class AliITSTPArrayFit: public TObject

 default constructor

 constructor with booking of np points

 copy constructor

 destructor

 reset to process new track

 create from piece of AliTrackPointArray

 set first and last point to fit

 invert the cov.matrices of the points

 init auxiliary space

 perform linear fit w/o accounting the errors
 fit is done in the parameterization
 x = res[0] + res[1]*z
 y = res[2] + res[3]*z
 where x,y,z are NOT the lab axes but z is the lab axis along which the track
 has the largest lever arm and x,y are the remaining 2 axis in
 the order of fgkAxisID[z][0], fgkAxisID[z][1]

 select the axis which will be used as a parameter for the line: longest baseline

 select the variable with largest base as a parameter

 calculate the position of the track at PCA to xyz

 calculate the position of the track at PCA to pntCovInv
 NOTE: the covariance matrix of the point must be inverted

 calculate the residuals  of the track at PCA to pntCovInv
 NOTE: the covariance matrix of the point must be inverted

 calculate the residuals of the track at PCA to xyz

 get parameter for the point with least weighted distance to the point

 calculate detivative of the PCA residuals vs point position and fill in user provide
 array in the format {dXdXp,dY/dXp,dZdXp, ... dXdZp,dYdZp,dZdZp}

 calculate detivative of the PCA residuals vs line parameters and fill in user provide
 array in the format {dXdP0,dYdP0,dZdP0, ... dXdPn,dYdPn,dZdPn}

 get t-param detivative over the parameters for the point with least weighted distance to the point

 get t-param detivative over the parameters for the point with least weighted distance to the point

  Double_t rhs;
  Double_t dx = fParams[kA0]-xyz[ fkAxID[kX] ];
  Double_t dy = fParams[kA1]-xyz[ fkAxID[kY] ];
  Double_t dz =             -xyz[ fkAxID[kZ] ];

 calculate detivative of the PCA residuals vs line parameters and fill in user provide
 array in the format {dXdP0,dYdP0,dZdP0, ... dXdPn,dYdPn,dZdPn}

 calculate detivative of the PCA residuals vs point position and fill in user provide
 array in the format {dXdXp,dY/dXp,dZdXp, ... dXdZp,dYdZp,dZdZp}

 calculate detivative of the PCA residuals vs track parameters and fill in user provide
 array in the format {dXdP0,dYdP0,dZdP0, ... dXdPn,dYdPn,dZdPn}

 calculate detivative of the PCA residuals vs point position and fill in user provide
 array in the format {dXdXp,dY/dXp,dZdXp, ... dXdZp,dYdZp,dZdZp}

 get residual detivatives over the track parameters for the point with least weighted distance to the point

 get residuals detivative over the point position for the point with least weighted distance to the point

 find track parameter t (eq.2) corresponding to point of closest approach to xyz

 find track parameter t (eq.2) corresponding to point on the circle with closest approach to x,y

 find helix parameter t (eq.2) corresponding to point on the circle of radius t

 calculate fit chi2/ndf

 calculate residuals at point

 calculate track position for parameter value t

 calculate track direction cosines for parameter value t

 estimate machine precision

 crude estimate of helix parameters, w/o errors and Eloss.
 Fast Riemann fit: Comp.Phy.Comm.131 (2000) 95

 if charge is not imposed (extQ==0) then it will be determined from the collision type

 fit by helix accounting for the errors of all coordinates (and energy loss if requested)

 If extQ is non-0, its sign is imposed as a charge of the track
 If extPT>0 and extPTerr>=0, constrain to measured tr.momentum PT
 with corresponding error (err=0 -> rel.err=1e-6)

 fit by helix accounting for the errors of all coordinates (and energy loss if requested)

 obtain the lab normal vector to the sensor from the covariance matrix
 in such a way that when the local frame of the sensor coincides with
 the lab frame, the vector {0,1,0} is obtained

 Calculate energy loss of the particle at given t-param on the layer with rhoL (thickness*density) with
 normal vector normS in the lab. The particle before eloss has energy "e" and momentum "p"
 cdip = cosine of the dip angle = 1/sqrt(1+tgL^2)
 Return the change DR of the radius due to the ELoss

 NOTE: with B>0 the negative particles propagate along increasing t-param and positive
 particles - against.
 t-param = 0 corresponds to the point of closest approach of the track to the beam.
 Since the fitted helix parameters of the track are defined in this PCA point, when the correction
 is applied upstream of the PCS, the energy must be increased (DR>0) rather than decreased (DR<0)

 return intercept of the parameterization coord = intercept + slope*t for given axis

 return intercept of the parameterization coord = intercept + slope*t for given axis

 print results of the fit

 Fill a look-up table with mean material a la AliITSTrackerMI

 calculate the PCA to plane normal ti axis and crossing it at axval
 fill the position and direction cosines at this point

 get direction cosines for t = 0;

 compute covariance matrix in lenear approximation of residuals on parameters

 get parameter for the point with least weighted distance to the point