Snell’s law revisited and generalized via Finsler geometry

Steen Markvorsen, Enrique Pendás-Recondo

Research output: Contribution to journalJournal articleResearchpeer-review

86 Downloads (Pure)


We study the variational problem of finding the fastest path between two points that belong to different anisotropic media, each with a prescribed speed profile and a common interface. The optimal curves are Finsler geodesics that are refracted – broken – as they pass through the interface, due to the discontinuity of their velocities. This “breaking” must satisfy a specific condition in terms of the Finsler metrics defined by the speed profiles, thus establishing the generalized Snell’s law. In the same way, optimal paths bouncing off the interface – without crossing into the second domain – provide the generalized law of reflection. The classical Snell’s and reflection laws are recovered in this setting when the velocities are isotropic. If one considers a wave that propagates in all directions from a given ignition point, the trajectories that globally minimize the travel time generate the wave front at each instant of time. We study in detail the global properties of such wavefronts in the Euclidean plane with anisotropic speed profiles. Like the individual rays, they break when they encounter the discontinuity interface. But they are also broken due to the formation of cut loci – stemming from the self-intersection of the wave fronts – which typically appear when they approach a high-speed profile domain from a low-speed profile.
Original languageEnglish
Article number2350138
JournalInternational Journal of Geometric Methods in Modern Physics
Issue number8
Number of pages26
Publication statusPublished - 2023


  • Snell's law
  • Fermat’s principle
  • Huygens’ principle
  • Zermelo navigation
  • Finsler metrics
  • Anisotropic discontinuous media
  • Rays with least traveltime
  • Wave propagation


Dive into the research topics of 'Snell’s law revisited and generalized via Finsler geometry'. Together they form a unique fingerprint.

Cite this