Basically D-modules are the differential operator analogue of affine varieties. Just like how a variety is the quotient of an algebra by some collection of polynomials, D-modules are quotients by collections of differential operators.
Linearity: D-modules can only handle
linear pdes
Global: Traditional analysis of PDEs
has difficulty with extending solutions globally on a manifold, because
it tries to find explicit solutions. One way of fixing this is to use
sheafs of D-modules; that way, you don't have to rely
on finding any explicit formula.
Riemann-Hilbert: This theorem tells you
that certain linear PDE systems (holonomic with regular singularities),
the solution space is determined by its topological
data. This allows you to replace a PDE by a perverse
sheaf and study the perverse sheaf instead.
Characteristic variety:
Monodromy:
Localization: Representations of Lie
algebras turn out to be the same as D-modules on flag varieties.
High-intensity lasers change the refractive index of the medium they're travelling through
Quantum gravity: Should behave somewhat
like non-linear optics — the energy of particles should alter the
spacetime background
Navier-Stokes
D-modules weren't directly applied to Navier-Stokes for a few reasons: most obviously, because the NS equations are non-linear.
More importantly, D-module theory only tells you if solutions exist formally, not whether those solutions are stable
A holonomic D-module with regular singularities is equivalent to a representation of the fundamental group
Let An(k) be the Weyl algebra in n dimensions over the field k
An example of an algebraic D-module on ℂn is a left An(ℂ)-module.
If we take some system of equations that we want to investigate, we consider the ideal they generate, I, and then take the quotient An(k)/I
Now let's say that we want to find solutions to the equations in some space of functions F on which the Weyl algebra acts (ie on some D-module). We just look at An(k)-module maps An(k)/I → F
The quotient An(k)/I is actually generated by 1 + I, since any x + I = x(1+I). So given a map ϕ : An(k)/I → F, we have ϕ(x+I) = ϕ(x(1+I)) = xϕ(1+I), using the fact that this is a morphism of An(k)-modules.
Now, taking x ∈ I, we must have ϕ(x+I) = ϕ(0+I) = 0, so xϕ(1+I) = 0 for all x ∈ I.
So the choice of image of 1 + I corresponds to a solution to the system of equations captured by I.
We replace the Weyl algebra by a sheaf of differential operators
An(k)/⟨∂2 + m2⟩