Martina studies reduced model equations in various contexts where the underlying model consists of partial differential equations. Such reduction procedures come into play, e.g. at the onset of Turing patterns with the Ginzburg-Landau equation as the so-called amplitude equation. Her main interest is the development of extensions of such wide-spread model reduction techniques to scenarios that incorporate more realistic modeling assumptions such as time- or space-dependent coefficients (think of a spatially varying terrain for vegetation patterns or time-varying rainfall). Martina has worked on various types of models (reaction-diffusion systems, nonlinear wave equations, lattice equations) and has developed reduction techniques based on finite and infinite-dimensional dynamical systems machinery such as center manifold theory, the theory of modulation equations and geometric singular perturbation theory. Her recent work explores the use of inverse spectral methods and the quest of rigorous justification of chaotic dynamics.