Geometrical theory of diffraction (GTD) is an alternative model of diffraction propounded first by Thomas Young in 1802. GTD has a long history of nearly 150 years over which many eminent people enriched this model which has now become an accepted tool in the calculation of diffraction patterns. In the conventional Helmholtz-Kirchhoff theory the diffracted field is obtained by computing the net effect of the waves emitted by all points within the area of the aperture. But GTD reduces this problem to one of computing the net effect of waves from a few points on the boundary of the aperture or obstacle, thus simplifying considerably the labour involved in computations. Also the theory can easily be modified to include polarization effects. This has been done specifically by Keller (1962) who exploited the Sommerfeld solution of diffraction of electromagnetic waves at a half plane, making the theory more versatile than the Kirchhoff scalar wave theory. Interestingly the geometry of difffracted rays is predictable from a generalized Fermat principle. According to this the total path chosen by light from the source to the point of observation via the diffracting boundary is an extremum. Historically it should be stated that many of the salient features of GTD were established by a school led by Raman which was active from 1919–1945. Later when Keller (1962) revived GTD independently, he and others who followed him rediscovered many of the results of the Raman school. We have stressed wherever necessary the contributions of the Indian School. We have also discussed certain geometries where GTD can be effectively used. We get some new and interesting results, which can be easily understood on GTD, but are difficult to interpret on the conventional theory of diffraction.

The curvature-sensitive localisation of proteins on membranes is vital for many cell biological processes. Coarse-grained models are routinely employed to study the curvature-sensing phenomena and membrane morphology at the length scale of a few micrometres. Two prevalent phenomenological models exist for modelling the experimental observations of curvature sensing: (1) the spontaneous curvature (SC) model and (2) the curvature mismatch (CM) model, which differ in their treatment of the change in elastic energy due to the binding of proteins on the membrane. In this work, the prediction of sensing and generation behaviour by these two models are investigated using analytical calculations as well as dynamic triangulation Monte Carlo simulations of quasispherical vesicles. While the SC model yields a monotonically decreasing sensing curve as a function of the vesicle radius, the CM model results in a non-monotonic sensing curve. We highlight the main differences in the interpretation of the protein-related parameters in the two models. We further propose that the SC model is appropriate for modelling peripheral proteins employing the hydrophobic insertion mechanism, with minimal modification of membrane rigidity, while the CM model is appropriate for modelling curvature generation using scaffolding mechanism where there is significant stiffening of the membrane due to protein binding.