Predicted state covariances (state_dim × state_dim × h). C[i][j][k] = covariance between states i and j at forecast step k+1.
Forecast horizon
State dimension
Predicted state means (state_dim × h). x[i][k] = i-th state component at forecast step k+1.
Predicted state standard deviations (h × state_dim). xstd[k][i] = sqrt(C[i][i][k])
Predicted observation means (h×1). yhat[k] = F · x_pred[k]
Predicted observation standard deviations (h×1). ystd[k] = sqrt(F · C_pred[k] · F' + s²)
Result from dlmForecast: h-step-ahead predictions with uncertainty.
All arrays have length h (the forecast horizon).