FUNCTION AVG,ARRAY,DIMENSION, NAN = NAN, DOUBLE = DOUBLE ;+ ; NAME: ; AVG ; PURPOSE: ; Return the average value of an array, or 1 dimension of an array ; EXPLANATION: ; Calculate the average value of an array (in which case AVG is identical ; to the RSI procedure mean.pro), or calculate the average ; value over one dimension of an array as a function of all the other ; dimensions. ; ; CALLING SEQUENCE: ; RESULT = AVG( ARRAY, [ DIMENSION, /NAN, /DOUBLE ] ) ; ; INPUTS: ; ARRAY = Input array. May be any type except string. ; ; OPTIONAL INPUT PARAMETERS: ; DIMENSION = Optional dimension to do average over, integer scalar ; ; OPTIONAL KEYWORD INPUT: ; /NAN - Set this keyword to cause the routine to check for occurrences of ; the IEEE floating-point value NaN in the input data. Elements with ; the value NaN are treated as missing data. ; /DOUBLE - By default, if the input Array is double-precision, complex, ; or double complex, the result is of the same type; 64 bit ; integers are also returned as double. Otherwise the result ; the result is floating-point. Use of the /DOUBLE keyword ; forces a double precision output. Note that internal ; computations are always done in double precision. ; OUTPUTS: ; The average value of the array when called with one parameter. ; ; If DIMENSION is passed, then the result is an array with all the ; dimensions of the input array except for the dimension specified, ; each element of which is the average of the corresponding vector ; in the input array. ; ; For example, if A is an array with dimensions of (3,4,5), then the ; command B = AVG(A,1) is equivalent to ; ; B = FLTARR(3,5) ; FOR J = 0,4 DO BEGIN ; FOR I = 0,2 DO BEGIN ; B[I,J] = TOTAL( A[I,*,J] ) / 4. ; ENDFOR ; ENDFOR ; ; RESTRICTIONS: ; Dimension specified must be valid for the array passed; otherwise the ; input array is returned as the output array. ; PROCEDURE: ; AVG(ARRAY) = TOTAL(ARRAY, /DOUBLE)/N_ELEMENTS(ARRAY) when called with ; one parameter. ; MODIFICATION HISTORY: ; William Thompson Applied Research Corporation ; July, 1986 8201 Corporate Drive ; Landover, MD 20785 ; Converted to Version 2 July, 1990 ; Replace SUM call with TOTAL W. Landsman May, 1992 ; Converted to IDL V5.0 W. Landsman September 1997 ; Added /NAN keyword W. Landsman July 2000 ; Accept a scalar input value W. Landsman/jimm@berkeley November 2000 ; Internal calculations always in double precision W. Landsman March 2002 ; Return NAN if all values in array are NAN W. Landsman April 2002 ; Fixed coding bug if all values in array are NAN W. Landsman Jan 2004 ;- ON_ERROR,2 S = SIZE(ARRAY,/STR) IF S.N_ELEMENTS EQ 1 THEN RETURN, array[0] IF S.N_ELEMENTS EQ 0 THEN $ MESSAGE,'Variable must be an array, name= ARRAY' ; IF N_PARAMS() EQ 1 THEN BEGIN IF KEYWORD_SET(NAN) THEN NPTS = TOTAL(FINITE(ARRAY) ) $ ELSE NPTS = N_ELEMENTS(ARRAY) IF NPTS EQ 0 THEN AVERAGE = !VALUES.F_NAN ELSE $ AVERAGE = TOTAL(ARRAY, NAN=NAN,/DOUBLE) / NPTS ENDIF ELSE BEGIN IF ((DIMENSION GE 0) AND (DIMENSION LT S.N_DIMENSIONS)) THEN BEGIN AVERAGE = TOTAL(ARRAY,DIMENSION+1,NAN=NAN,/DOUBLE) ; Install a bug workaround since TOTAL(A,/NAN) returns 0 rather than NAN if ; all A values are NAN. IF KEYWORD_SET(NAN) THEN BEGIN NPTS = TOTAL(FINITE(ARRAY),DIMENSION+1 ) BAD = WHERE(NPTS EQ 0, NBAD) AVERAGE = AVERAGE/(NPTS>1) IF NBAD GT 0 THEN AVERAGE[BAD] = !VALUES.D_NAN ENDIF ELSE AVERAGE = AVERAGE/S.DIMENSIONS[DIMENSION] END ELSE $ MESSAGE,'*** Dimension out of range, name= ARRAY' ENDELSE ; Convert to floating point unless of type double, complex, or L64, or ; if /DOUBLE is set. IF NOT KEYWORD_SET(DOUBLE) THEN BEGIN CASE S.TYPE OF 5: RETURN, AVERAGE 6: RETURN, COMPLEXARR( FLOAT(AVERAGE), FLOAT(IMAGINARY(AVERAGE)) ) 9: RETURN, AVERAGE 14: RETURN, AVERAGE 15: RETURN, AVERAGE ELSE: RETURN, FLOAT(AVERAGE) ENDCASE ENDIF ELSE RETURN, AVERAGE END