Upward Planar Drawings of Series-Parallel Digraphs with Maximum Degree Three



Similar documents
Reading. Minimum Spanning Trees. Outline. A File Sharing Problem. A Kevin Bacon Problem. Spanning Trees. Section 9.6

Link-Disjoint Paths for Reliable QoS Routing

Back left Back right Front left Front right. Blue Shield of California. Subscriber JOHN DOE. a b c d

Schedule C. Notice in terms of Rule 5(10) of the Capital Gains Rules, 1993

11 + Non-verbal Reasoning

Graph Theoretical Analysis and Design of Multistage Interconnection Networks

A122 MARION COUNTY HEALTH BUILDING HVAC, GLAZING AND LIGHTING RENOVATION 75% DOCUMENTS 08/31/2015

Where preparation meets opportunity. My Academic Planner. Early Academic Outreach Program (EAOP)

Usability Test Checklist

Process Mining Making Sense of Processes Hidden in Big Event Data

Hospitals. Internal Revenue Service Information about Schedule H (Form 990) and its instructions is at

Distributed Systems Principles and Paradigms. Chapter 11: Distributed File Systems. Distributed File Systems. Example: NFS Architecture

Operational Procedure: ACNC Data Breach Response Plan

Diagram Editing with Hypergraph Parser Support

Uses for Binary Trees -- Binary Search Trees

Algorithmic Aspects of Access Networks Design in B3G/4G Cellular Networks

MANAGEMENT OF INFORMATION SECURITY AND FORENSICS

SKILL TEST IR(H) HELICOPTER SE ME Application and report form A. Udfyldes af ansøgeren/to be filled out by the applicant:

CompactPCI Connectors acc. to PIGMG 2.0 Rev. 3.0

A simple algorithm to generate the minimal separators and the maximal cliques of a chordal graph

How To Understand The Rules Of A Game Of Chess

Oracle PL/SQL Programming Advanced

Change Your History How Can Soccer Knowledge Improve Your Business Processes?

1.- L a m e j o r o p c ió n e s c l o na r e l d i s co ( s e e x p li c a r á d es p u é s ).

Network Decoupling for Secure Communications in Wireless Sensor Networks

Othello: A Minute to Learn... A Lifetime to Master. Brian Rose

Predicting Current User Intent with Contextual Markov Models

- ASSEMBLY AND INSTALLATION -

Recall from Last Time: Disjoint Set ADT

Distributed Process Discovery and Conformance Checking

Discovering Petri Nets From Event Logs

/*

SCO TT G LEA SO N D EM O Z G EB R E-

Victims Compensation Claim Status of All Pending Claims and Claims Decided Within the Last Three Years

Revised Conditions (January 2009) LLOYDS BANKING GROUP SHARE ISA CONDITIONS

AdvancedTCA Connectors acc. to PICMG 3.0

Cloud and Big Data Summer School, Stockholm, Aug., 2015 Jeffrey D. Ullman

Binary Search Trees. Definition Of Binary Search Tree. Complexity Of Dictionary Operations get(), put() and remove()

SEE PAGE 2 FOR BRUSH MOTOR WIRING SEE PAGE 3 FOR MANUFACTURER SPECIFIC BLDC MOTOR WIRING EXAMPLES A

PRESENTED TO. Data Leakage Worldwide: The Effectiveness of Corporate Security Policies

B a rn e y W a r f. U r b a n S tu d ie s, V o l. 3 2, N o. 2, ±3 7 8

Form: Parental Consent for Blood Donation

Last time Interprocedural analysis Dimensions of precision (flow- and context-sensitivity) Flow-Sensitive Pointer Analysis

Important result on the first passage time and its integral functional for a certain diffusion process

WAVEGUIDES (& CAVITY RESONATORS)

Operation Transform Formulae for the Generalized. Half Canonical Sine Transform

PIN #1 ID FIDUCIAL LOCATED IN THIS AREA TOP VIEW. ccc C SIDE VIEW

The Swedish Radiation Protection Institute s Regulations on X- ray Diagnostics;

MODULE 3. 0, y = 0 for all y

5.4 Exponential Functions: Differentiation and Integration TOOTLIFTST:

A New Efficient Distributed Load Balancing Algorithm for OTIS-Star Networks

H ig h L e v e l O v e r v iew. S te p h a n M a rt in. S e n io r S y s te m A rc h i te ct

Discovering Block-Structured Process Models From Event Logs Containing Infrequent Behaviour

Binary Search Trees. Definition Of Binary Search Tree. The Operation ascend() Example Binary Search Tree

Homeomorphic Alignment of Weighted Trees

Matching Execution Histories of Program Versions

Scholarship Help for Technology Students

1. Number of questions to be answered: ALL Multiple Choice (Section A) and 3 from 5 of the short answer questions (Section B)

Using Predictive Modeling to Reduce Claims Losses in Auto Physical Damage

Fundamentals of Tensor Analysis

Ethical and Professional Standards

Dinh Hong Giang 1,2, Ed Sarobol 2, * and Sutkhet Nakasathien 2 ABSTRACT

Reasoning to Solve Equations and Inequalities

fun

4.1 Interval Scheduling. Chapter 4. Greedy Algorithms. Interval Scheduling. Interval Scheduling: Greedy Algorithms

B rn m e d s rlig e b e h o v... 3 k o n o m i S s k e n d e tils k u d o g k o n o m is k frip la d s... 7 F o r ld re b e ta lin g...

Learning Schemas for Unordered XML

MPLS FOR MISSION-CRITICAL MICROWAVE NETWORKS BUILDING A HIGHLY RESILIENT MICROWAVE NETWORK WITH MULTI-RING TOPOLOGY

Campus Sustainability Assessment and Related Literature

P U B L I C A T I O N I N T E R N E 1800 PARTIAL ORDER TECHNIQUES FOR DISTRIBUTED DISCRETE EVENT SYSTEMS: WHY YOU CAN T AVOID USING THEM

Maximum area of polygon

SUPERCEDED BY T1-Q12

Summary of changes to Regulations recommended to the Senate by Graduate School Management Committee. Changed wording is shown in bold italics.

Functional Valuation of Ecosystem Services on Bonaire

SEE PAGE 2 FOR BRUSH MOTOR WIRING SEE PAGE 3 FOR MANUFACTURER SPECIFIC BLDC MOTOR WIRING EXAMPLES

SecurView Antivirus Software Installation

A n d r e w S P o m e r a n tz, M D

One Ring to Rule them All: Service Discovery and Binding in Structured Peer-to-Peer Overlay Networks

T c k D E GR EN S. R a p p o r t M o d u le Aa n g e m a a k t o p 19 /09 /2007 o m 09 :29 u u r BJB M /V. ja a r.

The Fundamental Theorem of Calculus

EM EA. D is trib u te d D e n ia l O f S e rv ic e

SPRINGWELLS AND NORTHEAST WATER TREATMENT PLANTS SLUDGE REMOVAL AND DISPOSAL SERVICES

The art of Paperarchitecture (PA). MANUAL

i n g S e c u r it y 3 1B# ; u r w e b a p p li c a tio n s f r o m ha c ke r s w ith t his å ] í d : L : g u id e Scanned by CamScanner

w ith In fla m m a to r y B o w e l D ise a se. G a s tro in te s tin a l C lin ic, , K a s h iw a z a, A g e o C ity, S a ita m a

Erfa rin g fra b y g g in g a v

Economics 340: International Economics Andrew T. Hill Nontariff Barriers to Trade

A Quick Guide to Colleges. Offering Engineering Degrees

Paper Technics Orientation Course in Papermaking 2009:

Roof Terraces. Structural assemblies

motori asincroni monofase asynchronous single phase motors moteurs asynchrones monophasés einphasige Asynchronmotoren

Harvard College. Math 21a: Multivariable Calculus Formula and Theorem Review

Transcription:

Upwr Plnr Drwins of ris-prlll Dirps wit Mximum Dr Tr (Extn Astrt) M. Aul Hssn m n M. iur Rmn Dprtmnt of Computr in n Eninrin, Bnls Univrsity of Eninrin n Tnoloy (BUET). {sm,siurrmn}@s.ut.. Astrt. An upwr plnr rwin of irp G is plnr rwin of G wr vry is rwn s simpl urv monoton in t vrtil irtion. A irp is upwr plnr if it s n min tt mits n upwr plnr rwin. T prolm of tstin wtr irp is upwr plnr is NP-omplt. In tis ppr w iv linr-tim loritm to tst t upwr plnrity of sris-prlll irp G wit mximum r tr n otin n upwr plnr rwin of G if G mits on. Ky wors: Grp Drwin, Upwr Plnr Drwin, Dirt Ayli Grp, ris-prlll Grp, Aloritm, PQ-tr. 1 Introution In n upwr plnr rwin of irp, vry vrtx is mpp to point in t Eulin pln n vry is rwn s simpl urv monoton in t vrtil irtion witout prouin ny rossin wit otr s, s illustrt in Fi. 1(). Upwr plnr rwins of irps fin importnt pplitions in visuliztion of t irril ntwork struturs wi frquntly ris in softwr ninrin, projt mnmnt n visul lnus [BDMT98]. Unfortuntly, not ll irps v upwr plnr rwin. On n sily unrstn tt if irp ontins yl, tn on of t s on t yl nnot rwn monotonilly in t upwr irtion (s t yl inu y vrtis,, n of irp G in Fi. 1()). A irp is upwr plnr if it s n min wi mits n upwr plnr rwin. Ayliity is nssry onition for irp to upwr plnr. Trouout tis ppr, wrvr w rfr to irp, w mn n yli irp. Howvr, yliity is not suffiint onition for upwr plnrity. For xmpl, t yli irp G in Fi. 1() is not upwr plnr; tr r four possil upwr plnr M. Kyko n M. iur Rmn (Es.): WALCOM 2007, pp. 28 45, 2007.

Upwr Plnr Drwins 29 f f f G G () () () G Fi.1. () An upwr plnr irp G, () irp G wi ontins yl n trfor is not upwr plnr n () n yli irp G wi is not upwr plnr. () () () () Fi.2. T four possil mins of f F 1 of irp G. mins of t unirt yl inu y t vrtis,, n (s Fi. 2), n strtin wit ny of ts four mins, t rminin s nnot in ny wy to otin n upwr plnr rwin of G. T prolm of tstin upwr plnrity of irp is on of t most llnin prolms in t r of rp rwin n s n stui wit xtnsiv ffort. Linr-tim loritms r known for tstin wtr irp mits plnr rwin [HT74,BL76]. Tstin wtr irp mits n upwr rwin n lso solv in linr-tim usin t wll-known topoloil sortin tniqu [CLR01]. Nvrtlss, ominin ts two proprtis mks t prolm NP-r [GT94]. T prolm n stui ot in t fix min sttin n in t vril min sttin. In t fix min sttin, t loritm nnot ltr t ivn min, n if tt prtiulr min is not upwr plnr, t output must ntiv ltou som otr min of t sm rp oul upwr plnr. Fi. 3() sows n upwr plnr min of irp G wos min in Fi. 3() is not upwr plnr. Fi. 3() is notr upwr plnr min of G. Brtolzzi t l. [BDLM94] v ivn n loritm to tst upwr plnrity in tim O(n 2 ) in t fix min sttin. In t vril min sttin, t loritm n iv ntiv output only if tr is no upwr plnr min of t input rp. Gr n Tmssi [GT01] prov tt it is n NP-omplt prolm to trmin wtr irp s n upwr plnr rwin in t vril min sttin. Nvrtlss, t prolm s n stui in t vril min sttin for som rstrit lsss of irps [Pp95,HL96,BDMT98]. Gr n Tmssi

30 M. A. H. m n M. iur Rmn i f j () i j () f j () i f Fi. 3. () An upwr plnr irp G n n upwr plnr min of G, () non-upwr plnr min of G, n () notr upwr plnr min of G. [GT95] prov tt sris-prlll irp wit sinl sour n sinl sink is lwys upwr plnr. Unfortuntly, sris-prlll irp G wit multipl sours n sinks my not upwr plnr, n tstin upwr plnrity for irps wit multipl sours n sinks is mor iffiult. Rntly, Diimo t l. [DGL06] provi n loritm tt tsts upwr plnrity of sris-prlll irps in tim O(n 4 ) in t vril min sttin. In tis ppr, w stuy upwr plnr rwins of sris-prlll irps of 3 wit multipl sours n sinks in t vril min sttin. For su irp G, w iv linr-tim loritm to onstrut n upwr plnr rwin of G if G mits on. T ppro of our loritm is iffrnt from t on prsnt in [DGL06] n t loritm in [DGL06] rquirs tim O(n 4 ) vn for sris-prlll irp wit t mximum r tr. T min i of our loritm is s follows. Our loritm works in two pss, nmly tstin ps n onstrution ps. W in wit omposition tr ll PQ-tr T of G. In t tstin ps, w trvrs T ottom-up n tst t fsiility of otinin n upwr plnr rwin of G. If tis ps fils, w lr tt G is not n upwr plnr irp. If t tstin ps sus, w strt t onstrution ps n usin t informtion otin in t tstin ps, w otin n upwr plnr min of G in top-own trvrsl of T. T rst of t ppr is orniz s follows. tion 2 sris som finitions n prsnts prliminry rsults. In tion 3 w sri our primry finins on upwr plnrity of sris-prlll irps wit 3. tion 4 prsnts our loritm to tst upwr plnrity n fin n upwr plnr rwin of ionnt sris-prlll irp wit 3. Finlly, tion 5 is onlusion. 2 Prliminris In tis stion w iv som finitions n prsnt prliminry rsults. Lt G = (V, E) onnt rp wit vrtx st V n st E. T r of vrtx v, (v) is t numr of s inint to v in G. W not

Upwr Plnr Drwins 31 t mximum of t r of t vrtis of G y (G). T onntivity κ(g) of rp G is t minimum numr of vrtis wos rmovl rsults in isonnt rp or sinl-vrtx rp K 1. W sy tt G is k-onnt if κ(g) k. A plnr rwin of G prtitions t pln into topoloilly onnt rions ll fs. T unoun f is ll t outr f, t rminin fs r ll innr fs. A pt in G is n orr list of istint vrtis v 1, v 2,...,v q V su tt (v i 1, v i ) E for ll i, 2 i q. A pt P is ll u, v-pt if u n v r t first n lst vrtis in P rsptivly. A rp G = (V, E) is ll sris-prlll rp (wit sour s n sink t) if itr G onsists of pir of vrtis onnt y sinl or tr xist two sris-prlll rps G i = (V i, E i ), i = 1, 2, wit sour s i n sink t i su tt V = V 1 V 2, E = E 1 E 2, n itr s = s 1, t 1 = s 2 n t = t 2 or s = s 1 = s 2 n t = t 1 = t 2 [REN05]. A pir {u, v} of vrtis of onnt rp G is split pir if tr xist two surps G 1 = (V 1, E 1 ) n G 2 = (V 2, E 2 ) stisfyin t followin two onitions: 1. V = V 1 V 2, V 1 V 2 = {u, v}; n 2. E = E 1 E 2, E 1 E 2 =, E 1 1, E 2 1. Tus vry pir of jnt vrtis is split pir. A split omponnt of split pir {u, v} is itr n (u, v) or mximl onnt surp H of G su tt {u, v} is not split pir of H. Lt G ionnt sris-prlll rp. Lt (s, t) n of G. T PQ-tr T of G wit rspt to rfrn = (s, t) sris rursiv omposition of G inu y its split pirs [GL99]. Tr T is root orr tr wos nos r of tr typs:, P n Q. E no x of T orrspons to surp of G, ll its prtinnt rp G(x). E no x of T s n ssoit ionnt multirp, ll t sklton of x n not y sklton(x). Tr T is rursivly fin s follows. Trivil Cs: In tis s, G onsists of xtly two prlll s n joinin s n t. T onsists of sinl Q-no x, n t sklton of x is G itslf. T prtinnt rp G(x) onsists of only t. Prlll Cs: In tis s, t split pir {s, t} s tr or mor split omponnts G 0, G 1,, G k, k 2, n G 0 onsists of only rfrn = (s, t). T root of T is P-no x. T sklton(x) onsists of k + 1 prlll s 0, 1,, k joinin s n t, wr 0 = = (s, t) n i, 1 i k, orrspons to G i. T prtinnt rp G(x) = G 1 G 2 G k is union of G 1, G 2,, G k. As n xmpl, t sklton of P-no P 2 in Fi. 4 onsists of tr prlll s joinin vrtis n n Fiur 4() pits t prtinnt rp of P 2. ris Cs: In tis s t split pir {s, t} s xtly two split omponnts, n on of tm onsists of t rfrn. On my ssum tt t otr split omponnt s ut-vrtis 1, 2,, k 1, k 2, tt prtition t omponnt into its loks G 1, G 2,, G k in tis orr from s to t. Tn t root of T is n -no x. T sklton of x is yl 0, 1,, k wr 0 =, 0 = s, k = t, n i joins i 1 n i, 1 i k. T prtinnt rp G(x) of no x is union of G 1, G 2,, G k. For xmpl, t sklton of -no 3 in

32 M. A. H. m n M. iur Rmn m n l k i j () f () i l G () f f f () f (f) j k (i,n) i n i P1 1 m l 2 3 l P1 (n,m) (l,m) i 1 i 2 3 P2 (,l) (j,i) (k,j)(l,k) (,i) x l (,l) l P2 4 5 (i,n)(n,m)(l,m) (j,i) (k,j)(l,k) (,i) x P3 4 5 (,)(,)(,) (,) 6 7 P3 (,)(,)(,) (,) (,)(,) (,)(f,)(f,) 6 7 () (,)(,) (,) (f,)(f,) f () Fi.4. ()A ionnt sris-prlll rp G wit = 3, () PQ-tr T of G wit rspt to rfrn (i, n), n skltons of P- n -nos, () t prtinnt rp G( 3) of -no 3, () t prtinnt rp G(P 2) of P-no P 2, () t prtinnt rp G( 5) of -no 5, (f) t prtinnt rp G(P 3) of P-no P 3, () PQ-tr T of G wit P-no P 1 s t root Fi. 4 is t yl, i, l,, n Fiur 4() pits t prtinnt rp G( 3 ) of 3. In of t ss mntion ov, w ll t t rfrn of no x. Expt for t trivil s, no x of T s ilrn x 1, x 2,, x k in tis orr; x i is t root of t PQ-tr of rp G(x i ) i wit rspt to t rfrn i, 1 i k. W ll i t rfrn of no x i, n ll t npoints of i t pols of no x i. T tr otin so fr s Q-no ssoit wit of G, xpt t rfrn. W omplt t PQ-tr T y in Q-no, rprsntin t rfrn, n mkin it t prnt of x so tt it oms t root of T. An xmpl of t PQ-tr of ionnt sris-prlll rp in Fi. 4() is illustrt in Fi. 4(), wr t rwn y tik lin in sklton is t rfrn of t sklton. T PQ-tr T fin ov is t on us in [REN05] n is spil s of n PQR-tr [DT96,GL99] wr tr is no R-no n t root of t tr is Q-no orrsponin to t rfrn. On n sily moify T to n PQ-tr T wit n ritrry P-no s t root s illustrt in Fi. 4(). Lt G plnr irp. G is sris-prlll irp if t unrlyin unirt rp of G is sris-prlll rp. T PQ-tr of sris-prlll irp G is xtly t sm s t on of t unrlyin unirt srisprlll rp of G. In t rminr of tis ppr, w onsir n PQ-tr

Upwr Plnr Drwins 33 T of sris-prlll irp G wit P-no s t root. If (G) = 2, tn t unrlyin unirt rp of G is yl n E V = 0. It s n sown in [HL05] tt ll yli irps wit E V < 2 r upwr plnr. Hn, for (G) = 2, G is lwys upwr plnr. On my tus ssum tt (G) 3, n tt t root P-no of T s tr or mor ilrn. Tn t prtinnt irp G(x) of no x is t surp of G inu y t s orrsponin to ll snnt Q-no of x. Bs on t ssumption tt (G) = 3, t followin fts ol [REN05]. Ft 1 Lt (s, t) t rfrn of n -no x of T, n lt x 1, x 2,, x k t ilrn of x in tis orr from s to t. Tn (i) il x i of x is itr P-no or Q-no; (ii) ot x 1 n x k r Q-nos; n (iii) x i 1 n x i+1 must Q-nos if x i is P-no wr 2 i k 1. Ft 2 E non-root P-no of T s xtly two ilrn. A il of nonroot P-no n n - or Q-no. A P-no in n PQ-tr T is primitiv if it os not v ny snnt P-no in T. Lt x primitiv P-no in T. Lt x l n x r t lft n rit il of x in T rsptivly. Tn t unrlyin unirt rp of G(x) = G(x l ) G(x r ) is yl n n t irp G(x) is upwr plnr [HL05]. Trfor, t prtinnt irp G(x) of vry primitiv P-no x in T is lwys upwr plnr. W fin tt t it of primitiv P-no is zro. T it of ny otr P-no is (i + 1) if t mximum of t its of its snnt P-nos is i. T P-no P 3 in Fi. 4() is primitiv P-no. T its of t otr two P-nos P 2 n P 1 in Fi. 4() r 1 n 2 rsptivly. Lt G plnr irp. A rwin Γ of G is n upwr plnr rwin if it s no -rossin n ll t s of G r rwn s simpl urvs monotonilly inrsin in t vrtil irtion. G is n upwr plnr irp if G mits n upwr plnr rwin. On n osrv tt t followin lmm ols for n upwr plnr irp G. Lmm 1. A irp G is upwr plnr if n only if vry surp H of G is upwr plnr. Lt G irp wit fix plnr min. A vrtx v of G is imol if t irulr list of s inint to v n prtition into two (possily mpty) lists, on onsistin of inomin s n t otr onsistin of outoin s. If ll vrtis of G r imol tn G is ll imol. Ayliity n imolity r nssry onitions for t upwr plnrity of n m plnr irp [BDLM94]. Howvr, ty r not suffiint onitions. Lt f f of n m plnr imol irp G n suppos tt t ounry of f is visit lokwis if f is n innr f, n ountrlokwis if f is t outr f. Lt α = ( 1, v, 2 ) triplt su tt v is vrtx of t ounry of f n 1, 2 r two inint s of v tt r onsutiv on t ounry of f. Triplt α is ll n nl of f. W ll n nl α swit

34 M. A. H. m n M. iur Rmn nl of f if itr t irtion of 1 is opposit to t irtion of 2 on t ounry of f or 1 n 2 oini. Not tt if 1 n 2 oini tn G is not ionnt. If 1 n 2 r ot inomin in v, tn α is sink-swit of f n if ty r ot outoin, tn α is sour-swit of f. A sour or sink of G is ll swit vrtx of G n vrtx tt is not swit vrtx is ll n orinry vrtx of G. In t rminr of tis ppr w rfr to swit nl of f f y llin it simply swit of f. Lt G n m plnr irp. Lt Γ n upwr plnr rwin of G n lt α n nl of f f of G. W ssin ll F to t nl α in f f if α is not swit of f. Otrwis α is swit of f, n w ll α in f wit lttr L if α s vlu rtr tn π in Γ n wit lttr if t vlu of α in Γ is lss tn π. W ssin lls to ll nls of G s mntion ov n otin ll m irp. W ll tis ll m irp n upwr plnr rprsnttion of G n not it y U G. T rwin Γ is si to n upwr plnr rwin tt prsrvs U G [DGL06]. Fiur 5() illustrts U G of t rp G in Fi. 5() for t upwr plnr rwin Γ in Fi. 5(). It is mntionl tt ny ritrry llin of t swits of G my not v orrsponin upwr plnr rwin n n my not rr s n upwr plnr rprsnttion of G. T onitions wi must mt in orr to otin U G r sri low. v1 v7 v2 v5 v3 v6 v4 v11 v8 () v10 v9 v7 L F L v2 v5 F F v1 F () F v3 F v6 L F F v4 Lv11 F F v8 L v10 v9 L v1 v2 v5 v6 () v10 v3 v11 v4 v8 v9 v7 Fi.5. Illustrtion of upwr plnr rprsnttion n upwr plnr rwin. Lt G irp wit fix plnr min. Lt Φ n ssinmnt wi ssins ll of itr L,, or F to nl of vry f of G. For vrtx v of G, w not y L(v), (v), n F(v) t numr of nls t v tt Φ lls wit L,, n F rsptivly. For f f of G, w not y L(f), (f), n F(f) t numr of nls of f tt r ll y Φ wit L,, n F rsptivly. W ll t ssinmnt Φ n upwr onsistnt ssinmnt if t followin two onitions ol for Φ: (i) for swit vrtx v of G, L(v) = 1, (v) = (v) 1, F(v) = 0, n for orinry vrtx v of G, L(v) = 0, (v) = (v) 2, F(v) = 2; n (ii) for innr f f, (f) L(f) = 2 n for t outr f f, (f) L(f) = 2.

Upwr Plnr Drwins 35 1 Pl (y p ) : Pr (y p ) : 2 9 10 8 13 11 4 6 P l (x 1 ) : Pr (x 1 ) : P l (r) : 14 Pr (r) = P l (y p ) : 12 15 17 () 3 5 16 18 19 7 8 9 10 11 12 13 8 14 () 4 5 9 6 7 1 () (f) 15 2 8 10 16 3 4 5 9 10 8 y 1 P r y 2 1 2 3 P P 4 x 1 5 6 x 2 7 9 10 11 12 13 17 18 19 14 15 16 y 3 x 3 x 4 x 5 x 6 T () y 1 r y p y 2 y 3 T () Fi.6. () A ionnt sris-prlll rp G, () n PQ-tr T of G, () T wit ummy P-no y p, () (f) pol-pts of no x 1, y p n r rsptivly. On n intuitivly unrstn t nssity of onitions (i) n (ii) for n upwr plnr rwin. Conition (i) must ol u to t nssity of imolity, wil onition (ii) must ol u to si omtri onsirtion for upwr plnrity. Conitions (i) n (ii) r lso suffiint for upwr plnrity of G s stt in [DGL06]. Hn t followin lmm ols. Lmm 2. Lt G n yli plnr imol m irp su tt nl of G is ssin ll L, or F unr som ssinmnt Φ. Tn t lllins of t nls in G fin n upwr plnr rprsnttion U G of G if n only if Φ is upwr onsistnt. Givn n upwr plnr rprsnttion U G of G, it is lwys possil to onstrut n upwr plnr strit-lin rwin of G in linr-tim [DGL06]. 3 Fsil Lllins In tis stion, w introu t notion of fsil st of vlus for lllin t nls of G to otin n upwr plnr min of G. Lt T ivn PQ-tr of G wit P-no r s its root. Tn r s tr ilrn n vry otr P-no of T s xtly two ilrn. Lt y 1, y 2 n y 3 t tr ilrn of r. W insrt ummy P-no y p s il of r n mk y 2 n y 3 t two ilrn of y p. Lt T t rsultin tr (s Fi. 6() n ()). Evry P-no of T s xtly two ilrn n t pols of r n y p in T r t sm now. For ny two P-nos z 1 n z 2 in T, w sy tt z 1 is t prnt P-no of z 2 (quivlntly, z 2 is il P-no of z 1 ) if

36 M. A. H. m n M. iur Rmn z 2 is snnt of z 1 n tr is no otr P-no twn z 1 n z 2 on t z 1, z 2 -pt in T. By tis finition, y p is il P-no of r. Lt x P-no wit pols u, v. Lt x l n x r t lft n rit il of x rsptivly. W fin two -isjoint u, v-pts P l (x) n P r (x) s follows: (i) if x is primitiv, tn of G(x l ) n G(x r ) is pt. In tis s, P l (x) = G(x l ), P r (x) = G(x r ) (s Fi. 6()). (ii) if x is not primitiv n is not t root of T, tn lt y not il P-no of x in t lft sutr of x n y not il P-no of x in t rit sutr of x. In tis s, P l (x) will onsist of ll t il Q-nos of x l n P l (y), for il P-no y in t lft sutr. imilrly, P r (x) will onsist of ll t il Q-nos of x r n P l (y ), for il P-no y in t rit sutr (s Fi. 6()). (iii) if x is t root of T, tn x l is itr n - or Q-no n x r is t ummy P-no. In tis s, P l (x) is fin s in s (ii) ov n P r (x) = P l (x r ). (s Fi. 6(f)). W ll of P l (x) n P r (x) pol pt of x. In t rminr of tis ppr, w us C(x) to not t yl P l (x) P r (x) n F(x) to not t f oun y C(x). Lt x non-primitiv P-no in T n y not il P-no of x. W ll swit of F(x) fr swit of F(x) if t swit is nitr on P l (y) nor t t pols of y for ny il P-no y of x. W now introu t notion of fsil llin of P l (x) n t fsil st of P-no x in T. Lt U G(x) n upwr plnr rprsnttion of G(x). Tn n upwr plnr rprsnttion of C(x) n otin from U G(x). Tis n on y simply ltin ll tos vrtis n s of G(x) wi r not in C(x). Lt U C(x) not tis upwr plnr rprsnttion of C(x) n Φ t orrsponin upwr onsistnt ssinmnt. Lt L(x) n (x) not t numr of L- n - lls ssin y Φ to tos swits of F(x) wi r on t pt P l (x). If L(x) = p n (x) = p + q, tn (x) L(x) = q. W sy tt q is fsil vlu of L for llin t swits of F(x) on pt P l (x). Lt Fsil(x) not t st of ll fsil vlus of L for llin t swits on pt P l (x). W rr Fsil(x) s t fsil st of t P- no x. An ssinmnt of lls to t swits on P l (x) is fsil llin of P l (x) if (x) L(x) = q for som q Fsil(x). T followin ft follows t finition of fsil llin. Ft 3 For ivn fsil vlu q Fsil(x), tr is orrsponin upwr plnr rprsnttion of G(x). Lt x non-primitiv P-no in T n y il P-no of x in T. in G(y) is surp of G(x), ivn U G(x) w n otin n upwr plnr rprsnttion U G(y) of G(y) n n t followin ft ols. Ft 4 For ivn fsil vlu q Fsil(x), tr is orrsponin fsil vlu q Fsil(y) for il P-no y of x. W now v t followin lmm.

Upwr Plnr Drwins 37 Lmm 3. G(x) s n upwr plnr rprsnttion if n only if P l (x) n ivn fsil llin. Proof. Nssity. T nssity of t onition follows from t finition of fsil llin. uffiiny. Lt us ssum tt P l (x) is ivn fsil llin orin to fsil vlu q Fsil(x). Tn it follows from Ft 3 n Ft 4 tt tr xists n ssinmnt Φ of - n L- lls to t swits of F(x) su tt ) P l (x) is ll orin to t fsil vlu q, ) for il P-no y of x, P l (y) rivs fsil llin, ) for y, swits of F(x) t t pols of y (if ny) r ssin lls in su wy tt P r (y) n m insi t f wr P l (y) is ivn fsil llin, n ) (F(x)) L(F(x)) = 2 from t finition of upwr onsistnt ssinmnt. in Φ xists, on n fin it y tryin vlu from t fsil st Fsil(y) of il P-no y. Tn tis sm pross my ppli rursivly on P l (y) for il P-no y of x n finlly U G(x) n omput. Q.E.D. W immitly t t followin orollry from Lmm 3. Corollry 1. Lt x P-no in T. Tn t prtinnt irp G(x) of x s no upwr plnr rprsnttion if n only if t st Fsil(x) is mpty. Lt u pol of y su tt tr is swit nl u x of F(x) t vrtx u. As mntion in onition () in t proof of Lmm 3, w n to ll u x in su wy tt P r (y) n m insi t f in wi P l (y) is ivn fsil llin. At t sm tim, w must lso nsur tt t llin of t swits t vrtx u stisfis onition (i) of n upwr onsistnt ssinmnt. in (G) = 3 n G is ionnt, t r of vry vrtx in G is itr two or tr. Any ll ssin to swit t vrtx of r two of G lwys stisfis onition (i) of upwr onsistnt ssinmnt. W trfor n to onntrt on llin t swits t t vrtis of r tr of G. in G is sris-prlll irp, only t pols of t P-nos of T n v r tr. In rr to llin t swits t t pols of t P-nos of T, w v t followin lmm. Lmm 4. Lt x n y two P-nos in T su tt x is t prnt of y. Lt u pol of y su tt tr is swit nl u x of F(x) t vrtx u. Tn u x must ll wit n L-ll wn P r (y) is m insi t f F(x) n wit n -ll wn P r (y) is m in t xtrior of t f F(x) if t followin () or () ol. () u is n orinry vrtx of G n, () u is swit vrtx of G n itr P r (y) is m insi t f F(x) wit lr nl t t swit of F(y) t pol u or P r (y) is m in t xtrior of t f F(x) wit smll nl t t swit of F(y) t pol u. W v omitt t proof of Lmm 4 in tis xtn strt. In of t two ss of Lmm 4, w fin u s n L-pol of no y. T followin lmm is irt onsqun of Lmm 4.

38 M. A. H. m n M. iur Rmn Lmm 5. Lt x n y two P-nos in T su tt x is t prnt of y in T. If ot t pols of y r L-pols, tn in orr to otin n upwr plnr rwin, ot t swits of F(x) t t pols of y must ssin t sm ll. T lls ssin to t swits (if ny) of F(y) t t pols of y ply importnt rols in upwr plnrity tstin s sri in t followin fts. Ft 5 Lt swit vrtx u of G pol of no y, n lt u y t swit nl of F(y) t pol u. Tn u is n L-pol of no y if () u y is ll wit L n F(y) is m s n innr f, n () u y is ll wit n F(y) is m s t outr f. Ft 6 Lt n orinry vrtx u of G pol of no y. Lt u ontin swit nl of F(y). Tn () F(y) must m s t outr f if u y is ll wit L, n () F(y) must m s n innr f if u y is ll wit. W omit t proofs of Ft 5 n Ft 6 r sin t fts r intuitiv onsquns of t finition of L-pols. Lt x P-no of T n y il P-no of x. W now fin litimt llin for no y wi will us xtnsivly trouout t rminr of tis ppr. A llin of t swits of F(x) on P l (y) n t t pols of y is ll litimt llin for no y if t followin () n () ol: () P l (y) is ivn fsil llin n () if pol u of y is n L-pol tn t llin of u x stisfis Lmm 4, otrwis ot n L lls r onsir for llin u x. In t rminr of tis ppr, w us Litimt(y) to not t st of vlus of L insi F(x) orrsponin to litimt llin for no y. W fin llin of t swits on P r (x) to litimt llin of P r (x) if it prforms litimt llin for il P-no y of x in t rit sutr of x n it onsirs ot n L lls for t fr swits on P r (x). W similrly fin litimt llin of P l (x). In t rminr of tis ppr, w us q r n q l to not t vlu of L insi F(x) for litimt llin of P r (x) n P l (x), rsptivly, n w us q pol to not t vlu of L insi F(x) for llin t swits (if ny) of F(x) t t two pols of x. W sy tt 2 (q pol + q r ) is possil fsil vlu of x. If w n fin litimt llin of P l (x) for wi q l = 2 (q pol + q r ), tn q l will fsil vlu of no x n will inlu in t fsil st Fsil(x) of x. As stt rlir, our ojtiv is to omput Fsil(x) for P-no x in T. Givn Fsil(y) for il P-no y of x, w n lwys omput Fsil(x). For tis purpos, on soul fin possil ssinmnt of lls to t swits of F(x) tt nsurs t onitions () () mntion in t proof of Lmm 3. Any loritm followin rut-for ppro to omplis tis woul yil xponntil tim omplxity. T ppro for onstrutin U G(x) outlin in t proof of Lmm 3 woul lso yil xponntil tim omplxity. In our loritm wi w sri in t nxt stion, w sow tt w n omput Fsil(x) in linr-tim y usin t onpts introu tus fr. Furtrmor, if G(x) is upwr plnr, tn w n lso otin U G(x) in linr-tim.

4 An Upwr Plnr Drwin Aloritm Upwr Plnr Drwins 39 In tis stion w iv linr-tim loritm to tst t upwr plnrity of ionnt sris-prlll irp G wit (G) = 3. If G is upwr plnr, tn w lso onstrut n upwr plnr rprsnttion of G in linr-tim. An outlin of our loritm is ivn low. Our loritm onsists of two pss, nmly, t tstin ps n t onstrution ps. In t tstin ps, w trvrs t P-nos of T in ottom-up fsion n t P-no x, w tst t upwr plnrity of G(x). For tis purpos, w omput t fsil st Fsil(x) of no x. If x is primitiv, tn omputin Fsil(x) is quit strit forwr. On t otr n, if x is non-primitiv, tn w n omput Fsil(x) from t fsil sts of t il P-nos of x. If w su in tis ottom-up trvrsl to fin Fsil(r), wr r is t root of T, tn w lr G s n upwr plnr irp n strt our son ps in wi w onstrut n upwr plnr rprsnttion of G. On t otr n, if w fin tt Fsil(x)= for ny P-no x in T, tn from Corollry 1 n Lmm 1, w lr tt G is not upwr plnr. On n sily unrstn tt if w onsir only t omintoril mins of t sklton of P-no of T, tn our ision rrin t upwr plnrity of G tt w mk in sinl trvrsl of T my inorrt. Trfor, w nsur tt our mto onsirs vry plnr min of t sklton of P-no of T ; nvrtlss, our loritm ivs linr-tim s w will sow in tis stion. In t onstrution ps, w prform top-own trvrsl of t P-nos of T. W strt t onstrution ps wit fsil llin of P l (r) wr r is t root of T. Tn in top-own trvrsl of T, t P-no x w ssin lls to t swits of F(x) su tt t ssinmnt stisfis onitions () () ivn in t proof of Lmm 3. Tis prour is rri on t sis of informtion tr in t tstin ps. At t n of tis trvrsl w otin t finl upwr plnr rprsnttion U G of G. W now strt wit t sription of our prour to trmin t fsil st of primitiv P-no. Lt x P-no in T. In t rminr of tis ppr, w us t symols n r, n l n n x to not t numr of swits of F(x) on t pt P r (x), on t pt P l (x) n t t two pols of x, rsptivly. W lso opt t nottion [low.. i] to not t st of intrs in wi t numrs r list in snin orr n t first numr is low, t lst numr is i n if not mntion xpliitly, t prioiity of t numrs is 2. Lt x primitiv P-no n q possil fsil vlu of x. Tn q will fsil vlu of x n inlu in Fsil(x) if q n l. Lt I 0 = [ n r + 2.. n r + 2], I = [ n r +(2 n x ).. n r +(2 n x )] n I + = [ n r +(2+n x ).. n r +(2+n x )]. Tn w v t followin lmms rrin t possil fsil vlus of primitiv P-no x in T wos proofs r omitt in tis xtn strt. Lmm 6. Lt x primitiv P-no of T. Lt I innr n I outr t st of possil fsil vlus of no x for min F(x) s n innr f n t outr f, rsptivly. Tn t followin () n () ol.

40 M. A. H. m n M. iur Rmn () I innr = I outr = I 0 = [ n r + 2.. n r + 2], if n x = 0; n () I innr = I = [ n r + (2 n x ).. n r + (2 n x )] n I outr = I + = [ n r + (2 + n x ).. n r + (2 + n x )], if n x > 0 n of ts n x pols of x is n orinry vrtx of G. Lmm 7. Lt x primitiv P-no of T. Lt I innr n I outr t st of possil fsil vlus of no x for min F(x) s n innr f n t outr f, rsptivly. If n x = 2 n on of t two pols of x is swit vrtx of G n t otr is n orinry vrtx of G, tn t followin () n () ol. () I innr = I I 0 = [ n r.. n r + 2] n I outr = I + I 0 = [ n r + 2.. n r + 4]; n () For vry q I 0, t pol of x wi is lso swit vrtx of G is n L-pol. Lmm 8. Lt x primitiv P-no of T. Lt I innr n I outr t st of possil fsil vlus of no x for min F(x) s n innr f n t outr f, rsptivly. If n x > 0 n of ts n x pols of x is swit vrtx of G, tn t followin () () ol. () For n x = 1, I innr = I outr = I I + n for n x = 2, I innr = I outr = I I + I 0. () For vry q I, of t n x pols of x is n L-pol if F(x) is m s t outr f. () For vry q I +, of t n x pols of x is n L-pol if F(x) is m s n innr f. () For vry q I 0, on of t pols of x is n L-pol if n x = 2. Hvin omput t fsil st for primitiv P-no, w now pro towrs t omputtion of fsil st of ny P-no in T. To omput fsil vlu for non-primitiv P-no x in T, w must nsur tt for vry il P-no y of x, w r prformin litimt llin. Trfor, w first monstrt ow w n fin t st Litimt(y) for il P-no y of x. Lt q Fsil(y). By ivin fsil llin to P l (y) stisfyin q Fsil(y), w n otin tr possil mins of F(y) s sown in Fi. 7() (). In Fi. 7() n () F(y) is m s n innr f. Hn, in ts two ss, xtly on of P r (y) n P l (y), ut not ot, pprs t t outr f. On t otr n, in Fi. 7(), F(y) is m s t outr f, n n ot P l (y) n P r (y) v n rwn rwn t t outr f. From mor tortil point of viw, t tr fiurs in Fi. 7() () tully orrspon to tr possil plnr mins of t sklton of no y. W n isr t otr tr possil plnr mins of t sklton of no y us ty r just t mirror rfltions of t tr mins sown. In orr to otin n min s sown in Fi. 7() w v to ll t swits on P l (y) in su wy tt t lls of ts swits yil L = q insi f F(x). imilrly, in orr to otin t mins sown in Fi. 7() n ()

Upwr Plnr Drwins 41 u F(x) v () u u u u u u u u u u u v Pl(y) v Pl(y) Pr(y) Pr(y) Pl(y) F(y) Pr(y) v v v v v v v v () () () v Pl(y) Fi.7. Tr possiilitis to onsir from P-no x to m t fil yl F(y). t lls of t swits on P l (y) soul yil L = q insi f F(x). W n trmin t litimt vlus of L insi F(x) for t tr snrios sown in Fi. 7 y onsirin t followin tr possil ss: (i) Bot t pols of y r L-pols: for t tr mins in Fi. 7() (), w soul v 2 q, 2 + q, 2 + q rsptivly s t vlu of L insi F(x). (ii) Extly on of t pols of y is n L-pol: w soul v 1 q, 1+q, 1+q rsptivly for t mins in Fi. 7() (). (iii) Non of t pols of y is n L-pol: w soul v q, +q, +q rsptivly for t mins in Fi. 7() (). For t ss (ii) n (iii), if pol of y ontins fr swit of F(x), tn t vlu of L insi F(x) for llin tt swit woul ±1, sin it n ssin itr of t two possil lls. W now sow ow w n omput Litimt(y) wn y is primitiv P-no. In t followin w first onsir only t litimt vlus rsultin from min F(y) s n innr f. Lt q m not t mximum of ts litimt vlus. In Lmm 10 w sow tt, if F(y) is m s t outr f, tn t most two nw litimt vlus n otin, nmly, q m +2 n q m + 4. W v sn in Lmm 6, 7 n 8 tt t st of fsil vlus wi n stisfi for min F(y) s n innr f is of t form: [lo.. i]. W now v t followin lmm rrin t litimt vlus for primitiv P-no y wn F(y) is m s n innr f. Lmm 9. Lt y primitiv P-no in T n x t prnt P-no of y in T. Lt n Lpol not t numr of L-pols of no y. Lt Fsil(y)= [lo.. i] n k not t numr of swits of F(x) t tos pols of y wi r not L-pols. Tn for min F(y) s n innr f t followin () n () ol. () If lo = i = 0 n n Lpol = 2 tn Litimt(y)= { 2, +2}; n () Otrwis, Litimt(y)= [ (mx + k).. (mx + k)], wr mx is t mximum of n Lpol lo n n Lpol i. W now v t followin lmm rrin t litimt vlus of primitiv P-no x wn F(x) is m s t outr f.

42 M. A. H. m n M. iur Rmn Lmm 10. Lt x primitiv P-no in T. Lt q m t mximum of ll t litimt vlus otin y min F(x) s n innr f. Tn t most two nw litimt vlus, nmly, q m + 2 n q m + 4 n otin y min F(x) s t outr f. In t followin lmm, w rss t issu of omputin Fsil(x) for non-primitiv P-no x in T. Lmm 11. Lt x non-primitiv P-no in T. Lt y 1,...,y l t il P-nos of x in T. Tn Fsil(x) n omput from Fsil(y 1 ),..., Fsil(y l ). Proof. Lt y 1, y 2,..., y lft t il P-nos in t lft sutr of x n y 1, y 2,...,y rit t il P-nos in t rit sutr of x. Lt not t it of t P-no x in T. W prov t lim y inution on. W first ssum tt = 1. Tn vry il P-no of x is primitiv. Lt y il P-no of x. Aorin to Lmm 6, 7 n 8, y s two typs of fsil vlus, nmly, t fsil vlus stisfyin wi F(y) n m s n innr f n t fsil vlus stisfyin wi F(y) n m s t outr f. Amon ll t il P-nos y of x, for t most on y, w n m F(y) s t outr f sin plnrity woul violt otrwis. Hn, for y, w first onsir tos vlus from Fsil(y) stisfyin wi w n m F(y) s n innr f, ltr w nl t fsil vlus stisfyin wi w n m F(y) s t outr f. As w v sown in Lmm 9, if y i is il P-no of x in t lft sutr of x, tn t st of litimt vlus for min F(y) s n innr f is [ p i.. p i ] wit prioiity of itr 2 or 4 for som intr p i (1 i lft). imilrly, if y i is il P-no of x in t rit sutr of x, tn t st of litimt vlus for min F(y) s n innr f is [ p i.. p i ] wit prioiity of itr 2 or 4 for som intr p i (1 i rit). Lt p not t numr of fr swits of F(x) on pt P r (x). in llin of of ts fr swits n yil L = ±1 insi F(x), w woul v [ p.. p ] s t st of vlus for llin ts swits. Trfor, t st of litimt vlus for llin of P r (x) will of t form [ (p + p i ).. (p + p i )] wit prioiity of itr 2 or 4. Tkin k = p + p i, w n otin [ k.. k] s t st of litimt vlus for llin of P r (x). From tis st of vlus, w n otin t st of possil fsil vlus of x (i.., w n otin I innr n I outr ) xtly in t sm wy s w v sri in Lmm 6, 7 n 8. It now rmins to trmin wi of ts possil fsil vlus will t fsil vlus of x. For tis purpos, w first trmin t litimt llin of P l (x) xtly in t sm wy s w trmin t litimt vlus for llin of P r (x). Evry possil fsil vlu of x wi is lso litimt vlu for llin of P l (x), will fsil vlu of x n n, will inlu in Fsil(x). It is mntionl tt, in tis omputtion, w o not n to k vry vlu from t formr st wit vry vlu of t lttr st. Rtr, w n otin t wol informtion in tim O(1) from t prioiity n t first n lst vlus of ts two sts. Hvin omput Fsil(x), w

Upwr Plnr Drwins 43 n omput Litimt(x) for min F(x) s n innr f s illustrt in Lmm 9 for primitiv P-no n w n lso omput t possil ns in ts litimt vlus if F(x) is m s t outr f s illustrt in Lmm 10 for primitiv P-no. W now onsir tos fsil vlus of il P-no y of x stisfyin wi F(y) n m s t outr f. W si prviously tt ny su min n inrs t litimt vlus for y y t most +2 n +4. Hn, rrlss of t oi of y, ny su min n us n of i {±2, ±4} in t litimt vlus for x. Lt Extrnl(x) t st of possil ns in t litimt vlus for x if F(x) or F(z) is m s t outr f wr z is snnt P-no of x. Tn Extrnl(x) {i + j : i {±2, ±4} n j {2, 4}} = [ 2.. + 8]. Alon wit Litimt(x), w pss tis st of possil ns to t prnt P-no of x in T. W nxt ssum tt > 1 n tt t ilrn of x v n t followin two quntitis to x. (i) Litimt(y) for vry il P-no y of x n (ii) t possil ns in t litimt vlus for no y tt n otin y min itr F(y) s t outr f, or F(z) s t outr f wr z is snnt P-no of y. In mnnr xtly similr to t s for = 1, w n omput t fsil vlus of x first y onsirin only tos fsil vlus of il P-no y of x wi n stisfi wil min F(y) s n innr f. From tis w trmin t st Litimt(x). Nxt w trmin t possil ns in t litimt vlus for no x tt r otinl y min itr F(x) s t outr f, or F(z) s t outr f wr z is snnt P-no of x. By oin so, w v in quipp ourslvs wit Fsil(x) n lso wit t ov two quntitis () n () wi w woul pss to t prnt of x if x is not t root of T, or if x is t root of T, tn w n strt our son ps of onstrutin U G y usin fsil vlu from Fsil(x). Q.E.D. W ll t loritm outlin in t proofs of Lmm 6, Lmm 7, Lmm 8 n Lmm 11 Aloritm UP-Tstr. W now v t followin torm. Torm 1. Lt G sris-prlll irp wit (G) 3. Tn t upwr plnrity of G n tst in linr-tim. Proof. Lt T t PQ-tr of G. From Lmm 6, Lmm 7, Lmm 8 n Lmm 11, w n fin Fsil(x) for ny P-no x in T y pplyin Aloritm UP-Tstr. If UP-Tstr fins Fsil(x) =, tn from Corollry 1, G(x) is not upwr plnr n n, from Lmm 1 G itslf is not upwr plnr. W n sow tt t oprtions rquir y Aloritm UP-Tstr to omput t fsil sts of ll t P-nos of T n prform in tim linr to t numr of P-nos in T. T tils of ts omputtions r omitt in tis xtn strt. in t numr of P-nos of T is linr in t numr of vrtis of G, t upwr plnrity of G n tst in tim O(n). Q.E.D. Finlly, w iv t followin torm rrin t onstrution of n upwr plnr rprsnttion of G, U G.

44 M. A. H. m n M. iur Rmn Torm 2. Lt r t root of t PQ-tr T of G. If G is upwr plnr, tn strtin wit fsil llin of P l (r), n upwr plnr rprsnttion of G n onstrut in linr-tim. Proof. Our proof is onstrutiv. W sow r ow w n prform fsil llin of P l (y) for il P-no y of x ivn tt P l (x) s n ll wit fsil vlu. Lt q t fsil vlu stisfi for llin P l (x). If q rquirs tt for som snnt P-no z of x, F(z) soul m s t outr f, tn t first w trmin tt P-no. Dpnin on t stisfi fsil vlu q, w know wt lls soul ssin to t swits (if ny) t t pols of x. Hn w ll t swits (if ny) t t pols of x. Lt L = q p insi F(x) for t lls ssin to t swits (if ny) t t pols of x. As sown in t proof of Lmm 11, w n sily omput t sts Litimt(y) tt ltotr yil t st Fsil(x). Lt t st of ll possil vlus for llin t swits on P l (x) [l.. ]. Tn lt i = q. W itrt trou of t omput litimt sts. Lt Litimt(y) =[l y.. y ]. If i > ( y l y ) tn w stisfy t fsil vlu orrsponin to l y for no y n ru i y y l y. Tn w pro wit t nxt no. Wn w fin i ( y l y ) t no y, w stisfy t fsil vlu orrsponin to ( y i) for tt P-no. For of t rminin P- nos in t lft sutr of x, w stisfy t fsil vlu orrsponin to y. W tn prform t sm oprtions in orr to stisfy t vlu 2 q q p for t swits on P r (x). Clrly, t wol tr n trvrs in linr-tim wil prformin ts oprtions t x wil t ompltion of t trvrsl inits tt w v otin n upwr plnr min of G in wi t swits v n ll orin to n upwr onsistnt ssinmnt. Tis omplts t proof of t lim. Q.E.D. 5 Conlusion In tis ppr, w v simpl linr-tim loritm to tst upwr plnrity n in t positiv s, otin n upwr plnr rwin of sris-prlll irp wit t mximum r tr. in our ttntion ws onfin to sris-prlll irps wit t mximum r tr, it looks iffiult to xtn tis loritm in strit forwr wy for mor nrl lsss of irps n lso for sris-prlll irps wit ir rs. It is lft s futur work to fin otr rtriztions of upwr plnrity of sris-prlll irps n vis ffiint loritms for upwr plnrity tstin n upwr plnr rwins of sris-prlll irps wit ir rs. Rfrns [BCDTT94] P. Brtolzzi, R. F. Con, G. Di Bttist, R. Tmssi, n I. G. Tollis, How to rw sris-prlll irp, Intrntionl Journl of Computtionl Gomtry n Applitions, 4 (4), pp. 385 402, 1994.

Upwr Plnr Drwins 45 [BDLM94] P. Brtolzzi, G. Di Bttist, G. Liott, n C. Mnnino, Upwr rwins of trionnt irps, Aloritmi, 6 (12), pp. 476 497, 1994. [BDMT98] P. Brtolzzi, G. Di Bttist, C. Mnnino n R. Tmssi Optiml upwr plnrity tstin of sinl-sour irps, IAM Journl on Computin, 27, pp. 132 169, 1998. [BL76] K.. Boot n G.. Lukr, Tstin t onsutiv ons proprty, intrvl rps, n rp plnrity usin PQ-tr loritms, J. Comp. ys. i., 13, pp. 335 379, 1976. [CDTT94] R. F. Con, G. Di Bttist, R. Tmssi, n I. G. Tollis, Dynmi rp rwin:trs, sris-prlll irps, n plnr st-irps, IAM Journl on Computin, 24 (5), pp. 970 1001, 1995. [CLR01] T. H. Cormn, C. E. Lisrson, R. L. Rivst, n C. tin, Introution to Aloritms, MGrw-Hill Book Compny, on Eition, 2001. [DGL06] W. Diimo, F. Giorno, G. Liott, Upwr spirlity n upwr plnrity tstin, Pro. of GD 05, Ltur Nots in Computr in, 3843, pp. 117 128. prinr-vrl, 2006. [DT96] G. Di Bttist n R. Tmssi, On-lin plnrity tstin, IAM J. Comput., 25(5), pp. 956 997, 1996. [GL99] A. Gr n G. Liott, Almost n-optiml plnr ortoonl rwins of ionnt r-3 plnr rps in qurti tim, Pro. of GD 99, Lt. Nots in Computr in, 1731, pp.38 48, 1999. [GT94] A. Gr n R. Tmssi, On t omputtionl omplxity of upwr n rtilinr plnrity tstin Pro. Grp Drwin 1994, Ltur Nots in Computr in, 894, pp. 286 297. prinr-vrl, 1995. [GT95] A. Gr n R. Tmssi, Upwr plnrity tstin, Orr, 12, pp. 109 133, 1995. [GT01] A. Gr n R. Tmssi, On t omputtionl omplxity of upwr n rtilinr plnrity tstin, IAM Journl on Computin, 31(2), pp. 601 625, 2001. [HL96] M. D. Hutton n A. Luiw, Upwr plnrity tstin of sinl-sour yli irps, IAM Journl on Computin, 25(2), pp. 291 311, 1996. [HL05] P. Hly n K. Lyn, Fix-prmtr trtl loritms for tstin upwr plnrity, Proins of OFEM 05, Ltur Nots in Computr in, 3381, pp. 199 208, 2005. [HT74] J. Hoproft n R. E. Trjn, Effiint plnrity tstin, Journl of t ACM, 21(4), pp. 549 568, 1974. [Pp95] A. Ppkosts, Upwr plnrity tstin of outrplnr s, Pro. of GD 95, Ltur Nots in Computr in, 894, pp. 298 306, 1995. [REN05] M.. Rmn, N. Ei, n T. Nisizki, No-n ortoonl rwins of sris-prlll rps, Pro. of GD 05, Ltur Nots in Computr in, 3843, pp. 409 420. prinr-vrl, 2006. [VTL82] J. Vls, R. E. Trjn n E. L. Lwlr, T Ronition of ris-prlll Dirps, IAM J. Comput., 11 (2), pp. 298 313, 1982.