Visualization module¶
The meep_adjoint package ships with a built-in visualization
module to facilitate graphical inspection of material geometries and
functional subregions (solver inputs) as well as EM field
configurations and design-function gradients (solver outputs).
A first goal of this module is to provide a go-to visualization facility
for designers to have “under their fingertips”, easily remembered and
invoked at any stage of the design process to lend visual insight
into what’s going on at that moment. This is realized by the visualize
https://meep.readthedocs.io/en/latest/Python_User_Interface/#data-visualization
More specifically, the visualization module knows how to produce three types of plots:
Geometry plots, showing the material configuration throughout the computational cell of the problem, the PML layers, the finite-element mesh used to define the basis set (for problems using a
FiniteElementBasis), and significant subregions such as source regions, objective regions, and the design region.Forward field plots, showing the results of forward timestepping calculations, particularly the spatial distribution of energy fluxes and energy densities used to compute the objective function \(f^\text{obj}\).
Adjoint derivative plots, showing the results of adjoint timestepping calculations—specifically, the quantity \(\frac{\partial f^\text{obj}}{\partial \epsilon(\mathbf x)}\) as a function of \(\mathbf{x}\) throughout the design region.
In generating these plots,
executive mode, in which the user simply invokes the
visualizemethod ofOptimizationProblem—with no arguments!—and the visualization module automatically chooses not only the best default options for the plot, but also which of the above three plots to produce, based on the current state ofexpert mode, in which the user configures the parameters to
visualizeto control the plot produced and its appearance.