Coverage for .tox/p311/lib/python3.10/site-packages/scicom/historicalletters/model.py: 94%
109 statements
« prev ^ index » next coverage.py v7.4.4, created at 2024-04-26 18:34 +0200
« prev ^ index » next coverage.py v7.4.4, created at 2024-04-26 18:34 +0200
1"""The model class for HistoricalLetters."""
2import random
3from pathlib import Path
5import mesa
6import mesa_geo as mg
7import networkx as nx
8import pandas as pd
9from numpy import mean
10from shapely import contains
11from tqdm import tqdm
13from scicom.historicalletters.agents import RegionAgent, SenderAgent
14from scicom.historicalletters.space import Nuts2Eu
15from scicom.historicalletters.utils import createData
16from scicom.utilities.statistics import prune
19def getPrunedLedger(model: mesa.Model) -> pd.DataFrame:
20 """Model reporter for simulation of archiving.
22 Returns statistics of ledger network of model run
23 and various iterations of statistics of pruned networks.
25 The routine assumes that the network contains fields of sender,
26 receiver and step information.
27 """
28 if model.runPruning is True:
29 ledgerColumns = ["sender", "receiver", "sender_location", "receiver_location", "topic", "step"]
30 modelparams = {
31 "population": model.population,
32 "moveRange": model.moveRange,
33 "letterRange": model.letterRange,
34 "useActivation": model.useActivation,
35 "useSocialNetwork": model.useSocialNetwork,
36 "similarityThreshold": model.similarityThreshold,
37 "longRangeNetworkFactor": model.longRangeNetworkFactor,
38 "shortRangeNetworkFactor": model.shortRangeNetworkFactor,
39 }
40 result = prune(
41 modelparameters=modelparams,
42 network=model.letterLedger,
43 columns=ledgerColumns,
44 iterations=10,
45 delAmounts=(0.1, 0.25, 0.5, 0.75, 0.9),
46 delTypes=("degree", "unif", "exp", "beta"),
47 )
48 else:
49 result = model.letterLedger
50 return result
53def getComponents(model: mesa.Model) -> int:
54 """Model reporter to get number of components.
56 The MultiDiGraph is converted to undirected,
57 considering only edges that are reciprocal, ie.
58 edges are established if sender and receiver have
59 exchanged at least a letter in each direction.
60 """
61 newg = model.socialNetwork.to_undirected(reciprocal=True)
62 return nx.number_connected_components(newg)
65def getScaledLetters(model: mesa.Model) -> float:
66 """Return relative number of send letters."""
67 return len(model.letterLedger)/model.schedule.time
70def getScaledMovements(model: mesa.Model) -> float:
71 """Return relative number of movements."""
72 return model.movements/model.schedule.time
75class HistoricalLetters(mesa.Model):
76 """A letter sending model with historical informed initital positions.
78 Each agent has an initial topic vector, expressed as a RGB value. The
79 initial positions of the agents is based on a weighted random draw
80 based on data from [1].
82 Each step, agents generate two neighbourhoods for sending letters and
83 potential targets to move towards. The probability to send letters is
84 a self-reinforcing process. During each sending the internal topic of
85 the sender is updated as a random rotation towards the receivers topic.
87 [1] J. Lobo et al, Population-Area Relationship for Medieval European Cities,
88 PLoS ONE 11(10): e0162678.
89 """
91 def __init__(
92 self,
93 population: int = 100,
94 moveRange: float = 0.05,
95 letterRange: float = 0.2,
96 similarityThreshold: float = 0.2,
97 longRangeNetworkFactor: float = 0.3,
98 shortRangeNetworkFactor: float = 0.4,
99 regionData: str = Path(Path(__file__).parent.parent.resolve(), "data/NUTS_RG_60M_2021_3857_LEVL_2.geojson"),
100 populationDistributionData: str = Path(Path(__file__).parent.parent.resolve(), "data/pone.0162678.s003.csv"),
101 *,
102 useActivation: bool = False,
103 useSocialNetwork: bool = False,
104 runPruning: bool = False,
105 debug: bool = False,
106 ) -> None:
107 """Initialize a HistoricalLetters model."""
108 super().__init__()
110 # Parameters for agents
111 self.population = population
112 self.moveRange = moveRange
113 self.letterRange = letterRange
114 # Parameters for model
115 self.runPruning = runPruning
116 self.useActivation = useActivation
117 self.similarityThreshold = similarityThreshold
118 self.useSocialNetwork = useSocialNetwork
119 self.longRangeNetworkFactor = longRangeNetworkFactor
120 self.shortRangeNetworkFactor = shortRangeNetworkFactor
121 # Initialize social network
122 self.socialNetwork = nx.MultiDiGraph()
123 # Output variables
124 self.letterLedger = []
125 self.movements = 0
126 # Internal variables
127 self.schedule = mesa.time.RandomActivation(self)
128 self.scaleSendInput = {}
129 self.updatedTopicsDict = {}
130 self.space = Nuts2Eu()
131 self.debug = debug
133 #######
134 # Initialize region agents
135 #######
137 # Set up the grid with patches for every NUTS region
138 # Create region agents
139 ac = mg.AgentCreator(RegionAgent, model=self)
140 self.regions = ac.from_file(
141 regionData,
142 unique_id="NUTS_ID",
143 )
144 # Add regions to Nuts2Eu geospace
145 self.space.add_regions(self.regions)
147 #######
148 # Initialize sender agents
149 #######
151 # Draw initial geographic positions of agents
152 initSenderGeoDf = createData(
153 population,
154 populationDistribution=populationDistributionData,
155 )
157 # Calculate mean of mean distances for each agent.
158 # This is used as a measure for the range of exchanges.
159 meandistances = []
160 for idx in initSenderGeoDf.index.to_numpy():
161 name = initSenderGeoDf.loc[idx, "unique_id"]
162 geom = initSenderGeoDf.loc[idx, "geometry"]
163 otherAgents = initSenderGeoDf.query(f"unique_id != '{name}'").copy()
164 geometries = otherAgents.geometry.to_numpy()
165 distances = [geom.distance(othergeom) for othergeom in geometries]
166 meandistances.append(mean(distances))
167 self.meandistance = mean(meandistances)
169 # Populate factors dictionary
170 self.factors = {
171 "similarityThreshold": similarityThreshold,
172 "moveRange": moveRange,
173 "letterRange": letterRange,
174 }
176 # Set up agent creator for senders
177 ac_senders = mg.AgentCreator(
178 SenderAgent,
179 model=self,
180 agent_kwargs=self.factors,
181 )
183 # Create agents based on random coordinates generated
184 # in the createData step above, see util.py file.
185 senders = ac_senders.from_GeoDataFrame(
186 initSenderGeoDf,
187 unique_id="unique_id",
188 )
190 # Create random set of initial topic vectors.
191 topics = [
192 tuple(
193 [random.random() for x in range(3)],
194 ) for x in range(self.population)
195 ]
197 # Setup senders
198 for idx, sender in enumerate(senders):
199 # Add to social network
200 self.socialNetwork.add_node(
201 sender.unique_id,
202 numLettersSend=0,
203 numLettersReceived=0,
204 )
205 # Give sender topic
206 sender.topicVec = topics[idx]
207 # Add current topic to dict
208 self.updatedTopicsDict.update(
209 {sender.unique_id: topics[idx]},
210 )
211 # Set random activation weight
212 if useActivation is True:
213 sender.activationWeight = random.random()
214 # Add sender to its region
215 regionID = [
216 x.unique_id for x in self.regions if contains(x.geometry, sender.geometry)
217 ]
218 try:
219 self.space.add_sender(sender, regionID[0])
220 except IndexError as exc:
221 text = f"Problem finding region for {sender.geometry}."
222 raise IndexError(text) from exc
223 # Add sender to schedule
224 self.schedule.add(sender)
226 # Create social network
227 if useSocialNetwork is True:
228 for agent in self.schedule.agents:
229 if isinstance(agent, SenderAgent):
230 self._createSocialEdges(agent, self.socialNetwork)
232 self.datacollector = mesa.DataCollector(
233 model_reporters={
234 "Ledger": getPrunedLedger,
235 "Letters": getScaledLetters ,
236 "Movements": getScaledMovements,
237 "Clusters": getComponents,
238 },
239 )
241 def _createSocialEdges(self, agent: SenderAgent, graph: nx.MultiDiGraph) -> None:
242 """Create social edges with the different wiring factors.
244 Define a close range by using the moveRange parameter. Among
245 these neighbors, create a connection with probability set by
246 the shortRangeNetworkFactor.
248 For all other agents, that are not in this closeRange group,
249 create a connection with the probability set by the longRangeNetworkFactor.
250 """
251 closerange = [x for x in self.space.get_neighbors_within_distance(
252 agent,
253 distance=self.moveRange * self.meandistance,
254 center=False,
255 ) if isinstance(x, SenderAgent)]
256 for neighbor in closerange:
257 if neighbor.unique_id != agent.unique_id:
258 connect = random.choices(
259 population=[True, False],
260 weights=[self.shortRangeNetworkFactor, 1 - self.shortRangeNetworkFactor],
261 k=1,
262 )
263 if connect[0] is True:
264 graph.add_edge(agent.unique_id, neighbor.unique_id, step=0)
265 longrange = [x for x in self.schedule.agents if x not in closerange and isinstance(x, SenderAgent)]
266 for neighbor in longrange:
267 if neighbor.unique_id != agent.unique_id:
268 connect = random.choices(
269 population=[True, False],
270 weights=[self.longRangeNetworkFactor, 1 - self.longRangeNetworkFactor],
271 k=1,
272 )
273 if connect[0] is True:
274 graph.add_edge(agent.unique_id, neighbor.unique_id, step=0)
276 def step(self) -> None:
277 """One simulation step."""
278 self.scaleSendInput.update(
279 **{x.unique_id: x.numLettersReceived for x in self.schedule.agents},
280 )
281 self.schedule.step()
282 self.datacollector.collect(self)
284 def step_no_data(self) -> None:
285 """One simulation step without datacollection."""
286 self.scaleSendInput.update(
287 **{x.unique_id: x.numLettersReceived for x in self.schedule.agents},
288 )
289 self.schedule.step()
291 def run(self, n:int) -> None:
292 """Run the model for n steps."""
293 if self.debug is True:
294 for _ in tqdm(range(n)):
295 self.step_no_data()
296 else:
297 for _ in range(n):
298 self.step_no_data()
299 self.datacollector.collect(self)