The paper studies a class of evolution equations where the future behavior of a system depends not just on its current state but on its entire past history. This "time nonlocal" feature is captured through mathematical objects called kernels, which encode how memory effects influence the system's development. The equations also include a nonlinear source term that grows very rapidly, meaning small solutions can potentially blow up or fail to exist altogether. The authors want to understand when solutions exist and when they do not.
The main findings concern the relationship between the initial conditions and whether a solution can be found. The authors identify sufficient conditions guaranteeing that a local solution exists and belongs to a reasonable function space, even when the nonlinearity is growing fast. They also find conditions under which no solution can exist at all. A key discovery is a critical dimension phenomenon, meaning that whether the problem is solvable depends sensitively on a specific numerical threshold tied to the initial data. Below that threshold, solutions may exist; above it, they may not. The proofs use two main technical tools: a method called subordination, which relates the nonlocal equation to simpler ones, and a monotone iterative scheme, which builds solutions by successively improving approximations.
The paper also provides several concrete examples demonstrating how broadly the theoretical results apply. These examples cover different types of memory kernels, including cases relevant to fractional calculus and other generalized models of diffusion. The work contributes to a growing body of research on equations that model anomalous diffusion, viscoelastic materials, and other physical systems where standard local-in-time models are insufficient. By pinpointing the boundary between existence and nonexistence, the results offer practical guidance for understanding when such models are mathematically well-posed.