GTSAM is structured in a number of directories, which are on a need to know basis. In the tree below, modules are supposed to only know about ancestor modules. E.g., slam is a nonlinear optimization framework that of course uses inference, and everyone knows about base.

• base
  • geometry
  • inference/symbolic
    • discrete
    • linear
    • nonlinear
      • slam
      • navigation

The modules inference and symbolic are co-dependent.