Skip to content

Commit

Permalink
Tests fixed
Browse files Browse the repository at this point in the history
  • Loading branch information
@yuvalTrip
yuvalTrip committed Aug 30, 2024
1 parent 9e5fc58 commit f7fdeef
Show file tree
Hide file tree
Showing 3 changed files with 98 additions and 169 deletions.
113 changes: 65 additions & 48 deletions fairpyx/algorithms/Optimization_Matching/FaStGen.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -19,7 +19,7 @@ def FaStGen(alloc: AllocationBuilder, items_valuations:dict)->dict:
:param alloc: an allocation builder, which tracks the allocation and the remaining capacity for items and agents.
:param items_valuations: a dictionary represents how items valuates the agents
>>> from fairpyx.adaptors import divide
>>> S = ["s1", "s2", "s3", "s4", "s5", "s6", "s7"]
>>> C = ["c1", "c2", "c3", "c4"]
Expand All @@ -31,10 +31,18 @@ def FaStGen(alloc: AllocationBuilder, items_valuations:dict)->dict:
{'c1': ['s1', 's2', 's3'], 'c2': ['s4'], 'c3': ['s5'], 'c4': ['s7', 's6']}
"""
logger.info("Starting FaStGen algorithm")

agents = alloc.instance.agents
items = alloc.instance.items
agents_valuations = alloc.instance._valuations

if agents == []:
raise ValueError(f"Agents list can't be empty")
if items == []:
raise ValueError(f"Items list can't be empty")
if items_valuations == {}:
raise ValueError(f"Items valuations list can't be empty")

#Convert the list of the agents and item to dictionary so that each agent\item will have its coresponding integer
student_name_to_int = generate_dict_from_str_to_int(agents)
college_name_to_int = generate_dict_from_str_to_int(items)
Expand All @@ -51,7 +59,7 @@ def FaStGen(alloc: AllocationBuilder, items_valuations:dict)->dict:
LowerFix = [college_name_to_int[items[len(items)-1]]] # Inidces of colleges from which we cannot remove any more students.
SoftFix = []
UnFixed = [item for item in college_name_to_int.values() if item not in UpperFix]


#creating a dictionary of vj(µ) = Pi∈µ(cj) for each j in C
matching_college_valuations = update_matching_valuations_sum(match=str_match, items_valuations=items_valuations)
Expand All @@ -62,10 +70,10 @@ def FaStGen(alloc: AllocationBuilder, items_valuations:dict)->dict:
up = min([j for j in college_name_to_int.values() if j not in LowerFix])
down = min(UnFixed, key=lambda j: matching_college_valuations[int_to_college_name[j]])
logger.debug(f"up: {up}, down: {down}")

SoftFix = [pair for pair in SoftFix if not (pair[1] <= up < pair[0])]
logger.debug(f"Updating SoftFix to {SoftFix}")

logger.debug(f"vup(mu)={matching_college_valuations[int_to_college_name[up]]}, vdown(mu)={matching_college_valuations[int_to_college_name[down]]}")
if (len(integer_match[up]) == 1) or (matching_college_valuations[int_to_college_name[up]] <= matching_college_valuations[int_to_college_name[down]]):
LowerFix.append(up)
Expand All @@ -84,25 +92,25 @@ def FaStGen(alloc: AllocationBuilder, items_valuations:dict)->dict:
logger.info(f"Old match leximin tuple: {match_leximin_tuple}")
new_match_leximin_tuple = create_leximin_tuple(match=new_match_str, agents_valuations=agents_valuations, items_valuations=items_valuations)
logger.info(f"New match leximin tuple: {new_match_leximin_tuple}")

#Extarcting from the SourceDec function the problematic item or agent, if there isn't one then it will be ""
problematic_component = sourceDec(new_match_leximin_tuple=new_match_leximin_tuple, old_match_leximin_tuple=match_leximin_tuple)
logger.info(f" problematic component: {problematic_component}")

if problematic_component == "":
logger.debug(f"New match is leximin-better than old match:")
integer_match = new_integer_match
integer_match = new_integer_match
str_match = new_match_str
matching_college_valuations = update_matching_valuations_sum(match=str_match,items_valuations=items_valuations)
logger.debug(f" Match updated to {str_match}")
logger.info(f" matching_college_valuations: {matching_college_valuations}")

elif problematic_component == int_to_college_name[up]:
logger.debug(f"New match is leximin-worse because of c_up = c_{up}:")
LowerFix.append(up)
UpperFix.append(up + 1)
logger.info(f" Updated LowerFix and UpperFix with {up}")

elif problematic_component in alloc.instance.agents:
sd = problematic_component
logger.debug(f"New match is leximin-worse because of student {sd}: ")
Expand All @@ -113,15 +121,15 @@ def FaStGen(alloc: AllocationBuilder, items_valuations:dict)->dict:
A = [j for j in UnFixed if (j > t + 1)]
SoftFix.extend((j, t+1) for j in A)
logger.debug(f" Updating SoftFix to {SoftFix}")

else:
logger.debug(f"New match is leximin-worse because of college {sourceDec(new_match_leximin_tuple, match_leximin_tuple)}: ")
str_match, LowerFix, UpperFix, SoftFix = LookAheadRoutine((agents, items, agents_valuations, items_valuations), integer_match, down, LowerFix, UpperFix, SoftFix)
logger.debug(f" LookAheadRoutine result: match={str_match}, LowerFix={LowerFix}, UpperFix={UpperFix}, SoftFix={SoftFix}")

UnFixed = [
j for j in college_name_to_int.values()
if (j not in UpperFix) or
j for j in college_name_to_int.values()
if (j not in UpperFix) or
any((j, _j) not in SoftFix for _j in college_name_to_int.values() if _j > j)
]
logger.debug(f" Updating UnFixed to {UnFixed}")
Expand All @@ -143,25 +151,34 @@ def LookAheadRoutine(I:tuple, integer_match:dict, down:int, LowerFix:list, Upper
:param UpperFix: The group of colleges whose upper limit is fixed.
:param SoftFix: A set of temporary upper limits.
*We will asume that in the colleges list in index 0 there is college 1 in index 1 there is coll
>>> from fairpyx.adaptors import divide
>>> S = ["s1", "s2", "s3", "s4", "s5"]
>>> C = ["c1", "c2", "c3", "c4"]
>>> V = {"c1" : {"s1":50,"s2":23,"s3":21,"s4":13,"s5":10},"c2" : {"s1":45,"s2":40,"s3":32,"s4":29,"s5":26},"c3" : {"s1":90,"s2":79,"s3":60,"s4":35,"s5":28},"c4" : {"s1":80,"s2":48,"s3":36,"s4":29,"s5":15}}
>>> U = {"s1" : {"c1":16,"c2":10,"c3":6,"c4":5},"s2" : {"c1":36,"c2":20,"c3":10,"c4":1},"s3" : {"c1":29,"c2":24,"c3":12,"c4":10},"s4" : {"c1":41,"c2":24,"c3":5,"c4":3},"s5" : {"c1":36,"c2":19,"c3":9,"c4":6}}
>>> match = {1 : [1,2],2 : [3],3 : [4],4 : [5]}
>>> V = {"c1" : {"s1":50,"s2":23,"s3":21,"s4":13,"s5":10},"c2" : {"s1":45,"s2":40,"s3":32,"s4":29,"s5":26},"c3" : {"s1":90,"s2":79,"s3":60,"s4":35,"s5":28},"c4" : {"s1":80,"s2":48,"s3":36,"s4":29,"s5":15}}
>>> U = {"s1" : {"c1":16,"c2":10,"c3":6,"c4":5},"s2" : {"c1":36,"c2":20,"c3":10,"c4":1},"s3" : {"c1":29,"c2":24,"c3":12,"c4":10},"s4" : {"c1":41,"c2":24,"c3":5,"c4":3},"s5" : {"c1":36,"c2":19,"c3":9,"c4":6}}
>>> match = {1: [1,2], 2: [3], 3: [4], 4: [5]}
>>> I = (S,C,U,V)
>>> down = 4
>>> LowerFix = [4]
>>> UpperFix = [1]
>>> SoftFix = []
>>> LookAheadRoutine(I, match, down, LowerFix, UpperFix, SoftFix)
({'c1': ['s1', 's2'], 'c2': ['s3'], 'c3': ['s4'], 'c4': ['s5']}, [4], [1, 4], [])
"""
"""
agents, items, agents_valuations, items_valuations = I

if agents == []:
raise ValueError(f"Agents list can't be empty")
if items == []:
raise ValueError(f"Items list can't be empty")
if agents_valuations == {}:
raise ValueError(f"Agents valuations list can't be empty")
if items_valuations == {}:
raise ValueError(f"Items valuations list can't be empty")

student_name_to_int = generate_dict_from_str_to_int(agents)
college_name_to_int = generate_dict_from_str_to_int(items)
college_name_to_int = generate_dict_from_str_to_int(items)
int_to_student_name = generate_dict_from_int_to_str(agents)
int_to_college_name = generate_dict_from_int_to_str(items)

Expand All @@ -187,13 +204,13 @@ def LookAheadRoutine(I:tuple, integer_match:dict, down:int, LowerFix:list, Upper

new_str_match = integer_to_str_matching(integer_match=new_integer_match, items_dict=college_name_to_int, agent_dict=student_name_to_int)
matching_college_valuations = update_matching_valuations_sum(match=new_str_match,items_valuations=items_valuations)

old_match_leximin_tuple = create_leximin_tuple(match=given_str_match, agents_valuations=agents_valuations, items_valuations=items_valuations)
new_match_leximin_tuple = create_leximin_tuple(match=new_str_match, agents_valuations=agents_valuations, items_valuations=items_valuations)
logger.info(f" Old match leximin tuple: {old_match_leximin_tuple}")
new_match_leximin_tuple = create_leximin_tuple(match=new_str_match, agents_valuations=agents_valuations, items_valuations=items_valuations)
logger.info(f" New match leximin tuple: {new_match_leximin_tuple}")

#Extarcting from the SourceDec function the problematic item or agent, if there isn't one then it will be ""
problematic_component = sourceDec(new_match_leximin_tuple=new_match_leximin_tuple, old_match_leximin_tuple=old_match_leximin_tuple)
logger.info(f" problematic component: {problematic_component}")
Expand Down Expand Up @@ -244,18 +261,18 @@ def create_leximin_tuple(match:dict, agents_valuations:dict, items_valuations:di
Example:
>>> match = {"c1":["s1","s2","s3","s4"], "c2":["s5"], "c3":["s6"], "c4":["s7"]}
>>> items_valuations = { #the colleges valuations
... "c1":{"s1":50, "s2":23, "s3":21, "s4":13, "s5":10, "s6":6, "s7":5},
... "c2":{"s1":45, "s2":40, "s3":32, "s4":29, "s5":26, "s6":11, "s7":4},
... "c3":{"s1":90, "s2":79, "s3":60, "s4":35, "s5":28, "s6":20, "s7":15},
... "c1":{"s1":50, "s2":23, "s3":21, "s4":13, "s5":10, "s6":6, "s7":5},
... "c2":{"s1":45, "s2":40, "s3":32, "s4":29, "s5":26, "s6":11, "s7":4},
... "c3":{"s1":90, "s2":79, "s3":60, "s4":35, "s5":28, "s6":20, "s7":15},
... "c4":{"s1":80, "s2":48, "s3":36, "s4":29, "s5":15, "s6":7, "s7":1},
... }
>>> agents_valuations = { #the students valuations
... "s1":{"c1":16, "c2":10, "c3":6, "c4":5},
... "s2":{"c1":36, "c2":20, "c3":10, "c4":1},
... "s3":{"c1":29, "c2":24, "c3":12, "c4":10},
... "s4":{"c1":41, "c2":24, "c3":5, "c4":3},
... "s5":{"c1":36, "c2":19, "c3":9, "c4":6},
... "s6":{"c1":39, "c2":30, "c3":18, "c4":7},
>>> agents_valuations = { #the students valuations
... "s1":{"c1":16, "c2":10, "c3":6, "c4":5},
... "s2":{"c1":36, "c2":20, "c3":10, "c4":1},
... "s3":{"c1":29, "c2":24, "c3":12, "c4":10},
... "s4":{"c1":41, "c2":24, "c3":5, "c4":3},
... "s5":{"c1":36, "c2":19, "c3":9, "c4":6},
... "s6":{"c1":39, "c2":30, "c3":18, "c4":7},
... "s7":{"c1":40, "c2":29, "c3":6, "c4":1}
... }
>>> create_leximin_tuple(match, agents_valuations, items_valuations)
Expand All @@ -264,7 +281,7 @@ def create_leximin_tuple(match:dict, agents_valuations:dict, items_valuations:di
>>> match = {"c1":["s1","s2","s3"], "c2":["s4"], "c3":["s5"], "c4":["s7","s6"]}
>>> create_leximin_tuple(match, agents_valuations, items_valuations)
[('s7', 1), ('s6', 7), ('c4', 8), ('s5', 9), ('s1', 16), ('s4', 24), ('c3', 28), ('c2', 29), ('s3', 29), ('s2', 36), ('c1', 94)]
"""
"""
leximin_tuple = []
matching_college_valuations = update_matching_valuations_sum(match=match, items_valuations=items_valuations)
logger.debug(f"matching_college_valuations: {matching_college_valuations}")
Expand All @@ -275,7 +292,7 @@ def create_leximin_tuple(match:dict, agents_valuations:dict, items_valuations:di
leximin_tuple.append((item, matching_college_valuations[item]))
for agent in match[item]:
leximin_tuple.append((agent,agents_valuations[agent][item]))

leximin_tuple = sorted(leximin_tuple, key=lambda x: (x[1], x[0]))
return leximin_tuple

Expand Down Expand Up @@ -307,7 +324,7 @@ def sourceDec(new_match_leximin_tuple:list, old_match_leximin_tuple:list)->str:
elif new_match_leximin_tuple[k][1] > old_match_leximin_tuple[k][1]:
return ""
else: #new_match_leximin_tuple[k][1] < old_match_leximin_tuple[k][1]
return new_match_leximin_tuple[k][0]
return new_match_leximin_tuple[k][0]
return ""

def get_lowest_ranked_student(item_int:int, match_int:dict, items_valuations:dict, int_to_college_name:dict, int_to_student_name:dict):
Expand All @@ -325,8 +342,8 @@ def get_lowest_ranked_student(item_int:int, match_int:dict, items_valuations:dic
# Example:
>>> match = {1:[1,2,3,4], 2:[5], 3:[6], 4:[7]}
>>> items_valuations = { #the colleges valuations
... "c1" : {"s1":50,"s2":23,"s3":21,"s4":13,"s5":10,"s6":6,"s7":5},
... "c2" : {"s1":45,"s2":40,"s3":32,"s4":29,"s5":26,"s6":11,"s7":4},
... "c1" : {"s1":50,"s2":23,"s3":21,"s4":13,"s5":10,"s6":6,"s7":5},
... "c2" : {"s1":45,"s2":40,"s3":32,"s4":29,"s5":26,"s6":11,"s7":4},
... "c3" : {"s1":90,"s2":79,"s3":60,"s4":35,"s5":28,"s6":20,"s7":15},
... "c4" : {"s1":80,"s2":48,"s3":36,"s4":29,"s5":15,"s6":6,"s7":1}
... }
Expand Down Expand Up @@ -357,8 +374,8 @@ def update_matching_valuations_sum(match:dict, items_valuations:dict)->dict:
Example:
>>> match = {"c1":["s1","s2","s3","s4"], "c2":["s5"], "c3":["s6"], "c4":["s7"]}
>>> items_valuations = { #the colleges valuations
... "c1" : {"s1":50,"s2":23,"s3":21,"s4":13,"s5":10,"s6":6,"s7":5},
... "c2" : {"s1":45,"s2":40,"s3":32,"s4":29,"s5":26,"s6":11,"s7":4},
... "c1" : {"s1":50,"s2":23,"s3":21,"s4":13,"s5":10,"s6":6,"s7":5},
... "c2" : {"s1":45,"s2":40,"s3":32,"s4":29,"s5":26,"s6":11,"s7":4},
... "c3" : {"s1":90,"s2":79,"s3":60,"s4":35,"s5":28,"s6":20,"s7":15},
... "c4" : {"s1":80,"s2":48,"s3":36,"s4":29,"s5":15,"s6":7,"s7":1}
... }
Expand All @@ -370,7 +387,7 @@ def update_matching_valuations_sum(match:dict, items_valuations:dict)->dict:
{'c1': 94, 'c2': 29, 'c3': 28, 'c4': 8}
"""
matching_valuations_sum = { #in the artical it looks like this: vj(mu)
colleague: sum(items_valuations[colleague][student] for student in students)
colleague: sum(items_valuations[colleague][student] for student in students)
for colleague, students in match.items()
}
return matching_valuations_sum
Expand Down Expand Up @@ -407,7 +424,7 @@ def create_stable_matching(agents, agents_dict, items, items_dict):
# Assign the remaining students to cj for j >= 2
for j in range(1, len(items)):
matching[items_dict[items[j]]] = [agents_dict[agents[len(agents) - (len(items) - j)]]]

return matching

def generate_dict_from_str_to_int(input_list:list)->dict:
Expand All @@ -431,7 +448,7 @@ def generate_dict_from_str_to_int(input_list:list)->dict:
{'c1': 1, 'c2': 2, 'c3': 3, 'c4': 4}
"""
return {item: index + 1 for index, item in enumerate(input_list)}

def generate_dict_from_int_to_str(input_list:list)->dict:
"""
Creating a dictionary that includes for each string item in the list an index representing it, key=integer.
Expand Down Expand Up @@ -476,7 +493,7 @@ def integer_to_str_matching(integer_match:dict, agent_dict:dict, items_dict:dict
# Reverse the s_dict and c_dict to map integer values back to their string keys
s_reverse_dict = {v: k for k, v in agent_dict.items()}
c_reverse_dict = {v: k for k, v in items_dict.items()}

# Create the new dictionary with string keys and lists of string values
return {c_reverse_dict[c_key]: [s_reverse_dict[s_val] for s_val in s_values] for c_key, s_values in integer_match.items()}

Expand All @@ -493,7 +510,7 @@ def get_match(match:dict, value:str)->any:
Example:
>>> match = {"c1":["s1","s2","s3","s4"], "c2":["s5"], "c3":["s6"], "c4":["s7"]}
>>> value = "c1"
>>> get_match(match, value)
['s1', 's2', 's3', 's4']
Expand Down Expand Up @@ -522,16 +539,16 @@ def Demote(matching:dict, student_index:int, down_index:int, up_index:int, item_
:param down_index: Index of the college to move the student to.
:param up_index: Index of the upper bound college.
#*** The test is the same as the running example we gave in Ex2.***
#* The test is the same as the running example we gave in Ex2.*
# ... valuations = {"Alice": {"c1": 11, "c2": 22}, "Bob": {"c1": 33, "c2": 44}},
>>> matching = {1: [1, 2, 3, 4], 2: [5], 3: [6], 4: [7]}
>>> UP = 1
>>> DOWN = 4
>>> student_index = 6
>>> items_valuations = { #the colleges valuations
... "c1" : {"s1":50,"s2":23,"s3":21,"s4":13,"s5":10,"s6":6,"s7":5},
... "c2" : {"s1":45,"s2":40,"s3":32,"s4":29,"s5":26,"s6":11,"s7":4},
... "c1" : {"s1":50,"s2":23,"s3":21,"s4":13,"s5":10,"s6":6,"s7":5},
... "c2" : {"s1":45,"s2":40,"s3":32,"s4":29,"s5":26,"s6":11,"s7":4},
... "c3" : {"s1":90,"s2":79,"s3":60,"s4":35,"s5":28,"s6":20,"s7":15},
... "c4" : {"s1":80,"s2":48,"s3":36,"s4":29,"s5":15,"s6":6,"s7":1}
... }
Expand Down Expand Up @@ -566,7 +583,7 @@ def Demote(matching:dict, student_index:int, down_index:int, up_index:int, item_

return matching #Return the matching after the change

if __name__ == "__main__":
if __name__ == "_main_":
import doctest, sys
console=logging.StreamHandler() #writes to stderr (= cerr)
logger.handlers=[console] # we want the logs to be written to console
Expand Down
Loading

0 comments on commit f7fdeef

Please sign in to comment.