def main():
print(datetime.datetime.now())
#reading config file
json_file = open(sys.argv[2],"r",encoding="utf-8")
config = json.load(json_file)
json_file.close()
# Define parameters
bet= config['bet']
tau = config['tau']
ph = config['ph']
init = config['init']
T=config['T']
sigma = config['sigma']
ndt = config['ndt']
n_runs = config['n_runs']
# contact matrix
C = matrixCM("graph_data_more.csv",config['scalarC'],config['scalarM'],config['maxClassSize'])
print(datetime.datetime.now())
#running the covid19_Net simulation, 30 repetitions
C_list = [C,C] #Weekday, weekend
C_ind = [0, 0, 0, 0, 0, 1, 1]
[n,cu,p,ou,r0,r0t,y,x] = repeated_runs(n_runs, bet,tau,ph,init,C_list,T,sigma,ndt,C_ind)
print(datetime.datetime.now())
state = {
'bet' : bet,
'tau' : tau,
'ph' : ph,
'init' : init,
'T' : T,
'sigma' : sigma,
'ndt' : ndt,
'n_runs' : n_runs,
'C' : C,
'n' : n,
'cu' : cu,
'p' : p,
'ou' : ou,
'r0' : r0,
'r0t' : r0t,
'y' : y,
'x' : x
}
sio.savemat(sys.argv[1],state)
print(datetime.datetime.now())
# This code would load the data and generate the plots.
# results = sio.loadmat('save_data.mat')
# viz.gen_plots(results, "testrun")
return state
def main():
edge_file = sys.argv[1]
n_runs = int(sys.argv[2])
output_file = sys.argv[3]
# Define parameters
# Note: from small run using weekdays:
# >>> np.mean(dt['r0theory'][:,0])
# (6.930544488810122+0j)
# >>> 0.0080 * 3.8/np.mean(dt['r0theory'][:,0])
# (0.004386379749683889+0j)
bet = np.array([0.0044, 0.0044, 0.0044])
tau = [5.35, 5.2, 5.0, 12.0, 12.0, 10.0]
ph = [.5, 1, 1]
init = 1
T = 30
sigma = [1, 1]
ndt = 1
mask_efficacy = 0.44
# https://www.publichealthontario.ca/-/media/documents/lab/covid-19-lab-testing-faq.pdf?la=en
# https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7240870/
# Note: we only test symptomatic onward, so that averages about 0.4 false negative rate
#facility = tf.testing_facility(0.4, 0.0001, [0.0, 0.5, 0.5])
print(datetime.datetime.now())
people = np.array([
3834, 164, 3530, 3811, 3337, 132, 832, 3682, 1853, 2433, 2466, 3191,
544, 2776, 2534, 2404, 3827, 1855, 1494, 2504, 2945, 1884, 1975, 1743,
2698, 3542, 2408, 3545, 2702, 2342, 3310, 3391, 276, 2427, 3010, 153,
2946, 2815, 922, 3772, 318, 1224, 1682, 3284, 2032, 1641, 595, 3474,
3805, 3278
])
# load and calculate contact matrix
edges = graph_ops.loadEdgeList(edge_file)
C, M, R, S = graph_ops.edgesAsMatrices(edges)
#contacts = 12/7*C + 1/420*M + 1/4*R + 1.0*S
Npop = C.shape[0]
matrices = {'C': C, 'M': M, 'R': R, 'S': S}
avg_friends = 5
lock_effectiveness = 0.7
weekday = {
'C': {
'weight': 15.81 / 5.0
}, # Assume 1/5 of weight per class
'M': {
'weight': 1.0 / 300.0
},
'R': {
'weight': 3.36 * 0.1
},
'S': {
'weight': 3.36 * 0.9
},
'Off': {
'weight': 3.36
},
'Soc': {
'weight': 10.73 / float(avg_friends)
}
}
weekend = {
'C': {
'weight': 5.76 / 5.0
}, # Assume 1/5 of weight per class
'R': {
'weight': 5.16 * 0.1
},
'S': {
'weight': 5.16 * 0.9
},
'Off': {
'weight': 5.16
},
'Soc': {
'weight': 13.50 / float(avg_friends)
}
}
lockdown = {
'R': {
'weight': lock_effectiveness * 5.16 * 0.1
},
'S': {
'weight': lock_effectiveness * 5.16 * 0.9
},
'Off': {
'weight': lock_effectiveness * 3.36
},
'Soc': {
'weight': lock_effectiveness * 13.50 / float(avg_friends)
}
}
filter_set = [[weekday, lockdown], [weekday,
lockdown], [weekday, lockdown],
[weekday, lockdown], [weekday, lockdown],
[weekend, lockdown], [weekend, lockdown]]
compliances = np.array([1.0, 0.8, 0.6])
days_delayed = [0, 1, 2, 3, 4, 5]
combs = pd.DataFrame(list(product(days_delayed, compliances)),
columns=['d', 'c'])
trials = pd.DataFrame({
'comp': combs['c'].values,
'delay': combs['d'].values,
'p_spread': 0.0,
'all_final_size_mean': 0.0,
'all_final_size_min': 0.0,
'all_final_size_max': 0.0,
'all_final_size_median': 0.0,
'all_final_size_pc5': 0.0,
'all_final_size_pc25': 0.0,
'all_final_size_pc75': 0.0,
'all_final_size_pc95': 0.0,
'spread_final_size_mean': 0.0,
'spread_final_size_min': 0.0,
'spread_final_size_max': 0.0,
'spread_final_size_median': 0.0,
'spread_final_size_pc5': 0.0,
'spread_final_size_pc25': 0.0,
'spread_final_size_pc75': 0.0,
'spread_final_size_pc95': 0.0,
'all_maxPrev_mean': 0.0,
'all_maxPrev_min': 0.0,
'all_maxPrev_max': 0.0,
'all_maxPrev_median': 0.0,
'all_maxPrev_pc5': 0.0,
'all_maxPrev_pc25': 0.0,
'all_maxPrev_pc75': 0.0,
'all_maxPrev_pc95': 0.0,
'spread_maxPrev_mean': 0.0,
'spread_maxPrev_min': 0.0,
'spread_maxPrev_max': 0.0,
'spread_maxPrev_median': 0.0,
'spread_maxPrev_pc5': 0.0,
'spread_maxPrev_pc25': 0.0,
'spread_maxPrev_pc75': 0.0,
'spread_maxPrev_pc95': 0.0,
'all_maxSymp_mean': 0.0,
'all_maxSymp_min': 0.0,
'all_maxSymp_max': 0.0,
'all_maxSymp_median': 0.0,
'all_maxSymp_pc5': 0.0,
'all_maxSymp_pc25': 0.0,
'all_maxSymp_pc75': 0.0,
'all_maxSymp_pc95': 0.0,
'spread_maxSymp_mean': 0.0,
'spread_maxSymp_min': 0.0,
'spread_maxSymp_max': 0.0,
'spread_maxSymp_median': 0.0,
'spread_maxSymp_pc5': 0.0,
'spread_maxSymp_pc25': 0.0,
'spread_maxSymp_pc75': 0.0,
'spread_maxSymp_pc95': 0.0,
'all_TimeToPeak_mean': 0.0,
'all_TimeToPeak_min': 0.0,
'all_TimeToPeak_max': 0.0,
'all_TimeToPeak_median': 0.0,
'all_TimeToPeak_pc5': 0.0,
'all_TimeToPeak_pc25': 0.0,
'all_TimeToPeak_pc75': 0.0,
'all_TimeToPeak_pc95': 0.0,
'spread_TimeToPeak_mean': 0.0,
'spread_TimeToPeak_min': 0.0,
'spread_TimeToPeak_max': 0.0,
'spread_TimeToPeak_median': 0.0,
'spread_TimeToPeak_pc5': 0.0,
'spread_TimeToPeak_pc25': 0.0,
'spread_TimeToPeak_pc75': 0.0,
'spread_TimeToPeak_pc95': 0.0,
'all_TimeToNotice_mean': 0.0,
'all_TimeToNotice_min': 0.0,
'all_TimeToNotice_max': 0.0,
'all_TimeToNotice_median': 0.0,
'all_TimeToNotice_pc5': 0.0,
'all_TimeToNotice_pc25': 0.0,
'all_TimeToNotice_pc75': 0.0,
'all_TimeToNotice_pc95': 0.0,
'spread_TimeToNotice_mean': 0.0,
'spread_TimeToNotice_min': 0.0,
'spread_TimeToNotice_max': 0.0,
'spread_TimeToNotice_median': 0.0,
'spread_TimeToNotice_pc5': 0.0,
'spread_TimeToNotice_pc25': 0.0,
'spread_TimeToNotice_pc75': 0.0,
'spread_TimeToNotice_pc95': 0.0,
'all_CasesAtNotice_mean': 0.0,
'all_CasesAtNotice_min': 0.0,
'all_CasesAtNotice_max': 0.0,
'all_CasesAtNotice_median': 0.0,
'all_CasesAtNotice_pc5': 0.0,
'all_CasesAtNotice_pc25': 0.0,
'all_CasesAtNotice_pc75': 0.0,
'all_CasesAtNotice_pc95': 0.0,
'spread_CasesAtNotice_mean': 0.0,
'spread_CasesAtNotice_min': 0.0,
'spread_CasesAtNotice_max': 0.0,
'spread_CasesAtNotice_median': 0.0,
'spread_CasesAtNotice_pc5': 0.0,
'spread_CasesAtNotice_pc25': 0.0,
'spread_CasesAtNotice_pc75': 0.0,
'spread_CasesAtNotice_pc95': 0.0
})
print(datetime.datetime.now())
for index, row in trials.iterrows():
#bet_eff = (1 - mask_efficacy*row['comp']) * bet
bet_eff = bet
comp = row['comp']
delay = row['delay']
print("Starting %.1f and %d" % (comp, delay))
# DO STATIC DELAY, OR POISSON DELAY ON FACILITY? Go with Poisson. That gives individual variability
fac_return = [0.0, 0.5, 0.5]
#if delay > 0:
# de = [0.0] * delay
# de.extend(fac_return)
# fac_return = de
facility = tf.testing_facility(0.4,
0.0001,
fac_return,
poisson_delay=delay)
quar = [14, 5]
#running the covid19_Net simulation, n_runs repetitions
[n, cu, p, ou, r0t, r0inf, y, x, r0theory, norms,
mts] = repeated_runs(n_runs,
Npop,
bet_eff,
tau,
ph,
matrices,
T,
sigma,
filter_set,
sim_offcampus_using=R,
sim_social={
'name': 'Soc',
'n': avg_friends,
'd': 0.15
},
symp_test_comp=comp,
quar=quar,
diagnostics=False,
facility=facility,
init_ind=people)
max_time_step = cu.shape[1] - 1
fs = cu[:, max_time_step, 0]
spread = fs > 2
n_spread = np.sum(spread)
trials.at[index, 'p_spread'] = np.mean(spread)
trials.at[index, 'all_final_size_mean'] = np.mean(fs)
trials.at[index, 'all_final_size_min'] = np.min(fs)
trials.at[index, 'all_final_size_max'] = np.max(fs)
trials.at[index, 'all_final_size_median'] = np.median(fs)
trials.at[index, 'all_final_size_pc5'] = np.percentile(fs, 5, axis=0)
trials.at[index, 'all_final_size_pc25'] = np.percentile(fs, 25, axis=0)
trials.at[index, 'all_final_size_pc75'] = np.percentile(fs, 75, axis=0)
trials.at[index, 'all_final_size_pc95'] = np.percentile(fs, 95, axis=0)
if n_spread > 0:
trials.at[index, 'spread_final_size_mean'] = np.mean(fs[spread])
trials.at[index, 'spread_final_size_min'] = np.min(fs[spread])
trials.at[index, 'spread_final_size_max'] = np.max(fs[spread])
trials.at[index,
'spread_final_size_median'] = np.median(fs[spread])
trials.at[index,
'spread_final_size_pc5'] = np.percentile(fs[spread],
5,
axis=0)
trials.at[index,
'spread_final_size_pc25'] = np.percentile(fs[spread],
25,
axis=0)
trials.at[index,
'spread_final_size_pc75'] = np.percentile(fs[spread],
75,
axis=0)
trials.at[index,
'spread_final_size_pc95'] = np.percentile(fs[spread],
95,
axis=0)
prevalence = p[:, :, 0] + p[:, :, 1] + p[:, :, 2] + p[:, :,
3] + p[:, :, 4]
tmp = np.max(prevalence, axis=1)
trials.at[index, 'all_maxPrev_mean'] = np.mean(tmp)
trials.at[index, 'all_maxPrev_min'] = np.min(tmp)
trials.at[index, 'all_maxPrev_max'] = np.max(tmp)
trials.at[index, 'all_maxPrev_median'] = np.median(tmp)
trials.at[index, 'all_maxPrev_pc5'] = np.percentile(tmp, 5, axis=0)
trials.at[index, 'all_maxPrev_pc25'] = np.percentile(tmp, 25, axis=0)
trials.at[index, 'all_maxPrev_pc75'] = np.percentile(tmp, 75, axis=0)
trials.at[index, 'all_maxPrev_pc95'] = np.percentile(tmp, 95, axis=0)
if n_spread > 0:
trials.at[index, 'spread_maxPrev_mean'] = np.mean(tmp[spread])
trials.at[index, 'spread_maxPrev_min'] = np.min(tmp[spread])
trials.at[index, 'spread_maxPrev_max'] = np.max(tmp[spread])
trials.at[index, 'spread_maxPrev_median'] = np.median(tmp[spread])
trials.at[index, 'spread_maxPrev_pc5'] = np.percentile(tmp[spread],
5,
axis=0)
trials.at[index,
'spread_maxPrev_pc25'] = np.percentile(tmp[spread],
25,
axis=0)
trials.at[index,
'spread_maxPrev_pc75'] = np.percentile(tmp[spread],
75,
axis=0)
trials.at[index,
'spread_maxPrev_pc95'] = np.percentile(tmp[spread],
95,
axis=0)
symp = p[:, :, 1] + p[:, :, 3] + p[:, :, 4]
tmp = np.max(symp, axis=1)
trials.at[index, 'all_maxSymp_mean'] = np.mean(tmp)
trials.at[index, 'all_maxSymp_min'] = np.min(tmp)
trials.at[index, 'all_maxSymp_max'] = np.max(tmp)
trials.at[index, 'all_maxSymp_median'] = np.median(tmp)
trials.at[index, 'all_maxSymp_pc5'] = np.percentile(tmp, 5, axis=0)
trials.at[index, 'all_maxSymp_pc25'] = np.percentile(tmp, 25, axis=0)
trials.at[index, 'all_maxSymp_pc75'] = np.percentile(tmp, 75, axis=0)
trials.at[index, 'all_maxSymp_pc95'] = np.percentile(tmp, 95, axis=0)
if n_spread > 0:
trials.at[index, 'spread_maxSymp_mean'] = np.mean(tmp[spread])
trials.at[index, 'spread_maxSymp_min'] = np.min(tmp[spread])
trials.at[index, 'spread_maxSymp_max'] = np.max(tmp[spread])
trials.at[index, 'spread_maxSymp_median'] = np.median(tmp[spread])
trials.at[index, 'spread_maxSymp_pc5'] = np.percentile(tmp[spread],
5,
axis=0)
trials.at[index,
'spread_maxSymp_pc25'] = np.percentile(tmp[spread],
25,
axis=0)
trials.at[index,
'spread_maxSymp_pc75'] = np.percentile(tmp[spread],
75,
axis=0)
trials.at[index,
'spread_maxSymp_pc95'] = np.percentile(tmp[spread],
95,
axis=0)
tmp = np.argmax(prevalence, axis=1)
trials.at[index, 'all_TimeToPeak_mean'] = np.mean(tmp)
trials.at[index, 'all_TimeToPeak_min'] = np.min(tmp)
trials.at[index, 'all_TimeToPeak_max'] = np.max(tmp)
trials.at[index, 'all_TimeToPeak_median'] = np.median(tmp)
trials.at[index, 'all_TimeToPeak_pc5'] = np.percentile(tmp, 5, axis=0)
trials.at[index, 'all_TimeToPeak_pc25'] = np.percentile(tmp,
25,
axis=0)
trials.at[index, 'all_TimeToPeak_pc75'] = np.percentile(tmp,
75,
axis=0)
trials.at[index, 'all_TimeToPeak_pc95'] = np.percentile(tmp,
95,
axis=0)
if n_spread > 0:
trials.at[index, 'spread_TimeToPeak_mean'] = np.mean(tmp[spread])
trials.at[index, 'spread_TimeToPeak_min'] = np.min(tmp[spread])
trials.at[index, 'spread_TimeToPeak_max'] = np.max(tmp[spread])
trials.at[index,
'spread_TimeToPeak_median'] = np.median(tmp[spread])
trials.at[index,
'spread_TimeToPeak_pc5'] = np.percentile(tmp[spread],
5,
axis=0)
trials.at[index,
'spread_TimeToPeak_pc25'] = np.percentile(tmp[spread],
25,
axis=0)
trials.at[index,
'spread_TimeToPeak_pc75'] = np.percentile(tmp[spread],
75,
axis=0)
trials.at[index,
'spread_TimeToPeak_pc95'] = np.percentile(tmp[spread],
95,
axis=0)
symp_cum = np.cumsum(cu[:, :, 6], axis=1)
ttn = np.sum(symp_cum == 0, axis=1)
tmp = ttn
trials.at[index, 'all_TimeToNotice_mean'] = np.mean(tmp)
trials.at[index, 'all_TimeToNotice_min'] = np.min(tmp)
trials.at[index, 'all_TimeToNotice_max'] = np.max(tmp)
trials.at[index, 'all_TimeToNotice_median'] = np.median(tmp)
trials.at[index, 'all_TimeToNotice_pc5'] = np.percentile(tmp,
5,
axis=0)
trials.at[index, 'all_TimeToNotice_pc25'] = np.percentile(tmp,
25,
axis=0)
trials.at[index, 'all_TimeToNotice_pc75'] = np.percentile(tmp,
75,
axis=0)
trials.at[index, 'all_TimeToNotice_pc95'] = np.percentile(tmp,
95,
axis=0)
if n_spread > 0:
trials.at[index, 'spread_TimeToNotice_mean'] = np.mean(tmp[spread])
trials.at[index, 'spread_TimeToNotice_min'] = np.min(tmp[spread])
trials.at[index, 'spread_TimeToNotice_max'] = np.max(tmp[spread])
trials.at[index,
'spread_TimeToNotice_median'] = np.median(tmp[spread])
trials.at[index,
'spread_TimeToNotice_pc5'] = np.percentile(tmp[spread],
5,
axis=0)
trials.at[index,
'spread_TimeToNotice_pc25'] = np.percentile(tmp[spread],
25,
axis=0)
trials.at[index,
'spread_TimeToNotice_pc75'] = np.percentile(tmp[spread],
75,
axis=0)
trials.at[index,
'spread_TimeToNotice_pc95'] = np.percentile(tmp[spread],
95,
axis=0)
basis_dates = np.clip(ttn, 0, max_time_step)
cases_at_notice = prevalence[range(0, n_runs), basis_dates]
tmp = cases_at_notice
trials.at[index, 'all_CasesAtNotice_mean'] = np.mean(tmp)
trials.at[index, 'all_CasesAtNotice_min'] = np.min(tmp)
trials.at[index, 'all_CasesAtNotice_max'] = np.max(tmp)
trials.at[index, 'all_CasesAtNotice_median'] = np.median(tmp)
trials.at[index, 'all_CasesAtNotice_pc5'] = np.percentile(tmp,
5,
axis=0)
trials.at[index, 'all_CasesAtNotice_pc25'] = np.percentile(tmp,
25,
axis=0)
trials.at[index, 'all_CasesAtNotice_pc75'] = np.percentile(tmp,
75,
axis=0)
trials.at[index, 'all_CasesAtNotice_pc95'] = np.percentile(tmp,
95,
axis=0)
if n_spread > 0:
trials.at[index,
'spread_CasesAtNotice_mean'] = np.mean(tmp[spread])
trials.at[index, 'spread_CasesAtNotice_min'] = np.min(tmp[spread])
trials.at[index, 'spread_CasesAtNotice_max'] = np.max(tmp[spread])
trials.at[index,
'spread_CasesAtNotice_median'] = np.median(tmp[spread])
trials.at[index,
'spread_CasesAtNotice_pc5'] = np.percentile(tmp[spread],
5,
axis=0)
trials.at[index,
'spread_CasesAtNotice_pc25'] = np.percentile(tmp[spread],
25,
axis=0)
trials.at[index,
'spread_CasesAtNotice_pc75'] = np.percentile(tmp[spread],
75,
axis=0)
trials.at[index,
'spread_CasesAtNotice_pc95'] = np.percentile(tmp[spread],
95,
axis=0)
print(datetime.datetime.now())
state = {
'bet': bet,
'tau': tau,
'ph': ph,
'init': init,
'T': T,
'sigma': sigma,
'n_runs': n_runs,
'n': n,
'cu': cu,
'p': p,
'ou': ou,
'r0t': r0t,
'r0inf': r0inf,
'y': y,
'x': x,
'r0theory': r0theory,
'norms': norms,
'mts': mts
}
sio.savemat(
"%s_%d_%.1f.mat" % (output_file, row['delay'], row['comp']), state)
trials.to_csv("%s_summ_tmp.csv" % (output_file), index=False)
print(datetime.datetime.now())
trials.to_csv("%s_summ.csv" % (output_file), index=False)
def main():
edge_file = sys.argv[1]
n_runs = int(sys.argv[2])
output_file = sys.argv[3]
# Define parameters
bet = np.array([0.0080, 0.0080, 0.0080])
tau = [5.35, 5.2, 5.0, 12.0, 12.0, 10.0]
ph = [.5, 1, 1]
init = 1
T = 100
sigma = [1, 1]
ndt = 1
mask_efficacy = 0.44
print(datetime.datetime.now())
# load and calculate contact matrix
edges = graph_ops.loadEdgeList(edge_file)
C, M, R, S = graph_ops.edgesAsMatrices(edges)
contacts = 12 / 7 * C + 1 / 420 * M + 1 / 4 * R + 1.0 * S
compliances = np.array(
[0., 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 1.])
trials = pd.DataFrame({
'comp': compliances,
'p_spread': 0.0,
'all_final_size_mean': 0.0,
'all_final_size_min': 0.0,
'all_final_size_max': 0.0,
'all_final_size_median': 0.0,
'all_final_size_pc5': 0.0,
'all_final_size_pc25': 0.0,
'all_final_size_pc75': 0.0,
'all_final_size_pc95': 0.0,
'spread_final_size_mean': 0.0,
'spread_final_size_min': 0.0,
'spread_final_size_max': 0.0,
'spread_final_size_median': 0.0,
'spread_final_size_pc5': 0.0,
'spread_final_size_pc25': 0.0,
'spread_final_size_pc75': 0.0,
'spread_final_size_pc95': 0.0,
'all_maxPrev_mean': 0.0,
'all_maxPrev_min': 0.0,
'all_maxPrev_max': 0.0,
'all_maxPrev_median': 0.0,
'all_maxPrev_pc5': 0.0,
'all_maxPrev_pc25': 0.0,
'all_maxPrev_pc75': 0.0,
'all_maxPrev_pc95': 0.0,
'spread_maxPrev_mean': 0.0,
'spread_maxPrev_min': 0.0,
'spread_maxPrev_max': 0.0,
'spread_maxPrev_median': 0.0,
'spread_maxPrev_pc5': 0.0,
'spread_maxPrev_pc25': 0.0,
'spread_maxPrev_pc75': 0.0,
'spread_maxPrev_pc95': 0.0,
'all_maxSymp_mean': 0.0,
'all_maxSymp_min': 0.0,
'all_maxSymp_max': 0.0,
'all_maxSymp_median': 0.0,
'all_maxSymp_pc5': 0.0,
'all_maxSymp_pc25': 0.0,
'all_maxSymp_pc75': 0.0,
'all_maxSymp_pc95': 0.0,
'spread_maxSymp_mean': 0.0,
'spread_maxSymp_min': 0.0,
'spread_maxSymp_max': 0.0,
'spread_maxSymp_median': 0.0,
'spread_maxSymp_pc5': 0.0,
'spread_maxSymp_pc25': 0.0,
'spread_maxSymp_pc75': 0.0,
'spread_maxSymp_pc95': 0.0,
'all_TimeToPeak_mean': 0.0,
'all_TimeToPeak_min': 0.0,
'all_TimeToPeak_max': 0.0,
'all_TimeToPeak_median': 0.0,
'all_TimeToPeak_pc5': 0.0,
'all_TimeToPeak_pc25': 0.0,
'all_TimeToPeak_pc75': 0.0,
'all_TimeToPeak_pc95': 0.0,
'spread_TimeToPeak_mean': 0.0,
'spread_TimeToPeak_min': 0.0,
'spread_TimeToPeak_max': 0.0,
'spread_TimeToPeak_median': 0.0,
'spread_TimeToPeak_pc5': 0.0,
'spread_TimeToPeak_pc25': 0.0,
'spread_TimeToPeak_pc75': 0.0,
'spread_TimeToPeak_pc95': 0.0,
'all_TimeToNotice_mean': 0.0,
'all_TimeToNotice_min': 0.0,
'all_TimeToNotice_max': 0.0,
'all_TimeToNotice_median': 0.0,
'all_TimeToNotice_pc5': 0.0,
'all_TimeToNotice_pc25': 0.0,
'all_TimeToNotice_pc75': 0.0,
'all_TimeToNotice_pc95': 0.0,
'spread_TimeToNotice_mean': 0.0,
'spread_TimeToNotice_min': 0.0,
'spread_TimeToNotice_max': 0.0,
'spread_TimeToNotice_median': 0.0,
'spread_TimeToNotice_pc5': 0.0,
'spread_TimeToNotice_pc25': 0.0,
'spread_TimeToNotice_pc75': 0.0,
'spread_TimeToNotice_pc95': 0.0,
'all_CasesAtNotice_mean': 0.0,
'all_CasesAtNotice_min': 0.0,
'all_CasesAtNotice_max': 0.0,
'all_CasesAtNotice_median': 0.0,
'all_CasesAtNotice_pc5': 0.0,
'all_CasesAtNotice_pc25': 0.0,
'all_CasesAtNotice_pc75': 0.0,
'all_CasesAtNotice_pc95': 0.0,
'spread_CasesAtNotice_mean': 0.0,
'spread_CasesAtNotice_min': 0.0,
'spread_CasesAtNotice_max': 0.0,
'spread_CasesAtNotice_median': 0.0,
'spread_CasesAtNotice_pc5': 0.0,
'spread_CasesAtNotice_pc25': 0.0,
'spread_CasesAtNotice_pc75': 0.0,
'spread_CasesAtNotice_pc95': 0.0
})
print(datetime.datetime.now())
appt_size_rng = graph_ops.student_roommates_dist()
for index, row in trials.iterrows():
bet_eff = (1 - mask_efficacy * row['comp']) * bet
#running the covid19_Net simulation, 30 repetitions
C_list = [contacts, contacts] #Weekday, weekend
C_ind = [0, 0, 0, 0, 0, 1, 1]
[n, cu, p, ou, r0t, r0inf, y, x, r0theory, norms,
mts] = repeated_runs(n_runs,
bet_eff,
tau,
ph,
init,
C_list,
T,
sigma,
ndt,
C_ind,
sim_offcampus_using=R,
off_rate=2.0)
max_time_step = cu.shape[1] - 1
fs = cu[:, max_time_step, 0]
spread = fs > 1
n_spread = np.sum(spread)
trials.at[index, 'p_spread'] = np.mean(spread)
trials.at[index, 'all_final_size_mean'] = np.mean(fs)
trials.at[index, 'all_final_size_min'] = np.min(fs)
trials.at[index, 'all_final_size_max'] = np.max(fs)
trials.at[index, 'all_final_size_median'] = np.median(fs)
trials.at[index, 'all_final_size_pc5'] = np.percentile(fs, 5, axis=0)
trials.at[index, 'all_final_size_pc25'] = np.percentile(fs, 25, axis=0)
trials.at[index, 'all_final_size_pc75'] = np.percentile(fs, 75, axis=0)
trials.at[index, 'all_final_size_pc95'] = np.percentile(fs, 95, axis=0)
if n_spread > 0:
trials.at[index, 'spread_final_size_mean'] = np.mean(fs[spread])
trials.at[index, 'spread_final_size_min'] = np.min(fs[spread])
trials.at[index, 'spread_final_size_max'] = np.max(fs[spread])
trials.at[index,
'spread_final_size_median'] = np.median(fs[spread])
trials.at[index,
'spread_final_size_pc5'] = np.percentile(fs[spread],
5,
axis=0)
trials.at[index,
'spread_final_size_pc25'] = np.percentile(fs[spread],
25,
axis=0)
trials.at[index,
'spread_final_size_pc75'] = np.percentile(fs[spread],
75,
axis=0)
trials.at[index,
'spread_final_size_pc95'] = np.percentile(fs[spread],
95,
axis=0)
prevalence = p[:, :, 0] + p[:, :, 1] + p[:, :, 2] + p[:, :,
3] + p[:, :, 4]
tmp = np.max(prevalence, axis=1)
trials.at[index, 'all_maxPrev_mean'] = np.mean(tmp)
trials.at[index, 'all_maxPrev_min'] = np.min(tmp)
trials.at[index, 'all_maxPrev_max'] = np.max(tmp)
trials.at[index, 'all_maxPrev_median'] = np.median(tmp)
trials.at[index, 'all_maxPrev_pc5'] = np.percentile(tmp, 5, axis=0)
trials.at[index, 'all_maxPrev_pc25'] = np.percentile(tmp, 25, axis=0)
trials.at[index, 'all_maxPrev_pc75'] = np.percentile(tmp, 75, axis=0)
trials.at[index, 'all_maxPrev_pc95'] = np.percentile(tmp, 95, axis=0)
if n_spread > 0:
trials.at[index, 'spread_maxPrev_mean'] = np.mean(tmp[spread])
trials.at[index, 'spread_maxPrev_min'] = np.min(tmp[spread])
trials.at[index, 'spread_maxPrev_max'] = np.max(tmp[spread])
trials.at[index, 'spread_maxPrev_median'] = np.median(tmp[spread])
trials.at[index, 'spread_maxPrev_pc5'] = np.percentile(tmp[spread],
5,
axis=0)
trials.at[index,
'spread_maxPrev_pc25'] = np.percentile(tmp[spread],
25,
axis=0)
trials.at[index,
'spread_maxPrev_pc75'] = np.percentile(tmp[spread],
75,
axis=0)
trials.at[index,
'spread_maxPrev_pc95'] = np.percentile(tmp[spread],
95,
axis=0)
symp = p[:, :, 1] + p[:, :, 3] + p[:, :, 4]
tmp = np.max(symp, axis=1)
trials.at[index, 'all_maxSymp_mean'] = np.mean(tmp)
trials.at[index, 'all_maxSymp_min'] = np.min(tmp)
trials.at[index, 'all_maxSymp_max'] = np.max(tmp)
trials.at[index, 'all_maxSymp_median'] = np.median(tmp)
trials.at[index, 'all_maxSymp_pc5'] = np.percentile(tmp, 5, axis=0)
trials.at[index, 'all_maxSymp_pc25'] = np.percentile(tmp, 25, axis=0)
trials.at[index, 'all_maxSymp_pc75'] = np.percentile(tmp, 75, axis=0)
trials.at[index, 'all_maxSymp_pc95'] = np.percentile(tmp, 95, axis=0)
if n_spread > 0:
trials.at[index, 'spread_maxSymp_mean'] = np.mean(tmp[spread])
trials.at[index, 'spread_maxSymp_min'] = np.min(tmp[spread])
trials.at[index, 'spread_maxSymp_max'] = np.max(tmp[spread])
trials.at[index, 'spread_maxSymp_median'] = np.median(tmp[spread])
trials.at[index, 'spread_maxSymp_pc5'] = np.percentile(tmp[spread],
5,
axis=0)
trials.at[index,
'spread_maxSymp_pc25'] = np.percentile(tmp[spread],
25,
axis=0)
trials.at[index,
'spread_maxSymp_pc75'] = np.percentile(tmp[spread],
75,
axis=0)
trials.at[index,
'spread_maxSymp_pc95'] = np.percentile(tmp[spread],
95,
axis=0)
tmp = np.argmax(prevalence, axis=1)
trials.at[index, 'all_TimeToPeak_mean'] = np.mean(tmp)
trials.at[index, 'all_TimeToPeak_min'] = np.min(tmp)
trials.at[index, 'all_TimeToPeak_max'] = np.max(tmp)
trials.at[index, 'all_TimeToPeak_median'] = np.median(tmp)
trials.at[index, 'all_TimeToPeak_pc5'] = np.percentile(tmp, 5, axis=0)
trials.at[index, 'all_TimeToPeak_pc25'] = np.percentile(tmp,
25,
axis=0)
trials.at[index, 'all_TimeToPeak_pc75'] = np.percentile(tmp,
75,
axis=0)
trials.at[index, 'all_TimeToPeak_pc95'] = np.percentile(tmp,
95,
axis=0)
if n_spread > 0:
trials.at[index, 'spread_TimeToPeak_mean'] = np.mean(tmp[spread])
trials.at[index, 'spread_TimeToPeak_min'] = np.min(tmp[spread])
trials.at[index, 'spread_TimeToPeak_max'] = np.max(tmp[spread])
trials.at[index,
'spread_TimeToPeak_median'] = np.median(tmp[spread])
trials.at[index,
'spread_TimeToPeak_pc5'] = np.percentile(tmp[spread],
5,
axis=0)
trials.at[index,
'spread_TimeToPeak_pc25'] = np.percentile(tmp[spread],
25,
axis=0)
trials.at[index,
'spread_TimeToPeak_pc75'] = np.percentile(tmp[spread],
75,
axis=0)
trials.at[index,
'spread_TimeToPeak_pc95'] = np.percentile(tmp[spread],
95,
axis=0)
symp_cum = np.cumsum(symp, axis=1)
ttn = np.sum(symp_cum == 0, axis=1)
tmp = ttn
trials.at[index, 'all_TimeToNotice_mean'] = np.mean(tmp)
trials.at[index, 'all_TimeToNotice_min'] = np.min(tmp)
trials.at[index, 'all_TimeToNotice_max'] = np.max(tmp)
trials.at[index, 'all_TimeToNotice_median'] = np.median(tmp)
trials.at[index, 'all_TimeToNotice_pc5'] = np.percentile(tmp,
5,
axis=0)
trials.at[index, 'all_TimeToNotice_pc25'] = np.percentile(tmp,
25,
axis=0)
trials.at[index, 'all_TimeToNotice_pc75'] = np.percentile(tmp,
75,
axis=0)
trials.at[index, 'all_TimeToNotice_pc95'] = np.percentile(tmp,
95,
axis=0)
if n_spread > 0:
trials.at[index, 'spread_TimeToNotice_mean'] = np.mean(tmp[spread])
trials.at[index, 'spread_TimeToNotice_min'] = np.min(tmp[spread])
trials.at[index, 'spread_TimeToNotice_max'] = np.max(tmp[spread])
trials.at[index,
'spread_TimeToNotice_median'] = np.median(tmp[spread])
trials.at[index,
'spread_TimeToNotice_pc5'] = np.percentile(tmp[spread],
5,
axis=0)
trials.at[index,
'spread_TimeToNotice_pc25'] = np.percentile(tmp[spread],
25,
axis=0)
trials.at[index,
'spread_TimeToNotice_pc75'] = np.percentile(tmp[spread],
75,
axis=0)
trials.at[index,
'spread_TimeToNotice_pc95'] = np.percentile(tmp[spread],
95,
axis=0)
basis_dates = np.clip(ttn + 2, 0, max_time_step)
cases_at_notice = prevalence[range(0, n_runs), basis_dates]
tmp = cases_at_notice
trials.at[index, 'all_CasesAtNotice_mean'] = np.mean(tmp)
trials.at[index, 'all_CasesAtNotice_min'] = np.min(tmp)
trials.at[index, 'all_CasesAtNotice_max'] = np.max(tmp)
trials.at[index, 'all_CasesAtNotice_median'] = np.median(tmp)
trials.at[index, 'all_CasesAtNotice_pc5'] = np.percentile(tmp,
5,
axis=0)
trials.at[index, 'all_CasesAtNotice_pc25'] = np.percentile(tmp,
25,
axis=0)
trials.at[index, 'all_CasesAtNotice_pc75'] = np.percentile(tmp,
75,
axis=0)
trials.at[index, 'all_CasesAtNotice_pc95'] = np.percentile(tmp,
95,
axis=0)
if n_spread > 0:
trials.at[index,
'spread_CasesAtNotice_mean'] = np.mean(tmp[spread])
trials.at[index, 'spread_CasesAtNotice_min'] = np.min(tmp[spread])
trials.at[index, 'spread_CasesAtNotice_max'] = np.max(tmp[spread])
trials.at[index,
'spread_CasesAtNotice_median'] = np.median(tmp[spread])
trials.at[index,
'spread_CasesAtNotice_pc5'] = np.percentile(tmp[spread],
5,
axis=0)
trials.at[index,
'spread_CasesAtNotice_pc25'] = np.percentile(tmp[spread],
25,
axis=0)
trials.at[index,
'spread_CasesAtNotice_pc75'] = np.percentile(tmp[spread],
75,
axis=0)
trials.at[index,
'spread_CasesAtNotice_pc95'] = np.percentile(tmp[spread],
95,
axis=0)
print(datetime.datetime.now())
state = {
'bet': bet,
'tau': tau,
'ph': ph,
'init': init,
'T': T,
'sigma': sigma,
'ndt': ndt,
'n_runs': n_runs,
'C': contacts,
'n': n,
'cu': cu,
'p': p,
'ou': ou,
'r0t': r0t,
'r0inf': r0inf,
'y': y,
'x': x,
'r0theory': r0theory,
'norms': norms,
'mts': mts
}
sio.savemat("%s_%d.mat" % (output_file, row['sz']), state)
print(datetime.datetime.now())
trials.to_csv("%s_summ.csv" % (output_file), index=False)
def main():
edge_file = sys.argv[1]
output_file = sys.argv[3]
#reading config file
json_file = open(sys.argv[2], "r", encoding="utf-8")
config = json.load(json_file)
json_file.close()
# Define parameters
bet = config['bet']
tau = config['tau']
ph = config['ph']
init = config['init']
T = config['T']
sigma = config['sigma']
ndt = config['ndt']
n_runs = config['n_runs']
scalarC = config['scalarC']
scalarM = config['scalarM']
scalarR = config['scalarR']
scalarS = config['scalarS']
print(datetime.datetime.now())
# load and calculate contact matrix
edges = graph_ops.loadEdgeList(edge_file)
C, M, R, S = graph_ops.edgesAsMatrices(edges)
contacts = scalarC * C + scalarM * M + scalarR * R + scalarS * S
class_sizes = graph_ops.calculateClassSizes(edges)
size_limits = np.sort(
np.concatenate((np.arange(10, 301, 10), np.arange(5, 101, 5))))
#size_limits = np.array([10, 30, 50])
trials = pd.DataFrame({
'sz': size_limits,
'p_spread': 0.0,
})
su.add_summary_cols(trials, 'all_final_size', 0.0)
su.add_summary_cols(trials, 'spread_final_size', 0.0)
su.add_summary_cols(trials, 'all_maxPrev', 0.0)
su.add_summary_cols(trials, 'spread_maxPrev', 0.0)
su.add_summary_cols(trials, 'all_maxSymp', 0.0)
su.add_summary_cols(trials, 'spread_maxSymp', 0.0)
su.add_summary_cols(trials, 'all_TimeToPeak', 0.0)
su.add_summary_cols(trials, 'spread_TimeToPeak', 0.0)
su.add_summary_cols(trials, 'all_TimeToNotice', 0.0)
su.add_summary_cols(trials, 'spread_TimeToNotice', 0.0)
su.add_summary_cols(trials, 'all_CasesAtNotice', 0.0)
su.add_summary_cols(trials, 'spread_CasesAtNotice', 0.0)
print(datetime.datetime.now())
n_students = C.shape[0] - 1
for index, row in trials.iterrows():
include = class_sizes['Info'][
class_sizes['Size'] <= row['sz']].to_list()
edges_pruned = edges[np.logical_and(
edges['Type'] == 'C', edges['Info'].isin(include))].copy()
C2, M2, R2, S2 = graph_ops.edgesAsMatrices(edges_pruned, n_students)
contacts = scalarC * C2 + scalarM * M + scalarR * R + scalarS * S
#running the covid19_Net simulation, 30 repetitions
C_list = [contacts, contacts] #Weekday, weekend
C_ind = [0, 0, 0, 0, 0, 1, 1]
[n, cu, p, ou, r0t, r0inf, y, x, r0theory, norms,
mts] = repeated_runs(n_runs,
bet,
tau,
ph,
init,
C_list,
T,
sigma,
ndt,
C_ind,
sim_offcampus_using=R,
off_rate=2.0)
max_time_step = cu.shape[1] - 1
fs = cu[:, max_time_step, 0]
spread = fs > 1
n_spread = np.sum(spread)
trials.at[index, 'p_spread'] = np.mean(spread)
su.calculate(trials, index, 'all_final_size', fs)
if n_spread > 0:
su.calculate(trials, index, 'spread_final_size', fs[spread])
prevalence = p[:, :, 0] + p[:, :, 1] + p[:, :, 2] + p[:, :,
3] + p[:, :, 4]
tmp = np.max(prevalence, axis=1)
su.calculate(trials, index, 'all_maxPrev', tmp)
if n_spread > 0:
su.calculate(trials, index, 'spread_maxPrev', tmp[spread])
symp = p[:, :, 1] + p[:, :, 3] + p[:, :, 4]
tmp = np.max(symp, axis=1)
su.calculate(trials, index, 'all_maxSymp', tmp)
if n_spread > 0:
su.calculate(trials, index, 'spread_maxSymp', tmp[spread])
tmp = np.argmax(prevalence, axis=1)
su.calculate(trials, index, 'all_TimeToPeak', tmp)
if n_spread > 0:
su.calculate(trials, index, 'spread_TimeToPeak', tmp[spread])
symp_cum = np.cumsum(symp, axis=1)
ttn = np.sum(symp_cum == 0, axis=1)
tmp = ttn
su.calculate(trials, index, 'all_TimeToNotice', tmp)
if n_spread > 0:
su.calculate(trials, index, 'spread_TimeToNotice', tmp[spread])
basis_dates = np.clip(ttn + 2, 0, max_time_step)
cases_at_notice = prevalence[range(0, n_runs), basis_dates]
tmp = cases_at_notice
su.calculate(trials, index, 'all_CasesAtNotice', tmp)
if n_spread > 0:
su.calculate(trials, index, 'spread_CasesAtNotice', tmp[spread])
print(datetime.datetime.now())
state = {
'bet': bet,
'tau': tau,
'ph': ph,
'init': init,
'T': T,
'sigma': sigma,
'ndt': ndt,
'n_runs': n_runs,
'C': contacts,
'n': n,
'cu': cu,
'p': p,
'ou': ou,
'r0t': r0t,
'r0inf': r0inf,
'y': y,
'x': x,
'r0theory': r0theory,
'norms': norms,
'mts': mts,
'mask_efficacy': config['mask_efficacy'],
'mask_compliance': config['mask_compliance']
}
sio.savemat("%s_%d.mat" % (output_file, row['sz']), state)
print(datetime.datetime.now())
trials.to_csv("%s_summ.csv" % (output_file), index=False)
def main():
edge_file = sys.argv[1]
output_file = sys.argv[3]
#reading config file
json_file = open(sys.argv[2], "r", encoding="utf-8")
config = json.load(json_file)
json_file.close()
# Define parameters
bet = config['bet']
tau = config['tau']
ph = config['ph']
init = config['init']
T = config['T']
sigma = config['sigma']
ndt = config['ndt']
n_runs = config['n_runs']
scalarC = config['scalarC']
scalarM = config['scalarM']
scalarR = config['scalarR']
scalarS = config['scalarS']
print(datetime.datetime.now())
# load and calculate contact matrix
edges = graph_ops.loadEdgeList(edge_file)
C, M, R, S = graph_ops.edgesAsMatrices(edges)
contacts = scalarC * C + scalarM * M + scalarR * R + scalarS * S
# print("Debug")
soc = graph_ops.generate_random_social_graph(C, R, 4, 0.1, 2)
# print(soc)
# print("soc shape:",soc.shape)
# print("soc type",type(soc))
# soc = np.array(soc)
# print("soc type",type(soc))
# for rows in soc:
# print("Testing")
# for elem in rows:
# print(type(elem))
contacts = soc + contacts
'''
print(contacts)
print("contacts shape:",contacts.shape)
print("contacts type",type(contacts))
for rows in contacts:
for elem in rows:
print(type(elem))
test = soc + contacts
print(test)
print("test shape:",test.shape)
for rows in test:
for elem in rows:
print(type(elem))
contacts = soc + contacts
print(contacts)
print("new contacts shape:",contacts.shape)
'''
print(datetime.datetime.now())
#running the covid19_Net simulation, 30 repetitions
C_list = [contacts, contacts] #Weekday, weekend
C_ind = [0, 0, 0, 0, 0, 1, 1]
[n, cu, p, ou, r0t, r0inf, y, x, r0theory, norms,
mts] = repeated_runs(n_runs,
bet,
tau,
ph,
init,
C_list,
T,
sigma,
ndt,
C_ind,
sim_offcampus_using=R,
off_rate=2.0)
print(datetime.datetime.now())
state = {
'bet': bet,
'tau': tau,
'ph': ph,
'init': init,
'T': T,
'sigma': sigma,
'ndt': ndt,
'n_runs': n_runs,
'C': contacts,
'n': n,
'cu': cu,
'p': p,
'ou': ou,
'r0t': r0t,
'r0inf': r0inf,
'y': y,
'x': x,
'r0theory': r0theory,
'norms': norms,
'mts': mts,
'mask_efficacy': config['mask_efficacy'],
'mask_compliance': config['mask_compliance']
}
sio.savemat(output_file, state)
print(datetime.datetime.now())