120 lines
3.3 KiB
Python
120 lines
3.3 KiB
Python
import torch
|
|
import torch.nn as nn
|
|
from torch_geometric.nn import MessagePassing
|
|
|
|
|
|
class FiniteDifferenceStep(MessagePassing):
|
|
"""
|
|
TODO: add docstring.
|
|
"""
|
|
|
|
def __init__(self, aggr: str = "add", root_weight: float = 1.0):
|
|
super().__init__(aggr=aggr)
|
|
assert (
|
|
aggr == "add"
|
|
), "Per somme pesate, l'aggregazione deve essere 'add'."
|
|
# self.root_weight = float(root_weight)
|
|
self.p = torch.nn.Parameter(torch.tensor(1.0))
|
|
self.a = root_weight
|
|
|
|
def forward(self, x, edge_index, edge_attr, deg):
|
|
"""
|
|
TODO: add docstring.
|
|
"""
|
|
out = self.propagate(edge_index, x=x, edge_attr=edge_attr, deg=deg)
|
|
return out
|
|
|
|
def message(self, x_j, edge_attr):
|
|
"""
|
|
TODO: add docstring.
|
|
"""
|
|
p = torch.clamp(self.p, 0.0, 1.0)
|
|
return p * edge_attr.view(-1, 1) * x_j
|
|
|
|
def aggregate(self, inputs, index, deg):
|
|
"""
|
|
TODO: add docstring.
|
|
"""
|
|
out = super().aggregate(inputs, index)
|
|
deg = deg + 1e-7
|
|
return out / deg.view(-1, 1)
|
|
|
|
def update(self, aggr_out, x):
|
|
"""
|
|
TODO: add docstring.
|
|
"""
|
|
return aggr_out
|
|
|
|
|
|
class GraphFiniteDifference(nn.Module):
|
|
"""
|
|
TODO: add docstring.
|
|
"""
|
|
|
|
def __init__(self, max_iters: int = 5000, threshold: float = 1e-4):
|
|
"""
|
|
TODO: add docstring.
|
|
"""
|
|
super().__init__()
|
|
self.max_iters = max_iters
|
|
self.threshold = threshold
|
|
self.fd_step = FiniteDifferenceStep(aggr="add", root_weight=1.0)
|
|
|
|
@staticmethod
|
|
def _compute_deg(edge_index, edge_attr, num_nodes):
|
|
"""
|
|
TODO: add docstring.
|
|
"""
|
|
deg = torch.zeros(num_nodes, device=edge_index.device)
|
|
deg = deg.scatter_add(0, edge_index[1], edge_attr)
|
|
return deg + 1e-7
|
|
|
|
@staticmethod
|
|
def _compute_c_ij(c, edge_index):
|
|
"""
|
|
TODO: add docstring.
|
|
"""
|
|
return (0.5 * (c[edge_index[0]] + c[edge_index[1]])).squeeze()
|
|
|
|
def forward(
|
|
self,
|
|
x,
|
|
edge_index,
|
|
edge_attr,
|
|
c,
|
|
boundary_mask,
|
|
boundary_values,
|
|
**kwargs,
|
|
):
|
|
"""
|
|
TODO: add docstring.
|
|
"""
|
|
edge_attr = 1 / edge_attr[:, -1]
|
|
c_ij = self._compute_c_ij(c, edge_index)
|
|
edge_attr = edge_attr * c_ij
|
|
deg = self._compute_deg(edge_index, edge_attr, x.size(0))
|
|
|
|
# Calcola la soglia staccando x dal grafo
|
|
conv_thres = self.threshold * torch.norm(x.detach())
|
|
|
|
for _i in range(self.max_iters):
|
|
out = self.fd_step(x, edge_index, edge_attr, deg)
|
|
out[boundary_mask] = boundary_values.unsqueeze(-1)
|
|
|
|
# Controllo convergenza senza tracciamento gradienti
|
|
with torch.no_grad():
|
|
residual_norm = torch.norm(out - x)
|
|
|
|
if residual_norm < conv_thres:
|
|
break
|
|
|
|
# --- OTTIMIZZAZIONE CHIAVE ---
|
|
# Stacca 'out' dal grafo prima della prossima iterazione
|
|
# per evitare BPTT e risparmiare memoria.
|
|
x = out.detach()
|
|
|
|
# Il 'out' finale restituito mantiene i gradienti
|
|
# dell'ULTIMA chiamata a fd_step, permettendo al modello
|
|
# di apprendere correttamente.
|
|
return out, _i + 1
|