out_species.writerow(out_row_species)
file_species.close()
stop_sim = datetime.now()
elapsed_sim = stop_sim - start_sim
print 'time for simulation', elapsed_sim
#ecosystem.draw_species_distribution()
#layout = nx.graphviz_layout(net, prog='dot', args='-Gnodesep=.07, -Granksep=.1, -Grankdir=BT')
layout = nx.circular_layout(net)
#fig = figure()
#network_plot = fig.add_subplot(111)
#nx.draw_networkx(net, layout, ax=network_plot)
print 'S original =', net.order(), 'L original =', net.size(
), 'C original =', net.connectance()
#
# net_final = ecosystem.realised_net.copy()
# layout2 = nx.circular_layout(net_final)
#
# fig2 = figure()
# network_plot2 = fig2.add_subplot(111)
# nx.draw_networkx(net_final, layout2, ax=network_plot2)
#
# print 'S realised =', net_final.order(), 'L realised =', net_final.size(), 'C realised =', net_final.connectance()
#show()
class NetworkCreator():
'''
classdocs
'''
def __init__(self):
'''
Constructor
'''
self.rnd = Random()
self.rnd_uniform = uniform()
self.net = Network()
# self.invaders = []
def reset_state(self):
self.rnd = Random()
self.rnd_uniform = uniform()
self.net.clear()
def create_niche_model_network(self, S, C, prob=False):
'''
prob = False; whether to create the network following the probabilistic niche model
Niche model
This is an implementation of the niche model
'''
ns = self.rnd_uniform.rvs(size=S+POSSIBLE_INVADERS)
bet = (1/(2*C)) - 1
self.beta_dist = beta(1,bet)
producers = int(math.floor(S*PRIMARY_PRODS))
current_producers = 0
herbivores = int(math.floor(S*HERBIVORY))
current_herbivores = 0
predators = int(math.floor(S*TOP_PREDATORS))
current_predators = 0
#basal_to_top_links = 1 ## New constraint that this must be zero!
while self.net.size() == 0 or not nx.is_connected(self.net.to_undirected()) or current_producers < producers or current_herbivores < herbivores or current_predators < predators: #or max(self.net.get_shortest_chain_length().values()) < 3:
self.net.clear()
# self.invaders = []
basal = None
smallest_n = None
#here we obtain the fundamental niche values for each one of the species in the network
# n, c, r
for i in xrange(1,S+1):
self.net.add_node(i)
self.net.node[i]['n'] = float(ns[i-1]) #self.rnd_uniform.rvs()
self.net.node[i]['r'] = self.beta_dist.rvs() * self.net.node[i]['n']
self.net.node[i]['c'] = self.rnd.uniform((self.net.node[i]['r']/2), min(self.net.node[i]['n'], (1-(self.net.node[i]['r']/2))))
#the original value of c (commented bit) as originally presented in the Nature 2000 paper
#was changed after reading the niche model specification presented in the JAE 2008 paper
#the new specification ensures that the r of all the species always lie within the niche interval [0,1]
#self.net.node[i]['c'] = self.rnd.uniform(self.net.node[i]['r']/2, self.net.node[i]['n'])
self.net.node[i]['invader'] = False
if smallest_n == None or self.net.node[i]['n'] < smallest_n:
smallest_n = self.net.node[i]['n']
basal = i
#if INVASION:
# for i in xrange(S+1, (S+1)+(POSSIBLE_INVADERS)):
## invader = dict()
## invader['n'] = self.rnd_uniform.rvs()
## invader['r'] = self.beta_dist.rvs() * self.net.node[i]['n']
## invader['c'] = self.rnd.uniform(self.net.node[i]['r']/2, self.net.node[i]['n'])
## invader['invader'] = True
#
# self.net.add_node(i)
# self.net.node[i]['n'] = float(ns[i-1]) #self.rnd_uniform.rvs()
# self.net.node[i]['r'] = self.beta_dist.rvs() * self.net.node[i]['n']
# self.net.node[i]['c'] = self.rnd.uniform((self.net.node[i]['r']/2), min(self.net.node[i]['n'], (1-(self.net.node[i]['r']/2))))
# self.net.node[i]['invader'] = False
# self.invaders.append(invader)
self.net.node[basal]['r'] = 0.0
self._create_links_based_on_fundamental_niche()
tls = self.net.get_trophic_levels()
# replacement of links between trophic levels 0 and 3 (does not conform to niches):
top, top_preds = self.net.top_predators()
basal, basal_sps = self.net.basal()
#basal_to_top_links = 0
to_replace = []
for u,v in self.net.edges():
if u in basal_sps and v in top_preds and tls[v] == 3:
#basal_to_top_links += 1
self.net.remove_edge(u,v)
to_replace.append((u,v))
for pair in to_replace:
new_u = pair[1]
while new_u in basal_sps or new_u==pair[1]:
new_u = np.random.randint(1,61) # species indexing starts at 1
self.net.add_edge(new_u, pair[1])
# calculations to check trophic constraints:
current_herbivores = 0
current_producers = 0
current_predators = 0
for k in tls.keys():
if tls[k] == 1:
current_herbivores += 1
elif tls[k] == 0:
current_producers += 1
elif tls[k] == 3:
current_predators += 1
print("replaced %d links" %(len(to_replace)))
return self.net
def _create_links_based_on_fundamental_niche(self, nodes_to_link=None):
#based on the fundamental niche values obtained above we construct the network by adding
#the corresponding links according to the species' niche and feeding range
if nodes_to_link == None:
nodes_to_link = set(self.net.nodes())
for i in self.net.nodes():
if self.net.node[i]['r'] == 0.0:
continue
r_lower_bound = self.net.node[i]['c'] - (self.net.node[i]['r']/2)
r_upper_bound = self.net.node[i]['c'] + (self.net.node[i]['r']/2)
for j in self.net.nodes():
if i not in nodes_to_link and j not in nodes_to_link:
continue
if self.net.node[j]['n'] >= r_lower_bound and self.net.node[j]['n'] <= r_upper_bound:
self.net.add_edge(j,i)
#for disconnected or duplicated nodes
disc_nodes = set()
self_loops = set(self.net.selfloop_edges())
for i in nodes_to_link:
disconnected = False
if self.net.degree(i) == 0:
disconnected = True
elif self.net.in_degree(i) == 1 and (i,i) in self_loops:
# print 'producer with selfloop'
if self.net.out_degree(i) > 1:
self.net.remove_edge(i,i)
else:
disconnected = True
attrs = self.net.node[i]
self.net.remove_node(i)
self.net.add_node(i, attr_dict=attrs)
else:
i_succs = set(self.net.successors(i))
i_predecs = set(self.net.predecessors(i))
for j in self.net.nodes():
j_succs = self.net.successors(j)
j_predecs = self.net.predecessors(j)
if i_succs == j_succs and i_predecs == j_predecs:
disconnected = True
attrs = self.net.node[i]
self.net.remove_node(i)
self.net.add_node(i, attr_dict=attrs)
print 'duplicated node'
break
#we reassign the fundamental niche values to nodes that are disconnected or duplicated
if disconnected:
self.net.node[i]['n'] = self.rnd_uniform.rvs()
self.net.node[i]['r'] = self.beta_dist.rvs() * self.net.node[i]['n']
self.net.node[i]['c'] = self.rnd.uniform(self.net.node[i]['r']/2, self.net.node[i]['n'])
disc_nodes.add(i)
if len(disc_nodes) > 0:
self._create_links_based_on_fundamental_niche(disc_nodes)
return
class NetworkCreator():
'''
classdocs
'''
def __init__(self):
'''
Constructor
'''
self.rnd = Random()
self.rnd_uniform = uniform()
self.net = Network()
# self.invaders = []
def reset_state(self):
self.rnd = Random()
self.rnd_uniform = uniform()
self.net.clear()
def create_niche_model_network(self, S, C, prob=False):
'''
prob = False; whether to create the network following the probabilistic niche model
Niche model
This is an implementation of the niche model
'''
ns = self.rnd_uniform.rvs(size=S + POSSIBLE_INVADERS)
bet = (1 / (2 * C)) - 1
self.beta_dist = beta(1, bet)
producers = int(math.floor(S * PRIMARY_PRODS))
current_producers = 0
herbivores = int(math.floor(S * HERBIVORY))
current_herbivores = 0
predators = int(math.floor(S * TOP_PREDATORS))
current_predators = 0
while self.net.size() == 0 or not nx.is_connected(
self.net.to_undirected()
) or current_herbivores < herbivores or current_producers < producers or current_predators < predators: #or max(self.net.get_shortest_chain_length().values()) < 3:
self.net.clear()
# self.invaders = []
basal = None
smallest_n = None
#here we obtain the fundamental niche values for each one of the species in the network
# n, c, r
for i in range(1, S + 1):
self.net.add_node(i)
self.net.nodes[i]['n'] = float(ns[i -
1]) #self.rnd_uniform.rvs()
self.net.nodes[i]['r'] = self.beta_dist.rvs(
) * self.net.nodes[i]['n']
self.net.nodes[i]['c'] = self.rnd.uniform(
(self.net.nodes[i]['r'] / 2),
min(self.net.nodes[i]['n'],
(1 - (self.net.nodes[i]['r'] / 2))))
#the original value of c (commented bit) as originally presented in the Nature 2000 paper
#was changed after reading the niche model specification presented in the JAE 2008 paper
#the new specification ensures that the r of all the species always lie within the niche interval [0,1]
#self.net.node[i]['c'] = self.rnd.uniform(self.net.node[i]['r']/2, self.net.node[i]['n'])
self.net.nodes[i]['invader'] = False
if smallest_n == None or self.net.nodes[i]['n'] < smallest_n:
smallest_n = self.net.nodes[i]['n']
basal = i
#if INVASION:
# for i in xrange(S+1, (S+1)+(POSSIBLE_INVADERS)):
## invader = dict()
## invader['n'] = self.rnd_uniform.rvs()
## invader['r'] = self.beta_dist.rvs() * self.net.node[i]['n']
## invader['c'] = self.rnd.uniform(self.net.node[i]['r']/2, self.net.node[i]['n'])
## invader['invader'] = True
#
# self.net.add_node(i)
# self.net.node[i]['n'] = float(ns[i-1]) #self.rnd_uniform.rvs()
# self.net.node[i]['r'] = self.beta_dist.rvs() * self.net.node[i]['n']
# self.net.node[i]['c'] = self.rnd.uniform((self.net.node[i]['r']/2), min(self.net.node[i]['n'], (1-(self.net.node[i]['r']/2))))
# self.net.node[i]['invader'] = False
# self.invaders.append(invader)
self.net.nodes[basal]['r'] = 0.0
self._create_links_based_on_fundamental_niche()
current_herbivores = 0
current_producers = 0
current_predators = 0
tls = self.net.get_trophic_levels()
for k in tls.keys():
if tls[k] == 1:
current_herbivores += 1
elif tls[k] == 0:
current_producers += 1
elif tls[k] == 3:
current_predators += 1
return self.net
def _create_links_based_on_fundamental_niche(self, nodes_to_link=None):
#based on the fundamental niche values obtained above we construct the network by adding
#the corresponding links according to the species' niche and feeding range
if nodes_to_link == None:
nodes_to_link = set(self.net.nodes())
for i in self.net.nodes():
if self.net.nodes[i]['r'] == 0.0:
continue
r_lower_bound = self.net.nodes[i]['c'] - (self.net.nodes[i]['r'] /
2)
r_upper_bound = self.net.nodes[i]['c'] + (self.net.nodes[i]['r'] /
2)
for j in self.net.nodes():
if i not in nodes_to_link and j not in nodes_to_link:
continue
if self.net.nodes[j]['n'] >= r_lower_bound and self.net.nodes[
j]['n'] <= r_upper_bound:
self.net.add_edge(j, i)
#for disconnected or duplicated nodes
disc_nodes = set()
self_loops = set(nx.selfloop_edges(self.net))
for i in nodes_to_link:
disconnected = False
if self.net.degree(i) == 0:
disconnected = True
elif self.net.in_degree(i) == 1 and (i, i) in self_loops:
# print 'producer with selfloop'
if self.net.out_degree(i) > 1:
self.net.remove_edge(i, i)
else:
disconnected = True
attrs = self.net.nodes[i]
self.net.remove_node(i)
self.net.add_node(i, attr_dict=attrs)
else:
i_succs = set(self.net.successors(i))
i_predecs = set(self.net.predecessors(i))
for j in self.net.nodes():
j_succs = self.net.successors(j)
j_predecs = self.net.predecessors(j)
if i_succs == j_succs and i_predecs == j_predecs:
disconnected = True
attrs = self.net.nodes[i]
self.net.remove_node(i)
self.net.add_node(i, attr_dict=attrs)
print('duplicated node')
break
#we reassign the fundamental niche values to nodes that are disconnected or duplicated
if disconnected:
self.net.nodes[i]['n'] = self.rnd_uniform.rvs()
self.net.nodes[i]['r'] = self.beta_dist.rvs(
) * self.net.nodes[i]['n']
self.net.nodes[i]['c'] = self.rnd.uniform(
self.net.nodes[i]['r'] / 2, self.net.nodes[i]['n'])
disc_nodes.add(i)
if len(disc_nodes) > 0:
self._create_links_based_on_fundamental_niche(disc_nodes)
return
tps = cumulative_sps_stats[sp]['tps']
if len(tps) == 0:
mean_tps = 'N/A'
sd_tps = 'N/A'
else:
mean_tps = sum(tps) / len(tps)
sd_tps = 0.0
for n in tps:
sd_tps += (n - mean_tps)**2
sd_tps = math.sqrt(sd_tps / len(tps))
out_row_species['mean_tp'] = mean_tps
out_row_species['tp_sd'] = sd_tps
out_row_species['individuals'] = series_counts[ITERATIONS][sp]
out_row_species['immigrants'] = cumulative_sps_stats[sp]['immigrants']
out_row_species['born'] = cumulative_sps_stats[sp]['born']
out_row_species['dead'] = cumulative_sps_stats[sp]['dead']
out_species.writerow(out_row_species)
file_species.close()
stop_sim = datetime.now()
elapsed_sim = stop_sim - start_sim
print('time for simulation', elapsed_sim)
print('S original =', net.order(), 'L original =', net.size(),
'C original =', net.connectance())