Input for trunk net in PI DeepONet

[voloddia]

In all examples I've seen for PI DeepONet, input for trunk net are often coordinates of space or time or both.
Is there any example where PDE coefficients are fed into trunk net?

[lululxvi]

You can do that. Just consider the coefficient as another dimension.

[voloddia]

Thank you, professor Lu.
Take the advection_aligned_pideeponet example, if I want to add a coefficient, I guess the pde and geom will become

dim_x = dim_x + 1

def pde(x, y, v):
    dy_x = dde.grad.jacobian(y, x, j=0)
    dy_t = dde.grad.jacobian(y, x, j=1)
    k = x[:, 2] # not sure about index?
    return dy_t + k * dy_x

geom = dde.geometry.Rectangle([0, 0], [1, 1])

Is that correct? Are there other parts that need to change?