functions in histo.i -

             n= _histo_sum(data, x, order, hx, h0, h1, h2);  
 
     Worker for histo_stat routine: sum moments of response DATA with respect  
     to explanatory variable X.  See documentation of histo_stat for  
     further explanations and description of keywords.  Return number of  
     elements in output arrays (if N=0, outputs can be discarded...).  
   ARGUMENTS:  
     DATA ->  Response data to average.  
     X    ->  Value of explanatory variable for each data measurement (same  
              geometry as DATA).  
     HX   <-  Value of X in each bin.  
     H0   <-  Sum of WEIGHT in each bin.  
     H1   <-  Sum of WEIGHT*DATA in each bin.  
     H2   <-  Sum of WEIGHT*DATA^2 in each bin.  
     H3   <-  Sum of WEIGHT*WEIGHT in each bin.  
     ORDER->  0 to compute only HX and H0,  
              >=1 to also compute H1  
	      >=2 to also compute H2  
	      >=3 to also compute H3.  
   KEYWORDS: binsize, xmin, xmax, weight, interp.  

             hy= histo2(data, hx);  
 
     Returns histogram of DATA, the histogram abscissae get stored into  
     pure output HX (don't forget to declare it as `local', unless you don't  
     care of side effects).  
     The size of the bins can be given by keyword BINSIZE (default 1.0).  
     The minimum and maximum abscissae to consider can specified with  
     keywords BINMIN and BINMAX.  
     Keyword WEIGHT gives data point weights (default all 1).  If keyword  
     INTERP is set with a non-zero value, linear interpolation is used to  
     get the weights of the data points to each bin.  
     If keyword AVERAGE is set with a non-zero value, returns an average  
     rather than a sum of the weithed data points.  Note that with no weights  
     or uniform weights, the average value will be an array of 1's (now  
     you know!).  Averaging is useful for instance to compute the radial  
     profile (average along azimuth direction):  
       x= indgen(xdim)-x0;     // abscissa  
       y= indgen(ydim)-y0;     // ordinate  
       r= abs(x, y(-,));       // radius  
       a= exp(-r*r/50.0)+(random(xdim, ydim) - 0.5)/10.0;  
       local px;  
       py= histo2(r, px, weight=a, average=1, interp=1);  
       plg, py, px, color="red";  
       plg, exp(-px*px/50.0), px, color="green";  
     You can call histo2 one more time to compute the sample noise:  
       py_sigma= sqrt(histo2(r, weight=a*a, average=1, interp=1) - py*py);  
   

             histo_plot, data;  
 
     Compute (with  histo2) an  histogram of DATA  and plot it  (with plh).  
     All  keywords  of  plh  and  histo2 can  be  used.   Further  keywords  
     PRENORMALIZE and POSTNORMALIZE can  be used to normalize the histogram  
     so  that  its sum  is  equal  to  PRE/POSTNORMALIZE: normalization  to  
     PRENORMALIZE  (POSTNORMALIZE)   is  performed  *BEFORE*  (respectively  
     *AFTER*) data values less than BINMIN or greater than BINMAX have been  
     rejected.   Keyword  SCALE  can  be  used  to  multiply  ordinates  of  
     histogram  by  SCALE (default  SCALE=1);  this  is  useful to  compare  
     histograms of the same data set with different BINSIZE.  
   KEYWORDS: weight, binsize, binmin, binmax, interp, average,  
             just, legend, hide, type, width, color, marks, marker,  
             mspace, mphase, prenormalize, postnormalize, scale.  

             hs= histo_stat(data, x);  
 
     Compute statistics of response DATA with respect to explanatory  
     variable X.  DATA measurements are grouped in bins by rounding X  
     to the nearest integer multiple of BINSIZE and the statistical moments  
     are computed whithin each bin.  If you need unevenly spaced bins, you  
     may either operate a change of explicit variable or use Yorick's  
     `digitize' and `histogram' routines.  
   ARGUMENTS:  
     HS   <-  Result:  
                HS(,1)= value of explanatory variable X in each bin  
                HS(,2)= sum of weights for each bin  
                HS(,3)= average of DATA in each bin  
              only if keyword STD/AVG_STD is non-nil and non-zero:  
                HS(,4)= standard deviation of DATA in each bin  
              only if keyword AVG_STD is non-nil and non-zero:  
                HS(,5)= standard deviation of HS(,3)  
     DATA ->  Response data to average.  
     X    ->  Value of explanatory variable for each data measurement (same  
              geometry as DATA).  
   KEYWORDS:  
     STD=     Compute standard deviation per data sample in HS(,4).  
     AVG_STD= Compute standard deviation of average in HS(,5) (Note:  
              this also implies STD=1).  
     WEIGHT=  Statistical weight (e.g., 0.0 where data is not significant).  
     XMAX=    Maximum X value to account for (default max(X)).  
     XMIN=    Minimum X value to account for (default min(X)).  
     BINSIZE= Resolution for data bins, i.e. value of HS(,1)(dif) (default 1.0)  
     INTERP=  Use linear interpolation, instead of rounding to the nearest  
              integer multiple of BINSIZE?