spec['det_model']['constant'] = 0.2
model_input = ModelInput(spec)
# List of observed household compositions
composition_list = read_csv('inputs/uk_composition_list.csv',
header=None).to_numpy()
# Proportion of households which are in each composition
comp_dist = read_csv('inputs/uk_composition_dist.csv',
header=None).to_numpy().squeeze()
# With the parameters chosen, we calculate Q_int:
Q_int, states, which_composition, \
system_sizes, cum_sizes, \
inf_event_row, inf_event_col \
= build_household_population(composition_list, model_input)
rhs = RateEquations(model_input, Q_int, composition_list, which_composition,
states, inf_event_row, inf_event_col)
fully_sus = where(rhs.states_sus_only.sum(axis=1) == states.sum(axis=1))[0]
i_is_one = where((rhs.states_det_only +
rhs.states_undet_only).sum(axis=1) == 1)[0]
H = zeros(len(which_composition))
# Assign probability of 1e-5 to each member of each composition being sole infectious person in hh
H[i_is_one] = (1e-5) * comp_dist[which_composition[i_is_one]]
# Assign rest of probability to there being no infection in the household
H[fully_sus] = (1 -
1e-5 * sum(comp_dist[which_composition[i_is_one]])) * comp_dist
Q_ext_det, Q_ext_undet = rhs.external_matrices(H)
from os.path import isfile
if isfile('vars.pkl') is True:
with open('vars.pkl', 'rb') as f:
Q_int, states, which_composition, \
system_sizes, cum_sizes, \
inf_event_row, inf_event_col \
= load(f)
else:
# With the parameters chosen, we calculate Q_int:
Q_int, states, which_composition, \
system_sizes, cum_sizes, \
inf_event_row, inf_event_col \
= build_household_population(
composition_list,
params['sigma'],
params['det'],
params['tau'],
params['k_home'],
params['alpha'],
params['gamma'])
with open('vars.pkl', 'wb') as f:
dump((Q_int, states, which_composition, system_sizes, cum_sizes,
inf_event_row, inf_event_col), f)
total_size = len(which_composition)
# To define external mixing we need to set up the transmission matrices:
det_trans_matrix = diag(params['sigma']).dot(
params['k_ext']) # Scale rows of contact matrix by
# age-specific susceptibilities
# Scale columns by asymptomatic reduction in transmission
undet_trans_matrix = diag(params['sigma']).dot(params['k_ext'].dot(