"""Routines and class for all types of diagrams, inherited by others."""

from builtins import range
from builtins import object, str

import copy
import numpy
import networkx as nx

[docs]def no_trace(matrices):
"""Select matrices with full 0 diagonal.

Args:
matrices (list): A list of adjacency matrices.

Returns:
(list): The adjacency matrices without non-zero diagonal elements.

>>> test_matrices = [[[0, 1, 2], [2, 0, 1], [5, 2, 0]], \
[[2, 2, 2], [1, 2, 3], [0, 0, 0]], \
[[0, 1, 3], [2, 0, 8], [2, 1, 0]]]
>>> no_trace(test_matrices)
[[[0, 1, 2], [2, 0, 1], [5, 2, 0]], [[0, 1, 3], [2, 0, 8], [2, 1, 0]]]

"""
traceless_matrices = []
for matrix in matrices:
for ind_i, line in enumerate(matrix):
if line[ind_i] != 0:
break
else:
traceless_matrices.append(matrix)
return traceless_matrices

[docs]def check_vertex_degree(matrices, three_body_use, nbody_max_observable,
canonical_only, vertex_id):
"""Check the degree of a specific vertex in a set of matrices.

Args:
three_body_use (bool): True if one uses three-body forces.
nbody_max_observable (int): Maximum body number for the observable.
canonical_only (bool): True if one draws only canonical diagrams.
vertex_id (int): The position of the studied vertex.

>>> test_matrices = [numpy.array([[0, 1, 2], [1, 0, 1], [0, 2, 0]]), \
numpy.array([[2, 0, 2], [1, 2, 3], [1, 0, 0]]), \
numpy.array([[0, 1, 3], [2, 0, 8], [2, 1, 0]])]
>>> check_vertex_degree(test_matrices, True, 3, False, 0)
>>> test_matrices #doctest: +NORMALIZE_WHITESPACE
[array([[0, 1, 2], [1, 0, 1], [0, 2, 0]]),
array([[2, 0, 2], [1, 2, 3], [1, 0, 0]])]
>>> check_vertex_degree(test_matrices, False, 2, False, 0)
>>> test_matrices #doctest: +NORMALIZE_WHITESPACE
[array([[0, 1, 2], [1, 0, 1], [0, 2, 0]])]

"""
authorized_deg = 
if three_body_use:
authorized_deg.append(6)
if not canonical_only or vertex_id == 0:
authorized_deg.append(2)
authorized_deg = tuple(authorized_deg)

for i_mat, matrix in reversed_enumerate(matrices):
vertex_degree = sum(matrix[index][vertex_id] + matrix[vertex_id][index]
for index in list(range(matrix.shape)))
vertex_degree -= matrix[vertex_id][vertex_id]

if (vertex_id != 0 and vertex_degree not in authorized_deg) \
or (vertex_id == 0 and vertex_degree > 2*nbody_max_observable):
del matrices[i_mat]

[docs]def topologically_distinct_diagrams(diagrams):
"""Return a list of diagrams all topologically distinct.

Args:
diagrams (list): The Diagrams of interest.

Returns:
(list): Topologically unique diagrams.

"""
iso = nx.algorithms.isomorphism
op_nm = iso.categorical_node_match('operator', False)
anom_em = iso.categorical_multiedge_match('anomalous', False)
for i_diag, diag in reversed_enumerate(diagrams):
graph = diag.graph
diag_io_degrees = diag.io_degrees
for i_comp_diag, comp_diag in reversed_enumerate(diagrams[:i_diag]):
if diag_io_degrees == comp_diag.io_degrees:
# Check anomalous character of props for PBMBPT
doubled_graph = create_checkable_diagram(graph)
doubled_comp_diag = create_checkable_diagram(comp_diag.graph)
matcher = iso.DiGraphMatcher(doubled_graph,
doubled_comp_diag,
node_match=op_nm,
edge_match=anom_em)
# Check for topologically equivalent diags considering vertex
# properties but not edge attributes, i.e. anomalous character
else:
matcher = iso.DiGraphMatcher(graph,
comp_diag.graph,
node_match=op_nm)
if matcher.is_isomorphic():
# Store the set of permutations to recover the original TSD
diag.perms.update(
update_permutations(comp_diag.perms,
comp_diag.tags,
matcher.mapping)
)
diag.tags += comp_diag.tags
del diagrams[i_comp_diag]
break
return diagrams

[docs]def update_permutations(comp_graph_perms, comp_graph_tag, mapping):
"""Update permutations associated to the BMBPT diags for a shared TSD.

Args:
comp_graph_perms (dict): Permutations to be updated.
comp_graph_tag (int): The tag associated to the TSD configuration.
mapping (dict): permutations to go from previous ref TSD to new one.

"""
identity = {key: key for key in comp_graph_perms[comp_graph_tag]}
# Do permutations only when necessary
if mapping != identity:
for graph_id in comp_graph_perms:
# Create a dummy dictionary to avoid overwriting some nodes
dummy_nodes = copy.deepcopy(comp_graph_perms[graph_id])
# Permute the nodes according to the new mapping
for node in comp_graph_perms[graph_id]:
comp_graph_perms[graph_id][node] = dummy_nodes[mapping[node]]

return comp_graph_perms

[docs]def create_checkable_diagram(pbmbpt_graph):
"""Return a graph with anomalous props going both ways for topo check.

Args:
pbmbpt_graph (NetworkX MultiDiGraph): The graph to be copied.

Returns:
(NetworkX MultiDiGraph): Graph with double the anomalous props.

"""
doubled_graph = copy.deepcopy(pbmbpt_graph)
props_to_add = [(prop, prop) for prop
in doubled_graph.edges(keys=True, data='anomalous')
if prop and not prop == prop]
return doubled_graph

[docs]def label_vertices(graphs_list, theory_type, switch_flag):
"""Account for different status of vertices in operator diagrams.

Args:
graphs_list (list): The Diagrams of interest.
theory_type (str): The name of the theory of interest.
switch_flag (int): When to switch A and B operators for BIMSRG.

"""
if theory_type != 'BIMSRG':
for graph in graphs_list:
nx.set_node_attributes(graph, False, 'operator')
if theory_type in ("BMBPT", "PBMBPT"):
graph.nodes['operator'] = True
else:
for idx, graph in enumerate(graphs_list):
nx.set_node_attributes(graph, '', 'operator')
if idx < switch_flag:
graph.nodes['operator'] = 'B'
graph.nodes['operator'] = 'A'
else:
graph.nodes['operator'] = 'A'
graph.nodes['operator'] = 'B'

[docs]def feynmf_generator(graph, theory_type, diagram_name):
"""Generate the feynmanmp instructions corresponding to the diagram.

Args:
graph (NetworkX MultiDiGraph): The graph of interest.
theory_type (str): The name of the theory of interest.
diagram_name (str): The name of the studied diagram.

"""
p_order = graph.number_of_nodes()
diag_size = 20*p_order

theories = ["MBPT", "BMBPT", "PBMBPT", "BIMSRG"]
prop_types = ["half_prop", "prop_pm", "prop_pm", "half_prop"]
propa = prop_types[theories.index(theory_type)]

fmf_file = open(diagram_name + ".tex", 'w')
fmf_file.write("\\parbox{40pt}{\\begin{fmffile}{%s}\n" % diagram_name
+ "\\begin{fmfgraph*}(40,%i)\n" % diag_size)

# Define the appropriate line propagator_style
fmf_file.write(propagator_style(propa))
if theory_type == "PBMBPT":
fmf_file.write(propagator_style("prop_mm"))

# Set the position of the vertices
if theory_type == "BIMSRG":
bimsrg_diagram_internals(graph, fmf_file, propa)

else:
fmf_file.write(vertex_positions(graph, p_order))

# Special config for self-contraction
if theory_type == "PBMBPT":
fmf_file.write(self_contractions(graph))

# Loop over all elements of the graph to draw associated propagators
for vert_i in graph:
for vert_j in list(graph.nodes())[vert_i+1:]:
props_to_draw = [edge for edge in graph.edges([vert_i, vert_j],
data=True,
keys=True)
if edge in (vert_i, vert_j)
and edge != edge]
# Set the list of propagators directions to use
props_dir = prop_directions(vert_j - vert_i, len(props_to_draw))
# Draw the diagrams
key = 0
for _ in (p for p in props_to_draw
if p < p
and not ('anomalous' in p and p['anomalous'])):
fmf_file.write("\\fmf{%s%s}{v%i,v%i}\n"
% (propa, props_dir[key], vert_j, vert_i))
key += 1
# Reinitialise the drawing configuration as we change direction
key = 0
for _ in (p for p in props_to_draw
if ('anomalous' in p and p['anomalous'])):
fmf_file.write("\\fmf{prop_mm%s}{v%i,v%i}\n"
% (props_dir[key], vert_i, vert_j))
key += 1
for _ in (p for p in props_to_draw
if p < p
and not ('anomalous' in p and p['anomalous'])):
fmf_file.write("\\fmf{%s%s}{v%i,v%i}\n"
% (propa, props_dir[key], vert_i, vert_j))
key += 1

fmf_file.write("\\end{fmfgraph*}\n\\end{fmffile}}\n")
fmf_file.close()

[docs]def prop_directions(vert_distance, nb_props):
"""Return a list of possible propagators directions.

Args:
vert_distance (int): Distance between the two connected vertices.
nb_props (int): Number of propagators to be drawn.

Returns:
(list): Propagators directions stored as strings.

"""
if nb_props < 7:
directions = [",right=0.9", ",right=0.75", ",right=0.6", ",right=0.5",
"", ",left=0.5", ",left=0.6", ",left=0.75", ",left=0.9"]

if vert_distance != 1:
props_dir = directions[:3] + directions[-3:]
else:
props_dir = directions[:2] + directions[3:6] + directions[-2:]
if nb_props % 2 != 1:
props_dir = props_dir[:3] + props_dir[-3:]
else:
props_dir = props_dir[1:]
if nb_props < 5:
props_dir = props_dir[1:-1]
if nb_props < 3:
props_dir = props_dir[1:-1]

elif vert_distance == 1:
directions = [",right=0.%i" % angle for angle in range(90, 0, -10)] \
+ [",left=0.%i" % angle for angle in range(10, 100, 10)]

props_dir = directions[nb_props//4:-(nb_props//4)]

return props_dir

[docs]def propagator_style(prop_type):
"""Return the FeynMF definition for the appropriate propagator type.

Args:
prop_type (str): The type of propagators used in the diagram.

Returns:
(str): The FeynMF definition for the propagator style used.

"""
line_styles = {}

line_styles['prop_pm'] = "\\fmfcmd{style_def prop_pm expr p =\n" \
+ "draw_plain p;\nshrink(.7);\n" \
+ "\tcfill (marrow (p, .25));\n" \
+ "\tcfill (marrow (p, .75))\n" \
+ "endshrink;\nenddef;}\n"

line_styles['prop_mm'] = "\\fmfcmd{style_def prop_mm expr p =\n" \
+ "draw_plain p;\nshrink(.7);\n" \
+ "\tcfill (marrow (p, .75));\n" \
+ "\tcfill (marrow (reverse p, .75))\n" \
+ "endshrink;\nenddef;}\n"

line_styles['half_prop'] = "\\fmfcmd{style_def half_prop expr p =\n" \
+ "draw_plain p;\nshrink(.7);\n" \
+ "\tcfill (marrow (p, .5))\n" \
+ "endshrink;\nenddef;}\n"

return line_styles[prop_type]

[docs]def vertex_positions(graph, order):
"""Return the positions of the graph's vertices.

Args:
graph (NetworkX MultiDiGraph): The graph of interest.
order (int): The perturbative order of the graph.

Returns:
(str): The FeynMP instructions for positioning the vertices.

"""
positions = "\\fmftop{v%i}\\fmfbottom{v0}\n" % (order-1)
for vert in range(order-1):
positions += "\\fmf{phantom}{v%i,v%i}\n" % (vert, (vert+1)) \
+ ("\\fmfv{d.shape=square,d.filled=full,d.size=3thick"
if 'operator' in graph.nodes[vert] and graph.nodes[vert]['operator']
else "\\fmfv{d.shape=circle,d.filled=full,d.size=3thick") \
+ "}{v%i}\n" % vert
positions += "\\fmfv{d.shape=circle,d.filled=full,d.size=3thick}{v%i}\n" \
% (order-1) + "\\fmffreeze\n"
return positions

[docs]def self_contractions(graph):
"""Return the instructions for drawing the graph's self-contractions.

Args:
graph (NetworkX MultiDiGraph): The graph being drawn.

Returns:
(str): FeynMF instructions for drawing the self-contractions.

"""
instructions = ""
# Check for self-contractions before going further
if [nx.selfloop_edges(graph)]:
instructions += propagator_style("half_prop")
for vert in graph:
props_to_draw = [edge for edge
in nx.selfloop_edges(graph, data=True, keys=True)
if edge == vert]
positions = ["15pt", "-15pt"]
key = 0
for prop in props_to_draw:
if prop['anomalous']:
a_name = "a%i%i" % (vert, key)
instructions += ("\\fmfv{}{%s}\n" % a_name
+ "\\fmffixed{(%s,0)}{v%i,%s}\n"
% (positions[key], vert, a_name)
+ "\\fmf{half_prop,right}{%s,v%i}\n"
% (a_name, vert)
+ "\\fmf{half_prop,left}{%s,v%i}\n"
% (a_name, vert))
key += 1
instructions += "\\fmffreeze\n"
return instructions

[docs]def bimsrg_diagram_internals(graph, fmf_file, prop_type):
"""Write to file the vertices and propagators of BIMSRG diagrams.

Args:
graph (NetworkX MultiDiGraph): The graph to be drawn.
fmf_file (file): The FeymanMF file to be written to.
prop_type (str): The FeymanMF type for drawing the propagators.

"""
nbs_out_edges = (sum(1 for p in graph.in_edges(3, keys=True) if p == 1),
sum(1 for p in graph.in_edges(3, keys=True) if p == 2))
nbs_in_edges = (sum(1 for p in graph.out_edges(0, keys=True) if p == 1),
sum(1 for p in graph.out_edges(0, keys=True) if p == 2))

nb_out_edges = len(graph.in_edges(3, keys=True))
nb_in_edges = len(graph.out_edges(0, keys=True))

nb_top_vertices = nb_out_edges if nb_out_edges == 1 \
else (max(2*nbs_out_edges, 2*nbs_out_edges) + 1)
nb_bot_vertices = nb_in_edges if nb_in_edges == 1 \
else (max(2*nbs_in_edges, 2*nbs_in_edges) + 1)

positions = "\\fmfstraight\n" \
+ "\\fmftopn{t}{%i}" % nb_top_vertices \
+ "\\fmfbottomn{b}{%i}\n" % nb_bot_vertices \
+ "\\fmf{phantom}{b%i,v1}\n" % (nb_bot_vertices//2 + 1) \
+ "\\fmf{phantom}{v1,v2}\n" \
+ "\\fmf{phantom}{v2,t%i}\n" % (nb_top_vertices//2 + 1) \
+ "\\fmfv{d.shape=circle,d.filled=%s,d.size=3thick}{v1}\n" \
% ('full' if graph.nodes['operator'] == 'A' else 'empty') \
+ "\\fmfv{d.shape=circle,d.filled=%s,d.size=3thick}{v2}\n" \
% ('full' if graph.nodes['operator'] == 'A' else 'empty') \
+ "\\fmffreeze\n"
fmf_file.write(positions)

# Internal lines
nb_props = sum(1 for edge in graph.edges(1, keys=True) if edge == 2)
# Set the list of propagators directions to use
props_dir = prop_directions(1, nb_props)
# Draw the propagators
for idx in range(nb_props):
fmf_file.write("\\fmf{%s%s}{v1,v2}\n" % (prop_type, props_dir[idx]))

# Incoming external line
for vertex in (1, 2):
nb_props = sum(1 for edge in graph.edges(0, keys=True)
if edge == vertex)
if (nb_bot_vertices == 1) and (vertex == 1):
orientation = ""
else:
orientation = ",left=0.4" if vertex == 2 else ",right=0.3"
# Draw the propagators
for key in range(nb_props):
fmf_file.write("\\fmf{%s%s}{b%i,v%i}\n"
% (prop_type,
orientation,
key+1 if vertex == 2 else nb_bot_vertices - key,
vertex))

# Outgoing external lines
for vertex in (1, 2):
nb_props = sum(1 for edge in graph.in_edges(3, keys=True)
if edge == vertex)
if (nb_top_vertices == 1) and (vertex == 2):
orientation = ""
else:
orientation = ",right=0.4" if vertex == 1 else ",left=0.3"
# Draw the propagators
for key in range(nb_props):
fmf_file.write("\\fmf{%s%s}{v%i,t%i}\n"
% (prop_type,
orientation,
vertex,
key+1 if vertex == 2 else nb_top_vertices - key))

[docs]def draw_diagram(directory, result_file, diagram_index, diag_type):
"""Copy the diagram feynmanmp instructions in the result file.

Args:
directory (str): The path to the output folder.
result_file (file): The LaTeX ouput file of the program.
diagram_index (str): The number associated to the diagram.
diag_type (str): The type of diagram used here.

"""
with open(directory+"/Diagrams/%s_%s.tex"
% (diag_type, diagram_index)) as diag_file:

[docs]def to_skeleton(graph):
"""Return the bare skeleton of a graph, i.e. only non-redundant links.

Args:
graph (NetworkX MultiDiGraph): The graph to be turned into a skeleton.

Returns:
(NetworkX MultiDiGraph): The skeleton of the initial graph.

"""
for vertex_a in graph:
for vertex_b in graph:
while graph.number_of_edges(vertex_a, vertex_b) > 1:
graph.remove_edge(vertex_a, vertex_b)
if len(list(nx.all_simple_paths(graph, vertex_a, vertex_b))) > 1:
while len(nx.shortest_path(graph, vertex_a, vertex_b)) == 2:
graph.remove_edge(vertex_a, vertex_b)
return graph

[docs]def extract_denom(start_graph, subgraph):
"""Extract the appropriate denominator using the subgraph rule.

Args:
start_graph (NetworkX MultiDiGraph): The studied graph.
subgraph (NetworkX MultiDiGraph): The subgraph used for this particular
denominator factor.

Returns:
(str): The denominator factor for this subgraph.

"""
denomin = r"\epsilon^{" \
+ "".join("%s"
% prop['qp_state']
for prop
in start_graph.out_edges(subgraph, keys=True, data=True)
if not subgraph.has_edge(prop, prop, prop)
and not ('anomalous' in prop and prop['anomalous'])) \
+ "}_{" \
+ "".join("%s"
% prop['qp_state']
for prop
in start_graph.in_edges(subgraph, keys=True, data=True)
if not subgraph.has_edge(prop, prop, prop)
and not ('anomalous' in prop and prop['anomalous'])) \
+ "".join("%s"
% prop['qp_state']
for prop
in start_graph.in_edges(subgraph, keys=True, data=True)
if subgraph.has_edge(prop, prop, prop)
and ('anomalous' in prop and prop['anomalous'])) \
+ "".join("%s"
% (prop['qp_state'].split("}") + "}")
for prop
in start_graph.in_edges(subgraph, keys=True, data=True)
if not subgraph.has_edge(prop, prop, prop)
and ('anomalous' in prop and prop['anomalous'])) \
+ "".join("%s"
% (prop['qp_state'].split("}") + "}")
for prop
in start_graph.out_edges(subgraph, keys=True, data=True)
if not subgraph.has_edge(prop, prop, prop)
and ('anomalous' in prop and prop['anomalous'])) \
+ "}"
return denomin

Args:
directory (str): The path to the output directory.
diagrams (list): All the diagrams.

"""
for idx, diagram in enumerate(diagrams):
mat_file.write("Diagram n: %i\n" % (idx + 1))
numpy.savetxt(mat_file,
nx.to_numpy_matrix(diagram.graph, dtype=int),
fmt='%d')
mat_file.write("\n")

[docs]class Diagram(object):
"""Describes a diagram with its related properties.

Attributes:
graph (NetworkX MultiDiGraph): The actual graph.
unsorted_degrees (tuple): The degrees of the graph vertices
degrees (tuple): The ascendingly sorted degrees of the graph vertices.
unsort_io_degrees (tuple): The list of in- and out-degrees for each
vertex of the graph, stored in a (in, out) tuple.
io_degrees (tuple): The sorted version of unsort_io_degrees.
max_degree (int): The maximal degree of a vertex in the graph.
tags (list): The tag numbers associated to a diagram.

"""

__slots__ = ('graph', 'unsort_degrees', 'degrees', 'unsort_io_degrees',
'io_degrees', 'max_degree', 'tags')

def __init__(self, nx_graph):
"""Generate a Diagram object starting from the NetworkX graph.

Args:
nx_graph (NetworkX MultiDiGraph): The graph of interest.

"""
self.graph = nx_graph
self.unsort_degrees = tuple(nx_graph.degree(node) for node in nx_graph)
self.degrees = tuple(sorted(self.unsort_degrees))
self.unsort_io_degrees = tuple((nx_graph.in_degree(node),
nx_graph.out_degree(node))
for node in nx_graph)
self.io_degrees = tuple(sorted(self.unsort_io_degrees))
self.max_degree = self.degrees[-1]
self.tags = 

[docs]    def write_graph(self, latex_file, directory, write_time):
"""Write the graph of the diagram to the LaTeX file.

Args:
latex_file (file): The LaTeX ouput file of the program.
directory (str): Path to the result folder.
write_time (bool): (Here to emulate polymorphism).

"""
latex_file.write('\n\\begin{center}\n')
draw_diagram(directory, latex_file, str(self.tags), 'diag')
latex_file.write('\n\\end{center}\n\n')