.))
+             if(rr <= 1.)
                {
                  cluster[i][0]++;
                  i1             = cluster[i][0];
                  cluster[i][i1] = j;
                }
            }
+         // next nearest neighbour
+         if(cluster[i][0] == 0)
+           {
+             for(j=0; j<=nclust; j++)
+               {
+                 x2 = xc[j];
+                 y2 = yc[j];
+                 rr = Distance(x1, y1, x2, y2);
+                 if(rr <= TMath::Sqrt(3.))
+                   {
+                     cluster[i][0]++;
+                     i1             = cluster[i][0];
+                     cluster[i][i1] = j;
+                   }
+               }
+           }
+         // next-to-next nearest neighbour
+         if(cluster[i][0] == 0)
+           {
+             for(j=0; j<=nclust; j++)
+               {
+                 x2 = xc[j];
+                 y2 = yc[j];
+                 rr = Distance(x1, y1, x2, y2);
+                 if(rr <= 2.)
+                   {
+                     cluster[i][0]++;
+                     i1             = cluster[i][0];
+                     cluster[i][i1] = j;
+                   }
+               }
+           }
+         // one more
+         if(cluster[i][0] == 0)
+           {
+             for(j = 0; j <= nclust; j++)
+               {
+                 x2 = xc[j];
+                 y2 = yc[j];
+                 rr = Distance(x1, y1, x2, y2);
+                 if(rr <= 2.7)
+                   {
+                     cluster[i][0]++;
+                     i1             = cluster[i][0];
+                     cluster[i][i1] = j;
+                   }
+               }
+           }
        }
-      // next-to-next nearest neighbour
-      if(cluster[i][0] == 0)
+      
+      // computing cluster strength. Some cells are shared.
+      for(i = 0; i <= ncell; i++)
        {
-         for(j=0; j<=nclust; j++)
+         if(cluster[i][0] != 0)
            {
-             x2 = xxc[j];
-             y2 = yyc[j];
-             rr = Distance(x1, y1, x2, y2);
-             if(rr <= 2.)
+             i1 = cluster[i][0];
+             for(j = 1; j <= i1; j++)
                {
-                 cluster[i][0]++;
-                 i1 = cluster[i][0];
-                 cluster[i][i1] = j;
+                 i2       = cluster[i][j];
+                 str[i2] += z[i]/i1;
                }
            }
        }
-      // one more
-      if(cluster[i][0] == 0)
+      
+      for(k = 0; k < 5; k++)
        {
-         for(j=0; j<=nclust; j++)
+         for(i = 0; i <= ncell; i++)
            {
-             x2 = xxc[j];
-             y2 = yyc[j];
-             rr = Distance(x1, y1, x2, y2);
-             if(rr <= 2.7)
+             if(cluster[i][0] != 0)
                {
-                 cluster[i][0]++;
-                 i1 = cluster[i][0];
-                 cluster[i][i1] = j;
+                 i1=cluster[i][0];
+                 sum=0.;
+                 for(j = 1; j <= i1; j++)
+                   {
+                     sum += str[cluster[i][j]];
+                   }
+                 
+                 for(j = 1; j <= i1; j++)
+                   {
+                     i2               = cluster[i][j]; 
+                     str1[i2]        +=  z[i]*str[i2]/sum;
+                     clustcell[i2][i] = z[i]*str[i2]/sum;
+                   }
                }
            }
+         
+         
+         for(j = 0; j <= nclust; j++)
+           {
+             str[j]  = str1[j];
+             str1[j] = 0.;
+           }
        }
-    }
-
+      
+      for(i = 0; i <= nclust; i++)
+       {
+         sumx = 0.;
+         sumy = 0.;
+         sum  = 0.;
+         sum1 = 0.;
+         for(j = 0; j <= ncell; j++)
+           {
+             if(clustcell[i][j] != 0)
+               {
+                 sumx  +=  clustcell[i][j]*x[j];
+                 sumy  +=  clustcell[i][j]*y[j];
+                 sum   +=  clustcell[i][j];
+                 sum1  +=  clustcell[i][j]/z[j];
+               }
+           }
+         //** xcl and ycl are cluster centroid positions ( center of gravity )
+         
+         xcl[i] = sumx/sum;
+         ycl[i] = sumy/sum;
+         cln[i] = sum1;
+         sumxx = 0.;
+         sumyy = 0.;
+         sumxy = 0.;
+         for(j = 0; j <= ncell; j++)
+           {
+             sumxx += clustcell[i][j]*(x[j]-xcl[i])*(x[j]-xcl[i])/sum;
+             sumyy += clustcell[i][j]*(y[j]-ycl[i])*(y[j]-ycl[i])/sum;
+             sumxy += clustcell[i][j]*(x[j]-xcl[i])*(y[j]-ycl[i])/sum;
+           }
+         b = sumxx+sumyy;
+         c = sumxx*sumyy-sumxy*sumxy;
+         // ******************r1 and r2 are major and minor axes ( r1 > r2 ). 
+         r1 = b/2.+TMath::Sqrt(b*b/4.-c);
+         r2 = b/2.-TMath::Sqrt(b*b/4.-c);
+         // final assignments to proper external variables
+         xc[i]    = xcl[i];
+         yc[i]    = ycl[i];
+         zc[i]    = str[i];
+         cells[i] = cln[i];
+         rcl[i]   = r1;
+         rcs[i]   = r2;
 
-    // computing cluster strength. Some cells are shared.
-    for(i=0; i<=ncell; i++){
-      if(cluster[i][0] != 0){
-       i1 = cluster[i][0];
-       for(j=1; j<=i1; j++){
-         i2      = cluster[i][j];
-         str[i2] = str[i2]+zz[i]/i1;
        }
-      }
-    }
-
-    for(k=0; k<5; k++)
-      {
-       for(i=0; i<=ncell; i++)
-         {
-           if(cluster[i][0] != 0)
-             {
-               i1=cluster[i][0];
-               sum=0.;
-               for(j=1; j<=i1; j++)
-                 {
-                   sum=sum+str[cluster[i][j]];
-                 }
-
-               for(j=1; j<=i1; j++)
-                 {
-                   i2 = cluster[i][j]; 
-                   str1[i2]         = str1[i2] + zz[i]*str[i2]/sum;
-                   clustcell[i2][i] = zz[i]*str[i2]/sum;
-                 }
-             }
-         }
-
-
-         for(j=0; j<=nclust; j++)
+      
+      //To get the cell position in a cluster
+      
+      for(Int_t ii=0; ii<= ncell; ii++)
+       {
+         Int_t jj = cluster[ii][0]; 
+         for(Int_t kk=1; kk<= jj; kk++)
            {
-             str[j]=str1[j];
-             str1[j]=0.;
+             Int_t ll = cluster[ii][kk];
+             cell[ll][0]++;
+             cell[ll][cell[ll][0]] = ii;
            }
-      }
-
-    for(i=0; i<=nclust; i++){
+       }
+      
+      testncl.Set(nclust+1);
+      Int_t counter = 0;
+      
+      for(Int_t ii=0; ii <= nclust; ii++)
+       {
+         testncl[ii] =  cell[ii][0];
+         counter += testncl[ii];
+       }
+      testindex.Set(counter);
+      Int_t ll = 0;
+      for(Int_t ii=0; ii<= nclust; ii++)
+       {
+         for(Int_t jj = 1; jj<= testncl[ii]; jj++)
+           {
+             Int_t kk = cell[ii][jj];
+             testindex[ll] = kk;
+             ll++;
+           }
+       }
+      
+    }
+  else if(nclust == 0)
+    {
       sumx = 0.;
       sumy = 0.;
       sum  = 0.;
       sum1 = 0.;
-      for(j=0; j<=ncell; j++){
-       if(clustcell[i][j] != 0){
-         sumx = sumx+clustcell[i][j]*xx[j];
-         sumy = sumy+clustcell[i][j]*yy[j];
-         sum  = sum+clustcell[i][j];
-         sum1 = sum1+clustcell[i][j]/zz[j];
+      i    = 0;
+      for(j = 0; j <= ncell; j++)
+       {
+         sumx += z[j]*x[j];
+         sumy += z[j]*y[j];
+         sum  += z[j];
+         sum1++;
        }
-      }
-      //***** xcl and ycl are cluster centroid positions ( center of gravity )
-
       xcl[i] = sumx/sum;
       ycl[i] = sumy/sum;
       cln[i] = sum1;
-      sumxx = 0.;
-      sumyy = 0.;
-      sumxy = 0.;
-      for(j=0; j<=ncell; j++){
-       sumxx = sumxx+clustcell[i][j]*(xx[j]-xcl[i])*(xx[j]-xcl[i])/sum;
-       sumyy = sumyy+clustcell[i][j]*(yy[j]-ycl[i])*(yy[j]-ycl[i])/sum;
-       sumxy = sumxy+clustcell[i][j]*(xx[j]-xcl[i])*(yy[j]-ycl[i])/sum;
-      }
-      b = sumxx+sumyy;
-      c = sumxx*sumyy-sumxy*sumxy;
-      // ******************r1 and r2 are major and minor axes ( r1 > r2 ). 
-      r1 = b/2.+sqrt(b*b/4.-c);
-      r2 = b/2.-sqrt(b*b/4.-c);
-      // final assignments to proper external variables
-      *(&xc + i) = xcl[i];
-      *(&yc + i) = ycl[i];
-      *(&zc + i) = str[i];
-      *(&cells + i) = cln[i];
-      *(&rcl+i) = r1;
-      *(&rcs+i) = r2;
+      sumxx  = 0.;
+      sumyy  = 0.;
+      sumxy  = 0.;
+      for(j = 0; j <= ncell; j++)
+       {
+         sumxx += clustcell[i][j]*(x[j]-xcl[i])*(x[j]-xcl[i])/sum;
+         sumyy += clustcell[i][j]*(y[j]-ycl[i])*(y[j]-ycl[i])/sum;
+         sumxy += clustcell[i][j]*(x[j]-xcl[i])*(y[j]-ycl[i])/sum;
+       }
+      b  = sumxx+sumyy;
+      c  = sumxx*sumyy-sumxy*sumxy;
+      r1 = b/2.+ TMath::Sqrt(b*b/4.-c);
+      r2 = b/2.- TMath::Sqrt(b*b/4.-c);
+      
+      // To get the cell position in a cluster
+      testncl.Set(nclust+1);
+      testindex.Set(ncell+1);
+      cell[0][0] = ncell + 1;
+      testncl[0] = cell[0][0];
+      Int_t ll   = 0;
+      for(Int_t ii = 1; ii <= ncell; ii++)
+       {
+         cell[0][ii]=ii;
+         Int_t kk = cell[0][ii];
+         testindex[ll] = kk;
+         ll++;
+       }
+      // final assignments
+      xc[i]    = xcl[i];
+      yc[i]    = ycl[i];
+      zc[i]    = sum;
+      cells[i] = cln[i];
+      rcl[i]   = r1;
+      rcs[i]   = r2;
     }
-  }else{
-    sumx = 0.;
-    sumy = 0.;
-    sum  = 0.;
-    sum1 = 0.;
-    i    = 0;
-    for(j=0; j<=ncell; j++){
-      sumx = sumx+zz[j]*xx[j];
-      sumy = sumy+zz[j]*yy[j];
-      sum  = sum+zz[j];
-      sum1 = sum1+1.;
+  for(i = 0; i < kndim3; i++)
+    {
+      delete [] clustcell[i];
+      delete [] cell[i];
     }
-    xcl[i] = sumx/sum;
-    ycl[i] = sumy/sum;
-    cln[i] = sum1;
-    sumxx  = 0.;
-    sumyy  = 0.;
-    sumxy  = 0.;
-    for(j=0; j<=ncell; j++){
-      sumxx = sumxx+clustcell[i][j]*(xx[j]-xcl[i])*(xx[j]-xcl[i])/sum;
-      sumyy = sumyy+clustcell[i][j]*(yy[j]-ycl[i])*(yy[j]-ycl[i])/sum;
-      sumxy = sumxy+clustcell[i][j]*(xx[j]-xcl[i])*(yy[j]-ycl[i])/sum;
+  delete [] clustcell;
+  delete [] cell;
+  for(i = 0; i <kndim1 ; i++)
+    {
+      delete [] cluster[i];
     }
-    b  = sumxx+sumyy;
-    c  = sumxx*sumyy-sumxy*sumxy;
-    r1 = b/2.+sqrt(b*b/4.-c);
-    r2 = b/2.-sqrt(b*b/4.-c);
-    // final assignments
-    *(&xc + i)    = xcl[i];
-    *(&yc + i)    = ycl[i];
-    *(&zc + i)    = str[i];
-    *(&cells + i) = cln[i];
-    *(&rcl+i)     = r1;
-    *(&rcs+i)     = r2;
-  }
+  delete [] cluster;
+  delete [] str;
+  delete [] str1;
+  delete [] xcl;
+  delete [] ycl;
+  delete [] cln;
 }
 
 // ------------------------------------------------------------------------ //
 Double_t AliPMDClusteringV2::Distance(Double_t x1, Double_t y1,
                                      Double_t x2, Double_t y2)
 {
-  return sqrt((x1-x2)*(x1-x2) + (y1-y2)*(y1-y2));
+  return TMath::Sqrt((x1-x2)*(x1-x2) + (y1-y2)*(y1-y2));
 }
 // ------------------------------------------------------------------------ //
 void AliPMDClusteringV2::SetEdepCut(Float_t decut)
@@ -834,3 +1074,8 @@ void AliPMDClusteringV2::SetEdepCut(Float_t decut)
   fCutoff = decut;
 }
 // ------------------------------------------------------------------------ //
+void AliPMDClusteringV2::SetClusteringParam(Int_t cluspar)
+{
+  fClusParam = cluspar;
+}
+// ------------------------------------------------------------------------ //