Sie sind auf Seite 1von 9

best h highest ranked se 5 suchthat

hs is va hatched in Any
stable math
h bestChl ht H y
let Mt

Claim M't is a perfect matching

pf last
the

Them GS algo always produces MY


h
not this means some hospital
PI Suppose
valid
rejected by student is for which h s is
is a
11

best h best h

We may assume that this is the first time this


happens in the execution of algo

Since1h s valid can choose stable watching M


with his c M Sina s rejects L s holds an

offer from a hospital sa k that s prefers to h

Because h s is FIRST

rejection h prefers to t
valid partner of 6 and hasn't
why t
i
is a

but he made an offer to s


rejected k yet
blocking pair in
M Ba
fo k s is a Oops

1st 2nd
h did Neil Max Ibrahim
best h

valid parthen

Mt said to be hospital optimal


be the worst valid hospital for s
Let horsts

Mt worsts s se s
Cos

n't said to be student pessimal

PI Enough to show 2 Assure not

assume Ch s t M but ht worst s

K
Means there il a stable matching missuch that

but s prefers h to h
h stem

Look at the student t matched with h is M


because best h
h prefers s t t
M
th in M
s blocking pair
B h t
Oops
Free cultural diversion
Fix pref lists Cost funtion for a stable watching

M defined by

Effrank
inhi pref list of its
calm
math in MJ

Crank in si pref list of its


Cs LM's 5
match in M

of GS shows that
Our characterization matching
it minimizes Ct maxinites Cm

W A B C D A Z Y X W
X B A b C B Y z w X
Y c 0 A B c
X w Z Y
t B A
D e b w X Y Z
l l l l
41 4
Tons of stable matchings w X Y 7

i is
it
32 5 16 4 4 44

W
2 3

X
CtiV
Y Z
10

b t Id
3 2 2 3 to
Cs

CH 14
w X Y z
I l I i

C A B c 6
p 22
i

Afidenatdftsis called

n'iiiin c
a

it calm tGlm M
galiano
balanced Sms is NP
Finding SH equal

hard.BA SFin
or

its known how to find egalitarian

polynomial time

0 notation tractability
Big

algorithms to
solve computational
God Find efficient

problems
in some
problems are
discrete implicit search large

set usually finite think all perfect matchings


have n
e.g fr IsHH u perfect matchings
all possibilities
Brute force search through
until a solution I found terrible

is the
focus for now on runningtie resource

as a function at the input N


case running time ie
we analyze worst worst

possible running time on all inputs of size N


p ng

Gian yl
au in
Fy
m
y
n
points
the plane find the closest pair of points

intern there are


pain
Z
h
L T h

Compute ALL distances distfni yd Lai y D


and find the minimum

can do better

better
Vague goal find algorithms qualitatively

than brute force

Def An algorithm
has
polynomial running
the

saw
precise
def
It
I such
effrient

an algo doubling the input Nts 2N


constant
results in multiplying running time by a

d zd Cnd
2N

Oder of growth
Not going to prove running the B exactly

e I3 n t 3h t 81
g don't
when
There is no sense in being precise you

even know what you're talking about


Neumann
John von

Def For functions


fT N N say

1 n is 01km o TH B order
4ns
if there are C 0 NoC N fr which
An No Tcu c flu
T is dominated by a constant multiple
off
Example Tcn an'thonte a b c 20

is 042 sina.hr h't

Thi t an tbh ten tbt c n

T
constant

Das könnte Ihnen auch gefallen