import numpy as np
from functools import partial
import json
import csv
import math
import pickle
import matplotlib.pyplot as plt
from gerrychain.random import random
from gerrychain import Graph, Partition, Election, MarkovChain, GeographicPartition, accept, constraints
from gerrychain.updaters import Tally, cut_edges, county_splits, Tally, boundary_nodes, cut_edges_by_part, exterior_boundaries, interior_boundaries, perimeter
from gerrychain.constraints import single_flip_contiguous, Validator, within_percent_of_ideal_population, refuse_new_splits
from gerrychain.proposals import recom
from gerrychain.accept import always_accept
from gerrychain.tree import recursive_tree_part
from gerrychain.metrics.compactness import compute_polsby_popper, polsby_popper
random.seed(20210809)
np_load_old = np.load
np.load = lambda *a,**k: np_load_old(*a, allow_pickle=True, **k)
## ## ## ## ## ## ## ## ## ## ##
## set-up argparse!
## ## ## ## ## ## ## ## ## ## ##
parser = argparse.ArgumentParser(description="MO ensemble",
prog="sampling.py")
parser.add_argument("map", metavar="map", type=str,
choices=["state_house", "state_senate"],
help="the map to redistrict")
parser.add_argument("eps", metavar="epsilon", type=float,
# choices=[0.01, 0.03, 0.05, .08],
help="the number of plans to sample")
parser.add_argument(
"popcol",
metavar="population column",
type=str,
choices=["TOTPOP", "VAP", "CPOP", "CVAP"],
help="the population column by which to balance redistricting")
parser.add_argument("--i",
metavar="run number",
type=int,
default=0,
help="which chain run is this?")
args = parser.parse_args()
from gerrychain.random import random
random.seed(args.i)
from gerrychain import Graph, GeographicPartition, Partition, Election, accept
from gerrychain.updaters import Tally, cut_edges
from gerrychain import MarkovChain
from gerrychain.proposals import recom
from gerrychain.accept import always_accept
from gerrychain import constraints
from gerrychain.tree import recursive_tree_part
num_districts_in_map = {
"congress": 14,
"congress_2000": 13,
"state_senate": 56,
"state_house": 180
}
def generator(stop_event,
partition_queue,
_type,
step_count=1000,
burn_in=1000):
"""
Creates and runs generator for a proposal type.
Parameters
----------
_type : string
either "chunk" or "random"
step_count : int, Default 1000
steps of chain to run (after burn-in)
burn_in : int, Default 1000
steps to burn-in for (not to collect data)
stop_event: multiprocessing.Event
tells chain's workers if generation has stopped
partition_queue: multiprocessing.Queue
structure that takes each partition as it is generated
"""
# FOR REPRODUCIBILITY
from gerrychain.random import random
random.seed(2020)
init_partition = Partition(
graph,
assignment=race_matrix.to_dict()["partition"],
updaters={"population": Tally("population")},
)
if _type == "chunk":
chain = MarkovChain(
proposal=propose_chunk_flip,
constraints=is_valid,
accept=always_accept,
initial_state=init_partition,
total_steps=step_count + burn_in,
)
elif _type == "random":
chain = MarkovChain(
proposal=propose_random_flip,
constraints=is_valid,
accept=always_accept,
initial_state=init_partition,
total_steps=step_count + burn_in,
)
else:
raise ValueError("Wrong chain type given. Give 'chunk' or 'random'.")
iter(chain)
if _type == "chunk":
burn_bar = trange(burn_in,
desc=f"{_type} flip Burn-in",
leave=True,
position=0)
pbar = trange(step_count,
desc=f"Generating {_type} flip",
leave=True,
position=0)
else:
burn_bar = trange(burn_in,
desc=f"{_type} flip Burn-in",
leave=True,
position=1)
pbar = trange(step_count,
desc=f"Generating {_type} flip",
leave=True,
position=1)
# burn-in
for _ in burn_bar:
next(chain)
for i in pbar:
partition_queue.put((i, dict(next(chain).assignment)))
stop_event.set()
# # send a text when done (SEE FIELDS)
# client.messages.create(
# to=<YOUR PHONE NUMBER>,
# from_=<TWILIO SOUCE NUMBER>,
# body=f"{_type.capitalize()} flip for {CITY_NAME} completed.",
# )
print(f"{_type.capitalize()} Generator: {stop_event.is_set()}")
def generator(
stop_event,
partition_queue,
step_count=1000,
burn_in_ratio=0.1,
thinning=5,
seed=2020,
):
"""
Creates and runs generator of map proposals
Parameters
----------
step_count : int, Default 1000
steps of chain to run (after burn-in)
burn_in_ratio : float, Default 0.1
steps to burn-in for, as a ratio of the total step count (not to collect data)
stop_event: multiprocessing.Event
tells chain's workers if generation has stopped
partition_queue: multiprocessing.Queue
structure that takes each partition as it is generated
thinning: int, Default 5
Take every <thinning>th result from the chain to minimize dependence
seed: int, Default 2020
Random seed for reproducibility
"""
# FOR REPRODUCIBILITY
from gerrychain.random import random
random.seed(seed)
init_partition = Partition(
graph,
assignment=race_matrix.to_dict()["partition"],
updaters={"population": Tally("population")},
)
chain = MarkovChain(
proposal=propose_chunk_flip,
constraints=is_valid,
accept=always_accept,
initial_state=init_partition,
total_steps=step_count + burn_in_ratio * step_count,
)
iter(chain)
burn_bar = trange(int(burn_in_ratio * step_count),
desc=f"Burn-in",
leave=True,
position=0)
pbar = trange(
int(burn_in_ratio * step_count) + step_count,
desc=f"Generating maps",
leave=True,
position=0,
)
# burn-in
# for _ in burn_bar:
# next(chain)
for i in pbar:
map_proposal = (i, dict(next(chain).assignment))
# only push proposal to queue if it is <thinning>th proposal
if i % thinning == 0:
partition_queue.put(map_proposal)
stop_event.set()
# # send a text when done (SEE FIELDS)
# client.messages.create(
# to=<YOUR PHONE NUMBER>,
# from_=<TWILIO SOUCE NUMBER>,
# body=f"{_type.capitalize()} flip for {CITY_NAME} completed.",
# )
print(f"{CITY_NAME} Generator: {stop_event.is_set()}")
import argparse
import geopandas as gpd
import pandas as pd
import matplotlib.pyplot as plt
from gerrychain.random import random
random.seed(20210511)
from gerrychain import Graph, Partition, Election, MarkovChain, GeographicPartition, accept, constraints
from gerrychain.updaters import Tally, cut_edges
from gerrychain.constraints import single_flip_contiguous, Validator, within_percent_of_ideal_population, refuse_new_splits
from gerrychain.proposals import recom
from gerrychain.accept import always_accept
from gerrychain.tree import recursive_tree_part
import numpy as np
from functools import partial
import json
import csv
import pickle
## ## ## ## ## ## ## ## ## ## ##
## set-up argparse!
## ## ## ## ## ## ## ## ## ## ##
parser = argparse.ArgumentParser(description="MO ensemble", prog="sampling.py")
parser.add_argument("map",
metavar="map",
type=str,
choices=["state_house", "state_senate"],
help="the map to redistrict")
parser.add_argument("eps",
metavar="epsilon",
# FOR REPRODUCIBILITY
import time
from gerrychain.random import random
random.seed(2020)
import os
import argparse
from twilio.rest import Client
import geopandas as gpd
import matplotlib.pyplot as plt
import networkx as nx
import numpy as np
import pandas as pd
import seaborn as sns
from gerrychain import (
GeographicPartition,
Graph,
MarkovChain,
Partition,
accept,
constraints,
proposals,
updaters,
)
from gerrychain.accept import always_accept
from gerrychain.constraints import (
LowerBound,
UpperBound,
Validator,
WithinPercentRangeOfBounds,