Page 368 - Phase Space Optics Fundamentals and Applications
P. 368

Phase Space in Ultrafast Optics    349


               (s = 0, Fig. 11.2a) to the spectrogram (s = 2, Fig. 11.2c) and is positive
               for values of s larger than 2. As s increases, the Wigner function is
               smoothed out, but the resulting time-frequency distribution loses its
               ability to display chirp.
                 In fact, it is possible to cast all measurement strategies in terms of
               phase-space distributions, in a form that ensures the positivity of the
               signal distribution. The requirement that the signal be positive arises
               from the way in which optical detectors respond to the field. They
               are square-law detectors in which the intensity or energy (depending
               on the response time of the detector) gives rise to a photocurrent or
               charge. This implies that the detected signal is positive. As we have
               seen, not all phase-space distributions are positive, but a theorem due
               to Jordan shows that the distribution

                       S(T, 	) =  d  dt W(t,  )W M (t − T,   − 	)  (11.26)

               with W M (t,  ) the Wigner representation of the measurement appa-
               ratus, is always positive. Apparatuses for which the measurement
               function is a convolution, no matter how complex and negative the
               Wigner function of the pulse and the measurement apparatus are,
               yield a signal that is positive.
                 In general the measurement function may be written in the form

                           S(T, 	) =  d  dt W(t,  )W M (t,  ; 	,T)  (11.27)
               where the measurement Wigner function contains the apparatus pa-
               rameters. This cannot always be written as a convolution, as in
               Eq. (11.26), but nonetheless is always constrained to give a positive
               signal. Further, it is clear from these formulas that any measurement
               technique attempting to reconstruct either the Wigner or the ambigu-
               ity function must be capable of exploring the entire two-dimensional
               chronocyclic phase space.

               11.2.4 Phase-Space Representation
                        of Paraxial Optical Systems
               The representation of optical pulses in phase space can be understood
               by analogy to the representation of optical ray trajectories in geometri-
               cal optics. 34  This provides an important first-order design framework
               for ultrafast optical systems as well as an intuitive appreciation of the
               more formal representations of the chronocyclic distributions. Parax-
               ial optical systems are specified by a 2 × 2 real transfer matrix T,

                                            A  B
                                      T =                          (11.28)
                                            C  D
   363   364   365   366   367   368   369   370   371   372   373