Add time varying parameters

mobility/__init__.py

@@ -46,6 +46,7 @@
 
 from mobility.choice_models.population_trips import PopulationTrips
 from mobility.choice_models.population_trips_parameters import PopulationTripsParameters
+from .simulation_profile import ParameterProfile, SimulationStep
 
 
 from mobility.transport_graphs.speed_modifier import (

mobility/choice_models/destination_sequence_sampler.py

@@ -5,6 +5,7 @@
 from scipy.stats import norm
 
 from mobility.choice_models.add_index import add_index
+from mobility.simulation_profile import SimulationStep
 
 class DestinationSequenceSampler:
     """Samples destination sequences for trip chains.
@@ -17,9 +18,9 @@ class DestinationSequenceSampler:
     def run(
             self,
             motives,
+            step: SimulationStep,
             transport_zones,
             remaining_sinks,
-            iteration,
             chains,
             demand_groups,
             costs,
@@ -39,7 +40,6 @@ def run(
             transport_zones: Transport zone container used by motives.
             remaining_sinks (pl.DataFrame): Current sink state per (motive, to),
                 including capacity and saturation utility penalty.
-            iteration (int): Iteration index (>=1).
             chains (pl.DataFrame): Chain steps with
                 ["demand_group_id","motive_seq_id","motive","is_anchor","seq_step_index"].
             demand_groups (pl.DataFrame): ["demand_group_id","home_zone_id"] (merged for origins).
@@ -55,6 +55,7 @@ def run(
 
         utilities = self.get_utilities(
             motives,
+            step,
             transport_zones,
             remaining_sinks,
             costs,
@@ -64,6 +65,7 @@ def run(
         dest_prob = self.get_destination_probability(
             utilities,
             motives,
+            step,
             parameters.dest_prob_cutoff
         )
 
@@ -103,13 +105,13 @@ def run(
                 on=["demand_group_id", "motive_seq_id"]
             )
             .drop(["home_zone_id", "motive"])
-            .with_columns(iteration=pl.lit(iteration).cast(pl.UInt32))
+            .with_columns(iteration=pl.lit(step.iteration).cast(pl.UInt32))
         )
 
         return chains
 
 
-    def get_utilities(self, motives, transport_zones, sinks, costs, cost_uncertainty_sd):
+    def get_utilities(self, motives, step: SimulationStep, transport_zones, sinks, costs, cost_uncertainty_sd):
 
         """Assemble per-(from,to,motive) utility with cost uncertainty.
 
@@ -131,7 +133,7 @@ def get_utilities(self, motives, transport_zones, sinks, costs, cost_uncertainty
                 - cost_bin_to_dest: ["motive","from","cost_bin","to","p_to"].
         """
 
-        utilities = [(m.name, m.get_utilities(transport_zones)) for m in motives]
+        utilities = [(m.name, m.get_utilities(transport_zones, parameters=m.get_parameters_at_step(step))) for m in motives]
         utilities = [u for u in utilities if u[1] is not None]
         utilities = [u[1].with_columns(motive=pl.lit(u[0])) for u in utilities]
 
@@ -196,7 +198,7 @@ def offset_costs(costs, delta, prob):
         return costs_bin, cost_bin_to_dest
 
 
-    def get_destination_probability(self, utilities, motives, dest_prob_cutoff):
+    def get_destination_probability(self, utilities, motives, step: SimulationStep, dest_prob_cutoff):
 
         """Compute P(destination | from, motive) via a radiation-style model.
 
@@ -220,7 +222,10 @@ def get_destination_probability(self, utilities, motives, dest_prob_cutoff):
         costs_bin = utilities[0]
         cost_bin_to_dest = utilities[1]
 
-        motives_lambda = {motive.name: motive.inputs["parameters"].radiation_lambda for motive in motives}
+        motives_lambda = {
+            motive.name: motive.get_parameters_at_step(step).radiation_lambda
+            for motive in motives
+        }
 
         prob = (
 

mobility/choice_models/population_trips.py

@@ -32,6 +32,7 @@
     prune_tmp_artifacts,
     rehydrate_congestion_snapshot,
 )
+from mobility.simulation_profile import SimulationStep
 
 class PopulationTrips(FileAsset):
     """
@@ -183,8 +184,7 @@ def __init__(
         }
 
         super().__init__(inputs, cache_path)
-
-
+
     def validate_motives(self, motives: List[Motive]) -> None:
 
         if not motives:
@@ -303,7 +303,7 @@ def run_model(self, is_weekday: bool) -> Tuple[pl.DataFrame, pl.DataFrame, pl.Da
                 str(is_weekday),
                 str(resume_plan.resume_from_iter),
             )
-        tmp_folders = self.prepare_tmp_folders(cache_path, resume=(resume_plan.resume_from_iter is not None))
+        tmp_folders = self.get_tmp_folders(cache_path, resume=(resume_plan.resume_from_iter is not None))
 
         chains_by_motive, chains, demand_groups = self.state_initializer.get_chains(
             population,
@@ -317,11 +317,15 @@ def run_model(self, is_weekday: bool) -> Tuple[pl.DataFrame, pl.DataFrame, pl.Da
             chains_by_motive,
             demand_groups
         )
+
+        step = SimulationStep(iteration=1)
+        home_motive = [m for m in motives if m.name == "home"][0]
 
         stay_home_state, current_states = self.state_initializer.get_stay_home_state(
             demand_groups,
             home_night_dur,
-            motives,
+            home_motive,
+            step,
             parameters.min_activity_time_constant,
         )
 
@@ -374,13 +378,14 @@ def run_model(self, is_weekday: bool) -> Tuple[pl.DataFrame, pl.DataFrame, pl.Da
             logging.info(f"Iteration n°{iteration}")
 
             seed = self.rng.getrandbits(64)
+            step = SimulationStep(iteration=iteration)
 
             ( 
                 self.destination_sequence_sampler.run(
                     motives,
+                    step,
                     population.transport_zones,
                     remaining_sinks,
-                    iteration,
                     chains_by_motive,
                     demand_groups,
                     costs,
@@ -409,12 +414,12 @@ def run_model(self, is_weekday: bool) -> Tuple[pl.DataFrame, pl.DataFrame, pl.Da
                 costs_aggregator,
                 remaining_sinks,
                 motive_dur,
-                iteration,
+                step,
                 tmp_folders,
                 home_night_dur,
                 stay_home_state,
                 parameters,
-                motives
+                motives,
             )
             transition_events_per_iter.append(transition_events)
 
@@ -430,7 +435,8 @@ def run_model(self, is_weekday: bool) -> Tuple[pl.DataFrame, pl.DataFrame, pl.Da
             remaining_sinks = self.state_updater.get_new_sinks(
                 current_states_steps,
                 sinks,
-                motives
+                motives,
+                step,
             )
 
             # Save per-iteration checkpoint after all state has been advanced.
@@ -449,14 +455,15 @@ def run_model(self, is_weekday: bool) -> Tuple[pl.DataFrame, pl.DataFrame, pl.Da
         # If we resumed after completing all iterations (or start_iteration > n_iterations),
         # rebuild step-level flows from cached artifacts for final output.
         if "current_states_steps" not in locals():
+            step = SimulationStep(iteration=parameters.n_iterations)
             possible_states_steps = self.state_updater.get_possible_states_steps(
                 current_states,
                 demand_groups,
                 chains_by_motive,
                 costs_aggregator,
                 remaining_sinks,
                 motive_dur,
-                parameters.n_iterations,
+                step,
                 motives,
                 parameters.min_activity_time_constant,
                 tmp_folders
@@ -519,11 +526,12 @@ def remove(self, remove_checkpoints: bool = True):
                 logging.info("Removed %s checkpoint files for run_key=%s", str(removed), str(run_key))
 
 
-    def prepare_tmp_folders(self, cache_path, resume: bool = False):
-        """Create per-run temp folders next to the cache path.
+    def get_tmp_folders(self, cache_path: pathlib.Path, resume: bool = False) -> dict[str, pathlib.Path]:
+        """Return per-run temp folders next to the cache path.
 
         Args:
           cache_path (pathlib.Path): Target cache file used to derive temp roots.
+          resume (bool): When False, clears existing temp folders before reuse.
 
         Returns:
           dict[str, pathlib.Path]: Mapping of temp folder names to paths.

mobility/choice_models/state_initializer.py

@@ -1,6 +1,9 @@
 import math
 import polars as pl 
 
+from mobility.motives.motive import Motive
+from mobility.simulation_profile import SimulationStep
+
 class StateInitializer:
     """Builds initial chain demand, averages, and capacities for the model.
 
@@ -272,7 +275,8 @@ def get_stay_home_state(
             self,
             demand_groups,
             home_night_dur,
-            motives,
+            home_motive: Motive,
+            step: SimulationStep,
             min_activity_time_constant: float,
         ):
 
@@ -295,8 +299,8 @@ def get_stay_home_state(
                 - current_states: A clone of `stay_home_state` for iteration start.
         """
 
-        home_motive = [m for m in motives if m.name == "home"][0]
-        
+        value_of_time_stay_home = home_motive.get_parameters_at_step(step).value_of_time_stay_home
+
         stay_home_state = (
 
             demand_groups.select(["demand_group_id", "csp", "n_persons"])
@@ -309,13 +313,23 @@ def get_stay_home_state(
             .join(home_night_dur, on="csp")
             .with_columns(
                 utility=( 
-                    home_motive.inputs["parameters"].value_of_time_stay_home 
+                    value_of_time_stay_home
                     * pl.col("mean_home_night_per_pers")
                     * (pl.col("mean_home_night_per_pers")/pl.col("mean_home_night_per_pers")/math.exp(-min_activity_time_constant)).log().clip(0.0)
                 )
             )
 
-            .select(["demand_group_id", "iteration", "motive_seq_id", "mode_seq_id", "dest_seq_id", "utility", "n_persons"])
+            .select([
+                "demand_group_id",
+                "csp",
+                "mean_home_night_per_pers",
+                "iteration",
+                "motive_seq_id",
+                "mode_seq_id",
+                "dest_seq_id",
+                "utility",
+                "n_persons",
+            ])
         )
 
         current_states = (