Add T-PE and V-PE methods

Author: habibirani

src/positional_encodings.py

@@ -175,4 +175,32 @@ def forward(self, x):
         x = x.permute(0, 2, 1)  # Change shape to (batch_size, in_features, seq_len)
         x = self.conv(x)  # Apply convolution
         x = x.permute(0, 2, 1)  # Change shape back to (batch_size, seq_len, in_features)
-        return x    
+        return x    
+    
+
+# Temporal Positional Encoding (T-PE)
+class TemporalPositionalEncoding(nn.Module):
+    def __init__(self, d_model, max_len=896):  # Assuming 896 timesteps
+        super(TemporalPositionalEncoding, self).__init__()
+        pe = torch.zeros(max_len, d_model)
+        position = torch.arange(0, max_len, dtype=torch.float).unsqueeze(1)
+        div_term = torch.exp(torch.arange(0, d_model, 2).float() * (-math.log(10000.0) / d_model))
+
+        pe[:, 0::2] = torch.sin(position * div_term)
+        pe[:, 1::2] = torch.cos(position * div_term)
+        self.register_buffer('pe', pe)
+
+    def forward(self, x):
+        seq_len = x.size(1)
+        return self.pe[:seq_len, :].unsqueeze(0).expand(x.size(0), -1, -1)
+    
+
+# Variable Positional Encoding for handling multivariate data
+class VariablePositionalEncoding(nn.Module):
+    def __init__(self, d_model, num_variables):
+        super(VariablePositionalEncoding, self).__init__()
+        self.variable_embedding = nn.Embedding(num_variables, d_model)
+
+    def forward(self, x, variable_idx):
+        variable_embed = self.variable_embedding(variable_idx)
+        return x + variable_embed.unsqueeze(0)