2.20. Custom Class for coupled PDEs

This example shows how to solve a set of coupled PDEs, the spatially coupled FitzHugh–Nagumo model, which is a simple model for the excitable dynamics of coupled Neurons:

\[\begin{split}\partial_t u &= \nabla^2 u + u (u - \alpha) (1 - u) + w \\ \partial_t w &= \epsilon u\end{split}\]

Here, \(\alpha\) denotes the external stimulus and \(\epsilon\) defines the recovery time scale. We implement this as a custom PDE class below.

  0%|          | 0/100.0 [00:00<?, ?it/s]
Initializing:   0%|          | 0/100.0 [00:00<?, ?it/s]
  0%|          | 0/100.0 [00:00<?, ?it/s]
  0%|          | 0.23/100.0 [00:00<01:42,  1.03s/it]
  1%|          | 0.69/100.0 [00:00<00:40,  2.44it/s]
  3%|2         | 2.84/100.0 [00:00<00:16,  5.87it/s]
  8%|7         | 7.63/100.0 [00:00<00:11,  8.15it/s]
 15%|#4        | 14.75/100.0 [00:01<00:09,  9.23it/s]
 24%|##3       | 23.5/100.0 [00:02<00:07,  9.66it/s]
 33%|###3      | 33.09/100.0 [00:03<00:06,  9.95it/s]
 43%|####3     | 43.22/100.0 [00:04<00:05, 10.14it/s]
 54%|#####3    | 53.7/100.0 [00:05<00:04, 10.24it/s]
 64%|######4   | 64.28/100.0 [00:06<00:03, 10.34it/s]
 75%|#######4  | 74.99/100.0 [00:07<00:02, 10.39it/s]
 86%|########5 | 85.69/100.0 [00:08<00:01, 10.44it/s]
 96%|#########6| 96.45/100.0 [00:09<00:00, 10.47it/s]
 96%|#########6| 96.45/100.0 [00:09<00:00, 10.11it/s]
100%|##########| 100.0/100.0 [00:09<00:00, 10.48it/s]
100%|##########| 100.0/100.0 [00:09<00:00, 10.48it/s]

from pde import FieldCollection, PDEBase, UnitGrid


class FitzhughNagumoPDE(PDEBase):
    """FitzHugh–Nagumo model with diffusive coupling"""

    def __init__(self, stimulus=0.5, τ=10, a=0, b=0, bc="auto_periodic_neumann"):
        super().__init__()
        self.bc = bc
        self.stimulus = stimulus
        self.τ = τ
        self.a = a
        self.b = b

    def evolution_rate(self, state, t=0):
        v, w = state  # membrane potential and recovery variable

        v_t = v.laplace(bc=self.bc) + v - v**3 / 3 - w + self.stimulus
        w_t = (v + self.a - self.b * w) / self.τ

        return FieldCollection([v_t, w_t])


grid = UnitGrid([32, 32])
state = FieldCollection.scalar_random_uniform(2, grid)

eq = FitzhughNagumoPDE()
result = eq.solve(state, t_range=100, dt=0.01)
result.plot()