Commit 0a0536c3 authored by Jonas Boysen's avatar Jonas Boysen
Browse files

improve sa

parent bf075a4c

AgRobotSwarmForOptimization/field_optimization.ipynb

+5 −3
Original line number Diff line number Diff line 
%% Cell type:code id: tags:



``` python

%matplotlib ipympl

from field_optimization import Simulation, FieldVisualizer, IPyUI



def random_search_successor(current_state: list, neighbor_states: list[list], neighbor_costs: list) -> [list, bool]:

    terminate = True

    return current_state, terminate



def stochastic_hc_successor(current_state: list, neighbor_states: list[list], neighbor_costs: list) -> [list, bool]:

    terminate = True

    return current_state, terminate



def t_schedule(t: int) -> float:

    t_max = 100

    return t_max - t



MAX_ITERATIONS = 50



def t_schedule(t: int) -> float:

    return (MAX_ITERATIONS - t) / MAX_ITERATIONS







optimize_sim = Simulation(field_save_dir="field", t_schedule=t_schedule, stochastic_hc=stochastic_hc_successor, random_search=random_search_successor)

optimize_sim.prepare_complete_state_representation()

field_vis = FieldVisualizer(optimize_sim, canvas_size=600)

ui = IPyUI(field_vis=field_vis, optimize_sim=optimize_sim, max_iterations=MAX_ITERATIONS)

ui.optimization_ui()

```

AgRobotSwarmForOptimization/field_optimization.py

+6 −2
Original line number Diff line number Diff line
@@ -314,7 +314,7 @@ class Simulation: 
        # get t

        t = len(self.cost_history)

        # calculate temperature with function of user

        temperature = self.t_schedule(t)

        temperature = self.t_schedule(t) * 100

        if temperature <= 0:

            return self.state, True

        # choose random successor
@@ -327,10 +327,14 @@ class Simulation: 
        cost = self.evaluate_state(state=neighbor_candidate_state)



        delta_e = cost - self.cost_history[-1]



        if delta_e < 0:

            return neighbor_candidate_state, False

        else:

            propability = np.exp(delta_e / temperature)

            propability = np.exp(-delta_e / temperature)

            print(

                f"Keine Verbesserung mit delta_e {delta_e:.2f}, temperature {temperature / 100:.2f} und Annahmewahrscheinlichkeit {propability * 100:.2f}%"

            )

            if random.random() <= propability:

                return neighbor_candidate_state, False

            else:

AgRobotSwarmForOptimization/graph_optimization.ipynb

+2 −3
Original line number Diff line number Diff line 
%% Cell type:code id: tags:



``` python

%matplotlib ipympl

from graph_optimization import Simulation, Visualizer, IPyUI



###### Dieser Teil gehört zur optionalen Programmieraufgabe ######

def random_search_successor(current_state: list, neighbor_states: list[list], neighbor_costs: list) -> [list, bool]:

    terminate = True

    return current_state, terminate



def stochastic_hc_successor(current_state: list, neighbor_states: list[list], neighbor_costs: list) -> [list, bool]:

    terminate = True

    return current_state, terminate

###### Teil-Ende ######





###### In diesem Teil könnt ihr Änderungen vornehmen ######

MAX_ITERATIONS = 50

def t_schedule(t: int) -> float:

    t_max = 100

    return t_max - t

    return (MAX_ITERATIONS - t) / MAX_ITERATIONS



MAX_ITERATIONS = 50

MANUAL_STATE_TEST = ["Hof", 0, 1, 2, 3, "Hof", 4, 5, 6, 7]

###### Teil-Ende ######



###### Der restliche Teil startet das Programm ######

optimize_sim = Simulation(graph_load_dir="graph", t_schedule=t_schedule, stochastic_hc=stochastic_hc_successor, random_search=random_search_successor, dummy=True)

print(f"Die manuell eingegebne complete state formulation wird mit {optimize_sim.evaluate_state(MANUAL_STATE_TEST)} bewertet.")

del optimize_sim

optimize_sim = Simulation(graph_load_dir="graph", t_schedule=t_schedule, stochastic_hc=stochastic_hc_successor, random_search=random_search_successor)

field_vis = Visualizer(optimize_sim, canvas_size=550)

ui = IPyUI(field_vis=field_vis, optimize_sim=optimize_sim, max_iterations=MAX_ITERATIONS)

ui.optimization_ui()

```

AgRobotSwarmForOptimization/graph_optimization.py

+5 −2
Original line number Diff line number Diff line
@@ -229,7 +229,7 @@ class Simulation: 
        # get t

        t = len(self.cost_history)

        # calculate temperature with function of user

        temperature = self.t_schedule(t)

        temperature = self.t_schedule(t) * 100

        if temperature <= 0:

            return self.state, True

        # choose random successor
@@ -245,7 +245,10 @@ class Simulation: 
        if delta_e < 0:

            return neighbor_candidate_state, False

        else:

            propability = np.exp(delta_e / temperature)

            propability = np.exp(-delta_e / temperature)

            print(

                f"Keine Verbesserung mit delta_e {delta_e:.2f}, temperature {temperature / 100:.2f} und Annahmewahrscheinlichkeit {propability * 100:.2f}%"

            )

            if random.random() <= propability:

                return neighbor_candidate_state, False

            else: