20
Lecture 19 P-Complete Problems

Lecture 19 P-Complete Problems

Embed Size (px)

DESCRIPTION

Lecture 19 P-Complete Problems. Log-space many-one reduction. DEFINITION. PROPERTY. P-complete. Circuit Value Problem (CVP). THEOREM. PROOF. SIMULATE DTM COMPUTATION BY A CIRCUIT. Planar Circuit Value (PCV). THEOREM. PROOF. Monotone Circuit Value (MCV). THEOREM. PROOF. - PowerPoint PPT Presentation

Citation preview

Page 1: Lecture 19  P-Complete Problems

Lecture 19

P-Complete Problems

Page 2: Lecture 19  P-Complete Problems

Log-space many-one reduction

. and

. and

space. clogarithmiin computable function a via if

log

logloglog

log

NCANCBBA

CACBBA

fBABA

m

mmm

mm

DEFINITION

PROPERTY

Page 3: Lecture 19  P-Complete Problems

P-complete

n.computatio parallelfor hard are problems complete-

. ,any for and if complete- is log

P

ABPBPAPA m

Page 4: Lecture 19  P-Complete Problems

Circuit Value Problem (CVP)

complete.- is

.1| whether determine , assignmentan and circuit aGiven

PCVP

CC

THEOREM

PROOF

Page 5: Lecture 19  P-Complete Problems

SIMULATE DTM COMPUTATION BY A CIRCUIT

Page 6: Lecture 19  P-Complete Problems

Planar Circuit Value (PCV)

PCVCVP

complete.- is PCV

.1| whether determine , assignmentan and circuit planar aGiven

logm

P

CC

THEOREM

PROOF

Page 7: Lecture 19  P-Complete Problems

.for circuit planar A yx .crosserA

Page 8: Lecture 19  P-Complete Problems

Monotone Circuit Value (MCV)

MCVCVP

complete.- is MCV

.1| whether determine , assignmentan and circuit monotone aGiven

logm

P

CC

THEOREM

PROOF

Page 9: Lecture 19  P-Complete Problems

Odd Maximum Flow (OMF)

OMF.MCV

complete.- is OMF

number.

oddan is of flow maximum he whether tdetermine ,capacity edge

integer positive and sink , source a with ),(network aGiven

logm

P

Nc

tsEVN

PROOF

THEOREM

Page 10: Lecture 19  P-Complete Problems

time.polynomialin computed becan flow Maximum

Page 11: Lecture 19  P-Complete Problems

gate. OR is

gateoutput the(b)

two,

most at fanout has

gate ORevery (a)

:assumemay WeReduce fanout

Add OR gate as output gate

Page 12: Lecture 19  P-Complete Problems

label.smaller a with oneanother frominput an gets node no

:order icalin topologly consequent gatesother and 0h output wit Label (1)

Page 13: Lecture 19  P-Complete Problems

.2capacity assign we, label with node from coming edgeany For (2) ii

22

42 42

32 32

6252

2

Page 14: Lecture 19  P-Complete Problems

capacity. going-out

capacity comming-in total

make ocapacity t with togate

ANDany from edge Assign the

. togateoutput

from edge the to1capacity Assign

).deg(

where label with node to from

edge the to2capacity Assign (4)

gates. AND all and gateoutput

from conneting sink Add

1. e with valuableevery vari

toconnecting source Add (3)

t

t

ioutd

is

d

t

s

i

Page 15: Lecture 19  P-Complete Problems

edges. incomming

ofcapacity total toequal

capacity with , togate OReach

from edgeimaginary an Add

s

Page 16: Lecture 19  P-Complete Problems

. allfor 0),(set , variablefalseeach For

. allfor ),(),(set , vertex variableeach trueFor )2(

).,(),( ),,( edgeeach For (1)

:follows as flow aConstruct

jjifi

jjicjifi

iscisfis

f

Page 17: Lecture 19  P-Complete Problems

. allfor 0),( , gate OR falseeach For

.),(),(),(

for ),(),(, gates OR each trueFor

:follows as edges outgoing itsfor set is then

gate, a of edges incoming allfor set been has If (3)

jjifi

jicikfsif

sjjicjifi

f

f

sjk

Page 18: Lecture 19  P-Complete Problems

.for 0),(

),(),( , gate AND falseeach For

. allfor ),(),( , gate AND each tureFor

tjjif

ikftifi

jjicjifi

k

Page 19: Lecture 19  P-Complete Problems

).,(),(,any for then vertex, truea is If (b)

.0),(,any for then vertex,false a is If (a)

:properties following thehas flow This

jicjifsji

jiftji

vertex.falseany through

passnot does edgeimaginary an through passing flow the(a),By

Page 20: Lecture 19  P-Complete Problems