Directed homotopy is a variant of homotopy theory in which paths have a preferred direction and cannot be freely reversed.
In classical homotopy theory, a path from x to y can always be traversed backward to give a path from y to x, and two paths are homotopic if one can be continuously deformed into the other through a family of paths. Directed homotopy drops the reversibility assumption. A directed path (or d-path) from x to y does not imply a d-path from y to x. A directed homotopy between two d-paths must itself respect the direction — intermediate paths in the deformation must all go from x to y.
The motivating examples come from concurrent computation: the state space of a concurrent system has a natural partial order (time flows forward, resource usage is irreversible), and the execution traces are directed paths. Two executions that reach the same result by different interleavings are d-homotopic — they represent the same concurrent computation up to scheduling differences. The d-holes (directed obstructions to homotopy) correspond to deadlocks and unreachable states.
Formally, directed homotopy works with d-spaces (topological spaces equipped with a distinguished set of directed paths closed under concatenation and reparametrization) or with more algebraic structures like directed cubical sets. The fundamental category replaces the fundamental group: instead of a group of loops up to homotopy, one has a category of directed paths up to directed homotopy, where composition is concatenation but inverses need not exist.
In the semiotic universe formalism, directed homotopy captures the irreversibility of semiotic interpretation: a sign’s interpretive history runs forward through the trace comonad, and higher directed paths record coherent reinterpretations that respect this directionality.