How to convert the values in dynamic Snode into numpy? to_numpy() will get the full array

[Jianghanxiao]

For the dynamic SNode, how to get the values into numpy or other things. If calling to_numpy(), it will get the full-size stuff, however, I just want the dynamic size stuff. Below stuff works, however, seems not clever and efficient, is there more proper way?

collisions_pairs = []
for i in range(collision_len):
    collisions_pairs.append(self.collisions[i])
return collisions_pairs

[Jianghanxiao]

I find the below code may be the most efficient and proper way

@ti.kernel
def extract_collisions(self, buffer: ti.types.ndarray(ti.i32, ndim=1)):
    print(self.collisions.length())
    for i in range(self.collisions.length()):
        buffer[i] = self.collisions[i]

def get_collisions(self, collision_len):
    collisions_torch = torch.zeros(collision_len, dtype=torch.int32, device="cuda")
    self.extract_collisions(collisions_torch)
    return collisions_torch