Feature request: Changing the model parameters locally

[bjornverstraeten]

Specifically, I want to create a tissue with a region that has a longer APD.
This could be done be changing the model parameters for this region only,
but currently I don't think that is possible.

Could this feature be implemented?
am happy to help, but I am not sure where to start.

[TiNezlobinsky]

We know about this kind of tasks and consider this feature to be implemented in future updates.

But is is not difficult to create custom implementation.

Let's say, we have AP model and want to add different parameters to different nodes. Here: finitewave/cpuwave2D/model/aliev_panfilov_2d.py, in ionic_kernel_2d(u_new, u, v, indexes, dt, a, k, eap, mu_1, mu_2) you can see that we use both arrays (u_new, u, v) and variables (others) as arguments. You can create your own subclass and make a, k, eap, mu_1, mu_2 as arrays too - just change their definition in init and use them as arrays inside ionic_kernel_2d.

[bjornverstraeten]

Thanks, The following code worked for me

class AlievPanfilov2D(fw.AlievPanfilov2D):
    """Extension of AlievPanfilov model to locally change APD."""

    def initialize(self):
        super().initialize()
        self.a = self.a * np.ones_like(self.u, dtype=self.npfloat)
        self.k = self.k * np.ones_like(self.u, dtype=self.npfloat)
        self.eap = self.eap * np.ones_like(self.u, dtype=self.npfloat)
        self.mu_1 = self.mu_1 * np.ones_like(self.u, dtype=self.npfloat)
        self.mu_2 = self.mu_2 * np.ones_like(self.u, dtype=self.npfloat)

    def run_ionic_kernel(self):
        """
        Executes the ionic kernel for the Aliev-Panfilov model.
        """
        ionic_kernel_2d(
            self.u_new,
            self.u,
            self.v,
            self.cardiac_tissue.myo_indexes,
            self.dt,
            self.a,
            self.k,
            self.eap,
            self.mu_1,
            self.mu_2,
        )


@njit(parallel=True)
def ionic_kernel_2d(u_new, u, v, indexes, dt, a, k, eap, mu_1, mu_2):
    """
    Computes the ionic kernel for the Aliev-Panfilov 2D model.

    Parameters
    ----------
    u_new : np.ndarray
        Array to store the updated action potential values.
    u : np.ndarray
        Current action potential array.
    v : np.ndarray
        Recovery variable array.
    indexes : np.ndarray
        Array of indices where the kernel should be computed (``mesh == 1``).
    dt : float
        Time step for the simulation.
    """

    n_j = u.shape[1]

    for ind in prange(len(indexes)):
        ii = indexes[ind]
        i = int(ii / n_j)
        j = ii % n_j

        v[i, j] = fw.cpuwave2D.model.aliev_panfilov_2d.calc_v(
            v[i, j], u[i, j], dt, a[i, j], k[i, j], eap[i, j], mu_1[i, j], mu_2[i, j]
        )

        u_new[i, j] += dt * (
            -k[i, j] * u[i, j] * (u[i, j] - a[i, j]) * (u[i, j] - 1.0)
            - u[i, j] * v[i, j]
        )

I saw a small dip in performance when running on a 128x128 tissue, but nothing drastic.

[TiNezlobinsky]

Good. I leave this issue opened, till we add this functionality officially.