Theoretical Computer Science. Optimal algorithms for online scheduling with bounded rearrangement at the end

Size: px
Start display at page:

Download "Theoretical Computer Science. Optimal algorithms for online scheduling with bounded rearrangement at the end"

Transcription

1 Theoretical Computer Science 4 (0) Content lit available at SciVere ScienceDirect Theoretical Computer Science journal homepage: Optimal algorithm for online cheduling with bounded rearrangement at the end Xin Chen a, Yan Lan b, Attila Benko c, György Dóa c, Xin Han a, a Software School, Dalian Univerity of Technology, China b Dalian Neuoft Intitute of Information, China c Department of Mathematic, Univerity of Pannonia, Vezprém, Hungary a r t i c l e i n f o a b t r a c t Article hitory: Received 30 May 0 Received in revied form 5 July 0 Accepted 9 July 0 Communicated by D.-Z. Du Keyword: Online cheduling Bounded rearrangement In thi paper, we conider an online non-preemptive cheduling problem on two related machine, where at mot K job are allowed to be rearranged, but only after all job have been revealed and (temporarily) cheduled. We minimize the makepan, and we call the problem a Online cheduling with bounded rearrangement at the end (BRE), which i a emi-online problem. Job arrive one by one over lit. After all the job have been arrived and cheduled, we are informed that the input equence i over; then at mot K already cheduled job can be reaigned. With repect to the wort cae ratio, we cloe the gap between the lower bound and upper bound, improving the previou reult a well. Epecially, for the lower bound, (i) for an improved lower bound + i obtained, which i better than (+) ++ bound + (Liu et al. (009) [9]); (ii) for 5 <, an improved lower (+) i obtained, which i better than (Liu et al. (009) [9]). For the upper + ++ i obtained, which i optimal and bound, (i) for and K =, a new upper bound + + better than the one + bound in Liu et al. (009) [9]; (ii) for + 5 < and K =, an upper + i propoed, which i optimal and better than the previou one in Liu + et al. (009) [9]; (iii) for < + 5 alo optimal and better than the previou one min{ + + and K =, an upper bound (+) i obtained, which i ++, (+) } in Liu et al. (009) [9]. + 0 Elevier B.V. All right reerved.. Introduction In thi paper, we conider an online non-preemptive cheduling problem on two related machine with bounded rearrangement to minimize the completion time, called Online cheduling with bounded rearrangement at the end, BRE for hort, which i a emi-online problem. Job arrive one by one over lit, i.e., after the incoming job ha been aigned, the new job arrive. After all job have arrived, we are informed that there i no further job; then at mot K already cheduled job can be reaigned, where K 0 i a fixed integer. When K = 0 and =, thi problem degenerate into one of the mot fundamental cheduling problem on two machine, aigning job online to identical parallel machine to minimize the completion time [6]; the fundamental (offline) cheduling problem i trongly NP-hard, if there are m machine, and m i not part of the input. [5]. Correponding author. Tel.: addree: cx.dlut@gmail.com (X. Chen), lanyan@neuoft.edu.cn (Y. Lan), benko.attila@almo.vein.hu (A. Benko), doagy@almo.vein.hu (G. Dóa), hanxin.mail@gmail.com (X. Han) /$ ee front matter 0 Elevier B.V. All right reerved. doi:0.06/j.tc

2 670 X. Chen et al. / Theoretical Computer Science 4 (0) Table Lower bound. [, + 5 ) [ + 5, ) [, ) Previou reult Our reult (+) ++ (+) ++ (+) ++ + (+) Table Upper bound. [, + 5 ) [ + 5, ) [, ) Previou reult min{ +, (+) }, K = + +, K =, K = + (+) Our reult, K =, K = +, K = Fig.. The thick curve are for our upper bound which are equal to the lower bound of the problem, i.e., our upper bound are optimal. Related model: Our problem i related to three online model, conidering the poibility of reaigning ome job from ome machine to another. The following online model have been invetigated in the lat year: (i) cheduling with bounded migration [0]; (ii) cheduling with a buffer [7,,8,4,,]; (iii) cheduling with the poible rearrangement of any K job at any time when a new job come, (without knowledge that the equence i ended or not), denoted by BR for hort [3]. Tan and Yu [] defined three further imilar problem where the rearrangement can be done only after all job are revealed and cheduled: (i), the lat job of any machine can be rearranged to the other machine, (ii), the lat K job of the equence can be rearranged, (iii) any K job can be rearranged. In thi paper we will deal with the third problem, what we denote a BRE. Problem BR eem to be more flexible than BRE, ince in the latter cae the rearrangement can be done only once, at the end of the equence. But it i worthy to note an advantage of the latter condition comparing to the former one: in cae of BRE, when we do the rearrangement, we are already informed that there i no further job, while in cae of BR the rearrangement mut be done in uch a way, that we cannot know whether the equence i to be continued, or not. Thu, at thi moment we cannot tate that BR i really more flexible than BRE. Our contribution: In thi paper, we ue competitive ratio to evaluate online algorithm, which i one of the tandard meaure. If an online algorithm alway achieve a olution within a factor ρ of the offline optimum, we ay the online algorithm i ρ-competitive. Our reult are ummarized in Table and (refer to Fig. ).. Preliminarie Scheduling on two related machine Input: Given two machine M, M with peed and repectively, and a et of job J = {j,..., j n } aociated with proceing time p : J R +, Output: Schedule J on M and M uch that the maximal completion time of M and M i minimized. If all the job are known in advance, then we ay the problem i offline. If job are revealed incrementally, i.e., one by one, once the current job i given we have to immediately chedule or aign it and the aignment cannot be changed in the future, then thi verion of the problem i called online. Rearrangement: After all job have been aigned to the machine, we are informed that the equence i over, then at mot K 0 already cheduled job can be reaigned to other machine, where K i a contant. Then any algorithm conit of two main part, the cheduling phae, and then (after being informed that the equence i over), the reaignment phae. In the problem of online cheduling with rearrangement on two related machine, if K = 0, the problem i totally online, if K = n, where n i the number of job in the input, then the problem i offline. In thi paper, we mainly tudy the problem with K < n, which i between online and offline verion.

3 X. Chen et al. / Theoretical Computer Science 4 (0) Given an online algorithm A, if for any input J we have A(J) ρ OPT(J), where A(J) and OPT(J) are the cot by online algorithm A and an optimal algorithm repectively, then we ay online algorithm A i ρ-competitive. On the other ide, if there i an input J for any determinitic online algorithm A, we have A(J) ρ OPT(J), then we ay ρ i the lower bound of the problem. Furthermore ome online algorithm A if ρ = ρ, we ay algorithm A i optimal. In the following, we denote {j, j,..., j t } a J t for t. Let L i t be the load of machine M i after dealing with job j t for i in the cheduling phae. If time t i clear from the context or t i the current time, we ue L i to replace L i t. Function ρ() i defined a the competitive ratio of our online algorithm, for hort, we ue ρ. 3. Lower bound In thi ection, we give new lower bound for problem BRE. And the analyi i imilar with the one for problem BR in [3]. Lemma. For any K, we have the lower bound for problem BRE a below: (i) for + 5 no online algorithm ha it competitive ratio trictly le than (+) + ; (ii) for 5 no online algorithm ha it competitive ratio trictly le than ++ + ; (iii) for > no online algorithm ha it competitive ratio trictly le than. + + Proof. Let ϵ > 0 be a ufficiently mall number uch that /ϵ i integer. Let t be /ϵ. The firt t job are mall job, each one ha a proceing time exactly ϵ. Cae : + 5. The following lower bound wa firt given in [9] for problem BR. Our proof i impler than the one in [9]. For the ake of completion, the detail of the proof are given below. After the t job have been aigned on machine, if L t + or ++ L t +, we are informed that job j ++ t i the lat job. After the lat job j t i given, at mot K job can be reaigned, where K i a contant, we have L t + Kϵ or ++ L t + at mot + ϵ. So the competitive ratio i at leat + L L t (+) t min + ϵ, + ϵ Kϵ( + ) ++, + ϵ( + ) a ϵ goe to zero the lower bound (+) i implied. Ele we conider the next cenario: L t < + + +, + + L t < Kϵ. On the other hand, the optimal value i Then we are informed that job j t+ ha a proceing time and i the lat job. Then the optimal value OPT(J t+ ) =. If j t+ i aigned on M after the poible rearrangement, then L t+ + L t Kϵ Kϵ ele M accept j t+ then the completion time on M L t+ + L t Kϵ ( + ) + + Kϵ, Kϵ = ( + ) + + Kϵ. In both cae the lower bound (+) i implied a ϵ approache to zero. ++ Cae : + 5, aume L t, then the lat job j + t+ ha a proceing time p(j t+ ) =. Then OPT(J t+ ) =. The completion time of any online algorithm i min L +, L + t t Kϵ min, + L t Kϵ min +, + L t Kϵ = + L t Kϵ + Kϵ. In thi cae, when ϵ approache to zero, the competitive ratio approache to +. Ele if L t < +, then L t +. Next two job j t+ and j t+ with p(j t+ ) = + L t and p(j t+) = p(j t+ ) arrive and we are informed that job j t+ i the lat job. Note that p(j t+ ) p(j t+ ). The optimal value OPT(J t+ ) i equal to p(j t+ ) = + L t by the following aignment: j t+ M and J t+ \ {j t+ } M. If j t+ or j j+ i aigned on M then the completion time i at leat L t + p(j t+) Kϵ = L t Kϵ.

4 67 X. Chen et al. / Theoretical Computer Science 4 (0) Ele both j t+ and j t+ are aigned on M, then L t+ L + t p(j t+) + p(j t+ ) Kϵ = ( + ) + L t Kϵ = 3 + L t Kϵ The completion time L t+ / approache to zero. 3 L t Kϵ (by ). L t Kϵ. In both cae, the competitive ratio approache to f () = when ϵ + Cae 3: >, aume L t +, then job j t+ with p(j t+ ) = arrive and it i the lat job. Then OPT(J t+ ) =. The completion time of any online algorithm i min L t +, L t + Kϵ min, + Kϵ = + Kϵ (by > ). + + In thi cae, when ϵ approache to zero, the competitive ratio approache to + +. Ele if L t <, then + L t. Next two job j + t+ and j t+ with p(j t+ ) = ( + )L t and p(j t+ ) = p(j t+ ) arrive and the input end. Note that p(j t+ ) > p(j t+ ) for >. The optimal value OPT(J t+ ) = ( + )L t by the following aignment: j t+ M and J t+ \ {j t+ } M. If j t+ or j j+ i aigned on M, then the completion time i at leat L t + p(j t+) Kϵ = ( + )L t Kϵ. Ele both j t+ and j t+ are aigned to M, then L t+ L t + p(j t+) + p(j t+ ) Kϵ = ( + )L t Kϵ. The completion time L t+ / ( + )L t zero. + Kϵ. In both cae, the competitive ratio approache to when ϵ approache to + 4. Upper bound In thi ection, we give two optimal online algorithm for and repectively. In the optimal algorithm for, we allow to reaign only one job, i.e., K =. In the optimal algorithm for, we allow to reaign at mot two job, i.e., K =. We firt give ome definition and ome ueful obervation. Let t = l max t, P t +, which i a natural lower bound for the problem, where lt denote the ize of the longet job o far (among the firt t job) and P t denote the total ize o far. If t = l t we ay that the lower bound i determined by the longet ize (o far), otherwie we ay that the lower bound i determined by the total ize (o far). Obervation. P t ( + ) t. Given a function ρ() >, conider a chedule jut after aigning job j t, if L t > ρ() t then we ay that M i overloaded ele underloaded; if L t > ρ() t then we ay that M i overloaded, ele underloaded. If both machine are underloaded at time t, we ay the chedule i underloaded. Note that both machine cannot be overloaded at the ame time. Alo note that in cae the chedule i underloaded, then naturally the competitive ratio i not violated (at leat at the current point of the running). But in the oppoite cae, if the chedule i overloaded, then the competitive ratio till can be valid, if the optimum value i bigger than the lower bound. Lemma. If M i overloaded then L > ρ() + ρ() L and if M i overloaded then L > ρ() + ρ() L. Proof. If M i overloaded then L > ρ(), then uing Obervation, we have L ( + ) L ( + ρ()). Hence L > ρ() + ρ() L. If M i overloaded then L > ρ() and L ( + ) L ( + ρ()). Hence L > ρ() + ρ() L.

5 X. Chen et al. / Theoretical Computer Science 4 (0) An optimal algorithm for We give an online algorithm with a competitive ratio ρ() = + for all which algorithm ue only one arrangement, + i.e., K =, and which i optimal for all. The idea are a follow: before the input end, we keep the low machine underloaded all the time; when the input end, if neceary, we find an appreciate job in M and migrate it to M. Algorithm LC (Larget Change). Let job j with ize p be the incoming job. Then update the lower bound.. Aign j to M, if L + p ρ(), ele j M. 3. If the input doe not end, goto Step. 4. Ele if M i overloaded then migrate a job from M to M uch that the final makepan decreae a much a poible by the migration, (if there exit uch job). Theorem 3. Algorithm LC i ρ = + -competitive for any. + Proof. Let ρ = +. Let L + i be the load on M i when the input end for i before the migration. Let L i be the load on M i after the migration for i. It i trivial to ee if max{l, L /} ρ, thi lemma hold. On the other hand, according to the algorithm, if L ρ then max{l, L /} ρ. Hence we conider the cae: max{l, L /} > ρ. In thi cae, before the migration M i overloaded. Aume L = ρ + x with ome x > 0. Then ( + ) L ( + ) L = + x = x. () + + Then we have two lemma. Lemma 4. If max{l, L /} > ρ then there i no job of ize in [x, x + ] on M jut before the migration. Proof. Aume there i a job j with ize p [x, x + ]. Then from () we get L = L p ρ and L = L + p + = ρ, which contradict to the fact + max{l, L /} > ρ. Hence thi lemma hold. () + + = Lemma 5. If max{l, L /} > ρ then jut before the migration the total ize of all the job with ize in (0, x) on M i le than + + x. + + Proof. Let X be the total ize of all the job with ize in (0, x) on M. Aume thi lemma doe not hold, i.e, X > + +x. + + Let t be the time when the input end. Let j r with a ize p (0, x) be the lat job aigned on M, i.e., job j r arrived at time r t. Since X > 0, j r i well-defined, i.e. there exit uch job. According to the algorithm, if job j r i aigned on M, then M would be overloaded. So we have r L + r p > ρ r ρ P + ρ X + L +. + Since L r L x by () and p < x, we have X < L + p + + r L r ρ = ρ + < ρ + x + + ρ x ( + ) = x = + + L + + r ρ p + + x, which caue a contradiction. The aumption i not true and thi lemma hold. By (), Lemma 4 and 5, the total ize of all the job with ize larger than + x i larger than + ρ + x x ( + ) = ( + ) + = = + +. (3)

6 674 X. Chen et al. / Theoretical Computer Science 4 (0) Oberve that in an optimal chedule at leat one job of ize greater than + x i aigned on M or all the job with ize greater than + x are on M, hence we have OPT min + x, + + = + min x,. (4) ( + ) If OPT + (+) = (+)+ (+) max{l, L /} L P = (+), then ince P/( + ), we have (+) ( + ) = + ( + ) + ( + ) ρ OPT. Ele OPT + x, then by () and ince ρ >, we have max{l, L /} L = ρ + x = ρ + x ρ( + x) ρ OPT. Hence thi theorem hold. 4.. An optimal algorithm for We propoe an algorithm which rearrange at mot two job at the end, and it i ρ()-competitive, refer to the following table. The competitive ratio matche the lower bound of the problem, thu the algorithm i optimal. We need two technical ratio b() and c(), which are defined in the following table. We firt give ome lemma which will be ueful in analyzing our algorithm, then propoe and analyze the algorithm. ρ() b() c() I := < + 5 I := + 5 (+) We call a job big, if it ize i bigger than b() ele call it mall. Note that at any point of the execution, it i well defined whether a job i big or mall. If a job i mall at ome time, it never become big. But a big job can become mall at ome time later, ince the value of lower bound can increae. The next Obervation can be checked eaily for both cae regarding I or I. Obervation. For, we have b() = (ρ() )( + ) > ( + ). 4 Lemma 6. Any time, the number of big job i at three. Proof. It follow from that the total ize of all job i P ( + ), while the total ize of four big job would be more than 4b() > ( + ) by Obervation. Obervation 3. For, we have ρ() + = ( + c())b(). To make eay checking the validity of the next obervation we give here a table about the ued expreion. Then the obervation can be checked eaily by ome imple calculation. ρ() ( + )(3 ρ()) (+) I ++ I + (+)( +) ρ() ρ() + + ρ() + ρ() ( + ) 3 c() + Obervation 4. For [, + 5 ), we have ρ ( + ), c() = ρ + ρ > 0. Obervation 5. For, we have ( + )(3 ρ()) ρ(). Proof. For I, the olution of equation = are approximately a follow = , =.3473 and =.53, thu hold if I. Obervation 6. For, we have ρ() ρ() ρ() + ρ(). Proof. For I, to check + 3, one need to validate + 4 3, i.e = + 0 which hold in the conidered interval I. Obervation 7. For, we have ρ() + ρ() > c().

7 X. Chen et al. / Theoretical Computer Science 4 (0) The following table i for the next obervation. ρ() (+) I ++ I + +c() c() ρ() + ρ() ρ() + ρ() Obervation 8. For [ + 5, ], we have c() +c() = +. Obervation 9. For, we have ρ() +c() c() ( + ), or in equivalent form c() ρ() + ρ(). Obervation 0. For, we have ρ() + ρ() +. Proof. To check 3 + we need to ee that 4 ( + ) 3 + = hold, which i the ame 3 + a + which hold if I. Obervation. For, we have b() c()( + b()). Lemma 7. Aume M i i overloaded, for i =,. If M j i overloaded after a job migrate from M i to M j, where j = 3 i, then the job mut be a big job. Proof. Note that the total ize of proceing time i at mot ( + ). We have L > ρ() if M i overloaded, and L > ρ() if M i overloaded. Let j be the job migrated from an overloaded M i to M j, where j = 3 i. After migration, M j become overloaded. Then the ize of job j mut be bigger than ρ() + ρ() ( + ) = ( + ) (ρ() ) = b(), where the lat inequality hold from Obervation. Hence job j i big. Lemma 8. Suppoe that a big job i aigned on M (among other job or alone). Then M cannot be overloaded. Proof. If a big job i aigned to the low machine, then L > b(). We have L ( + ) L < ( + b()), by Obervation and 6, L /L + b() b() Thu M cannot be overloaded by Lemma. = ρ() ρ() ρ() + ρ() An online algorithm and it analyi In our algorithm, there are two phae, the cheduling phae and the reaignment phae. In the cheduling phae,we try to keep the next two propertie through the whole execution, which alo give u a help in the reaignment phae. Let l denote the total load of the mall job aigned on M, without taking into account the big job, we call it a the retricted load. P: there are two mall job on M with ize p and p uch that (L + p + p ) c()(l p p ), where p and p can be zero. It mean that after moving at mot two mall job from M to M, in the modified chedule L c() l hold. P: (L p) c() (l + p), where p i the ize of the lat job aigned on M (p = 0 if there i no job aigned on M ). Scheduling phae of algorithm SMF (low machine firt). Let job j with ize p be the incoming job. Then update the lower bound and alo l ince ome big job may become mall.. Job j i big: if (L + p) ρ() then j M ; otherwie j M. 3. Job j i mall: if L c() l then aign job j to M otherwie to M. 4. Update L, L, and l. If the input end, goto the reaignment phae, ele goto Step. Lemma 9. If there are three big job in the input, at leat one of the three job i aigned on M. Proof. Let j i, j k, j l be the three big job, which have ize larger than b()ρ. If one of the three job i aigned on M, then we are done. Without lo of generality, aume job j i and j k are aigned on M and job j l arrive later than the two other job. We prove that job j l mut be aigned to M at Step of the cheduling phae. According to the definition of the big job, after j l arrive, L p(j i )+p(j k ) > b(). If job j l i aigned on M, L < (+) b() = (+) ( + ) (ρ() ) = ( + )(3 ρ()) ρ() by Obervation and 5. Lemma 0. For, if propertie P and P keep holding during the whole execution of the chedule phae, SMF i ρ()-competitive after the reaignment phae. Proof. Suppoe that the tatement doe not hold. Since machine M and M cannot be overloaded at the ame time, we have the following two cae.

8 676 X. Chen et al. / Theoretical Computer Science 4 (0) Reaignment phae of algorithm SMF (low machine firt). If M i overloaded then move the lat job from M to M.. Ele if M i overloaded then {there are at mot two big job on M } (a) If there exit a job j uch that migrating j from M to M both machine are underloaded, then migrate job j from M to M. (b) Ele i. if there i at mot one big job on M, then move two mall job from M to M to enure L c() l in the modified chedule, ii. ele if there are two big job on M, A. if L > c() l, then move the lat mall job from M to M {to enure L c() l }, and B. move the maller big job from M to M. Cae A: M i overloaded at the end of the cheduling phae. Let t be the current time. In thi cae we perform Step of the reaignment phae. We prove that after the reaignment, both machine are underloaded. Let j k be the lat job aigned in the cheduling phae to M, where k t. If j k wa aigned on M a a big job, then M cannot be overloaded by the algorithm. Hence job j k wa aigned on M a a mall job. In the reaignment tep, job j k i migrated from M to M. After the migration, if M i overloaded then job j k would have been big by Lemma 7, which contradict with the fact that job j k i mall. Hence after the reaignment, M i till underloaded. Aume M i till overloaded after the reaignment. Then L k > ρ() t. Note that job j k wa aigned on M at Step 3 in the cheduling phae. Thu l L k k c() the job we get contradiction a follow: P L + k L > ρ() k t + ρ() t = ρ() + t ( + ) t P, c() c() > ρ() t. Then for the total ize of all c() where we ued Obervation 9. Thu, the aumption that M i overloaded doe not hold, i.e., both machine are underloaded after the reaignment tep in thi cae. Cae B: M i overloaded at the end of the cheduling phae. By Lemma 8, there i no big job on M after the cheduling phae. Let n b be the number of big job on the input. By Lemma 6 we have n b 3. Since M i overloaded, by Lemma 8 and 9, all the big job are on M, thu n b. If the execution pae through Step (a) of the reaignment phae, then both machine are underloaded. Next we conider the cae where the execution pae through Step (b) of the reaignment phae, there are the following three ubcae. Subcae : n b = 0. By property P we can reaign two mall job from M to M to make ure that L c()l. We claim that after the migration M i underloaded. By Lemma 7, after migrating the firt mall job from M to M, M i till underloaded. Since the execution doe not top at Step (a) in the reaignment, after migrating the firt mall item M i till overloaded. Again by Lemma 7, after the econd migration M i till underloaded. Moreover after the moving L l c() hold by the guarantee of property P. Since there i no big job, thi i the ame a L L c(). If thi chedule would be M overloaded, then L /L > ρ() ρ() by Lemma. However by Obervation 7, + ρ() + ρ(), which caue a contradiction. c() Hence after the reaignment M i underloaded. Subcae : n b =. We reaign two mall job from M to M to make ure that L c()l by property P. By the above argument, after the migration M i underloaded. Suppoe that the chedule i M -overloaded after the migration. We have L > ρ(). Since the ize of the big job i at mot, we have l > ρ() = (ρ ), and L c()l > c()(ρ ). Then by Obervation 9 the total ize i L + L > c()(ρ ) + ρ = (c()(ρ ) + ρ) ρ(ρ ) + ρ + ρ ρ = ( + ), + ρ where the lat inequality hold from Obervation 0, i.e. which caue contradiction. Hence M i underloaded after the reaignment. Subcae 3: n b =. Let the two big job be j x and j y, which are aigned on M, and both are bigger than b(). Aume job j x i not larger than j y, i.e., p(j x ) p(j y ). In thi cae, we reaign the lat job j k from M to M to get L c()l by property P, where L i the load of M jut after the firt migration and l i the retricted load of M jut after the firt migration, then reaign job j x from M to M. After migrating job j x on M, by Lemma 8 M i underloaded. If M i underloaded too, then we are done. Otherwie, aume that M i overloaded. Next we prove thi i not poible.

9 X. Chen et al. / Theoretical Computer Science 4 (0) If M i overloaded, then L + p(j x) > ρ. Becaue the total ize i P ( + ), we have l + p(j y) < ( + ) L p(j x) < ( + ) ρ. (5) For [, + 5 ), by Obervation 4 we have ρ < L + p(j x) c()l + p(j y) = (c() )l + l + p(j y) < ( + ρ) + (c() )l, (ρ ) < ρ + ρ l, thu we have l > ( + ρ), which contradict with (5). For [ + 5, ], i.e., > +, we are going to prove that after the rearrangement the load of M i at mot ρ()opt, where OPT i the value by an optimal olution. In thi cae, an optimal algorithm chedule two big job together on one machine or eparately on two machine, then we have a new lower bound, i.e., OPT min p(j x ), p(j x) + p(j y ) = p(j x ). Since L c()l, we have L L + l + c() c() ( + ) p(j x) p(j y ) c() ( + ) p(j x) c(). + c() + c() Uing thi inequality, if p(j x ) then by Obervation 8 L c() + c() ( ) = + = (ρ() ) (ρ() )p(j x) then L + p(j x) ρ()p(j x ) ρ()opt. Ele p(j x ) < then L + p(j x) ( + ) p(j x) + Hence M i underloaded. + p(j x ) = ( + ) + + p(j + x ) ( + ) = p(j ( + ) x) < = ρ(). Lemma. Propertie P and P keep holding through the whole execution in the cheduling phae of the algorithm. Proof. We ue induction approach to prove thi lemma. It i not difficult to ee jut after the firt job of the equence arrive, P and P hold. Aume propertie P and P hold at time t. Let job j t be the next job, with ize p. Regarding property P, if the current job j t i aigned to M by the algorithm, then property P till hold, ince the retricted load of M, i.e. the value of l can only increae. Thu conider the cae j t i aigned to M next. If j t i aigned on M a a big job, i.e., p > b(), then M i underloaded by the algorithm, i.e., L ρ(), where i the lower bound at time t. Thu we have L p < (ρ() b()) c()b() < c()p c()(l + p), ince ρ() b() c()b() hold by Obervation 3. Ele job j t i aigned a a mall job, i.e., it happen at Step 3 of cheduling phae. We have L p c()l c()(l + p). Hence P hold at time t. Let u conider the validity of property P at time t. (We uppoe again that propertie P and P hold at time t.) Firt we note that if the next job j t i mall and aigned to M by the algorithm (in Step 3), then property P trivially hold, ince reaigning only j t to the low machine, the inequality L + j t c() (l j t ) i atified by the algorithmic rule. Thu uppoe in the following that j t i big, or it i mall and aigned to M. Let S t be the et of all the mall job on M at time t. (Thu j t / S t follow by the previou note.) If S t = then property P trivially hold without any rearrangement ince l t = 0. Thu let u uppoe that S t i not empty. Let job j r S t be that job which i aigned to M at the latet time, where r t, i.e. j r i the lat job what i ever aigned to M and it i mall at moment t. Then, ince S t i not empty, and j t / S t, follow that r < t. It mean that all job which are aigned to M after time r are big job when they come, and they remain big until time t. Since any time at mot two big job can be aigned to M, at moment r there are at mot two big job on M. If there i no big job on M at time r, then the value of l r cannot change until time t, that i, l t = l r. If there i one big job on M at time r which become mall until time t, let thi job be denoted by j x. Then l t = l + r p(j x) hold. Finally, if there are two big job on M at time r which become mall until time t, let thee job be denoted by j x and j y, then l t = l r +p(j x)+p(j y ).

10 678 X. Chen et al. / Theoretical Computer Science 4 (0) Furthermore, by the definition of job j r, the inequality hold l t L r, ince all the job aigned on M after time r are big at time t. Now we ditinguih two cae. Cae. Job j r wa aigned on M a a big job. Thi mean that if j r i aigned to M, then M would become to be overloaded, i.e., L r + p(j r) > ρ r. By Lemma and Obervation 9, we have L + r p(j r) ρ > + ρ c(). L r Since L r = L r + p(j r), by (6) we have L t + p(j r) L r + p(j r) > c()l r = c()(l r p(j r)) c()(l t p(j r)). Cae. Job j r wa aigned on M a a mall job. Then, a we have een, there can be three cae, according to that how many big job are at moment r on the fat machine which job become to be mall till moment t. Cae.. If there i not uch job, then l t = l r. Then property P hold trivially, ince after time r the retricted load of the fat machine doe not change, and the load of the low machine can only increae, and at time r hold the next inequality, L + r p(j r) > c()l r, ince j r i aigned to the fat machine. Thu by reaigning j r to the low machine at moment t property P hold. Cae.. There i one big job on M, we denote it by j x, which become to be mall until time t. Then l t = l + r p(j x). Similarly to the previou cae, by reaigning j r and j x to the low machine, L t + p(j r ) + p(j x ) > c() l t p(j r) p(j x ) hold. Cae.3. There are two job on M, j x and j y, which job become to be mall until time t. Then l t = l + r p(j x) + p(j y ). Since job j x and j y wa big at time r, min{p(j x ), p(j y )} > b() r. Hence l r ( + ) r p(j x ) p(j y ) < ( + b()) r, and b() r < L t + p(j x) + p(j y ). By Obervation, we have b() c()( + b()). Thu we have L t + p(j x) + p(j y ) > c()(l t p(j x) p(j y )). Hence thi lemma hold. Remark: in thi paper we cloe the gap between the lower bound and upper bound for all cae except for K = and < <. So the open quetion i to cloe the gap for the cae K = and < <. (6) Acknowledgement The third author wa partially upported by Project TAMOP-4../B-0/ The lat author wa partially upported by the Fundamental Reearch Fund for the Central Univeritie. Reference [] G. Dóa, L. Eptein, Online cheduling with a buffer on related machine, J. Comb. Optim. 0 () (00) [] G. Dóa, L. Eptein, Preemptive online cheduling with reordering, in: ESA, 009, pp [3] G. Dóa, Y. Wang, X. Han, H. Guo, Online cheduling with rearrangement on two related machine, Theoret. Comput. Sci. 4 (8 0) (0) [4] M. Englert, D. Özmen, M. Wetermann, The power of reordering for online minimum makepan cheduling, in: Proc. 48th Symp. Foundation of Computer Science, FOCS, 008, [5] M.R. Garey, D.S. Johnon, Strong np-completene reult: motivation, example and implication, J. ACM 5 (978) [6] R.L. Graham, E.L. Lawler, J.K. Lentra, A.H.G. Rinnony Kan, Optimization and approximation in determinitic equencing and cheduling: a urvey, Ann. Dicrete Math. 5 (979) [7] H. Kellerer, V. Kotov, M.G. Speranza, Z. Tuza, Semi on-line algorithm for the partition problem, Oper. Re. Lett. (5) (997) [8] S. Li, Y. Zhou, G. Sun, G. Chen, Study on parallel machine cheduling problem with buffer, in: Proc. of the nd International Multiympoium on Computer and Computational Science, IMSCCS 007, 007, pp [9] M. Liu, Y. Xu, C. Chu, F. Zheng, Online cheduling on two uniform machine to minimize the makepan, Theoret. Comput. Sci. 40 ( 3) (009) [0] N. Sivadaan, P. Sander, M. Skutella, Online cheduling with bounded migration, Math. Oper. Re. 34 () (009) [] Z. Tan, S. Yu, Online cheduling with reaignment, Oper. Re. Lett. 36 () (008) [] G. Zhang, A imple emi on-line algorithm for p//c max with a buffer, Inform. Proce. Lett. 6 (997)

Preemptive scheduling on a small number of hierarchical machines

Preemptive scheduling on a small number of hierarchical machines Available online at www.ciencedirect.com Information and Computation 06 (008) 60 619 www.elevier.com/locate/ic Preemptive cheduling on a mall number of hierarchical machine György Dóa a, Leah Eptein b,

More information

Bogoliubov Transformation in Classical Mechanics

Bogoliubov Transformation in Classical Mechanics Bogoliubov Tranformation in Claical Mechanic Canonical Tranformation Suppoe we have a et of complex canonical variable, {a j }, and would like to conider another et of variable, {b }, b b ({a j }). How

More information

Social Studies 201 Notes for November 14, 2003

Social Studies 201 Notes for November 14, 2003 1 Social Studie 201 Note for November 14, 2003 Etimation of a mean, mall ample ize Section 8.4, p. 501. When a reearcher ha only a mall ample ize available, the central limit theorem doe not apply to the

More information

Computers and Mathematics with Applications. Sharp algebraic periodicity conditions for linear higher order

Computers and Mathematics with Applications. Sharp algebraic periodicity conditions for linear higher order Computer and Mathematic with Application 64 (2012) 2262 2274 Content lit available at SciVere ScienceDirect Computer and Mathematic with Application journal homepage: wwweleviercom/locate/camwa Sharp algebraic

More information

Problem Set 8 Solutions

Problem Set 8 Solutions Deign and Analyi of Algorithm April 29, 2015 Maachuett Intitute of Technology 6.046J/18.410J Prof. Erik Demaine, Srini Devada, and Nancy Lynch Problem Set 8 Solution Problem Set 8 Solution Thi problem

More information

Lecture 21. The Lovasz splitting-off lemma Topics in Combinatorial Optimization April 29th, 2004

Lecture 21. The Lovasz splitting-off lemma Topics in Combinatorial Optimization April 29th, 2004 18.997 Topic in Combinatorial Optimization April 29th, 2004 Lecture 21 Lecturer: Michel X. Goeman Scribe: Mohammad Mahdian 1 The Lovaz plitting-off lemma Lovaz plitting-off lemma tate the following. Theorem

More information

Competitive Analysis of Task Scheduling Algorithms on a Fault-Prone Machine and the Impact of Resource Augmentation

Competitive Analysis of Task Scheduling Algorithms on a Fault-Prone Machine and the Impact of Resource Augmentation Competitive Analyi of Tak Scheduling Algorithm on a Fault-Prone Machine and the Impact of Reource Augmentation Antonio Fernández Anta a, Chryi Georgiou b, Dariuz R. Kowalki c, Elli Zavou a,d,1 a Intitute

More information

List coloring hypergraphs

List coloring hypergraphs Lit coloring hypergraph Penny Haxell Jacque Vertraete Department of Combinatoric and Optimization Univerity of Waterloo Waterloo, Ontario, Canada pehaxell@uwaterloo.ca Department of Mathematic Univerity

More information

Social Studies 201 Notes for March 18, 2005

Social Studies 201 Notes for March 18, 2005 1 Social Studie 201 Note for March 18, 2005 Etimation of a mean, mall ample ize Section 8.4, p. 501. When a reearcher ha only a mall ample ize available, the central limit theorem doe not apply to the

More information

Multicolor Sunflowers

Multicolor Sunflowers Multicolor Sunflower Dhruv Mubayi Lujia Wang October 19, 2017 Abtract A unflower i a collection of ditinct et uch that the interection of any two of them i the ame a the common interection C of all of

More information

Chapter 4. The Laplace Transform Method

Chapter 4. The Laplace Transform Method Chapter 4. The Laplace Tranform Method The Laplace Tranform i a tranformation, meaning that it change a function into a new function. Actually, it i a linear tranformation, becaue it convert a linear combination

More information

arxiv: v1 [math.mg] 25 Aug 2011

arxiv: v1 [math.mg] 25 Aug 2011 ABSORBING ANGLES, STEINER MINIMAL TREES, AND ANTIPODALITY HORST MARTINI, KONRAD J. SWANEPOEL, AND P. OLOFF DE WET arxiv:08.5046v [math.mg] 25 Aug 20 Abtract. We give a new proof that a tar {op i : i =,...,

More information

SOME RESULTS ON INFINITE POWER TOWERS

SOME RESULTS ON INFINITE POWER TOWERS NNTDM 16 2010) 3, 18-24 SOME RESULTS ON INFINITE POWER TOWERS Mladen Vailev - Miana 5, V. Hugo Str., Sofia 1124, Bulgaria E-mail:miana@abv.bg Abtract To my friend Kratyu Gumnerov In the paper the infinite

More information

μ + = σ = D 4 σ = D 3 σ = σ = All units in parts (a) and (b) are in V. (1) x chart: Center = μ = 0.75 UCL =

μ + = σ = D 4 σ = D 3 σ = σ = All units in parts (a) and (b) are in V. (1) x chart: Center = μ = 0.75 UCL = Our online Tutor are available 4*7 to provide Help with Proce control ytem Homework/Aignment or a long term Graduate/Undergraduate Proce control ytem Project. Our Tutor being experienced and proficient

More information

TRIPLE SOLUTIONS FOR THE ONE-DIMENSIONAL

TRIPLE SOLUTIONS FOR THE ONE-DIMENSIONAL GLASNIK MATEMATIČKI Vol. 38583, 73 84 TRIPLE SOLUTIONS FOR THE ONE-DIMENSIONAL p-laplacian Haihen Lü, Donal O Regan and Ravi P. Agarwal Academy of Mathematic and Sytem Science, Beijing, China, National

More information

ALLOCATING BANDWIDTH FOR BURSTY CONNECTIONS

ALLOCATING BANDWIDTH FOR BURSTY CONNECTIONS SIAM J. COMPUT. Vol. 30, No. 1, pp. 191 217 c 2000 Society for Indutrial and Applied Mathematic ALLOCATING BANDWIDTH FOR BURSTY CONNECTIONS JON KLEINBERG, YUVAL RABANI, AND ÉVA TARDOS Abtract. In thi paper,

More information

Unbounded solutions of second order discrete BVPs on infinite intervals

Unbounded solutions of second order discrete BVPs on infinite intervals Available online at www.tjna.com J. Nonlinear Sci. Appl. 9 206), 357 369 Reearch Article Unbounded olution of econd order dicrete BVP on infinite interval Hairong Lian a,, Jingwu Li a, Ravi P Agarwal b

More information

Some Sets of GCF ϵ Expansions Whose Parameter ϵ Fetch the Marginal Value

Some Sets of GCF ϵ Expansions Whose Parameter ϵ Fetch the Marginal Value Journal of Mathematical Reearch with Application May, 205, Vol 35, No 3, pp 256 262 DOI:03770/jin:2095-26520503002 Http://jmredluteducn Some Set of GCF ϵ Expanion Whoe Parameter ϵ Fetch the Marginal Value

More information

Codes Correcting Two Deletions

Codes Correcting Two Deletions 1 Code Correcting Two Deletion Ryan Gabry and Frederic Sala Spawar Sytem Center Univerity of California, Lo Angele ryan.gabry@navy.mil fredala@ucla.edu Abtract In thi work, we invetigate the problem of

More information

Online Parallel Scheduling of Non-uniform Tasks: Trading Failures for Energy

Online Parallel Scheduling of Non-uniform Tasks: Trading Failures for Energy Online Parallel Scheduling of Non-uniform Tak: Trading Failure for Energy Antonio Fernández Anta a, Chryi Georgiou b, Dariuz R. Kowalki c, Elli Zavou a,d,1 a Intitute IMDEA Network b Univerity of Cypru

More information

LINEAR ALGEBRA METHOD IN COMBINATORICS. Theorem 1.1 (Oddtown theorem). In a town of n citizens, no more than n clubs can be formed under the rules

LINEAR ALGEBRA METHOD IN COMBINATORICS. Theorem 1.1 (Oddtown theorem). In a town of n citizens, no more than n clubs can be formed under the rules LINEAR ALGEBRA METHOD IN COMBINATORICS 1 Warming-up example Theorem 11 (Oddtown theorem) In a town of n citizen, no more tha club can be formed under the rule each club have an odd number of member each

More information

Balanced Network Flows

Balanced Network Flows revied, June, 1992 Thi paper appeared in Bulletin of the Intitute of Combinatoric and it Application 7 (1993), 17-32. Balanced Network Flow William Kocay* and Dougla tone Department of Computer cience

More information

arxiv: v2 [math.co] 11 Sep 2017

arxiv: v2 [math.co] 11 Sep 2017 The maximum number of clique in graph without long cycle Ruth Luo September 13, 017 arxiv:1701.0747v [math.co] 11 Sep 017 Abtract The Erdő Gallai Theorem tate that for k 3 every graph on n vertice with

More information

Electronic Theses and Dissertations

Electronic Theses and Dissertations Eat Tenneee State Univerity Digital Common @ Eat Tenneee State Univerity Electronic Thee and Diertation Student Work 5-208 Vector Partition Jennifer French Eat Tenneee State Univerity Follow thi and additional

More information

One Class of Splitting Iterative Schemes

One Class of Splitting Iterative Schemes One Cla of Splitting Iterative Scheme v Ciegi and V. Pakalnytė Vilniu Gedimina Technical Univerity Saulėtekio al. 11, 2054, Vilniu, Lithuania rc@fm.vtu.lt Abtract. Thi paper deal with the tability analyi

More information

c n b n 0. c k 0 x b n < 1 b k b n = 0. } of integers between 0 and b 1 such that x = b k. b k c k c k

c n b n 0. c k 0 x b n < 1 b k b n = 0. } of integers between 0 and b 1 such that x = b k. b k c k c k 1. Exitence Let x (0, 1). Define c k inductively. Suppoe c 1,..., c k 1 are already defined. We let c k be the leat integer uch that x k An eay proof by induction give that and for all k. Therefore c n

More information

New bounds for Morse clusters

New bounds for Morse clusters New bound for More cluter Tamá Vinkó Advanced Concept Team, European Space Agency, ESTEC Keplerlaan 1, 2201 AZ Noordwijk, The Netherland Tama.Vinko@ea.int and Arnold Neumaier Fakultät für Mathematik, Univerität

More information

Lecture 9: Shor s Algorithm

Lecture 9: Shor s Algorithm Quantum Computation (CMU 8-859BB, Fall 05) Lecture 9: Shor Algorithm October 7, 05 Lecturer: Ryan O Donnell Scribe: Sidhanth Mohanty Overview Let u recall the period finding problem that wa et up a a function

More information

Lecture 7: Testing Distributions

Lecture 7: Testing Distributions CSE 5: Sublinear (and Streaming) Algorithm Spring 014 Lecture 7: Teting Ditribution April 1, 014 Lecturer: Paul Beame Scribe: Paul Beame 1 Teting Uniformity of Ditribution We return today to property teting

More information

A PROOF OF TWO CONJECTURES RELATED TO THE ERDÖS-DEBRUNNER INEQUALITY

A PROOF OF TWO CONJECTURES RELATED TO THE ERDÖS-DEBRUNNER INEQUALITY Volume 8 2007, Iue 3, Article 68, 3 pp. A PROOF OF TWO CONJECTURES RELATED TO THE ERDÖS-DEBRUNNER INEQUALITY C. L. FRENZEN, E. J. IONASCU, AND P. STĂNICĂ DEPARTMENT OF APPLIED MATHEMATICS NAVAL POSTGRADUATE

More information

List Coloring Graphs

List Coloring Graphs Lit Coloring Graph February 6, 004 LIST COLORINGS AND CHOICE NUMBER Thomaen Long Grotzch girth 5 verion Thomaen Long Let G be a connected planar graph of girth at leat 5. Let A be a et of vertice in G

More information

Avoiding Forbidden Submatrices by Row Deletions

Avoiding Forbidden Submatrices by Row Deletions Avoiding Forbidden Submatrice by Row Deletion Sebatian Wernicke, Jochen Alber, Jen Gramm, Jiong Guo, and Rolf Niedermeier Wilhelm-Schickard-Intitut für Informatik, niverität Tübingen, Sand 13, D-72076

More information

Flag-transitive non-symmetric 2-designs with (r, λ) = 1 and alternating socle

Flag-transitive non-symmetric 2-designs with (r, λ) = 1 and alternating socle Flag-tranitive non-ymmetric -deign with (r, λ = 1 and alternating ocle Shenglin Zhou, Yajie Wang School of Mathematic South China Univerity of Technology Guangzhou, Guangdong 510640, P. R. China lzhou@cut.edu.cn

More information

MATEMATIK Datum: Tid: eftermiddag. A.Heintz Telefonvakt: Anders Martinsson Tel.:

MATEMATIK Datum: Tid: eftermiddag. A.Heintz Telefonvakt: Anders Martinsson Tel.: MATEMATIK Datum: 20-08-25 Tid: eftermiddag GU, Chalmer Hjälpmedel: inga A.Heintz Telefonvakt: Ander Martinon Tel.: 073-07926. Löningar till tenta i ODE och matematik modellering, MMG5, MVE6. Define what

More information

Convex Hulls of Curves Sam Burton

Convex Hulls of Curves Sam Burton Convex Hull of Curve Sam Burton 1 Introduction Thi paper will primarily be concerned with determining the face of convex hull of curve of the form C = {(t, t a, t b ) t [ 1, 1]}, a < b N in R 3. We hall

More information

7.2 INVERSE TRANSFORMS AND TRANSFORMS OF DERIVATIVES 281

7.2 INVERSE TRANSFORMS AND TRANSFORMS OF DERIVATIVES 281 72 INVERSE TRANSFORMS AND TRANSFORMS OF DERIVATIVES 28 and i 2 Show how Euler formula (page 33) can then be ued to deduce the reult a ( a) 2 b 2 {e at co bt} {e at in bt} b ( a) 2 b 2 5 Under what condition

More information

Clustering Methods without Given Number of Clusters

Clustering Methods without Given Number of Clusters Clutering Method without Given Number of Cluter Peng Xu, Fei Liu Introduction A we now, mean method i a very effective algorithm of clutering. It mot powerful feature i the calability and implicity. However,

More information

Unavoidable Cycles in Polynomial-Based Time-Invariant LDPC Convolutional Codes

Unavoidable Cycles in Polynomial-Based Time-Invariant LDPC Convolutional Codes European Wirele, April 7-9,, Vienna, Autria ISBN 978--87-4-9 VE VERLAG GMBH Unavoidable Cycle in Polynomial-Baed Time-Invariant LPC Convolutional Code Hua Zhou and Norbert Goertz Intitute of Telecommunication

More information

An Inequality for Nonnegative Matrices and the Inverse Eigenvalue Problem

An Inequality for Nonnegative Matrices and the Inverse Eigenvalue Problem An Inequality for Nonnegative Matrice and the Invere Eigenvalue Problem Robert Ream Program in Mathematical Science The Univerity of Texa at Dalla Box 83688, Richardon, Texa 7583-688 Abtract We preent

More information

Assignment for Mathematics for Economists Fall 2016

Assignment for Mathematics for Economists Fall 2016 Due date: Mon. Nov. 1. Reading: CSZ, Ch. 5, Ch. 8.1 Aignment for Mathematic for Economit Fall 016 We now turn to finihing our coverage of concavity/convexity. There are two part: Jenen inequality for concave/convex

More information

Online Appendix for Managerial Attention and Worker Performance by Marina Halac and Andrea Prat

Online Appendix for Managerial Attention and Worker Performance by Marina Halac and Andrea Prat Online Appendix for Managerial Attention and Worker Performance by Marina Halac and Andrea Prat Thi Online Appendix contain the proof of our reult for the undicounted limit dicued in Section 2 of the paper,

More information

CHAPTER 8 OBSERVER BASED REDUCED ORDER CONTROLLER DESIGN FOR LARGE SCALE LINEAR DISCRETE-TIME CONTROL SYSTEMS

CHAPTER 8 OBSERVER BASED REDUCED ORDER CONTROLLER DESIGN FOR LARGE SCALE LINEAR DISCRETE-TIME CONTROL SYSTEMS CHAPTER 8 OBSERVER BASED REDUCED ORDER CONTROLLER DESIGN FOR LARGE SCALE LINEAR DISCRETE-TIME CONTROL SYSTEMS 8.1 INTRODUCTION 8.2 REDUCED ORDER MODEL DESIGN FOR LINEAR DISCRETE-TIME CONTROL SYSTEMS 8.3

More information

Technical Appendix: Auxiliary Results and Proofs

Technical Appendix: Auxiliary Results and Proofs A Technical Appendix: Auxiliary Reult and Proof Lemma A. The following propertie hold for q (j) = F r [c + ( ( )) ] de- ned in Lemma. (i) q (j) >, 8 (; ]; (ii) R q (j)d = ( ) q (j) + R q (j)d ; (iii) R

More information

Discrete Mathematics

Discrete Mathematics Dicrete Mathematic 310 (010) 334 3333 Content lit available at ScienceDirect Dicrete Mathematic journal homepage: www.elevier.com/locate/dic Rank number of grid graph Hannah Alpert Department of Mathematic,

More information

Chapter Landscape of an Optimization Problem. Local Search. Coping With NP-Hardness. Gradient Descent: Vertex Cover

Chapter Landscape of an Optimization Problem. Local Search. Coping With NP-Hardness. Gradient Descent: Vertex Cover Coping With NP-Hardne Chapter 12 Local Search Q Suppoe I need to olve an NP-hard problem What hould I do? A Theory ay you're unlikely to find poly-time algorithm Mut acrifice one of three deired feature

More information

Hybrid Control and Switched Systems. Lecture #6 Reachability

Hybrid Control and Switched Systems. Lecture #6 Reachability Hbrid Control and Switched Stem Lecture #6 Reachabilit João P. Hepanha Univerit of California at Santa Barbara Summar Review of previou lecture Reachabilit tranition tem reachabilit algorithm backward

More information

A SIMPLE NASH-MOSER IMPLICIT FUNCTION THEOREM IN WEIGHTED BANACH SPACES. Sanghyun Cho

A SIMPLE NASH-MOSER IMPLICIT FUNCTION THEOREM IN WEIGHTED BANACH SPACES. Sanghyun Cho A SIMPLE NASH-MOSER IMPLICIT FUNCTION THEOREM IN WEIGHTED BANACH SPACES Sanghyun Cho Abtract. We prove a implified verion of the Nah-Moer implicit function theorem in weighted Banach pace. We relax the

More information

arxiv: v4 [math.co] 21 Sep 2014

arxiv: v4 [math.co] 21 Sep 2014 ASYMPTOTIC IMPROVEMENT OF THE SUNFLOWER BOUND arxiv:408.367v4 [math.co] 2 Sep 204 JUNICHIRO FUKUYAMA Abtract. A unflower with a core Y i a family B of et uch that U U Y for each two different element U

More information

ON A CERTAIN FAMILY OF QUARTIC THUE EQUATIONS WITH THREE PARAMETERS. Volker Ziegler Technische Universität Graz, Austria

ON A CERTAIN FAMILY OF QUARTIC THUE EQUATIONS WITH THREE PARAMETERS. Volker Ziegler Technische Universität Graz, Austria GLASNIK MATEMATIČKI Vol. 1(61)(006), 9 30 ON A CERTAIN FAMILY OF QUARTIC THUE EQUATIONS WITH THREE PARAMETERS Volker Ziegler Techniche Univerität Graz, Autria Abtract. We conider the parameterized Thue

More information

MULTIPLE POSITIVE SOLUTIONS OF BOUNDARY VALUE PROBLEMS FOR P-LAPLACIAN IMPULSIVE DYNAMIC EQUATIONS ON TIME SCALES

MULTIPLE POSITIVE SOLUTIONS OF BOUNDARY VALUE PROBLEMS FOR P-LAPLACIAN IMPULSIVE DYNAMIC EQUATIONS ON TIME SCALES Fixed Point Theory, 5(24, No. 2, 475-486 http://www.math.ubbcluj.ro/ nodeacj/fptcj.html MULTIPLE POSITIVE SOLUTIONS OF BOUNDARY VALUE PROBLEMS FOR P-LAPLACIAN IMPULSIVE DYNAMIC EQUATIONS ON TIME SCALES

More information

OPTIMAL STOPPING FOR SHEPP S URN WITH RISK AVERSION

OPTIMAL STOPPING FOR SHEPP S URN WITH RISK AVERSION OPTIMAL STOPPING FOR SHEPP S URN WITH RISK AVERSION ROBERT CHEN 1, ILIE GRIGORESCU 1 AND MIN KANG 2 Abtract. An (m, p) urn contain m ball of value 1 and p ball of value +1. A player tart with fortune k

More information

ONLINE APPENDIX: TESTABLE IMPLICATIONS OF TRANSLATION INVARIANCE AND HOMOTHETICITY: VARIATIONAL, MAXMIN, CARA AND CRRA PREFERENCES

ONLINE APPENDIX: TESTABLE IMPLICATIONS OF TRANSLATION INVARIANCE AND HOMOTHETICITY: VARIATIONAL, MAXMIN, CARA AND CRRA PREFERENCES ONLINE APPENDIX: TESTABLE IMPLICATIONS OF TRANSLATION INVARIANCE AND HOMOTHETICITY: VARIATIONAL, MAXMIN, CARA AND CRRA PREFERENCES CHRISTOPHER P. CHAMBERS, FEDERICO ECHENIQUE, AND KOTA SAITO In thi online

More information

Math Skills. Scientific Notation. Uncertainty in Measurements. Appendix A5 SKILLS HANDBOOK

Math Skills. Scientific Notation. Uncertainty in Measurements. Appendix A5 SKILLS HANDBOOK ppendix 5 Scientific Notation It i difficult to work with very large or very mall number when they are written in common decimal notation. Uually it i poible to accommodate uch number by changing the SI

More information

ON THE APPROXIMATION ERROR IN HIGH DIMENSIONAL MODEL REPRESENTATION. Xiaoqun Wang

ON THE APPROXIMATION ERROR IN HIGH DIMENSIONAL MODEL REPRESENTATION. Xiaoqun Wang Proceeding of the 2008 Winter Simulation Conference S. J. Maon, R. R. Hill, L. Mönch, O. Roe, T. Jefferon, J. W. Fowler ed. ON THE APPROXIMATION ERROR IN HIGH DIMENSIONAL MODEL REPRESENTATION Xiaoqun Wang

More information

arxiv: v2 [math.nt] 30 Apr 2015

arxiv: v2 [math.nt] 30 Apr 2015 A THEOREM FOR DISTINCT ZEROS OF L-FUNCTIONS École Normale Supérieure arxiv:54.6556v [math.nt] 3 Apr 5 943 Cachan November 9, 7 Abtract In thi paper, we etablih a imple criterion for two L-function L and

More information

V = 4 3 πr3. d dt V = d ( 4 dv dt. = 4 3 π d dt r3 dv π 3r2 dv. dt = 4πr 2 dr

V = 4 3 πr3. d dt V = d ( 4 dv dt. = 4 3 π d dt r3 dv π 3r2 dv. dt = 4πr 2 dr 0.1 Related Rate In many phyical ituation we have a relationhip between multiple quantitie, and we know the rate at which one of the quantitie i changing. Oftentime we can ue thi relationhip a a convenient

More information

Pacific Journal of Mathematics

Pacific Journal of Mathematics Pacific Journal of Mathematic OSCILLAION AND NONOSCILLAION OF FORCED SECOND ORDER DYNAMIC EQUAIONS MARIN BOHNER AND CHRISOPHER C. ISDELL Volume 230 No. March 2007 PACIFIC JOURNAL OF MAHEMAICS Vol. 230,

More information

A relationship between generalized Davenport-Schinzel sequences and interval chains

A relationship between generalized Davenport-Schinzel sequences and interval chains A relationhip between generalized Davenport-Schinzel equence and interval chain The MIT Faculty ha made thi article openly available. Pleae hare how thi acce benefit you. Your tory matter. Citation A Publihed

More information

Sampling and the Discrete Fourier Transform

Sampling and the Discrete Fourier Transform Sampling and the Dicrete Fourier Tranform Sampling Method Sampling i mot commonly done with two device, the ample-and-hold (S/H) and the analog-to-digital-converter (ADC) The S/H acquire a CT ignal at

More information

ON A CERTAIN FAMILY OF QUARTIC THUE EQUATIONS WITH THREE PARAMETERS

ON A CERTAIN FAMILY OF QUARTIC THUE EQUATIONS WITH THREE PARAMETERS ON A CERTAIN FAMILY OF QUARTIC THUE EQUATIONS WITH THREE PARAMETERS VOLKER ZIEGLER Abtract We conider the parameterized Thue equation X X 3 Y (ab + (a + bx Y abxy 3 + a b Y = ±1, where a, b 1 Z uch that

More information

Singular perturbation theory

Singular perturbation theory Singular perturbation theory Marc R. Rouel June 21, 2004 1 Introduction When we apply the teady-tate approximation (SSA) in chemical kinetic, we typically argue that ome of the intermediate are highly

More information

NCAAPMT Calculus Challenge Challenge #3 Due: October 26, 2011

NCAAPMT Calculus Challenge Challenge #3 Due: October 26, 2011 NCAAPMT Calculu Challenge 011 01 Challenge #3 Due: October 6, 011 A Model of Traffic Flow Everyone ha at ome time been on a multi-lane highway and encountered road contruction that required the traffic

More information

DIFFERENTIAL EQUATIONS

DIFFERENTIAL EQUATIONS DIFFERENTIAL EQUATIONS Laplace Tranform Paul Dawkin Table of Content Preface... Laplace Tranform... Introduction... The Definition... 5 Laplace Tranform... 9 Invere Laplace Tranform... Step Function...4

More information

TMA4125 Matematikk 4N Spring 2016

TMA4125 Matematikk 4N Spring 2016 Norwegian Univerity of Science and Technology Department of Mathematical Science TMA45 Matematikk 4N Spring 6 Solution to problem et 6 In general, unle ele i noted, if f i a function, then F = L(f denote

More information

The machines in the exercise work as follows:

The machines in the exercise work as follows: Tik-79.148 Spring 2001 Introduction to Theoretical Computer Science Tutorial 9 Solution to Demontration Exercie 4. Contructing a complex Turing machine can be very laboriou. With the help of machine chema

More information

On the chromatic number of a random 5-regular graph

On the chromatic number of a random 5-regular graph On the chromatic number of a random 5-regular graph J. Díaz A.C. Kapori G.D. Kemke L.M. Kiroui X. Pérez N. Wormald Abtract It wa only recently hown by Shi and Wormald, uing the differential equation method

More information

Math 273 Solutions to Review Problems for Exam 1

Math 273 Solutions to Review Problems for Exam 1 Math 7 Solution to Review Problem for Exam True or Fale? Circle ONE anwer for each Hint: For effective tudy, explain why if true and give a counterexample if fale (a) T or F : If a b and b c, then a c

More information

RELIABILITY OF REPAIRABLE k out of n: F SYSTEM HAVING DISCRETE REPAIR AND FAILURE TIMES DISTRIBUTIONS

RELIABILITY OF REPAIRABLE k out of n: F SYSTEM HAVING DISCRETE REPAIR AND FAILURE TIMES DISTRIBUTIONS www.arpapre.com/volume/vol29iue1/ijrras_29_1_01.pdf RELIABILITY OF REPAIRABLE k out of n: F SYSTEM HAVING DISCRETE REPAIR AND FAILURE TIMES DISTRIBUTIONS Sevcan Demir Atalay 1,* & Özge Elmataş Gültekin

More information

SOLUTIONS TO ALGEBRAIC GEOMETRY AND ARITHMETIC CURVES BY QING LIU. I will collect my solutions to some of the exercises in this book in this document.

SOLUTIONS TO ALGEBRAIC GEOMETRY AND ARITHMETIC CURVES BY QING LIU. I will collect my solutions to some of the exercises in this book in this document. SOLUTIONS TO ALGEBRAIC GEOMETRY AND ARITHMETIC CURVES BY QING LIU CİHAN BAHRAN I will collect my olution to ome of the exercie in thi book in thi document. Section 2.1 1. Let A = k[[t ]] be the ring of

More information

SPREAD: An Adaptive Scheme for Redundant and Fair Storage in Dynamic Heterogeneous Storage Systems

SPREAD: An Adaptive Scheme for Redundant and Fair Storage in Dynamic Heterogeneous Storage Systems SPREAD: An Adaptive Scheme for Redundant and Fair Storage in Dynamic Heterogeneou Storage Sytem Mario Mene Chritian Scheideler Abtract In thi paper we tudy the problem of deigning an adaptive hah table

More information

The Laplace Transform (Intro)

The Laplace Transform (Intro) 4 The Laplace Tranform (Intro) The Laplace tranform i a mathematical tool baed on integration that ha a number of application It particular, it can implify the olving of many differential equation We will

More information

Online scheduling of jobs with favorite machines

Online scheduling of jobs with favorite machines Online cheduling o job with avorite machine Cong Chen 1, Paolo Penna, and Yineng Xu 1,3 arxiv:181.01343v1 [c.ds] 4 Dec 018 1 School o Management, Xi an Jiaotong Univerity, Xi an, China Department o Computer

More information

Imperfect Signaling and the Local Credibility Test

Imperfect Signaling and the Local Credibility Test Imperfect Signaling and the Local Credibility Tet Hongbin Cai, John Riley and Lixin Ye* Abtract In thi paper we tudy equilibrium refinement in ignaling model. We propoe a Local Credibility Tet (LCT) which

More information

Optimal Coordination of Samples in Business Surveys

Optimal Coordination of Samples in Business Surveys Paper preented at the ICES-III, June 8-, 007, Montreal, Quebec, Canada Optimal Coordination of Sample in Buine Survey enka Mach, Ioana Şchiopu-Kratina, Philip T Rei, Jean-Marc Fillion Statitic Canada New

More information

Laplace Transformation

Laplace Transformation Univerity of Technology Electromechanical Department Energy Branch Advance Mathematic Laplace Tranformation nd Cla Lecture 6 Page of 7 Laplace Tranformation Definition Suppoe that f(t) i a piecewie continuou

More information

Chapter 2 Sampling and Quantization. In order to investigate sampling and quantization, the difference between analog

Chapter 2 Sampling and Quantization. In order to investigate sampling and quantization, the difference between analog Chapter Sampling and Quantization.1 Analog and Digital Signal In order to invetigate ampling and quantization, the difference between analog and digital ignal mut be undertood. Analog ignal conit of continuou

More information

2 Hatad, Jukna & Pudlak gate, namely we hall tudy the ize of depth-three circuit. The technique we hall ue ha two ource. The rt one i a \nite" verion

2 Hatad, Jukna & Pudlak gate, namely we hall tudy the ize of depth-three circuit. The technique we hall ue ha two ource. The rt one i a \nite verion TOP-DOWN LOWER BOUNDS FOR DEPTH-THREE CIRCUITS J. Hatad, S. Jukna and P. Pudlak Abtract. We preent a top-down lower bound method for depth-three ^ _ :-circuit which i impler than the previou method and

More information

IEOR 3106: Fall 2013, Professor Whitt Topics for Discussion: Tuesday, November 19 Alternating Renewal Processes and The Renewal Equation

IEOR 3106: Fall 2013, Professor Whitt Topics for Discussion: Tuesday, November 19 Alternating Renewal Processes and The Renewal Equation IEOR 316: Fall 213, Profeor Whitt Topic for Dicuion: Tueday, November 19 Alternating Renewal Procee and The Renewal Equation 1 Alternating Renewal Procee An alternating renewal proce alternate between

More information

into a discrete time function. Recall that the table of Laplace/z-transforms is constructed by (i) selecting to get

into a discrete time function. Recall that the table of Laplace/z-transforms is constructed by (i) selecting to get Lecture 25 Introduction to Some Matlab c2d Code in Relation to Sampled Sytem here are many way to convert a continuou time function, { h( t) ; t [0, )} into a dicrete time function { h ( k) ; k {0,,, }}

More information

Stochastic Optimization with Inequality Constraints Using Simultaneous Perturbations and Penalty Functions

Stochastic Optimization with Inequality Constraints Using Simultaneous Perturbations and Penalty Functions Stochatic Optimization with Inequality Contraint Uing Simultaneou Perturbation and Penalty Function I-Jeng Wang* and Jame C. Spall** The John Hopkin Univerity Applied Phyic Laboratory 11100 John Hopkin

More information

White Rose Research Online URL for this paper: Version: Accepted Version

White Rose Research Online URL for this paper:   Version: Accepted Version Thi i a repoitory copy of Identification of nonlinear ytem with non-peritent excitation uing an iterative forward orthogonal leat quare regreion algorithm. White Roe Reearch Online URL for thi paper: http://eprint.whiteroe.ac.uk/107314/

More information

Suggested Answers To Exercises. estimates variability in a sampling distribution of random means. About 68% of means fall

Suggested Answers To Exercises. estimates variability in a sampling distribution of random means. About 68% of means fall Beyond Significance Teting ( nd Edition), Rex B. Kline Suggeted Anwer To Exercie Chapter. The tatitic meaure variability among core at the cae level. In a normal ditribution, about 68% of the core fall

More information

Research Article Existence for Nonoscillatory Solutions of Higher-Order Nonlinear Differential Equations

Research Article Existence for Nonoscillatory Solutions of Higher-Order Nonlinear Differential Equations International Scholarly Reearch Network ISRN Mathematical Analyi Volume 20, Article ID 85203, 9 page doi:0.502/20/85203 Reearch Article Exitence for Nonocillatory Solution of Higher-Order Nonlinear Differential

More information

Moment of Inertia of an Equilateral Triangle with Pivot at one Vertex

Moment of Inertia of an Equilateral Triangle with Pivot at one Vertex oment of nertia of an Equilateral Triangle with Pivot at one Vertex There are two wa (at leat) to derive the expreion f an equilateral triangle that i rotated about one vertex, and ll how ou both here.

More information

Lecture 8: Period Finding: Simon s Problem over Z N

Lecture 8: Period Finding: Simon s Problem over Z N Quantum Computation (CMU 8-859BB, Fall 205) Lecture 8: Period Finding: Simon Problem over Z October 5, 205 Lecturer: John Wright Scribe: icola Rech Problem A mentioned previouly, period finding i a rephraing

More information

BUBBLES RISING IN AN INCLINED TWO-DIMENSIONAL TUBE AND JETS FALLING ALONG A WALL

BUBBLES RISING IN AN INCLINED TWO-DIMENSIONAL TUBE AND JETS FALLING ALONG A WALL J. Autral. Math. Soc. Ser. B 4(999), 332 349 BUBBLES RISING IN AN INCLINED TWO-DIMENSIONAL TUBE AND JETS FALLING ALONG A WALL J. LEE and J.-M. VANDEN-BROECK 2 (Received 22 April 995; revied 23 April 996)

More information

Long-term returns in stochastic interest rate models

Long-term returns in stochastic interest rate models Long-term return in tochatic interet rate model G. Deeltra F. Delbaen Vrije Univeriteit Bruel Departement Wikunde Abtract In thi paper, we oberve the convergence of the long-term return, uing an extenion

More information

arxiv: v1 [math.co] 17 Nov 2014

arxiv: v1 [math.co] 17 Nov 2014 Maximizing proper coloring on graph Jie Ma Humberto Nave arxiv:1411.4364v1 [math.co] 17 Nov 2014 Abtract The number of proper q-coloring of a graph G, denoted by P G q, i an important graph parameter that

More information

COHOMOLOGY AS A LOCAL-TO-GLOBAL BRIDGE

COHOMOLOGY AS A LOCAL-TO-GLOBAL BRIDGE COHOMOLOGY AS A LOCAL-TO-GLOBAL BRIDGE LIVIU I. NICOLAESCU ABSTRACT. I dicu low dimenional incarnation of cohomology and illutrate how baic cohomological principle lead to a proof of Sperner lemma. CONTENTS.

More information

UNIT 15 RELIABILITY EVALUATION OF k-out-of-n AND STANDBY SYSTEMS

UNIT 15 RELIABILITY EVALUATION OF k-out-of-n AND STANDBY SYSTEMS UNIT 1 RELIABILITY EVALUATION OF k-out-of-n AND STANDBY SYSTEMS Structure 1.1 Introduction Objective 1.2 Redundancy 1.3 Reliability of k-out-of-n Sytem 1.4 Reliability of Standby Sytem 1. Summary 1.6 Solution/Anwer

More information

Second order conditions for k means clustering: partitions vs. centroids

Second order conditions for k means clustering: partitions vs. centroids Second order condition for k mean clutering: partition v. centroid Ziqiu Su Jacob Kogan Abtract Familie of centroid baed clutering method are capable of handling high dimenional pare data ariing in Text

More information

Primitive Digraphs with the Largest Scrambling Index

Primitive Digraphs with the Largest Scrambling Index Primitive Digraph with the Larget Scrambling Index Mahmud Akelbek, Steve Kirkl 1 Department of Mathematic Statitic, Univerity of Regina, Regina, Sakatchewan, Canada S4S 0A Abtract The crambling index of

More information

Robustness analysis for the boundary control of the string equation

Robustness analysis for the boundary control of the string equation Routne analyi for the oundary control of the tring equation Martin GUGAT Mario SIGALOTTI and Mariu TUCSNAK I INTRODUCTION AND MAIN RESULTS In thi paper we conider the infinite dimenional ytem determined

More information

Estimation of Peaked Densities Over the Interval [0,1] Using Two-Sided Power Distribution: Application to Lottery Experiments

Estimation of Peaked Densities Over the Interval [0,1] Using Two-Sided Power Distribution: Application to Lottery Experiments MPRA Munich Peronal RePEc Archive Etimation of Peaed Denitie Over the Interval [0] Uing Two-Sided Power Ditribution: Application to Lottery Experiment Krzyztof Konte Artal Invetment 8. April 00 Online

More information

Comparing Means: t-tests for Two Independent Samples

Comparing Means: t-tests for Two Independent Samples Comparing ean: t-tet for Two Independent Sample Independent-eaure Deign t-tet for Two Independent Sample Allow reearcher to evaluate the mean difference between two population uing data from two eparate

More information

Problem 1. Construct a filtered probability space on which a Brownian motion W and an adapted process X are defined and such that

Problem 1. Construct a filtered probability space on which a Brownian motion W and an adapted process X are defined and such that Stochatic Calculu Example heet 4 - Lent 5 Michael Tehranchi Problem. Contruct a filtered probability pace on which a Brownian motion W and an adapted proce X are defined and uch that dx t = X t t dt +

More information

CONGESTION control is a key functionality in modern

CONGESTION control is a key functionality in modern IEEE TRANSACTIONS ON INFORMATION TEORY, VOL. X, NO. X, XXXXXXX 2008 On the Connection-Level Stability of Congetion-Controlled Communication Network Xiaojun Lin, Member, IEEE, Ne B. Shroff, Fellow, IEEE,

More information

Riemann s Functional Equation is Not Valid and its Implication on the Riemann Hypothesis. Armando M. Evangelista Jr.

Riemann s Functional Equation is Not Valid and its Implication on the Riemann Hypothesis. Armando M. Evangelista Jr. Riemann Functional Equation i Not Valid and it Implication on the Riemann Hypothei By Armando M. Evangelita Jr. On November 4, 28 ABSTRACT Riemann functional equation wa formulated by Riemann that uppoedly

More information

11.2 Stability. A gain element is an active device. One potential problem with every active circuit is its stability

11.2 Stability. A gain element is an active device. One potential problem with every active circuit is its stability 5/7/2007 11_2 tability 1/2 112 tability eading Aignment: pp 542-548 A gain element i an active device One potential problem with every active circuit i it tability HO: TABIITY Jim tile The Univ of Kana

More information

ON TESTING THE DIVISIBILITY OF LACUNARY POLYNOMIALS BY CYCLOTOMIC POLYNOMIALS Michael Filaseta* and Andrzej Schinzel 1. Introduction and the Main Theo

ON TESTING THE DIVISIBILITY OF LACUNARY POLYNOMIALS BY CYCLOTOMIC POLYNOMIALS Michael Filaseta* and Andrzej Schinzel 1. Introduction and the Main Theo ON TESTING THE DIVISIBILITY OF LACUNARY POLYNOMIALS BY CYCLOTOMIC POLYNOMIALS Michael Filaeta* and Andrzej Schinzel 1. Introduction and the Main Theorem Thi note decribe an algorithm for determining whether

More information