DECONVOLVE

        DSB2SSB\DECONVOLVE Tolerance  Itmax  [Lambda1  [Lambda2  [Lambda3]]]
    [/GAIN [First_Guess]] [/KEEP]

    Deconvolves  DSB dataset loaded with INITIALIZE into SSB spectrum. Mini-
    mizes the function:
            Chi^2 - lambda1 F1 - lambda2 F2
    until either the difference in Chi^2 between the iterations is less than
    the  Tolerance,  or  the  maximum  number  of  iterations Itmax has been
    reached.

    Here:
      Chi^2 = Sum (dsb - ssb_lsb*gain_lsb - ssb_usb*gain_usb)^2
              /(weight * nu)

      F1 : entropy of the channels.
      F1 = - Sum ssb/SSB * log ssb/SSB

      F2 : entropy of the gains.
      F2 = - Sum gain_ssb/e * log gain_ssb/e
    and
      dsb          - double sideband data
      ssb          - single sideband data
      gain         - gain of single sideband data
      weight       - T_sys^2/(integrationime * bandwidth)
      nu           - number of free parameters:
      dsb_channels - ssb_channels - sideband_gains
      SSB          - Sum ssb

    If Lambda1 or Lambda2 or Lambda3 are not specified, they are  set  to  0
    and the algorithm does NOT try to maximize the entropy term.

    By  default  the  deconvolution makes no attempt at fitting the sideband
    gains, unless the option /GAIN is specified.