Outline. Parallel Numerical Algorithms. Forward Substitution. Triangular Matrices. Solving Triangular Systems. Back Substitution. Parallel Algorithm

Size: px
Start display at page:

Download "Outline. Parallel Numerical Algorithms. Forward Substitution. Triangular Matrices. Solving Triangular Systems. Back Substitution. Parallel Algorithm"

Transcription

1 Outine Parae Numerica Agorithms Chapter 8 Prof. Michae T. Heath Department of Computer Science University of Iinois at Urbana-Champaign CS 554 / CSE Trianguar Matrices Michae T. Heath Parae Numerica Agorithms 1 / 66 Matrix L is ower trianguar if a entries above its main diagona are zero, ij = 0 for i < j Matrix U is upper trianguar if a entries beow its main diagona are zero, u ij = 0 for i > j Trianguar matrices are important because trianguar inear systems are easiy soved by successive substitution Most direct methods for soving genera inear systems first reduce matrix to trianguar form and then sove resuting equivaent trianguar system(s) Trianguar systems are aso frequenty used as preconditioners in iterative methods for soving inear systems Back Substitution Michae T. Heath Parae Numerica Agorithms 3 / 66 For upper trianguar system Ux = b, soution can be obtained by back substitution ( x i = b i for j = n to 1 x j = b j /u jj for i = 1 to j 1 b i = b i u ij x j n u ij x j )/u ii, i = n,..., 1 j=i+1 { compute son component } { update right-hand side } Forward Substitution Michae T. Heath Parae Numerica Agorithms 2 / 66 For ower trianguar system Lx = b, soution can be obtained by forward substitution ( i 1 x i = b i ij x j )/ ii, i = 1,..., n for j = 1 to n x j = b j / jj for i = j + 1 to n j=1 Soving { compute son component } { update right-hand side } Michae T. Heath Parae Numerica Agorithms 4 / 66 Forward or back substitution requires about n 2 /2 mutipications and simiar number of additions, so mode seria time as T 1 = t c n 2 /2 where t c is cost of paired mutipication and addition (we ignore cost of n divisions) We wi consider ony ower trianguar systems, as anaogous agorithms for upper trianguar systems are simiar Michae T. Heath Parae Numerica Agorithms 5 / 66 Michae T. Heath Parae Numerica Agorithms 6 / 66 Loop Orderings for Forward Substitution Parae Agorithm for j = 1 to n x j = b j / jj for i = j + 1 to n right-ooking immediate-update data-driven fan-out for i = 1 to n for j = 1 to i 1 x i = b i / ii eft-ooking deayed-update demand-driven fan-in Partition For i = 2,..., n, j = 1,..., i 1, fine-grain task (i, j) stores ij and computes product ij x j For i = 1,..., n, fine-grain task (i, i) stores ii and b i, coects sum t i = i 1 j=1 ij x j, and computes and stores x i = (b i t i )/ ii yieding 2-D trianguar array of n (n + 1)/2 fine-grain tasks Communicate For j = 1,..., n 1, task (j, j) broadcasts x j to tasks (i, j), i = j + 1,..., n For i = 2,..., n, sum reduction of products ij x j across tasks (i, j), j = 1,..., i, with task (i, i) as root Michae T. Heath Parae Numerica Agorithms 7 / 66 Michae T. Heath Parae Numerica Agorithms 8 / 66

2 Fine-Grain Tasks and Communication Fine-Grain Parae Agorithm if i = j then t = 0 if i > 1 then recv sum reduction of t across tasks (i, k), k = 1,..., i x i = (b i t)/ ii broadcast x i to tasks (k, i), k = i + 1,..., n ese recv broadcast of x j from task (j, j) t = ij x j reduce t across tasks (i, k), k = 1,..., i Michae T. Heath Parae Numerica Agorithms 9 / 66 Michae T. Heath Parae Numerica Agorithms 10 / 66 Aggomeration If communication is suitaby pipeined, then fine-grain agorithm can achieve Θ(n) execution time, but uses Θ(n 2 ) tasks, so it is inefficient If there are mutipe right-hand-side vectors b, then successive soutions can be pipeined to increase overa efficiency Aggomerating fine-grain tasks yieds more reasonabe number of tasks and improves ratio of computation to communication Aggomerate With n n array of fine-grain tasks, natura strategies are 2-D: combine k k subarray of fine-grain tasks to form each coarse-grain task, yieding (n/k) 2 coarse-grain tasks 1-D coumn: combine n fine-grain tasks in each coumn into coarse-grain task, yieding n coarse-grain tasks 1-D row: combine n fine-grain tasks in each row into coarse-grain task, yieding n coarse-grain tasks Michae T. Heath Parae Numerica Agorithms 11 / 66 Michae T. Heath Parae Numerica Agorithms 12 / 66 2-D Aggomeration 1-D Coumn Aggomeration Michae T. Heath Parae Numerica Agorithms 13 / 66 Michae T. Heath Parae Numerica Agorithms 14 / 66 1-D Row Aggomeration Mapping Map 2-D: assign (n/k) 2 /p coarse-grain tasks to each of p processes using any desired mapping in each dimension, treating target network as 2-D mesh 1-D: assign n/p coarse-grain tasks to each of p processes using any desired mapping, treating target network as 1-D mesh Michae T. Heath Parae Numerica Agorithms 15 / 66 Michae T. Heath Parae Numerica Agorithms 16 / 66

3 2-D Aggomeration, Bock Mapping 2-D Aggomeration, Cycic Mapping Michae T. Heath Parae Numerica Agorithms 17 / 66 Michae T. Heath Parae Numerica Agorithms 18 / 66 2-D Aggomeration, Refection Mapping 2 6 For 2-D aggomeration with (n/ p ) (n/ p ) subarray of fine-grain tasks per process, both vertica broadcasts and horizonta sum reductions are required to communicate soution components and accumuate inner products, respectivey If each process hods contiguous bock of rows and coumns, we obtain bock version of origina fine-grain agorithm, with poor concurrency and efficiency Moreover, this approach yieds ony (p + p )/2 non-nu processes, wasting amost haf of 2-D mesh of processors Michae T. Heath Parae Numerica Agorithms 19 / 66 Michae T. Heath Parae Numerica Agorithms 20 / 66 Cycic assignment of rows and coumns to processes yieds p non-nu processes, so fu 2-D mesh can be utiized But obvious impementation, ooping over successive soution components and performing corresponding horizonta sum reductions and vertica broadcasts, sti has imited concurrency because computation for each component invoves ony one process row and one process coumn Better agorithm can be obtained by computing soution components in groups of p, which permits a processes to perform resuting updating concurrenty Each step of resuting agorithm has four phases 1 Computation of next p soution components by processes in ower triange using 2-D fine-grain agorithm 2 Broadcast of resuting soution components verticay from processes on diagona to processes in upper triange 3 Computation of resuting updates (partia sums in inner products) by a processes 4 Horizonta sum reduction from processes in upper triange to processes on diagona Michae T. Heath Parae Numerica Agorithms 21 / 66 Michae T. Heath Parae Numerica Agorithms 22 / 66 Tota time required is approximatey T p = t c n 2 /(2p) + (4(t s + t w ) + 5 t c ) n 1. Fine-grain agorithm 2. Broadcast To determine isoefficiency function, set t c n 2 /2 E (t c n 2 /2 + (4(t s + t w ) + 5 t c ) p n) which hods for arge p if n = Θ(p), so isoefficiency function is Θ(p 2 ), since T 1 = Θ(n 2 ) 3. Update 4. Sum reduction Michae T. Heath Parae Numerica Agorithms 23 / 66 Michae T. Heath Parae Numerica Agorithms 24 / 66

4 1-D Coumn Aggomeration, Bock Mapping 1-D Coumn Aggomeration, Cycic Mapping Michae T. Heath Parae Numerica Agorithms 25 / 66 Michae T. Heath Parae Numerica Agorithms 26 / 66 1-D Coumn Aggomeration, Refection Mapping For 1-D aggomeration with n/p coumns of fine-grain tasks per process, vertica broadcasts of components of x are unnecessary because any given matrix coumn is entirey contained in ony one process But there is aso no paraeism in computing products resuting from given component of x Horizonta communication is required for sum reductions to accumuate inner products 2 6 Michae T. Heath Parae Numerica Agorithms 27 / 66 Michae T. Heath Parae Numerica Agorithms 28 / 66 1-D Coumn Fan-in Agorithm for i = 1 to n t = 0 for j mycos, j < i, t = t + ij x j if i mycos then recv sum reduction of t x i = (b i t)/ ii ese reduce t across processes Each process remains ide unti soution component corresponding to its first coumn is computed If each process hods contiguous bock of coumns, it may remain ide through most of computation Moreover, number of products computed invoving each component of x decines with increasing coumn number Concurrency and oad baance can be improved by assigning coumns to processes in cycic manner Other mappings may aso be usefu, such as bock-cycic or refection Michae T. Heath Parae Numerica Agorithms 29 / 66 Michae T. Heath Parae Numerica Agorithms 30 / 66 If successive steps (outer oop) are overapped, then approximate execution time is T p = t c (n 2 + 2n(p 1))/(2p) + (t s + t w ) (n 1) ignoring cost of additions in sum reductions Without such overapping, term representing communication cost is mutipied by factor of p 1 for 1-D mesh 2( p 1) for 2-D mesh og p for hypercube representing path ength for sum reduction To determine isoefficiency function, set t c n 2 /2 E (t c (n 2 + 2n(p 1))/2 + (t s + t w ) p (n 1)) which hods for arge p if n = Θ(p), so isoefficiency function is Θ(p 2 ), since T 1 = Θ(n 2 ) Without overapping of successive steps, isoefficiency function becomes p 4 for 1-D mesh p 3 for 2-D mesh p 2 (og p) 2 for hypercube Michae T. Heath Parae Numerica Agorithms 31 / 66 Michae T. Heath Parae Numerica Agorithms 32 / 66

5 1-D Row Aggomeration, Bock Mapping Overap achievabe is strongy affected by network topoogy and mapping of rows to processes For exampe, cycic mapping on ring network permits amost compete overap, whereas hypercube permits much ess overap Overap of successive steps can potentiay be enhanced by compute ahead strategy Process owning coumn i coud compute most of its contribution to inner product for step i + 1 whie waiting for contributions from other processes in step i, thereby avoiding being botteneck for next step (because it wi be ast to compete step i) Michae T. Heath Parae Numerica Agorithms 33 / 66 1-D Row Aggomeration, Cycic Mapping Michae T. Heath Parae Numerica Agorithms 34 / 66 1-D Row Aggomeration, Refection Mapping Michae T. Heath Parae Numerica Agorithms 35 / 66 Michae T. Heath Parae Numerica Agorithms 36 / 66 1-D Row Fan-out Agorithm For 1-D aggomeration with n/p rows of fine-grain tasks per process, communication for horizonta sum reductions across process rows is unnecessary because any given matrix row is entirey contained in ony one process But there is aso no paraeism in computing these sums Vertica broadcasts are required to communicate components of x for j = 1 to n if j myrows then x j = b j / jj broadcast x j for i myrows, i > j, Michae T. Heath Parae Numerica Agorithms 37 / 66 Michae T. Heath Parae Numerica Agorithms 38 / 66 Each process fas ide as soon as soution component corresponding to its ast row has been computed If each process hods contiguous bock of rows, it may become ide ong before overa computation is compete Moreover, computation of inner products across rows requires successivey more work with increasing row number Concurrency and oad baance can be improved by assigning rows to processes in cycic manner Other mappings may aso be usefu, such as bock-cycic or refection If successive steps (outer oop) are overapped, then approximate execution time is T p = t c (n 2 + 2n(p 1))/(2p) + (t s + t w ) (n 1) Without such overapping, term representing communication cost is mutipied by factor of p 1 for 1-D mesh 2( p 1) for 2-D mesh og p for hypercube representing path ength for broadcast Michae T. Heath Parae Numerica Agorithms 39 / 66 Michae T. Heath Parae Numerica Agorithms 40 / 66

6 To determine isoefficiency function, set t c n 2 /2 E (t c (n 2 + 2n(p 1))/2 + (t s + t w ) p (n 1)) which hods for arge p if n = Θ(p), so isoefficiency function is Θ(p 2 ), since T 1 = Θ(n 2 ) Without overapping of successive steps, isoefficiency function becomes p 4 for 1-D mesh p 3 for 2-D mesh p 2 (og p) 2 for hypercube Overap achievabe is aso strongy affected by network topoogy and mapping of rows to processes For exampe, cycic mapping on ring network permits amost compete overap, whereas hypercube permits much ess overap Overap of successive steps can potentiay be enhanced by s ahead strategy At step j, process owning row j + 1 coud compute x j+1 and broadcast it before competing remaining updating due to x j Michae T. Heath Parae Numerica Agorithms 41 / 66 Michae T. Heath Parae Numerica Agorithms 42 / 66 Fan-out and fan-in agorithms derive their paraeism from inner oop, whose work is partitioned and distributed across processes, whie outer oop is seria Conceptuay, fan-out and fan-in agorithms work on ony one component of soution at a time, though successive steps may be pipeined to some degree Wavefront agorithms expoit paraeism in outer oop expicity by working on mutipe components of soution simutaneousy Michae T. Heath Parae Numerica Agorithms 43 / 66 To formaize wavefront coumn agorithm we introduce z : vector in which to accumuate updates to right-hand-side segment : set containing at most s consecutive components of z Michae T. Heath Parae Numerica Agorithms 45 / 66 1-D coumn fan-out agorithm seems to admit no paraeism: after process owning coumn j computes x j, resuting updating of right-hand side cannot be shared with other processes because they have no access to coumn j Instead of performing a such updates immediatey, however, process owning coumn j coud compete ony first s components of update vector and forward them to process owning coumn j + 1 before continuing with next s components of update vector, etc. Upon receiving first s components of update vector, process owning coumn j + 1 can compute x j+1, begin further updates, forward its own contributions to next process, etc. Michae T. Heath Parae Numerica Agorithms 44 / 66 for j mycos for k = 1 to # segments recv segment if k = 1 then x j = (b j z j )/ jj segment = segment {z j } for z i segment z i = z i + ij x j if segment > 0 then s segment to process owning coumn j + 1 Michae T. Heath Parae Numerica Agorithms 46 / 66 Deping on segment size, coumn mapping, communication-to-computation speed ratio, etc., it may be possibe for a processes to become busy simutaneousy, each working on different component of soution Segment size is adjustabe parameter that contros tradeoff between communication and concurrency First segment for given coumn shrinks by one eement after each component of soution is computed, disappearing after s steps, when next segment becomes first segment, etc. At of computation ony one segment remains and it contains ony one eement Communication voume decines throughout agorithm As segment ength s increases, communication start-up cost decreases but computation cost increases, and vice versa as segment ength decreases Optima choice of segment ength s can be predicted from performance mode Michae T. Heath Parae Numerica Agorithms 47 / 66 Michae T. Heath Parae Numerica Agorithms 48 / 66

7 Approximate execution time is T p = ((t s /s) + t w + t c ) (n 2 + np + s(s 1)p 2 )/(2p) where s is segment ength To determine isoefficiency function, set t c n 2 /2 E (((t s /s) + t w + t c ) (n 2 + np + s(s 1)p 2 )/2) which hods for arge p if n = Θ(p), assuming s is constant, so isoefficiency function is Θ(p 2 ), since T 1 = Θ(n 2 ) Wavefront approach can aso be appied to 1-D row fan-in agorithm Computation of ith inner product cannot be shared because ony one process has access to row i of matrix Thus, work on mutipe components must be overapped to attain any concurrency Anaogous approach is to break soution vector x into segments that are pipeined through processes Michae T. Heath Parae Numerica Agorithms 49 / 66 Michae T. Heath Parae Numerica Agorithms 50 / 66 Initiay, process owning row 1 computes and ss it to process owning row 2, which computes resuting update and then This process continues (seriay at this eary stage) unti s components of soution have been computed Henceforth, receiving processes forward any fu-size segments before they are used in updating Forwarding of currenty incompete segment is deayed unti next component of soution is computed and apped to it Michae T. Heath Parae Numerica Agorithms 51 / 66 for i myrows for k = 1 to # segments 1 recv segment s segment to process owning row i + 1 for x j segment recv segment /* ast may be empty */ for x j segment x i = b i / ii segment = segment {x i } s segment to process owning row i + 1 Michae T. Heath Parae Numerica Agorithms 53 / 66 Michae T. Heath Parae Numerica Agorithms 52 / 66 Instead of starting with fu set of segments that shrink and eventuay disappear, segments appear and grow unti there is a fu set of them It may be possibe for a processes to be busy simutaneousy, each working on different segment Segment size is adjustabe parameter that contros tradeoff between communication and concurrency, and optima vaue of segment ength s can be predicted from performance mode Performance anaysis and resuting performance mode are more compicated than for 1-D coumn wavefront agorithm, but performance and scaabiity for 1-D row wavefront agorithm are nevertheess simiar Michae T. Heath Parae Numerica Agorithms 54 / 66 In wavefront agorithms, each segment is sent up to s times and may pass through same process repeatedy, deping on mapping of rows or coumns Cycic agorithms are somewhat simiar to wavefront agorithms, but they minimize communication by expoiting cycic mapping of rows or coumns Instead of having variabe number of segments of adjustabe ength, cycic agorithms circuate singe segment of ength p 1 In cycic 1-D coumn agorithm, segment of size p 1, containing partiay accumuated components of update vector z, passes from process to process, one step for each coumn of matrix, cycing through a p 1 other processes before returning to any given process At step j, process owning coumn j receives segment from process owning coumn j 1 and uses its first eement (which has accumuated a necessary prior updates) to compute x j Task owning coumn j then modifies segment by deeting first eement, updating remaining eements, and apping new eement to begin accumuation of z j+p 1 Michae T. Heath Parae Numerica Agorithms 55 / 66 Michae T. Heath Parae Numerica Agorithms 56 / 66

8 Segment is then sent to process owning coumn j + 1, where simiar procedure is repeated After forwarding modified segment, process owning coumn j then computes remaining updates resuting from x j, which wi be needed when segment returns to this process again Updating whie segment is esewhere provides a concurrency, since computations on segment are seria for j mycos recv segment x j = (b j z j t j )/ jj segment = segment {z j } for z i segment z i = z i + t i + ij x j z j+p 1 = t j+p 1 + j+p 1,j x j segment = segment {z j+p 1 } s segment to process owning coumn j + 1 for i = j + p to n t i = t i + ij x j Michae T. Heath Parae Numerica Agorithms 57 / 66 Michae T. Heath Parae Numerica Agorithms 58 / 66 Segment must pass through a other processes before returning to any given process, so correctness deps on use of cycic mapping Maps naturay to 1-D torus (ring) network, but since ony one pair of processes communicates at any given time, aso works we with bus network Attains minimum possibe voume of interprocessor communication to sove trianguar system using coumn-oriented agorithm For n p (t p + p), where t p = (t s + t w (p 1))/t c is cost, measured in fops, of sing message of ength p 1, execution time is determined by segment cyce time, so that T p = t c (n (t p + p) p (p 1)/2 t p ) For n > p (t p + p), execution time is dominated by cost of updating, so that T p = t c ((n 2 + n p)/(2p) + p ((t p + p) 2 t p p + 1)/2 t p ) Michae T. Heath Parae Numerica Agorithms 59 / 66 Michae T. Heath Parae Numerica Agorithms 60 / 66 Two-phase behavior compicates scaabiity anaysis, but tradeoff point between phases for n as function of p grows ike p 2, so isoefficiency function is at east Θ(p 4 ) Performance of both phases can be improved Segment cyce time can be reduced by breaking segment into smaer pieces and pipeining them through processes Updating work can be reorganized, deferring excessive work unti ater cyces, to obtain more even distribution throughout computation 1-D row cycic agorithm is simiar, except processes are aggomerated by rows and segment contains p 1 components of soution x At step i, process owning row i receives segment from process owning row i 1 and uses components of x it contains to compete ith inner product, so that x i can then be computed Task then modifies segment by deeting first eement and apping new eement x i just computed Segment is then sent to process owning row i + 1, where simiar procedure is repeated Michae T. Heath Parae Numerica Agorithms 61 / 66 After forwarding modified segment, process then computes partia inner products that use components of segment, which wi be further accumuated when segment returns to this process again Latter computations, which take pace whie segment passes through other processes, provide concurrency in agorithm, because computations on segment itsef are seria Again, correctness of agorithm deps on use of cycic mapping Performance and scaabiity are simiar to those for 1-D coumn cycic agorithm, athough detais differ Michae T. Heath Parae Numerica Agorithms 63 / 66 Michae T. Heath Parae Numerica Agorithms 62 / 66 for i myrows recv segment for x j segment x i = b i / ii segment = segment {x i p } {x i } s segment to process owning row i + 1 for m myrows, m > i, for x j segment b m = b m mj x j Michae T. Heath Parae Numerica Agorithms 64 / 66

9 References References R. H. Bisseing and J. G. G. van de Vorst, Parae trianguar system soving on a mesh network of Transputers, SIAM J. Sci. Stat. Comput. 12: , 1991 S. C. Eisenstat, M. T. Heath, C. S. Henke, and C. H. Romine, Modified cycic agorithms for soving trianguar systems on distributed-memory mutiprocessors, SIAM J. Sci. Stat. Comput. 9: , 1988 M. T. Heath and C. H. Romine, Parae soution of trianguar systems on distributed-memory mutiprocessors, SIAM J. Sci. Stat. Comput. 9: , 1988 N. J. Higham, Stabiity of parae trianguar system sovers, SIAM J. Sci. Comput. 16: , 1995 G. Li and T. F. Coeman, A parae trianguar sover for a distributed-memory mutiprocessor, SIAM J. Sci. Stat. Comput. 9: , 1988 G. Li and T. F. Coeman, A new method for soving trianguar systems on distributed-memory message-passing mutiprocessors, SIAM J. Sci. Stat. Comput. 10: , 1989 C. H. Romine and J. M. Ortega, Parae soution of trianguar systems of equations, Parae Computing 6: , 1988 E. E. Santos, On designing optima parae trianguar sovers, Information and Computation 161: , 2000 Michae T. Heath Parae Numerica Agorithms 65 / 66 Michae T. Heath Parae Numerica Agorithms 66 / 66

Parallel Numerical Algorithms

Parallel Numerical Algorithms Parallel Numerical Algorithms Chapter 3 Dense Linear Systems Section 3.3 Triangular Linear Systems Michael T. Heath and Edgar Solomonik Department of Computer Science University of Illinois at Urbana-Champaign

More information

Further Optimization of the Decoding Method for Shortened Binary Cyclic Fire Code

Further Optimization of the Decoding Method for Shortened Binary Cyclic Fire Code Further Optimization of the Decoding Method for Shortened Binary Cycic Fire Code Ch. Nanda Kishore Heosoft (India) Private Limited 8-2-703, Road No-12 Banjara His, Hyderabad, INDIA Phone: +91-040-3378222

More information

A Design Method for Optimal Truss Structures with Certain Redundancy Based on Combinatorial Rigidity Theory

A Design Method for Optimal Truss Structures with Certain Redundancy Based on Combinatorial Rigidity Theory 0 th Word Congress on Structura and Mutidiscipinary Optimization May 9 -, 03, Orando, Forida, USA A Design Method for Optima Truss Structures with Certain Redundancy Based on Combinatoria Rigidity Theory

More information

A Comparison of a Second-Order versus a Fourth- Order Laplacian Operator in the Multigrid Algorithm

A Comparison of a Second-Order versus a Fourth- Order Laplacian Operator in the Multigrid Algorithm A Comparison of a Second-Order versus a Fourth- Order Lapacian Operator in the Mutigrid Agorithm Kaushik Datta (kdatta@cs.berkeey.edu Math Project May 9, 003 Abstract In this paper, the mutigrid agorithm

More information

On-Chip CNN Accelerator for Image Super-Resolution

On-Chip CNN Accelerator for Image Super-Resolution On-Chip CNN Acceerator for Image Super-Resoution Jung-Woo Chang and Suk-Ju Kang Dept. of Eectronic Engineering, Sogang University, Seou, South Korea {zwzang91, sjkang}@sogang.ac.kr ABSTRACT To impement

More information

Parallel Numerical Algorithms

Parallel Numerical Algorithms Parallel Numerical Algorithms Chapter 5 Vector and Matrix Products Prof. Michael T. Heath Department of Computer Science University of Illinois at Urbana-Champaign CS 554 / CSE 512 Michael T. Heath Parallel

More information

MULTIGRID REDUCTION IN TIME FOR NONLINEAR PARABOLIC PROBLEMS: A CASE STUDY

MULTIGRID REDUCTION IN TIME FOR NONLINEAR PARABOLIC PROBLEMS: A CASE STUDY MULTIGRID REDUCTION IN TIME FOR NONLINEAR PARABOLIC PROBLEMS: A CASE STUDY R.D. FALGOUT, T.A. MANTEUFFEL, B. O NEILL, AND J.B. SCHRODER Abstract. The need for paraeism in the time dimension is being driven

More information

Searching, Sorting & Analysis

Searching, Sorting & Analysis Searching, Sorting & Anaysis Unit 2 Chapter 8 CS 2308 Fa 2018 Ji Seaman 1 Definitions of Search and Sort Search: find a given item in an array, return the index of the item, or -1 if not found. Sort: rearrange

More information

A Memory Grouping Method for Sharing Memory BIST Logic

A Memory Grouping Method for Sharing Memory BIST Logic A Memory Grouping Method for Sharing Memory BIST Logic Masahide Miyazai, Tomoazu Yoneda, and Hideo Fuiwara Graduate Schoo of Information Science, Nara Institute of Science and Technoogy (NAIST), 8916-5

More information

A Fast-Convergence Decoding Method and Memory-Efficient VLSI Decoder Architecture for Irregular LDPC Codes in the IEEE 802.

A Fast-Convergence Decoding Method and Memory-Efficient VLSI Decoder Architecture for Irregular LDPC Codes in the IEEE 802. A Fast-Convergence Decoding Method and Memory-Efficient VLSI Decoder Architecture for Irreguar LDPC Codes in the IEEE 82.16e Standards Yeong-Luh Ueng and Chung-Chao Cheng Dept. of Eectrica Engineering,

More information

Space-Time Trade-offs.

Space-Time Trade-offs. Space-Time Trade-offs. Chethan Kamath 03.07.2017 1 Motivation An important question in the study of computation is how to best use the registers in a CPU. In most cases, the amount of registers avaiabe

More information

Lecture outline Graphics and Interaction Scan Converting Polygons and Lines. Inside or outside a polygon? Scan conversion.

Lecture outline Graphics and Interaction Scan Converting Polygons and Lines. Inside or outside a polygon? Scan conversion. Lecture outine 433-324 Graphics and Interaction Scan Converting Poygons and Lines Department of Computer Science and Software Engineering The Introduction Scan conversion Scan-ine agorithm Edge coherence

More information

Nearest Neighbor Learning

Nearest Neighbor Learning Nearest Neighbor Learning Cassify based on oca simiarity Ranges from simpe nearest neighbor to case-based and anaogica reasoning Use oca information near the current query instance to decide the cassification

More information

Load Balancing by MPLS in Differentiated Services Networks

Load Balancing by MPLS in Differentiated Services Networks Load Baancing by MPLS in Differentiated Services Networks Riikka Susitaiva, Jorma Virtamo, and Samui Aato Networking Laboratory, Hesinki University of Technoogy P.O.Box 3000, FIN-02015 HUT, Finand {riikka.susitaiva,

More information

An Exponential Time 2-Approximation Algorithm for Bandwidth

An Exponential Time 2-Approximation Algorithm for Bandwidth An Exponentia Time 2-Approximation Agorithm for Bandwidth Martin Fürer 1, Serge Gaspers 2, Shiva Prasad Kasiviswanathan 3 1 Computer Science and Engineering, Pennsyvania State University, furer@cse.psu.edu

More information

Priority Queueing for Packets with Two Characteristics

Priority Queueing for Packets with Two Characteristics 1 Priority Queueing for Packets with Two Characteristics Pave Chuprikov, Sergey I. Nikoenko, Aex Davydow, Kiri Kogan Abstract Modern network eements are increasingy required to dea with heterogeneous traffic.

More information

Chapter Multidimensional Direct Search Method

Chapter Multidimensional Direct Search Method Chapter 09.03 Mutidimensiona Direct Search Method After reading this chapter, you shoud be abe to:. Understand the fundamentas of the mutidimensiona direct search methods. Understand how the coordinate

More information

University of Illinois at Urbana-Champaign, Urbana, IL 61801, /11/$ IEEE 162

University of Illinois at Urbana-Champaign, Urbana, IL 61801, /11/$ IEEE 162 oward Efficient Spatia Variation Decomposition via Sparse Regression Wangyang Zhang, Karthik Baakrishnan, Xin Li, Duane Boning and Rob Rutenbar 3 Carnegie Meon University, Pittsburgh, PA 53, wangyan@ece.cmu.edu,

More information

Special Edition Using Microsoft Excel Selecting and Naming Cells and Ranges

Special Edition Using Microsoft Excel Selecting and Naming Cells and Ranges Specia Edition Using Microsoft Exce 2000 - Lesson 3 - Seecting and Naming Ces and.. Page 1 of 8 [Figures are not incuded in this sampe chapter] Specia Edition Using Microsoft Exce 2000-3 - Seecting and

More information

CSE120 Principles of Operating Systems. Prof Yuanyuan (YY) Zhou Scheduling

CSE120 Principles of Operating Systems. Prof Yuanyuan (YY) Zhou Scheduling CSE120 Principes of Operating Systems Prof Yuanyuan (YY) Zhou Scheduing Announcement Homework 2 due on October 25th Project 1 due on October 26th 2 CSE 120 Scheduing and Deadock Scheduing Overview In discussing

More information

Mobile App Recommendation: Maximize the Total App Downloads

Mobile App Recommendation: Maximize the Total App Downloads Mobie App Recommendation: Maximize the Tota App Downoads Zhuohua Chen Schoo of Economics and Management Tsinghua University chenzhh3.12@sem.tsinghua.edu.cn Yinghui (Catherine) Yang Graduate Schoo of Management

More information

CSE120 Principles of Operating Systems. Prof Yuanyuan (YY) Zhou Lecture 4: Threads

CSE120 Principles of Operating Systems. Prof Yuanyuan (YY) Zhou Lecture 4: Threads CSE120 Principes of Operating Systems Prof Yuanyuan (YY) Zhou Lecture 4: Threads Announcement Project 0 Due Project 1 out Homework 1 due on Thursday Submit it to Gradescope onine 2 Processes Reca that

More information

DETERMINING INTUITIONISTIC FUZZY DEGREE OF OVERLAPPING OF COMPUTATION AND COMMUNICATION IN PARALLEL APPLICATIONS USING GENERALIZED NETS

DETERMINING INTUITIONISTIC FUZZY DEGREE OF OVERLAPPING OF COMPUTATION AND COMMUNICATION IN PARALLEL APPLICATIONS USING GENERALIZED NETS DETERMINING INTUITIONISTIC FUZZY DEGREE OF OVERLAPPING OF COMPUTATION AND COMMUNICATION IN PARALLEL APPLICATIONS USING GENERALIZED NETS Pave Tchesmedjiev, Peter Vassiev Centre for Biomedica Engineering,

More information

Solutions to the Final Exam

Solutions to the Final Exam CS/Math 24: Intro to Discrete Math 5//2 Instructor: Dieter van Mekebeek Soutions to the Fina Exam Probem Let D be the set of a peope. From the definition of R we see that (x, y) R if and ony if x is a

More information

Distance Weighted Discrimination and Second Order Cone Programming

Distance Weighted Discrimination and Second Order Cone Programming Distance Weighted Discrimination and Second Order Cone Programming Hanwen Huang, Xiaosun Lu, Yufeng Liu, J. S. Marron, Perry Haaand Apri 3, 2012 1 Introduction This vignette demonstrates the utiity and

More information

Performance of data networks with random links

Performance of data networks with random links Performance of data networks with random inks arxiv:adap-org/9909006 v2 4 Jan 2001 Henryk Fukś and Anna T. Lawniczak Department of Mathematics and Statistics, University of Gueph, Gueph, Ontario N1G 2W1,

More information

Alpha labelings of straight simple polyominal caterpillars

Alpha labelings of straight simple polyominal caterpillars Apha abeings of straight simpe poyomina caterpiars Daibor Froncek, O Nei Kingston, Kye Vezina Department of Mathematics and Statistics University of Minnesota Duuth University Drive Duuth, MN 82-3, U.S.A.

More information

Extended Node-Arc Formulation for the K-Edge-Disjoint Hop-Constrained Network Design Problem

Extended Node-Arc Formulation for the K-Edge-Disjoint Hop-Constrained Network Design Problem Extended Node-Arc Formuation for the K-Edge-Disjoint Hop-Constrained Network Design Probem Quentin Botton Université cathoique de Louvain, Louvain Schoo of Management, (Begique) botton@poms.uc.ac.be Bernard

More information

Proceedings of the International Conference on Systolic Arrays, San Diego, California, U.S.A., May 25-27, 1988 AN EFFICIENT ASYNCHRONOUS MULTIPLIER!

Proceedings of the International Conference on Systolic Arrays, San Diego, California, U.S.A., May 25-27, 1988 AN EFFICIENT ASYNCHRONOUS MULTIPLIER! [1,2] have, in theory, revoutionized cryptography. Unfortunatey, athough offer many advantages over conventiona and authentication), such cock synchronization in this appication due to the arge operand

More information

ECEn 528 Prof. Archibald Lab: Dynamic Scheduling Part A: due Nov. 6, 2018 Part B: due Nov. 13, 2018

ECEn 528 Prof. Archibald Lab: Dynamic Scheduling Part A: due Nov. 6, 2018 Part B: due Nov. 13, 2018 ECEn 528 Prof. Archibad Lab: Dynamic Scheduing Part A: due Nov. 6, 2018 Part B: due Nov. 13, 2018 Overview This ab's purpose is to expore issues invoved in the design of out-of-order issue processors.

More information

Joint Optimization of Intra- and Inter-Autonomous System Traffic Engineering

Joint Optimization of Intra- and Inter-Autonomous System Traffic Engineering Joint Optimization of Intra- and Inter-Autonomous System Traffic Engineering Kin-Hon Ho, Michae Howarth, Ning Wang, George Pavou and Styianos Georgouas Centre for Communication Systems Research, University

More information

Genetic Algorithms for Parallel Code Optimization

Genetic Algorithms for Parallel Code Optimization Genetic Agorithms for Parae Code Optimization Ender Özcan Dept. of Computer Engineering Yeditepe University Kayışdağı, İstanbu, Turkey Emai: eozcan@cse.yeditepe.edu.tr Abstract- Determining the optimum

More information

Formulation of Loss minimization Problem Using Genetic Algorithm and Line-Flow-based Equations

Formulation of Loss minimization Problem Using Genetic Algorithm and Line-Flow-based Equations Formuation of Loss minimization Probem Using Genetic Agorithm and Line-Fow-based Equations Sharanya Jaganathan, Student Member, IEEE, Arun Sekar, Senior Member, IEEE, and Wenzhong Gao, Senior member, IEEE

More information

17.3 Surface Area of Pyramids and Cones

17.3 Surface Area of Pyramids and Cones Name Cass Date 17.3 Surface Area of Pyramids and Cones Essentia Question: How is the formua for the atera area of a reguar pyramid simiar to the formua for the atera area of a right cone? Expore G.11.C

More information

l Tree: set of nodes and directed edges l Parent: source node of directed edge l Child: terminal node of directed edge

l Tree: set of nodes and directed edges l Parent: source node of directed edge l Child: terminal node of directed edge Trees & Heaps Week 12 Gaddis: 20 Weiss: 21.1-3 CS 5301 Fa 2016 Ji Seaman 1 Tree: non-recursive definition Tree: set of nodes and directed edges - root: one node is distinguished as the root - Every node

More information

Probabilistic Graphical Models

Probabilistic Graphical Models Schoo of Computer Science Probabiistic Graphica Modes Distributed Agorithms for ML David (Wei) Dai Lecture 21, Apri 5, 2017 Eric Xing @ CMU, 2005-2017 1 Massive Data 1B+ USERS 30+ PETABYTES 32 miion pages

More information

Further Concepts in Geometry

Further Concepts in Geometry ppendix F Further oncepts in Geometry F. Exporing ongruence and Simiarity Identifying ongruent Figures Identifying Simiar Figures Reading and Using Definitions ongruent Trianges assifying Trianges Identifying

More information

Replication of Virtual Network Functions: Optimizing Link Utilization and Resource Costs

Replication of Virtual Network Functions: Optimizing Link Utilization and Resource Costs Repication of Virtua Network Functions: Optimizing Link Utiization and Resource Costs Francisco Carpio, Wogang Bziuk and Admea Jukan Technische Universität Braunschweig, Germany Emai:{f.carpio, w.bziuk,

More information

A METHOD FOR GRIDLESS ROUTING OF PRINTED CIRCUIT BOARDS. A. C. Finch, K. J. Mackenzie, G. J. Balsdon, G. Symonds

A METHOD FOR GRIDLESS ROUTING OF PRINTED CIRCUIT BOARDS. A. C. Finch, K. J. Mackenzie, G. J. Balsdon, G. Symonds A METHOD FOR GRIDLESS ROUTING OF PRINTED CIRCUIT BOARDS A C Finch K J Mackenzie G J Basdon G Symonds Raca-Redac Ltd Newtown Tewkesbury Gos Engand ABSTRACT The introduction of fine-ine technoogies to printed

More information

AN EVOLUTIONARY APPROACH TO OPTIMIZATION OF A LAYOUT CHART

AN EVOLUTIONARY APPROACH TO OPTIMIZATION OF A LAYOUT CHART 13 AN EVOLUTIONARY APPROACH TO OPTIMIZATION OF A LAYOUT CHART Eva Vona University of Ostrava, 30th dubna st. 22, Ostrava, Czech Repubic e-mai: Eva.Vona@osu.cz Abstract: This artice presents the use of

More information

file://j:\macmillancomputerpublishing\chapters\in073.html 3/22/01

file://j:\macmillancomputerpublishing\chapters\in073.html 3/22/01 Page 1 of 15 Chapter 9 Chapter 9: Deveoping the Logica Data Mode The information requirements and business rues provide the information to produce the entities, attributes, and reationships in ogica mode.

More information

Analysis and parallelization strategies for Ruge-Stüben AMG on many-core processors

Analysis and parallelization strategies for Ruge-Stüben AMG on many-core processors Anaysis and paraeization strategies for Ruge-Stüben AMG on many-core processors P. Zaspe Departement Mathematik und Informatik Preprint No. 217-6 Fachbereich Mathematik June 217 Universität Base CH-451

More information

Complex Human Activity Searching in a Video Employing Negative Space Analysis

Complex Human Activity Searching in a Video Employing Negative Space Analysis Compex Human Activity Searching in a Video Empoying Negative Space Anaysis Shah Atiqur Rahman, Siu-Yeung Cho, M.K.H. Leung 3, Schoo of Computer Engineering, Nanyang Technoogica University, Singapore 639798

More information

Self-Control Cyclic Access with Time Division - A MAC Proposal for The HFC System

Self-Control Cyclic Access with Time Division - A MAC Proposal for The HFC System Sef-Contro Cycic Access with Time Division - A MAC Proposa for The HFC System S.M. Jiang, Danny H.K. Tsang, Samue T. Chanson Hong Kong University of Science & Technoogy Cear Water Bay, Kowoon, Hong Kong

More information

A Near-Optimal Distributed QoS Constrained Routing Algorithm for Multichannel Wireless Sensor Networks

A Near-Optimal Distributed QoS Constrained Routing Algorithm for Multichannel Wireless Sensor Networks Sensors 2013, 13, 16424-16450; doi:10.3390/s131216424 Artice OPEN ACCESS sensors ISSN 1424-8220 www.mdpi.com/journa/sensors A Near-Optima Distributed QoS Constrained Routing Agorithm for Mutichanne Wireess

More information

Functions. 6.1 Modular Programming. 6.2 Defining and Calling Functions. Gaddis: 6.1-5,7-10,13,15-16 and 7.7

Functions. 6.1 Modular Programming. 6.2 Defining and Calling Functions. Gaddis: 6.1-5,7-10,13,15-16 and 7.7 Functions Unit 6 Gaddis: 6.1-5,7-10,13,15-16 and 7.7 CS 1428 Spring 2018 Ji Seaman 6.1 Moduar Programming Moduar programming: breaking a program up into smaer, manageabe components (modues) Function: a

More information

AUTOMATIC IMAGE RETARGETING USING SALIENCY BASED MESH PARAMETERIZATION

AUTOMATIC IMAGE RETARGETING USING SALIENCY BASED MESH PARAMETERIZATION S.Sai Kumar et a. / (IJCSIT Internationa Journa of Computer Science and Information Technoogies, Vo. 1 (4, 010, 73-79 AUTOMATIC IMAGE RETARGETING USING SALIENCY BASED MESH PARAMETERIZATION 1 S.Sai Kumar,

More information

Delay Budget Partitioning to Maximize Network Resource Usage Efficiency

Delay Budget Partitioning to Maximize Network Resource Usage Efficiency Deay Budget Partitioning to Maximize Network Resource Usage Efficiency Kartik Gopaan Tzi-cker Chiueh Yow-Jian Lin Forida State University Stony Brook University Tecordia Technoogies kartik@cs.fsu.edu chiueh@cs.sunysb.edu

More information

Quality of Service Evaluations of Multicast Streaming Protocols *

Quality of Service Evaluations of Multicast Streaming Protocols * Quaity of Service Evauations of Muticast Streaming Protocos Haonan Tan Derek L. Eager Mary. Vernon Hongfei Guo omputer Sciences Department University of Wisconsin-Madison, USA {haonan, vernon, guo}@cs.wisc.edu

More information

Searching & Sorting. Definitions of Search and Sort. Other forms of Linear Search. Linear Search. Week 11. Gaddis: 8, 19.6,19.8.

Searching & Sorting. Definitions of Search and Sort. Other forms of Linear Search. Linear Search. Week 11. Gaddis: 8, 19.6,19.8. Searching & Sorting Week 11 Gaddis: 8, 19.6,19.8 CS 5301 Fa 2017 Ji Seaman 1 Definitions of Search and Sort Search: find a given item in a ist, return the position of the item, or -1 if not found. Sort:

More information

Defense against Low-rate TCP-targeted Denial-of-Service Attacks

Defense against Low-rate TCP-targeted Denial-of-Service Attacks Defense against Low-rate CP-targeted Denia-of-Service Attacks Guang Yang, Mario Gera and M. Y. Sanadidi Computer Science Department, UCLA Los Angees, CA 995 {yangg, gera, medy}@cs.uca.edu e: +-3-825-888,

More information

Insert the power cord into the AC input socket of your projector, as shown in Figure 1. Connect the other end of the power cord to an AC outlet.

Insert the power cord into the AC input socket of your projector, as shown in Figure 1. Connect the other end of the power cord to an AC outlet. Getting Started This chapter wi expain the set-up and connection procedures for your projector, incuding information pertaining to basic adjustments and interfacing with periphera equipment. Powering Up

More information

DISTANCE TRANSFORMATION FOR NETWORK DESIGN PROBLEMS

DISTANCE TRANSFORMATION FOR NETWORK DESIGN PROBLEMS DISTANCE TRANSFORMATION FOR NETWORK DESIGN PROBLEMS A Ridha Mahjoub, Michae Poss, Luidi Simonetti, Eduardo Uchoa To cite this version: A Ridha Mahjoub, Michae Poss, Luidi Simonetti, Eduardo Uchoa. DISTANCE

More information

A Fast Block Matching Algorithm Based on the Winner-Update Strategy

A Fast Block Matching Algorithm Based on the Winner-Update Strategy In Proceedings of the Fourth Asian Conference on Computer Vision, Taipei, Taiwan, Jan. 000, Voume, pages 977 98 A Fast Bock Matching Agorithm Based on the Winner-Update Strategy Yong-Sheng Chenyz Yi-Ping

More information

Register Allocation. Consider the following assignment statement: x = (a*b)+((c*d)+(e*f)); In posfix notation: ab*cd*ef*++x

Register Allocation. Consider the following assignment statement: x = (a*b)+((c*d)+(e*f)); In posfix notation: ab*cd*ef*++x Register Aocation Consider the foowing assignment statement: x = (a*b)+((c*d)+(e*f)); In posfix notation: ab*cd*ef*++x Assume that two registers are avaiabe. Starting from the eft a compier woud generate

More information

Discrete elastica model for shape design of grid shells

Discrete elastica model for shape design of grid shells Abstracts for IASS Annua Symposium 017 5 8th September, 017, Hamburg, Germany Annette Böge, Manfred Grohmann (eds.) Discrete eastica mode for shape design of grid shes Yusuke SAKAI* and Makoto OHSAKI a

More information

THE PERCENTAGE OCCUPANCY HIT OR MISS TRANSFORM

THE PERCENTAGE OCCUPANCY HIT OR MISS TRANSFORM 17th European Signa Processing Conference (EUSIPCO 2009) Gasgow, Scotand, August 24-28, 2009 THE PERCENTAGE OCCUPANCY HIT OR MISS TRANSFORM P. Murray 1, S. Marsha 1, and E.Buinger 2 1 Dept. of Eectronic

More information

Distributed Hierarchical Control for Parallel Processing

Distributed Hierarchical Control for Parallel Processing Distributed Hierarchica Contro for Parae Processing Dror G. Feiteson and Larry Rudoph Hebrew University of Jerusaem T he deveopment of operating systems for parae computers has cosey foowed that for seria

More information

IBC DOCUMENT PROG007. SA/STA SERIES User's Guide V7.0

IBC DOCUMENT PROG007. SA/STA SERIES User's Guide V7.0 IBC DOCUMENT SA/STA SERIES User's Guide V7.0 Page 2 New Features for Version 7.0 Mutipe Schedues This version of the SA/STA firmware supports mutipe schedues for empoyees. The mutipe schedues are impemented

More information

Parallel Numerical Algorithms

Parallel Numerical Algorithms Parallel Numerical Algorithms Chapter 3 Dense Linear Systems Section 3.1 Vector and Matrix Products Michael T. Heath and Edgar Solomonik Department of Computer Science University of Illinois at Urbana-Champaign

More information

QoS-Aware Data Transmission and Wireless Energy Transfer: Performance Modeling and Optimization

QoS-Aware Data Transmission and Wireless Energy Transfer: Performance Modeling and Optimization QoS-Aware Data Transmission and Wireess Energy Transfer: Performance Modeing and Optimization Dusit Niyato, Ping Wang, Yeow Wai Leong, and Tan Hwee Pink Schoo of Computer Engineering, Nanyang Technoogica

More information

A HIGH PERFORMANCE, LOW LATENCY, LOW POWER AUDIO PROCESSING SYSTEM FOR WIDEBAND SPEECH OVER WIRELESS LINKS

A HIGH PERFORMANCE, LOW LATENCY, LOW POWER AUDIO PROCESSING SYSTEM FOR WIDEBAND SPEECH OVER WIRELESS LINKS A HIGH PERFORMANCE, LOW LATENCY, LOW POWER AUDIO PROCESSING SYSTEM FOR WIDEBAND SPEECH OVER WIRELESS LINKS Etienne Cornu 1, Aain Dufaux 2, and David Hermann 1 1 AMI Semiconductor Canada, 611 Kumpf Drive,

More information

Optimized Base-Station Cache Allocation for Cloud Radio Access Network with Multicast Backhaul

Optimized Base-Station Cache Allocation for Cloud Radio Access Network with Multicast Backhaul Optimized Base-Station Cache Aocation for Coud Radio Access Network with Muticast Backhau Binbin Dai, Student Member, IEEE, Ya-Feng Liu, Member, IEEE, and Wei Yu, Feow, IEEE arxiv:804.0730v [cs.it] 28

More information

Fastest-Path Computation

Fastest-Path Computation Fastest-Path Computation DONGHUI ZHANG Coege of Computer & Information Science Northeastern University Synonyms fastest route; driving direction Definition In the United states, ony 9.% of the househods

More information

ACTIVE LEARNING ON WEIGHTED GRAPHS USING ADAPTIVE AND NON-ADAPTIVE APPROACHES. Eyal En Gad, Akshay Gadde, A. Salman Avestimehr and Antonio Ortega

ACTIVE LEARNING ON WEIGHTED GRAPHS USING ADAPTIVE AND NON-ADAPTIVE APPROACHES. Eyal En Gad, Akshay Gadde, A. Salman Avestimehr and Antonio Ortega ACTIVE LEARNING ON WEIGHTED GRAPHS USING ADAPTIVE AND NON-ADAPTIVE APPROACHES Eya En Gad, Akshay Gadde, A. Saman Avestimehr and Antonio Ortega Department of Eectrica Engineering University of Southern

More information

Arithmetic Coding. Prof. Ja-Ling Wu. Department of Computer Science and Information Engineering National Taiwan University

Arithmetic Coding. Prof. Ja-Ling Wu. Department of Computer Science and Information Engineering National Taiwan University Arithmetic Coding Prof. Ja-Ling Wu Department of Computer Science and Information Engineering Nationa Taiwan University F(X) Shannon-Fano-Eias Coding W..o.g. we can take X={,,,m}. Assume p()>0 for a. The

More information

CLOUD RADIO ACCESS NETWORK WITH OPTIMIZED BASE-STATION CACHING

CLOUD RADIO ACCESS NETWORK WITH OPTIMIZED BASE-STATION CACHING CLOUD RADIO ACCESS NETWORK WITH OPTIMIZED BASE-STATION CACHING Binbin Dai and Wei Yu Ya-Feng Liu Department of Eectrica and Computer Engineering University of Toronto, Toronto ON, Canada M5S 3G4 Emais:

More information

On Upper Bounds for Assortment Optimization under the Mixture of Multinomial Logit Models

On Upper Bounds for Assortment Optimization under the Mixture of Multinomial Logit Models On Upper Bounds for Assortment Optimization under the Mixture of Mutinomia Logit Modes Sumit Kunnumka September 30, 2014 Abstract The assortment optimization probem under the mixture of mutinomia ogit

More information

CSE120 Principles of Operating Systems. Prof Yuanyuan (YY) Zhou Synchronization: Semaphore

CSE120 Principles of Operating Systems. Prof Yuanyuan (YY) Zhou Synchronization: Semaphore CSE120 Principes of Operating Systems Prof Yuanyuan (YY) Zhou Synchronization: Synchronization Needs Two synchronization needs Mutua excusion Whenever mutipe threads access a shared data, you need to worry

More information

Basic segmentation from CT: proximal femur

Basic segmentation from CT: proximal femur Chapter 7 Basic segmentation from CT: proxima femur This tutoria wi demonstrate how to import DICOM data from a CT scan of a dry cadaveric femur and appy different fiters and segmentation toos. 7.1 What

More information

MCSE Training Guide: Windows Architecture and Memory

MCSE Training Guide: Windows Architecture and Memory MCSE Training Guide: Windows 95 -- Ch 2 -- Architecture and Memory Page 1 of 13 MCSE Training Guide: Windows 95-2 - Architecture and Memory This chapter wi hep you prepare for the exam by covering the

More information

Design of IP Networks with End-to. to- End Performance Guarantees

Design of IP Networks with End-to. to- End Performance Guarantees Design of IP Networks with End-to to- End Performance Guarantees Irena Atov and Richard J. Harris* ( Swinburne University of Technoogy & *Massey University) Presentation Outine Introduction Mutiservice

More information

Timing-Driven Hierarchical Global Routing with Wire-Sizing and Buffer-Insertion for VLSI with Multi-Routing-Layer

Timing-Driven Hierarchical Global Routing with Wire-Sizing and Buffer-Insertion for VLSI with Multi-Routing-Layer Timing-Driven Hierarchica Goba Routing with Wire-Sizing and Buffer-Insertion for VLSI with Muti-Routing-Layer Takahiro DEGUCHI y Tetsushi KOIDE z Shin ichi WAKABAYASHI y yfacuty of Engineering, Hiroshima

More information

Lecture Notes for Chapter 4 Part III. Introduction to Data Mining

Lecture Notes for Chapter 4 Part III. Introduction to Data Mining Data Mining Cassification: Basic Concepts, Decision Trees, and Mode Evauation Lecture Notes for Chapter 4 Part III Introduction to Data Mining by Tan, Steinbach, Kumar Adapted by Qiang Yang (2010) Tan,Steinbach,

More information

A Two-Step Approach for Interactive Pre-Integrated Volume Rendering of Unstructured Grids

A Two-Step Approach for Interactive Pre-Integrated Volume Rendering of Unstructured Grids A Two-Step Approach for Interactive Pre-Integrated Voume Rendering of Unstructured Grids Stefan Roettger and Thomas Ert Visuaization and Interactive Systems Group University of Stuttgart Abstract In the

More information

GPU Implementation of Parallel SVM as Applied to Intrusion Detection System

GPU Implementation of Parallel SVM as Applied to Intrusion Detection System GPU Impementation of Parae SVM as Appied to Intrusion Detection System Sudarshan Hiray Research Schoar, Department of Computer Engineering, Vishwakarma Institute of Technoogy, Pune, India sdhiray7@gmai.com

More information

Topology-aware Key Management Schemes for Wireless Multicast

Topology-aware Key Management Schemes for Wireless Multicast Topoogy-aware Key Management Schemes for Wireess Muticast Yan Sun, Wade Trappe,andK.J.RayLiu Department of Eectrica and Computer Engineering, University of Maryand, Coege Park Emai: ysun, kjriu@gue.umd.edu

More information

(0,l) (0,0) (l,0) (l,l)

(0,l) (0,0) (l,0) (l,l) Parae Domain Decomposition and Load Baancing Using Space-Fiing Curves Srinivas Auru Fatih E. Sevigen Dept. of CS Schoo of EECS New Mexico State University Syracuse University Las Cruces, NM 88003-8001

More information

More Relation Model: Functional Dependencies

More Relation Model: Functional Dependencies More Reation Mode: Functiona Dependencies Lecture #7 Autumn, 2001 Fa, 2001, LRX #07 More Reation Mode: Functiona Dependencies HUST,Wuhan,China 152 Functiona Dependencies X -> A = assertion about a reation

More information

Abstract. We investigate the parallel implementation of the diagonal{implicitly iterated Runge{

Abstract. We investigate the parallel implementation of the diagonal{implicitly iterated Runge{ Diagona{Impicity Iterated Runge{Kutta Methods on Distributed Memory Mutiprocessors Thomas Rauber Gudua Runger Computer Science Department Universitat des Saarandes Postfach 151150 66041 Saarbrucken, Germany

More information

As Michi Henning and Steve Vinoski showed 1, calling a remote

As Michi Henning and Steve Vinoski showed 1, calling a remote Reducing CORBA Ca Latency by Caching and Prefetching Bernd Brügge and Christoph Vismeier Technische Universität München Method ca atency is a major probem in approaches based on object-oriented middeware

More information

An Adaptive Two-Copy Delayed SR-ARQ for Satellite Channels with Shadowing

An Adaptive Two-Copy Delayed SR-ARQ for Satellite Channels with Shadowing An Adaptive Two-Copy Deayed SR-ARQ for Sateite Channes with Shadowing Jing Zhu, Sumit Roy zhuj@ee.washington.edu Department of Eectrica Engineering, University of Washington Abstract- The paper focuses

More information

Searching & Sorting. Definitions of Search and Sort. Other forms of Linear Search. Linear Search. Week 11. Gaddis: 8, 19.6,19.8. CS 5301 Spring 2017

Searching & Sorting. Definitions of Search and Sort. Other forms of Linear Search. Linear Search. Week 11. Gaddis: 8, 19.6,19.8. CS 5301 Spring 2017 Searching & Sorting Week 11 Gaddis: 8, 19.6,19.8 CS 5301 Spring 2017 Ji Seaman 1 Definitions of Search and Sort Search: find a given item in a ist, return the position of the item, or -1 if not found.

More information

Layout Conscious Approach and Bus Architecture Synthesis for Hardware-Software Co-Design of Systems on Chip Optimized for Speed

Layout Conscious Approach and Bus Architecture Synthesis for Hardware-Software Co-Design of Systems on Chip Optimized for Speed Layout Conscious Approach and Bus Architecture Synthesis for Hardware-Software Co-Design of Systems on Chip Optimized for Speed Nattawut Thepayasuwan, Member, IEEE and Aex Doboi, Member, IEEE Abstract

More information

index.pdf March 17,

index.pdf March 17, index.pdf March 17, 2013 1 ITI 1121. Introduction to omputing II Marce Turcotte Schoo of Eectrica Engineering and omputer Science Linked List (Part 2) Tai pointer ouby inked ist ummy node Version of March

More information

Computer Networks. College of Computing. Copyleft 2003~2018

Computer Networks. College of Computing.   Copyleft 2003~2018 Computer Networks Computer Networks Prof. Lin Weiguo Coege of Computing Copyeft 2003~2018 inwei@cuc.edu.cn http://icourse.cuc.edu.cn/computernetworks/ http://tc.cuc.edu.cn Attention The materias beow are

More information

Hiding secrete data in compressed images using histogram analysis

Hiding secrete data in compressed images using histogram analysis University of Woongong Research Onine University of Woongong in Dubai - Papers University of Woongong in Dubai 2 iding secrete data in compressed images using histogram anaysis Farhad Keissarian University

More information

Automatic Hidden Web Database Classification

Automatic Hidden Web Database Classification Automatic idden Web atabase Cassification Zhiguo Gong, Jingbai Zhang, and Qian Liu Facuty of Science and Technoogy niversity of Macau Macao, PRC {fstzgg,ma46597,ma46620}@umac.mo Abstract. In this paper,

More information

Application of Intelligence Based Genetic Algorithm for Job Sequencing Problem on Parallel Mixed-Model Assembly Line

Application of Intelligence Based Genetic Algorithm for Job Sequencing Problem on Parallel Mixed-Model Assembly Line American J. of Engineering and Appied Sciences 3 (): 5-24, 200 ISSN 94-7020 200 Science Pubications Appication of Inteigence Based Genetic Agorithm for Job Sequencing Probem on Parae Mixed-Mode Assemby

More information

Response Surface Model Updating for Nonlinear Structures

Response Surface Model Updating for Nonlinear Structures Response Surface Mode Updating for Noninear Structures Gonaz Shahidi a, Shamim Pakzad b a PhD Student, Department of Civi and Environmenta Engineering, Lehigh University, ATLSS Engineering Research Center,

More information

M. Badent 1, E. Di Giacomo 2, G. Liotta 2

M. Badent 1, E. Di Giacomo 2, G. Liotta 2 DIEI Dipartimento di Ingegneria Eettronica e de informazione RT 005-06 Drawing Coored Graphs on Coored Points M. Badent 1, E. Di Giacomo 2, G. Liotta 2 1 University of Konstanz 2 Università di Perugia

More information

Approximate Volume Rendering for Curvilinear and Unstructured Grids by Hardware-Assisted Polyhedron Projection

Approximate Volume Rendering for Curvilinear and Unstructured Grids by Hardware-Assisted Polyhedron Projection Approximate Voume Rendering for Curviinear and Unstructured Grids by Hardware-Assisted Poyhedron Projection Neson Max, 1 Peter Wiiams, 1 Caudio Siva 2 1 Lawrence Livermore Nationa Laboratory 2 AT&T Labs-Research

More information

arxiv: v1 [math.qa] 31 Aug 2018

arxiv: v1 [math.qa] 31 Aug 2018 arxiv:808.0575v [math.qa] 3 Aug 208 A new approach to the SL n spider Stephen Bigeow September 3, 208 Abstract The SL n spider gives adiagrammatic way toencode therepresentation category of the quantum

More information

ECE 172 Digital Systems. Chapter 5 Uniprocessor Data Cache. Herbert G. Mayer, PSU Status 6/10/2018

ECE 172 Digital Systems. Chapter 5 Uniprocessor Data Cache. Herbert G. Mayer, PSU Status 6/10/2018 ECE 172 Digita Systems Chapter 5 Uniprocessor Data Cache Herbert G. Mayer, PSU Status 6/10/2018 1 Syabus UP Caches Cache Design Parameters Effective Time t eff Cache Performance Parameters Repacement Poicies

More information

Modeling of Problems of Projection: A Non-countercyclic Approach * Jason Ginsburg Osaka Kyoiku University

Modeling of Problems of Projection: A Non-countercyclic Approach * Jason Ginsburg Osaka Kyoiku University Modeing of Probems of Projection: A Non-countercycic Approach * Jason Ginsburg Osaka Kyoiku University Abstract This paper describes a computationa impementation of the recent Probems of Projection (POP)

More information

Non-Lecture N: Convex Hulls

Non-Lecture N: Convex Hulls N Convex Hus N.1 Definitions We are given a set P of n oints in the ane. We want to comute something caed the convex hu of P. Intuitivey, the convex hu is what you get by driving a nai into the ane at

More information

Quality Assessment using Tone Mapping Algorithm

Quality Assessment using Tone Mapping Algorithm Quaity Assessment using Tone Mapping Agorithm Nandiki.pushpa atha, Kuriti.Rajendra Prasad Research Schoar, Assistant Professor, Vignan s institute of engineering for women, Visakhapatnam, Andhra Pradesh,

More information

Sensitivity Analysis of Hopfield Neural Network in Classifying Natural RGB Color Space

Sensitivity Analysis of Hopfield Neural Network in Classifying Natural RGB Color Space Sensitivity Anaysis of Hopfied Neura Network in Cassifying Natura RGB Coor Space Department of Computer Science University of Sharjah UAE rsammouda@sharjah.ac.ae Abstract: - This paper presents a study

More information

Endoscopic Motion Compensation of High Speed Videoendoscopy

Endoscopic Motion Compensation of High Speed Videoendoscopy Endoscopic Motion Compensation of High Speed Videoendoscopy Bharath avuri Department of Computer Science and Engineering, University of South Caroina, Coumbia, SC - 901. ravuri@cse.sc.edu Abstract. High

More information

An Introduction to Design Patterns

An Introduction to Design Patterns An Introduction to Design Patterns 1 Definitions A pattern is a recurring soution to a standard probem, in a context. Christopher Aexander, a professor of architecture Why woud what a prof of architecture

More information