2. The geometric theory of waves

Geometrical theory of waves. Fermat principle. Rays,  wave fronts. Singularities. Caustics. Huyghens principle.

The Fermat principle defines the rays.  To any pair of points A, B in a connected region we can assign a ray as follows:  given an arbitrary curve, C, starting from the initial point A and ending at the end point B, there belongs a "propagation time," T(C), defined by the formula:

T(C) = integral over the curve C (ds/v) eq 2.1

Fermat's principle states that this propagation time is a minimum for the actual path of propagation.

Ray:  A curve connecting A with B for which:

Tc = minimum eq 2.1

Fronts: surfaces orthogonal to rays.

Remarks:

1) the rays defined here are the extremals of the variational problem 2.2 (see: variational calculus)

2) the admissible set of curves should satisfy some mathematical requirements (rectifiability) and must lie entirely in the admissible region.

3) Between a given pair of points it is possible there exists a unique ray, but it is also possible to have several different paths which produce the same minimum. This problem is related to singularity theory (V.I. Arnold; Singularities of Caustics and Wave Fronts; Kluwer Academic Publishers, 1990) and conflict sets (Dirk Siersma)

4) In the geometric theory, rays are primary, while fronts are derived as surfaces orthogonal to the rays, compared to the  former structure where the wave function was given, the wave fronts were primary, and rays were defined as their orthogonal trajectories.

5) The eikonal defined here differs from the one used in optics [Max Born & Emil Wolf; Principles of Optics; Pergamon Press;Sixth (Corrected) Edition, 1980] g* with a constant factor c: g*=cg, where c is the light velocity in vacuum. Our eikonal  has the time dimension while the eikonal in optics has the length dimension.

6) As stated in many books Fermat principle requires that the propagation time must be stationary, that is the propagation time for paths closely adjacent to the actual path differs from that of the actual path by a second order term. This is the case in optics, where the laws of reflection and refraction satisfy this requirement. There are known examples when the propagation  time is stationary or even maximun for the actual path. E.g., for an elliptic mirror, if a point source is in one focal point F1, the image point will  be on the other focal point F2, consequently, the propagation  time is equal for any ray starting from F1, reflectiong at the mirror and reaching F2 /the propagation time is stationary/. If we use a mirror having common tangent with the elliptic mirror at a point, the actual ray reflected at this point is the same, but the adjacent rays have greater time or lesser time according to the ratio of curvatures of mirrors [Jenkins, F.A.  & H.E. White; Fundamentals of Optics; McGraw-Hill Publishing Co.  1951]. Notwithstanding this fact, we formulated the Fermat principle as minimum principle. Really, for waves of excitations as well as for chemical waves the minimum condition is necessarily satisfied. One can see the importance of minimum for the case of praire fire  model: the fire propagates according to the minimum  time requirement, since after the fire reaches a point, all the other paths ending at that point are irrelevant.
 

Huygens' principle

Any point of a given front at t0 can be considered a point source which radiates spherical waves propagating in any direction.  At a next time t there belongs a surface which is spherical if the medium is homogenious, otherwise it will be some distorted spheroid. Consider the family of those elementary spheroids starting from the points of the initial  front. The wave front at a given later time t will be the envelope of those spheroids belonging to time  t.
 

Remarks

1.  Huygens priciple is equivalent to the Fermat principle.  Either enables us to determine the evolution of fronts.
2.  Backward propagation should be prohibited:  this restriction is not involved in Huygens or Fermats principle but is an additional restriction.
3.  In the usual derivation of refraction law, the refracted front is constructed as the envelope of spheres with different radii originating from the region of the interface of the two media.  This example illustrates that Huygens principe is valid in a more general formulation: the set of point souces may be an arbitrary set [not only a front], if the time is adjusted properly.

Evolution of fronts

Given a front surface we can construct the successive fronts using Fermat's principle. From every point of the actual front rays starts at any direction. However, it is sufficient to