def test_extinction(rerun=0,measure=0,nsys=10,show=0):
#Comparison between competitive, trophic, disordered
path=Path('test_extinction')
try:
if not os.listdir(path):
rerun=1
except:
rerun =1
if rerun:
#If not already run before
kwargs={}
prm=deepcopy(dft_prm)
loop(path=path,axes=axes,check_key=0,print_msg=1,tmax=500000,tsample=10000.,
#keep='all',
parameters=prm,dynamics=dft_dyn,
**kwargs )
if rerun or measure:
rebuild_filelist(path)
make_measures(path,use_measures=['usual','effective','abundance',
'matrix',
'removal',
'testcavity',
measure_cavity,
])
if not show:
return
axes=['n_shape']
measures=open_measures(path)
measures=measures[measures['avgN']!=0]
show_hist(measures=measures,axes=axes, values=['n_abundance'] )
from cavity import cavity_distri
xs=np.linspace(0,0.5,100)
sigma,mu,gamma=[measures['n_couplings_'+z].mean() for z in ('sigma', 'mu','gamma') ]
dist=cavity_distri(S=2000,sigma=sigma,sigma_k=0,mu=mu,gamma=gamma)
plot(xs,[dist(x) for x in xs],hold=1,linewidth=2 )
plt.xlabel(r'Abundance $n_i$')
plot_main(measures=measures,axes=axes)
plot_cavity(measures=measures, axes=axes,)
def intersection(rerun=1, measure=0, nsys=100, show=0):
#Comparison between competitive trophic
path = Path('interp_intersection')
#(model_resource_generalists, {'vars':('R','X'), 'S':6, 'R':np.logspace(.8,.81,5), 'X':np.linspace(.9909,.991,5), 'sigma_rho':.25 } ),
#(model_trophic_satur, {'vars':('m0','e'),'sigma_k':0.05,'S':2000, 'm0':np.linspace(12.1,12.2,6),'e':np.linspace(0.01,.03,11),'s0':2, } ),
kwargs = {}
prm = deepcopy(dft_prm)
kwargs['r_shape'] = (6, )
kwargs['n_shape'] = (6, )
kwargs['r_std'] = 0.25
cmean = prm['consumption']['mean']
prm['consumption']['save'] = False
prm['r']['save'] = False
X = 0.991
kwargs['consumption_std'] = np.sqrt(cmean**2 * (1 / X - 1))
axes = [('sys', range(nsys))]
if rerun:
loop(
path=path,
axes=axes,
check_key=0,
print_msg=1,
converge=1,
tmax=500000,
tsample=15000.,
#keep='all',
parameters=prm,
dynamics=dft_dyn,
**kwargs)
if rerun or measure:
rebuild_filelist(path)
make_measures(path,
use_measures=[
'usual',
'effective',
'abundance',
measure_cavity,
],
update_prev=0)
measures = open_measures(path)
kwargs = {}
show_hist(measures=measures,
axes=['n_shape'],
values=['n_abundance'],
log='y',
bins=30)
from cavity import cavity_distri
xs = np.linspace(0, 0.5, 100)
sigma, mu, gamma = [
measures['n_couplings_' + z].mean() for z in ('sigma', 'mu', 'gamma')
]
dist = cavity_distri(S=2000, sigma=sigma, sigma_k=0, mu=mu, gamma=gamma)
plot(xs, [dist(x) for x in xs], hold=1, linewidth=2)
plt.xlabel(r'Abundance $n_i$')
plt.show()
def gamma_role(rerun=0, measure=0, nsys=10, show=0):
#Comparison between competitive, trophic, disordered
path = Path('interp_gamma_cavity')
try:
if not os.listdir(path):
rerun = 1
except:
rerun = 1
if rerun:
#If not already run before
kwargs = {}
prm = deepcopy(dft_prm)
prm['fluxes'] = {
'type': 'matrix',
'variables': [('n', 'com'), ('n', 'com')],
'role': 'fluxes',
'save': False,
'mean': 50. / 1600.,
'std': 0,
'sign': 1,
}
prm['commtroph'] = {
'type': 'matrix',
'variables': [('n', 'com'), ('n', 'com')],
'role': 'interactions',
'structure': 'trophic',
'requires': ['fluxes'],
#'efficiency':0.724,
'efficiency': 0.6,
'ordered': 0,
'dynamics': None,
}
prm['community'].update({
#'distribution':'exponential'
})
#specs=[ ((16,),None,'nlv'),((5625,),'nlv',None) ][1:]
specs = [((300, ), None, 'nlv'), ((400, ), None, 'nlv'),
((600, ), None, 'nlv'), ((1600, ), 'nlv', None)] #[:-1]
axes = [(('n_shape', 'commtroph_dynamics', 'community_dynamics'),
specs), ('sys', range(10))]
loop(
path=path,
axes=axes,
check_key=0,
print_msg=1,
tmax=500000,
tsample=10000.,
#keep='all',
parameters=prm,
dynamics=dft_dyn,
**kwargs)
if rerun or measure:
rebuild_filelist(path)
make_measures(path,
use_measures=[
'usual', 'effective', 'abundance', 'matrix',
measure_cavity, 'testcavity'
]) #,'trophic','assembly_corr'] )
if not show:
return
axes = ['n_shape']
measures = open_measures(path)
measures = measures[measures['avgN'] != 0]
show_hist(measures=measures, axes=axes, values=['n_abundance'])
from cavity import cavity_distri
xs = np.linspace(0, 0.5, 100)
sigma, mu, gamma = [
measures['n_couplings_' + z].mean() for z in ('sigma', 'mu', 'gamma')
]
dist = cavity_distri(S=2000, sigma=sigma, sigma_k=0, mu=mu, gamma=gamma)
plot(xs, [dist(x) for x in xs], hold=1, linewidth=2)
plt.xlabel(r'Abundance $n_i$')
plot_main(measures=measures, axes=axes)
plot_cavity(
measures=measures,
axes=axes,
)
if 0:
plot_data(
measures=measures,
axes=axes,
#split_by=['community_symmetry'],
values=[('press', {
'style': 'plot',
'linewidth': 2
}), ('n_matrix_press', {
'style': 'scatter'
})],
hold=1)
plt.ylabel(r'Response to press$')
plt.xlabel(r'Pool size $S$')
#show_data(measures=measures,axes=axes, values= ['n_%alive','phi','avgN','stdN','n_biomass','n_biomass_std','n_interactions_mu' ,'n_interactions_sigma'])
if 0:
compare = {
'phi': 'n_%alive',
'avgN': 'n_biomass',
'stdN': 'n_biomass_std',
'press': 'n_matrix_press',
}
for trait in ('phi', 'avgN', 'stdN', 'press'):
plot_data(measures=measures,
axes=axes,
hold=1,
newfig=1,
title=trait,
values=[(trait, {
'style': 'plot',
'log': 'x'
}), (compare[trait], {
'style': 'scatter'
})])
plt.show()
def intersection(path=Path('interp_intersection'),
rerun=0,
nsys=10,
measure=0,
show=0,
hold=0):
#Comparison between competitive, trophic
if 0:
compare_prms_models([
(model_trophic_satur, {
'vars': ('m0', 'e'),
'm0': np.linspace(20, 20.1, 10),
'e': np.linspace(0, 0.001, 10)[1:-1],
'S': 100,
's0': 0.,
'sigma_k': .3
}),
],
mode='scatter')
try:
if not os.listdir(path):
rerun = 1
except:
rerun = 1
if rerun:
#If not already run before
kwargs = {}
prm = deepcopy(dft_prm)
#(model_trophic_satur, {'vars':('m0','e'),'m0':np.linspace(20,20.1,10),'e':np.linspace(0,0.001,10)[1:-1],
#'S':100,'s0':1.,'sigma_k':.3 } ),
prm['fluxes'] = {
'type': 'matrix',
'structure': 'random',
#'distribution':'exponential',
'variables': [('n', 'com'), ('n', 'com')],
'role': 'fluxes',
'save': False,
'sign': 1,
'mean': 20. / 100.,
'std': 1. / 100,
}
prm['commtroph'] = {
'type': 'matrix',
'variables': [('n', 'com'), ('n', 'com')],
'role': 'interactions',
'structure': 'trophic',
'requires': ['fluxes'],
'efficiency': 0.,
'ordered': 0,
'dynamics': None,
'diagonal': 0,
'mean': 1,
}
prm['community'].update({
'mean': 10.,
'std': 1.,
'symmetry': -1.,
'diagonal': 0,
})
prm['growth']['std'] = 0
prm['n']['shape'] = (100, )
specs = [(None, 'nlv'), ('nlv', None)]
axes = [(('commtroph_dynamics', 'community_dynamics'), specs[1:]),
('sys', range(nsys))]
loop(path=path,
axes=axes,
check_key=0,
print_msg=1,
tmax=500000,
tsample=10000.,
converge=1,
parameters=prm,
dynamics=dft_dyn,
**kwargs)
if rerun or measure:
rebuild_filelist(path)
make_measures(
path,
use_measures=[
'usual',
'effective',
'abundance',
#'matrix',
measure_cavity, #'testcavity'
]) #,'trophic','assembly_corr'] )
if not show:
return
axes = ['commtroph_dynamics']
dico = {
'commtroph_dynamics': 'community',
0: 'competition',
1: 'predation',
'n_abundance': 'Species abundance'
}
measures = open_measures(path)
measures['commtroph_dynamics'][measures['commtroph_dynamics'].isnull()] = 0
measures['commtroph_dynamics'][measures['commtroph_dynamics'] == 'nlv'] = 1
hists = show_hist(measures=measures,
axes=axes,
values=['n_abundance'],
dictionary=dico,
log='',
bins=30)
from cavity import cavity_distri
xs = np.linspace(0, np.max([np.max(z) for z in measures['n_abundance']]),
100)
sigma, mu, gamma = [
measures['n_couplings_' + z].mean() for z in ('sigma', 'mu', 'gamma')
]
#sigma,mu,gamma=1,10,-1
dist = cavity_distri(S=2000,
sigma=sigma,
sigma_k=measures['n_capacity_std'].mean(),
mu=mu,
gamma=gamma)
plot(xs, [dist(x) for x in xs], hold=1, linewidth=2, log='')
#plt.xlabel(r'Abundance $n_i$')
# print np.sum(hists['n_abundance'][1][1:])*1./np.sum(hists['n_abundance'][1]),np.sum([dist(x) for x in xs[:-1]]*(xs[1:]-xs[:-1]) )
# plt.bar(0,np.sum(hists['n_abundance'][1][1])*(1-np.sum([dist(x) for x in xs[:-1]]*(xs[1:]-xs[:-1]) )), width=0.01 )
#plot_main(measures=measures,axes=axes)
#plot_cavity(measures=measures, axes=axes,)
if not hold:
plt.show()
def ordering(path=Path('ordering'),
mode='dft',
struct='all',
rerun=0,
measure=0,
nsys=3,
sysmin=0,
show=0,
funcresp=False,
**kw):
if struct == 'all':
struct = [
'triangular', 'nested', 'bipartite', 'scalefree', 'assortativity',
'clustering', 'connectivity'
]
if hasattr(struct, '__iter__'):
for st in struct:
ordering(path=path,
mode=mode,
struct=st,
rerun=rerun,
measure=0,
nsys=nsys,
sysmin=sysmin,
show=0,
funcresp=funcresp)
ordering(path=path,
mode=mode,
struct=None,
rerun=0,
measure=measure,
nsys=nsys,
sysmin=sysmin,
show=show,
funcresp=funcresp)
return
if mode != 'dft':
path = Path(path.strip() + '_{}'.format(mode))
try:
if not os.listdir(str(path)):
rerun = 1
except OSError as e:
print e
rerun = 1
kwargs = {}
prm = deepcopy(dft_prm)
prm['growth']['std'] = 0.25
prm['n']['shape'] = (120, )
prm['community']['stdrel'] = .3
prm['community']['diagonal'] = 0
prm['community']['structure'] = 'random'
prm['edges'] = {
'type': 'matrix',
'variables': [('n', 'com'), ('n', 'com')],
'mode': 'network',
'connectivity': .2,
'role': 'edges',
'save': False,
'structure': 'random',
}
if funcresp:
prm['threshold'] = {
'type': 'matrix',
'variables': ['n'],
'dynamics': 'nlv',
'role': 'threshold',
'mean': 5.,
'std': 0.,
'sign': 1,
}
dyn = {
'nlv': {
'type': 'simplefr',
'variables': [
('n', 'com'),
],
},
}
else:
dyn = deepcopy(dft_dyn)
if mode == 'competition':
prm['community']['mean'] = -0.04
prm['community']['symmetry'] = .9
prm['growth']['sign'] = 1
elif mode == 'predation':
prm['final'] = prm.pop('community')
prm['community'] = {
'type': 'matrix',
'variables': [('n', 'com'), ('n', 'com')],
'role': 'fluxes',
'mean': 0.05,
'stdrel': .3,
'save': False,
}
prm['final']['requires'] = ['community']
prm['final']['efficiency'] = 0.2
prm['final']['structure'] = 'trophic'
prm['final']['ordered'] = 0
prm['final'].pop('symmetry', None)
prm['growth']['sign'] = 0
elif mode == 'mutualism':
if not funcresp:
prm['community']['rowmax'] = 0.5
else:
prm['community']['mean'] = 0.05
prm['community']['symmetry'] = 0.3
prm['growth']['sign'] = 0
#prm['growth']['std']=0.25
prm['community']['requires'] = ['edges']
prm['community']['edgefrom'] = 'edges'
order = np.linspace(0, 1, 11)
axes = []
if struct == 'nested':
prm['edges']['structure'] = 'nested'
#prm['community']['connectivity']=.6
axes = [
('edges_nestedness', order),
]
elif struct == 'triangular':
prm['edges']['structure'] = 'triangular'
#prm['community']['connectivity']=.6
axes = [
('edges_triangular', order[6:]),
]
elif struct == 'bipartite':
prm['edges']['structure'] = 'multipartite'
axes = [
('edges_partition', order),
]
elif struct == 'scalefree':
prm['edges']['structure'] = 'barabasi'
axes = [
('edges_order', order),
]
prm['edges']['connectivity'] = prm['edges']['connectivity'] / 3.
prm['community']['mean'] = prm['community']['mean'] * 3.
elif struct == 'assortativity':
axes = [
('edges_assortativity', np.linspace(-1, 1, 2 * len(order))),
]
elif struct == 'clustering':
axes = [
('edges_clustering', np.linspace(-1, 1, 2 * len(order))),
]
elif struct == 'connectivity':
axes = [
('edges_connectivity', np.linspace(0.05, 1, len(order))),
]
elif not struct is None:
raise Exception('Unknown structure: {}'.format(struct))
axes += [('sys', range(sysmin, nsys + sysmin))]
if not struct is None:
axes[0] = (axes[0][0], 'community_order'), zip(axes[0][1],
np.abs(axes[0][1]))
kwargs['community_label'] = struct
if rerun:
loop(path=path,
axes=axes,
check_key=0,
print_msg=1,
tmax=500000,
ftol=0.0002,
tsample=10000.,
parameters=prm,
dynamics=dyn,
converge=1,
**kwargs)
if rerun or measure:
rebuild_filelist(path)
make_measures(
path,
use_measures=[
'usual',
'effective',
ifelse(funcresp, measure_funcresp, 'usual'),
ifelse(funcresp, measure_funcresp_cavity, measure_cavity),
#'matrix',
#'degree',
#measure_ordered,
#measure_ordered_cavity,
# measure_groups,measure_groups_cavity,
measure_cavity_compare,
#'abundance',
],
# read_filter=ifelse( struct is None,None,('community_label','==',struct)),
death=10**-10)
if not show:
return
axes = ['community_order']
split_by = ['community_label']
measures = open_measures(path)
dico = {
'community_edgemean': r'$s$',
'n_shape': r'$S$',
'community_nestedness': 'nestedness',
'community_partition': 'partition',
'edges_structure': 'network',
'community_structure': 'structure',
'cavity_compare': 'Distance to reference',
'community_order': 'Ordering',
'multipartite': 'bipartite',
'random': 'scale-free'
}
plot_modelprm(measures=measures,
axes=axes,
split_by=split_by,
dictionary=dico)
#plt.show()
plot_main(measures=measures,
axes=axes,
split_by=split_by,
dictionary=dico,
aggregator='mean',
log='')
plot_data(measures=measures,
axes=axes,
split_by=split_by,
hold=1,
values=[('n_couplings_n_std', {
'style': 'plot'
}), ('n_couplings_n_mean', {
'style': 'plot',
'color': 'r'
})])
plot_data(measures=measures,
axes=axes,
split_by=split_by,
hold=1,
values=['n_connectivity'])
if 0:
plot_data(measures=measures,
axes=axes,
split_by=split_by,
hold=1,
values=['n_degree_std'])
plot_data(
measures=measures,
axes=axes,
split_by=split_by,
hold=1,
values=[('n_biomass_tot', {
'style': 'scatter'
}), ('totN', {
'style': 'plot'
})],
dictionary=dico,
title='{}: Total biomass'.format(mode.capitalize()),
)
plot_data(
measures=measures,
axes=axes,
split_by=split_by,
hold=1,
values=['cavity_compare'],
dictionary=dico,
title=mode.capitalize(),
)
plt.ylim(ymin=0, ymax=.5)
if 0:
#ABUNDANCE HIST
if struct != 'all':
measures = measures[measures['community_structure'] == struct]
show_hist(measures=measures[measures['community_order'] == 1],
axes=axes,
values=['n_abundance'],
dictionary=dico,
log='y')
sigma, mu, gamma = [
measures['n_couplings_' + z].mean()
for z in ('sigma', 'mu', 'gamma')
]
zeta = measures['n_capacity_std'].mean()
from cavity import cavity_distri
print sigma, mu, gamma
dist = cavity_distri(S=measures['n_shape'][0][0],
sigma=sigma,
sigma_k=zeta,
mu=mu,
gamma=gamma)
xs = np.linspace(0,
np.max([np.max(z) for z in measures['n_abundance']]),
100)
plot(xs, [dist(x) for x in xs], hold=1, linewidth=2, log='y')
if not hold:
plt.show()