GPUMemSort: A High Performance Graphic Co-processors Sorting Algorithm for Large Scale In-Memory Data

Size: px
Start display at page:

Download "GPUMemSort: A High Performance Graphic Co-processors Sorting Algorithm for Large Scale In-Memory Data"

Transcription

1 GPUMemSort: A High Performace Graphic Co-processors Sortig Algorithm for Large Scale I-Memory Data Yi Ye 1, Zhihui Du 1+, David A. Bader 2 1 Natioal Laboratory for Iformatio Sciece ad Techology Departmet of Computer Sciece ad Techology, Tsighua Uiversity, Beijig, 184, Chia + Correspodig Author s [email protected] 2 College of Computig, Georgia Istitute of Techology, Atlata, GA, 3332, USA Abstract I this paper, we preset a GPU-based sortig algorithm, GPUMemSort, which achieves high performace i sortig large-scale i-memory data by exploitig highparallel GPU processors. It cosists of two algorithms: i-core algorithm, which is resposible for sortig data i GPU global memory efficietly, ad out-of-core algorithm, which is resposible for dividig large scale data ito multiple chuks that fit GPU global memory. GPUMemSort is implemeted based o NVIDIA CUDA framework ad some critical ad detailed optimizatio methods are also preseted. The tests of differet algorithms have bee ru o multiple data sets. The experimetal results show that our i-core sortig ca outperform other compariso-based algorithms ad GPUMemSort is highly effective i sortig large-scale i-memory data. Keywords: Parallel Sortig Algorithm, GPU, CUDA 1. Itroductio With the improvemet of CPU performace ad multicore CPU, badwidth betwee CPU ad memory becomes the bottleeck of large-scale computig. May hardware vedors, such as AMD, IBM, NVIDIA itegrate coprocessors to offload tasks from CPU ad this ca alleviate the effect caused by low CPU-memory badwidth. Meawhile, high performace computers ow much larger memory tha before, so it is very importat to develop efficiet co-processors algorithm to deal with large-scale i-memory data. Recetly, GPU has become the best-kow co-processor. It has bee utilized i may differet sorts of geeral purpose applicatios. GPU is suitable for highly parallel,compute itesive workloads, because of higher memory badwidth, thousads of hardware thread cotexts with This research is supported i part by Natioal Natural Sciece Foudatio of Chia((No , ad No.65339), Beijig Natural Sciece Foudatio(No ), NSF Grats CNS-7837, IIP , OCI-9446 (Bader), ad the Ceter for Adaptive Supercomputig Software for Multithreaded Architectures (CASS-MT). hudreds of parallel compute pipelies executig programs i a SIMD fashios. The peak performace of GPUs has bee icreasig at the rate of times a year, much faster tha the performace of CPUs based o Moore s law. Nowadays, several GPGPU (Geeral Purpose computig o GPUs) laguages,such as OpeCL[2] ad NVIDIA CUDA [1]are proposed for developers to use GPUs with exteded C programmig laguage, istead of graphics API. I CUDA, threads are orgaized i a hierarchy of grids, blocks, ad threads, which are executed i a SIMT (sigleistructio, multiple-thread) maer; threads are virtually mapped to a arbitrary umber of streamig multiprocessors (SMs) through warps. There exists several types of memory, such as register, local memory, shared memory, global memory, costat memory, etc. Differet type of memory ows differet characteristics. Therefore, how to orgaize the memory access hierarchy is very importat to improve programs performace. I this paper, the GPU part of our algorithm is implemeted with CUDA ad we will show how we desig ad optimize memory access patter i details. Mai Cotributio: We proposes a ovel graphics coprocessor sortig algorithm to sort large-scale i-memory data. Our idea is to split a large-scale sortig task ito a umber of disjoit oes which ca fit GPU memory. I geeral, our cotributios are as follows: (1) We provide the desig, detailed implemetatio ad tests of a graphics co-processors sortig algorithm, which ca sort large-scale i-memory data. (2) We ehace GPU Sample Sort[12] algorithm, the fastest compariso-based GPU sortig algorithm. The ehaced algorithm outperforms the others because it ca achieve better load balacig. The otatios i this paper are summarized i Table 1. The paper is orgaized as follows. Sectio 2 will itroduce the backgroud ad the related work. I sectio 3, the proposed algorithm is itroduced. Detailed implemetatio ad optimizatio will be preseted i sectio 4. Our experimetal results are show i sectio 5. I sectio 6, we will give the coclusio ad future work.

2 Table 1. NOTATIONS NOTATION DESCRIPTION N d s s[i] e[i] list[i] umber of elemets i the iput data set size of elemets which ca fit ito the global memory umber of chuks umber of sample poits the i th sample poit the i th iput elemet the i th sorted list 2. Backgroud ad Related Work 2.1. Parallel Sortig Algorithm Parallel sortig has bee studied extesively durig the past 3 years. Geerally, parallel sortig algorithms ca be divided ito two categories[3]: Partitio-based Sortig: First, use partitio keys to split the data ito disjoit buckets. Secod, sort each bucket idepedetly, the cocateate sorted buckets. Merge-based Sortig: First, partitio the iput data ito data chuks of approximately equal size ad sort these data chuks i differet processors. Secod, merge the data across all the processors. Each category has its ow potetial bottleeck. Partitiobased algorithms have to deal with problem of how to keep load balacig amog all the processors. Merge-based sortig algorithms perform well oly for a small umber of processors. To solve the load balace problem, Parallel Sortig by Regular Sample (PSRS)[5] guaratees that the size of data chuk assiged to processor is less tha ( 2 p p p + 1). 2 A ew oe [4] ca guaratee that each processor will have at most ( p + s p) elemets, where p s p ad s is 2 a parameter GPU Programmig with CUDA The NVIDIA CUDA programmig model is created for developig applicatios o GPU. Some major priciples [6] o this platform are: (1) Leverage zero-overhead thread schedulig to hide memory latecy. (2) Optimize the use of o-chip memory to reduce badwidth usage ad redudat executio. (3) Group threads to avoid SIMD pealties ad memory port/bak coflicts. (4) Threads withi a thread block ca commuicate via sychroizatio, but there is o built-i global commuicatio mechaism for all threads Parallel Sortig Algorithm based o GPU Sice most sortig algorithms are bouded by memory badwidth, sortig o the high-badwidth GPUs becomes a popular topic. Purcell[7] itroduced bitoic merge sort, while Kipfer ad Westerma [8]improved it to oddeve merge sort. Greβ ad Zachma[9] itroduced the GPUABiSort based o adaptive bitoic sortig. Naga K. Govidaraju[3] preseted a GPUTeraSort algorithm to sort billio record wide-key databases. Also, some CUDA-based sortig algorithms have bee proposed recetly. Erik Sitor [1]itroduced a hybrid algorithm combiig bucket sort ad merge sort, but ca oly sort floats as it uses a float4 i merge sort. Cederma [11]proposed Quicksort i CUDA, which is sesitive to the distributio of the iput data. The compariso-based Thrust Merge method by Nadathur Satish, etc combies odd-eve merge ad twoway merge to balace the load. Satishet.al.[13] preseted GPU radix sort for itegers. [12] is a radomized sample sort that sigificatly outperforms Thrust Merge. Because of its radom selectio, the load balacig is bad. However, most of them are desiged for small-scale data sortig ad are ieffective whe data size is larger tha the global memory size. 3. GPUMemSort Algorithm I this sectio, we will preset the two parts of GPUMemSort.The Out-of-core sortig ca divide largescale data ito multiple disjoited subsets ad assig them to GPU. The I-core sortig ca sort the subsets efficietly. 3.1 Out of core algorithm We adopt the idea of Determiistic Sample-based Parallel Sortig (DSPS) i out-of-core sortig algorithm. The idea behid DSPS algorithm is to fid s-1 samples to partitio the iput data set ito several data chuks. Elemets i (i+1)-th chuk are o less tha those i (i)-th chuk. The sizes of these chuks has a determiistic upper boud. They ca be put ito GPU global memory by adjustig parameter d i the algorithm accordig to the value of. The out-of-core algorithm ca be described as follows: Step 1: Divide the iput data set ito d chuks, each cotais d elemets. Without loss of geerality, we assume that d divides evely. Step 2: Copy the chuks to GPU global memory oe by oe, sort them by i-core algorithm. The split the chuk ito d buckets, bucket j of chuk i is called Bi[i][j]. The x th data i chuk i will be put ito bucket Bi[i][ x d ]. Copy these buckets back to mai memory. Step 3: Swap buckets amog chuks, i [, d-1], j (i, d-1], switch Bi[i][j] ad Bi[j][i]. So that ew chuk i cosists of {Bi[][i],Bi[1][i],...,Bi[d-1][i]}. Step 4: I the d-1 th chuk, i [, d-1],bi[i][d-1] selects the ((x+1) d 2 s ) th elemet as a sample cadidate. x s-1. The sample cadidates list should cotai s*d sample poits. Step 5: Sort the sample cadidate list, select each (k+1)*s sample poit, k [,d-2] as s[k], where s[d-1] is the largest. Copy the sample poits array from mai memory to GPU global memory.

3 Step 6: Copy each chuk to GPU global memory agai, split the chuk ito d buckets based o d sample poits. The bucket j of chuk i is called NS[i][j] j d-1. After splittig, all the elemets i NS[i][j] should be o larger tha s[j]. At last, copy these buckets back to mai memory. Step 7: Swap buckets amog chuks agai, ew chuk i cosists of {NS[][i],NS[1][i],...,NS[d-1][i]}. i [,d-1]. All the elemets i chuk i are o larger tha s[i]. Step 8: i [,d-1], calculate the total legth of chuk i. If the legth is less tha the threshold Θ, copy the whole chuk to GPU global memory, use icore sortig algorithm to sort it. Otherwise, we will copy NS[][i],NS[1][i],...,NS[d-1][i] to GPU oe by oe. For NS[j][i], split it ito two parts called part[j][i][] ad part[j][i][1], part[j][i][] cotais elemets equal to s[i] while part[j][i][1] cotais the rest. Copy back the part[j][i][1] to the mai memory, the merge all the part[j][i][1], j d-1 ito oe array. At last, sort this array by GPU ad write it back to the result set, fill out the rest part of result set usig s[i]. I step 8, the threshold Θ is the maximum size of array that ca be sorted o GPU. Accordig to coductio i [5], we ca easily get that: d θ, so d Θ 2 2s + 2 ( s Θ 4 +) d + s. This meas that if every chuk s size is guarateed to be less tha Θ, the umber of chuks splitted i step1 must be larger tha Θ 2 2s + 2 Θ ( s 4 +). Suppose that GPU is able to sort 128MB data set oce, the sample umber s=64, the N = 1 GB, accordig to the coduct above, d 8.47, so that d must be larger tha or equal to I core algorithm I-core algorithm is based o GPU Sample Sort, which is curretly the fastest compariso-based sortig algorithm. However, it ecouters load balacig problem. The key to make subsets well-balaced i sample sortig algorithm is to fid appropriate splitters, such as PSRS(Parallel Sortig by Regular Sample) ad DSPS. However, if they are directly ported to GPU, overhead of geeratig splitters will be eve much larger tha that of sortig imbalaced subsets. So it is importat to fid the tradeoff poit betwee them. Let s review the procedure of PSRS. Supposed that the size of data set is. First, split the data set ito p subsets. The, for each subset, select s-1 equidistat poits as sample cadidate poits. Fially, merge the (s-1)*p sample cadidate poits, sort them ad select s-1 equidistat poits as splitters. Overhead brought by splitters geeratio i PSRS is splittig the whole data set ad sortig all the subsets ad it is proportioal to the data size. I-core sortig algorithm uses a iovative strategy to select sample poits. First pick up a set from whole data set radomly. The size of set equals to (s-1)*k*m (k p), M is the maximum size of array that ca be sorted i share memory of oe SM. The, split the set ito k subsets ad assig k blocks to sort these subsets i parallel. Afterward, for each subset, select s-1 equidistat poits as sample cadidate poits. At last, merge the (s-1)*k samples, sort them ad select s-1 equidistat poits as splitters. The parameter k should be assiged at rutime depedig o data size. 4. Detailed Implemetatio ad Optimizatio Here we preset the detailed implemetatio ad optimizatio of GPUMemSort. First describe the task executio egie, which ca overlap data trasfer with GPU computatio based o pipelie. Secod idicate how to swap buckets betwee chuks. At last show the compesatio algorithm based o optimistic mechaism. 4.1 Task Executio Egie based o Pipelie The Data trasfer betwee CPU ad GPU is a sigificat overhead i GPUMemSort algorithm. Without optimizatio, more tha 3% of the time would be spet o data trasfer operatios betwee CPU ad GPU. O oe had, GPU ca ot be fully used because it will remai idle whe data trasfer operatios are performed. O the other had, sice the badwidth betwee CPU ad GPU is fully-duplex, oly 5% badwidth resource ca be used at most simultaeously. So Overlappig data trasfer from CPU to GPU, GPU computatio ad data trasfer from GPU to CPU will brig remarkable performace improvemet. Thus a task executio egie is implemeted based o pipelie mechaism. First, divide a sortig task ito three subtasks: CPU-GPU data trasfer, kerel sortig ad GPU- CPU data trasfer. The,pipelie these three type of subtasks based o streamig with asychroous memory copy techology proposed by CUDA.Streamig maitais the depedecy, while asychroous memory copy parallelizes data trasfer operatios ad sortig operatio. Fig. 1 shows the compariso betwee GPU classic computatio patter ad pipelie-based oe. 4.2 Implemetatio of Buckets Swap I the implemetatio of DSPS, differet buckets are swaped through etwork commuicatio because differet chuks are scattered i distributed memory. Poiters are used to avoid hard memory copy. I Algorithm 1, we preset the data structure of poiter array to swap buckets, ad the fuctio of data trasfer from mai memory to GPU global memory. Assig each data chuks TrasposeChuk structure, icludig a vector of TrasposeBlock to record the start address ad the size of a bucket. The swap the start address ad size i the correspodig TrasposeBlock structures. I the comig data trasfer, traverse the buckets ad copy them from mai memory to GPU global memory, thus hard memory copy is avoided.

4 Algorithm 1 Data Structure for buckets swap ad comig data trasfer algorithm Struct TrasposeBlock{ it* block ptr; log size; }; Struct TrasposeChuk{ TrasposeBlock blocks[d]; }; procedure memcpyf romhostt odevice(t raspose Chuk&chuk, it dvalue) offset ; for q = to d do TrasposeBlock& tmpblock chuk.blocks[q]; cudamemcpyhosttodevice (dvalue + offset,tmpblock.block ptr,sizeof(it) * tmpblock.size); offset offset + tmpblock.size; ed for Figure 1. Compariso betwee GPU classic computatio patter ad pipelie based computatio patter 4.3 Compesatio Algorithm based o Optimistic Machaism I the step 4 of DSPS, Est[i] is calculated to record the size of elemets equals to s[i] i sample cadidate list. I the comig splittig operatio for chuks, it should be guarateed that i NS[i], the umber of elemets equal to s[i] is p 2 s smaller tha Est[i]. If ot, we should try to shift this elemet to the adjacet buckets whe splittig. To add the compariso logic above ito chuks splittig module, a global variable should be maitaied for each bucket to keep record of the umber of elemets equal to correspodig splitter. Atomic FAA (Fetch Ad Add) method will be called a few times to keep cosistecy, thus deterioratig the performace. Otherwise, the size of chuks i the last step may exceed the threshold θ. I order to solve this problem, we propose a iovative compesatio algorithm based o optimistic mechaism. Assume that iϵ[, d-1], the umber of elemets i NS[i] equal to s[i] is o less tha Est[i] p 2 s is a small probability evet ad a compesatio logic is added i step 8. First, judge whether the size of each chuk is o larger tha the give θ. If yes, copy this chuk to GPU global memory ad sort it with i-core algorithm. Otherwise, copy NS[][i],NS[1][i],...,NS[d-1][i] to GPU oe by oe. For NS[j][i], split it ito two parts: part[j][i][] ad part[j][i][1], the former cotais elemets equal to s[i] while the latter cotais the rest. Copy back the part[j][i][1] to the mai memory, the merge all the part[j][i][1], j d-1 ito oe array, the sort this array employig GPU ad write it back to the result set. At last, fill the rest part of result set with s[i]. I Algorithm 2, preseted is the pseudo code of compesatio algorithm. 5. Experimetal Results I this sectio, we itroduce our hardware eviromet ad compare our i-core sortig with GPU, GPU ad Thrust Merge Sort based o six differet data sets ad show the performace ad scalability of GPUMemSort. 5.1 Hardware Eviromet Our system cosists of two GPU 26GTX co-processors, 16GB DDR3 mai memory ad a Itel Quad Core i5-75 CPU. Each GPU coects the mai memory through exclusive PCIe 16X data bus, providig 4GB/s badwidth with full duplex. Experimets have show that data trasmissios betwee each GPU ad mai memory will ot be affected too much. Also, time cosumed by data trasmissio betwee GPU ad mai memory ca be almost overlapped by GPU computatio. Table 1 shows the badwidth measuremet results i differet scearios. Table 2. GPU Memory badwidth measuremet results Test Cases Sigle GPU Two GPUs Device to Host 338.5MB/s MB/s Host to Device MB/s 282.1MB/s Device to Device MB/s MB/s GTX 26 with CUDA cosists of 16 SMs(Streamig Multiprocessor), each has 8 processors executig the same istructio o differet data. I CUDA, each SM supports up to 768 threads, ows 16KB of share memory ad has

5 Bucket ArraySize(MB) Gaussia ArraySize(MB) Algorithm 2 Compesatio algorithm i CPU Side # chuk: [iput] TrasposeChuk of chuk which will be processed, # splitter: [iput] the correspodig splitter value # outputblock: [output] the poiter of array i which results will be writte back # splittersize: [output] the umber of elemets which equals to splitter i the chuk procedure hadlelogarrayexceptio(cost TrasposeChuk& chuk, cost it splitter, it* & outputblock, it& splittersize) it boudary[d]; // splitter of each bucket struct TrasposeChuk m chuk; alloc memory whose size equal to d i dboudary ad copy boudary to dboudary; for q = to d do // hadle blocks i chuk oe by oe. it* dbucketvalue = NULL; it* dbucketoutputvalue = NULL; cost TrasposeBlock& tmpblock = chuk.blocks[q]; alloc memory whose size equal to tmpblock.size i dbucketvalue ad copy tmpblock.block ptr to device memory; malloc tmpblock.size legth array to dbucketoutput- Value; splitequality kerel <<<BLOCK NUM,THREADS NUM>>> (dbucketvalue, tmpblock.size, splitter, dboudary); boudary[q] ΣdBoudary[i]; iϵ[, BLOCK N U M); prefixsum(dboudary); divide kerel <<<BLOCK NUM,THREADS NUM>>> (dbucketvalue, tmpblock.size, splitter, dboudary); copy dbucketoutputvalue back to outputblock i mai memory; ed for copy all buckets i m chuk to global memory; employ icore sortig algorithm to sort them; copy sorted buckets back to outputblock; pad the rest of outputblock usig splitter; free memory i device ad mai memory; Sorted ArraySize(MB) Uiform ArraySize(MB) (MS) Staggered ArraySize(MB) Zero ArraySize(MB) Figure 2. Performace compariso betwee i core sort ad other existig sort algorithms 8192 available registers. Threads are logically divided ito blocks assiged to a specific SM. Depedig o how may registers ad how may local memory the block of threads requires, there could be multiple blocks assiged to a SM. GPU Data is stored i a 512MB global memory. Each block ca use share memory as cache. Hardware ca coalesce several read or write operatios ito a big oe, so it is ecessary to keep threads visitig memory cosecutively. 5.2 Performace Evaluatio I this sectio, we first compare the performace of icore sort, GPU, GPU ad Thrust Merge Sort based o differet data sets of usiged itegers. Six differet types of data sets icludig Uiform,Sorted Zero,Bucket,Gaussia,Staggered [11]. Fig. 2 shows the result o data of differet array sizes: i-core sortig outperforms the others because it ca achieve good load balacig with small cost. The performace evaluatio of out-of-core algorithm o sigle GPU is show i Fig. 3, idicatig that out-of-core algorithm is robust ad is capable of hadlig data efficietly with differet distributios ad sizes. At last, the compariso of the out-of-core algorithm performace betwee sigle GPU ad two GPUs is show i Fig.4. It is clear that out-of-core sortig algorithm ca reach ear-liear speedup i two GPUs, showig that our out-ofcore algorithm has good scalability whe the badwidth betwee mai memory ad GPU memory is ot a bottleeck.

6 Bucket Gaussia Sorted Stagger Uiform Zero 256MB 512MB 124MB 248MB ArraySize Figure 3. Performace for differet data distributios of out of core algorithm MB 512MB 124MB 248MB ArraySize Sigle GPU Two GPU Figure 4. Performace compariso of out ofcore betwee sigle GPU ad two GPUs 6. Coclusio ad Future Work I this paper, we preset GPUMemSort: a high performace graphics co-processor sortig framework for largescale i-memory data by exploitig high-parallel GPU processors. We test the performace of the algorithm based o multiple data sets ad it shows that GPUMemSort sortig algorithm outperforms other multi-core based parallel sortig algorithms. I the future, we will try to exted our algorithm to multiple GPUs augmeted cluster, implemet ad optimize this algorithm i distributed heterogeeous architecture. I additio, we will try to ehace i-core sortig algorithm to help GPUMemSort reach eve higher performace. A sigificat coclusio draw from this work, is that our GPUMemSort ca break through the limitatio of GPU global memory ad ca sort large-scale i-memory data efficietly. Refereces [1] NVIDIA CUDA (Compute Uified Device Architecture) [2] OPENCL, [3] N. Govidaraju, J. Gray, R. Kumar ad D. Maocha. GPUTeraSort: high performace graphics coprocessor sortig for large database maagemet. SIGMOD, 26 [4] D. R. Helma, J. JaJa, D. A. Bader. A New Determiistic Parallel Sortig Algorithm with a Experimetal Evaluatio. ACM Joural of Experimetal Algorithmics (JEA), September 1998, Volume 3. [5] H. Shi ad J. Schaeffer, Parallel Sortig by Regular samplig, Joural of Parallel ad Distributed Computig 14, pp , 1992 [6] Shae Ryoo, Christopher I. Rodrigues, Sara S. Baghsorkhi, Sam S. Stoe, David B. Kirk, We-mei W. Hwu: Optimizatio priciples ad applicatio performace evaluatio of a multithreaded GPU usig CUDA. PPOPP 28: [7] T.Purcell,C.Doer,M.Cammarao,H.Jese,ad P.Haraha.Photo mappig o programmable graphics hardware. ACMSIGGRAPH/Eurographics Coferece o Graphics Hardware,pages 41C5,23 [8] P.Kipfer,M.Segal,ad R.Westerma.Uberflow:A gpu-based particle egie. SIGGRAPH/Euro graphics Workshop o Graphics Hardware,24. [9] Gre,A., ad Zachma, 26,GPU-ABiSort: Optimal Parallel Sortig o Stream Architectures, The 2th IEEE Iteratioal Parallel ad Distributed Processig Symposium, Rhodes Islad, Greece, pp [1] Sitor, E., ad Assarsso, U. 27, Fast Parallel GPU-Sortig Usig a Hybrid Algorithm, Joural of Parallel ad Distributed Computig, pp [11] Cederma, D., ad Tsigas, P., 28, A Practical Quicksort Algorithm for Graphics Processors, Techical Report, Gotheburg, Swede, pp [12] N.Leischer,V.Osipov,ad P.Saders.GPU sample sort. I IEEE Iteratioal Parallel ad Distributed Processig Symposium,21 (curretly available at [13] N.Satish,M.Harris,adM.Garlad.Desigig efficiet sortig algorithms for may-core GPUs.I IEEE Iteratioal Parallel ad Distributed Processig Symposium,29. [14] Felix Putze, Peter Saders, Johaes Sigler. The Multi-Core Stadard Template Library (Exteded Poster Abstract). Symposium o Priciples ad Practice of Parallel Programmig (PPoPP) 27

Domain 1: Designing a SQL Server Instance and a Database Solution

Domain 1: Designing a SQL Server Instance and a Database Solution Maual SQL Server 2008 Desig, Optimize ad Maitai (70-450) 1-800-418-6789 Domai 1: Desigig a SQL Server Istace ad a Database Solutio Desigig for CPU, Memory ad Storage Capacity Requiremets Whe desigig a

More information

Recovery time guaranteed heuristic routing for improving computation complexity in survivable WDM networks

Recovery time guaranteed heuristic routing for improving computation complexity in survivable WDM networks Computer Commuicatios 30 (2007) 1331 1336 wwwelseviercom/locate/comcom Recovery time guarateed heuristic routig for improvig computatio complexity i survivable WDM etworks Lei Guo * College of Iformatio

More information

ODBC. Getting Started With Sage Timberline Office ODBC

ODBC. Getting Started With Sage Timberline Office ODBC ODBC Gettig Started With Sage Timberlie Office ODBC NOTICE This documet ad the Sage Timberlie Office software may be used oly i accordace with the accompayig Sage Timberlie Office Ed User Licese Agreemet.

More information

Modified Line Search Method for Global Optimization

Modified Line Search Method for Global Optimization Modified Lie Search Method for Global Optimizatio Cria Grosa ad Ajith Abraham Ceter of Excellece for Quatifiable Quality of Service Norwegia Uiversity of Sciece ad Techology Trodheim, Norway {cria, ajith}@q2s.tu.o

More information

Evaluating Model for B2C E- commerce Enterprise Development Based on DEA

Evaluating Model for B2C E- commerce Enterprise Development Based on DEA , pp.180-184 http://dx.doi.org/10.14257/astl.2014.53.39 Evaluatig Model for B2C E- commerce Eterprise Developmet Based o DEA Weli Geg, Jig Ta Computer ad iformatio egieerig Istitute, Harbi Uiversity of

More information

A Combined Continuous/Binary Genetic Algorithm for Microstrip Antenna Design

A Combined Continuous/Binary Genetic Algorithm for Microstrip Antenna Design A Combied Cotiuous/Biary Geetic Algorithm for Microstrip Atea Desig Rady L. Haupt The Pesylvaia State Uiversity Applied Research Laboratory P. O. Box 30 State College, PA 16804-0030 [email protected] Abstract:

More information

(VCP-310) 1-800-418-6789

(VCP-310) 1-800-418-6789 Maual VMware Lesso 1: Uderstadig the VMware Product Lie I this lesso, you will first lear what virtualizatio is. Next, you ll explore the products offered by VMware that provide virtualizatio services.

More information

Analyzing Longitudinal Data from Complex Surveys Using SUDAAN

Analyzing Longitudinal Data from Complex Surveys Using SUDAAN Aalyzig Logitudial Data from Complex Surveys Usig SUDAAN Darryl Creel Statistics ad Epidemiology, RTI Iteratioal, 312 Trotter Farm Drive, Rockville, MD, 20850 Abstract SUDAAN: Software for the Statistical

More information

Department of Computer Science, University of Otago

Department of Computer Science, University of Otago Departmet of Computer Sciece, Uiversity of Otago Techical Report OUCS-2006-09 Permutatios Cotaiig May Patters Authors: M.H. Albert Departmet of Computer Sciece, Uiversity of Otago Micah Colema, Rya Fly

More information

Reliability Analysis in HPC clusters

Reliability Analysis in HPC clusters Reliability Aalysis i HPC clusters Narasimha Raju, Gottumukkala, Yuda Liu, Chokchai Box Leagsuksu 1, Raja Nassar, Stephe Scott 2 College of Egieerig & Sciece, Louisiaa ech Uiversity Oak Ridge Natioal Lab

More information

C.Yaashuwanth Department of Electrical and Electronics Engineering, Anna University Chennai, Chennai 600 025, India..

C.Yaashuwanth Department of Electrical and Electronics Engineering, Anna University Chennai, Chennai 600 025, India.. (IJCSIS) Iteratioal Joural of Computer Sciece ad Iformatio Security, A New Schedulig Algorithms for Real Time Tasks C.Yaashuwath Departmet of Electrical ad Electroics Egieerig, Aa Uiversity Cheai, Cheai

More information

In nite Sequences. Dr. Philippe B. Laval Kennesaw State University. October 9, 2008

In nite Sequences. Dr. Philippe B. Laval Kennesaw State University. October 9, 2008 I ite Sequeces Dr. Philippe B. Laval Keesaw State Uiversity October 9, 2008 Abstract This had out is a itroductio to i ite sequeces. mai de itios ad presets some elemetary results. It gives the I ite Sequeces

More information

LECTURE 13: Cross-validation

LECTURE 13: Cross-validation LECTURE 3: Cross-validatio Resampli methods Cross Validatio Bootstrap Bias ad variace estimatio with the Bootstrap Three-way data partitioi Itroductio to Patter Aalysis Ricardo Gutierrez-Osua Texas A&M

More information

CS100: Introduction to Computer Science

CS100: Introduction to Computer Science Review: History of Computers CS100: Itroductio to Computer Sciece Maiframes Miicomputers Lecture 2: Data Storage -- Bits, their storage ad mai memory Persoal Computers & Workstatios Review: The Role of

More information

Domain 1 - Describe Cisco VoIP Implementations

Domain 1 - Describe Cisco VoIP Implementations Maual ONT (642-8) 1-800-418-6789 Domai 1 - Describe Cisco VoIP Implemetatios Advatages of VoIP Over Traditioal Switches Voice over IP etworks have may advatages over traditioal circuit switched voice etworks.

More information

Effective Techniques for Message Reduction and Load Balancing in Distributed Graph Computation

Effective Techniques for Message Reduction and Load Balancing in Distributed Graph Computation Effective Techiques for Message Reductio ad Load Balacig i Distributed Graph Computatio ABSTRACT Da Ya, James Cheg, Yi Lu Dept. of Computer Sciece ad Egieerig The Chiese Uiversity of Hog Kog {yada, jcheg,

More information

DAME - Microsoft Excel add-in for solving multicriteria decision problems with scenarios Radomir Perzina 1, Jaroslav Ramik 2

DAME - Microsoft Excel add-in for solving multicriteria decision problems with scenarios Radomir Perzina 1, Jaroslav Ramik 2 Itroductio DAME - Microsoft Excel add-i for solvig multicriteria decisio problems with scearios Radomir Perzia, Jaroslav Ramik 2 Abstract. The mai goal of every ecoomic aget is to make a good decisio,

More information

Effective Techniques for Message Reduction and Load Balancing in Distributed Graph Computation

Effective Techniques for Message Reduction and Load Balancing in Distributed Graph Computation Effective Techiques for Message Reductio ad Load Balacig i Distributed Graph Computatio ABSTRACT Da Ya, James Cheg, Yi Lu Dept. of Computer Sciece ad Egieerig The Chiese Uiversity of Hog Kog {yada, jcheg,

More information

Quantitative Computer Architecture

Quantitative Computer Architecture Performace Measuremet ad Aalysis i Computer Quatitative Computer Measuremet Model Iovatio Proposed How to measure, aalyze, ad specify computer system performace or My computer is faster tha your computer!

More information

Study on the application of the software phase-locked loop in tracking and filtering of pulse signal

Study on the application of the software phase-locked loop in tracking and filtering of pulse signal Advaced Sciece ad Techology Letters, pp.31-35 http://dx.doi.org/10.14257/astl.2014.78.06 Study o the applicatio of the software phase-locked loop i trackig ad filterig of pulse sigal Sog Wei Xia 1 (College

More information

Running Time ( 3.1) Analysis of Algorithms. Experimental Studies ( 3.1.1) Limitations of Experiments. Pseudocode ( 3.1.2) Theoretical Analysis

Running Time ( 3.1) Analysis of Algorithms. Experimental Studies ( 3.1.1) Limitations of Experiments. Pseudocode ( 3.1.2) Theoretical Analysis Ruig Time ( 3.) Aalysis of Algorithms Iput Algorithm Output A algorithm is a step-by-step procedure for solvig a problem i a fiite amout of time. Most algorithms trasform iput objects ito output objects.

More information

SYSTEM INFO. MDK - Multifunctional Digital Communications System. Efficient Solutions for Information and Safety

SYSTEM INFO. MDK - Multifunctional Digital Communications System. Efficient Solutions for Information and Safety Commuicatios Systems for Itercom, PA, Emergecy Call ad Telecommuicatios MDK - Multifuctioal Digital Commuicatios System SYSTEM INFO ms NEUMANN ELEKTRONIK GmbH Efficiet Solutios for Iformatio ad Safety

More information

Multi-server Optimal Bandwidth Monitoring for QoS based Multimedia Delivery Anup Basu, Irene Cheng and Yinzhe Yu

Multi-server Optimal Bandwidth Monitoring for QoS based Multimedia Delivery Anup Basu, Irene Cheng and Yinzhe Yu Multi-server Optimal Badwidth Moitorig for QoS based Multimedia Delivery Aup Basu, Iree Cheg ad Yizhe Yu Departmet of Computig Sciece U. of Alberta Architecture Applicatio Layer Request receptio -coectio

More information

Chapter 6: Variance, the law of large numbers and the Monte-Carlo method

Chapter 6: Variance, the law of large numbers and the Monte-Carlo method Chapter 6: Variace, the law of large umbers ad the Mote-Carlo method Expected value, variace, ad Chebyshev iequality. If X is a radom variable recall that the expected value of X, E[X] is the average value

More information

CS100: Introduction to Computer Science

CS100: Introduction to Computer Science I-class Exercise: CS100: Itroductio to Computer Sciece What is a flip-flop? What are the properties of flip-flops? Draw a simple flip-flop circuit? Lecture 3: Data Storage -- Mass storage & represetig

More information

Center, Spread, and Shape in Inference: Claims, Caveats, and Insights

Center, Spread, and Shape in Inference: Claims, Caveats, and Insights Ceter, Spread, ad Shape i Iferece: Claims, Caveats, ad Isights Dr. Nacy Pfeig (Uiversity of Pittsburgh) AMATYC November 2008 Prelimiary Activities 1. I would like to produce a iterval estimate for the

More information

Capacity of Wireless Networks with Heterogeneous Traffic

Capacity of Wireless Networks with Heterogeneous Traffic Capacity of Wireless Networks with Heterogeeous Traffic Migyue Ji, Zheg Wag, Hamid R. Sadjadpour, J.J. Garcia-Lua-Aceves Departmet of Electrical Egieerig ad Computer Egieerig Uiversity of Califoria, Sata

More information

A Distributed Dynamic Load Balancer for Iterative Applications

A Distributed Dynamic Load Balancer for Iterative Applications A Distributed Dyamic Balacer for Iterative Applicatios Harshitha Meo, Laxmikat Kalé Departmet of Computer Sciece, Uiversity of Illiois at Urbaa-Champaig {gplkrsh2,kale}@illiois.edu ABSTRACT For may applicatios,

More information

1 Computing the Standard Deviation of Sample Means

1 Computing the Standard Deviation of Sample Means Computig the Stadard Deviatio of Sample Meas Quality cotrol charts are based o sample meas ot o idividual values withi a sample. A sample is a group of items, which are cosidered all together for our aalysis.

More information

Report Documentation Page

Report Documentation Page Applyig performace models to uderstad data-itesive computig efficiecy Elie Krevat, Tomer Shira, Eric Aderso, Joseph Tucek, Jay J. Wylie, Gregory R. Gager Caregie Mello Uiversity HP Labs CMU-PDL-10-108

More information

Data Center Ethernet Facilitation of Enterprise Clustering. David Flynn, Linux Networx Orlando, Florida March 16, 2004

Data Center Ethernet Facilitation of Enterprise Clustering. David Flynn, Linux Networx Orlando, Florida March 16, 2004 Data Ceter Etheret Facilitatio of Eterprise Clusterig David Fly, Liux Networx Orlado, Florida March 16, 2004 1 2 Liux Networx builds COTS based clusters 3 Clusters Offer Improved Performace Scalability

More information

Incremental calculation of weighted mean and variance

Incremental calculation of weighted mean and variance Icremetal calculatio of weighted mea ad variace Toy Fich [email protected] [email protected] Uiversity of Cambridge Computig Service February 009 Abstract I these otes I eplai how to derive formulae for umerically

More information

A model of Virtual Resource Scheduling in Cloud Computing and Its

A model of Virtual Resource Scheduling in Cloud Computing and Its A model of Virtual Resource Schedulig i Cloud Computig ad Its Solutio usig EDAs 1 Jiafeg Zhao, 2 Wehua Zeg, 3 Miu Liu, 4 Guagmig Li 1, First Author, 3 Cogitive Sciece Departmet, Xiame Uiversity, Xiame,

More information

Chair for Network Architectures and Services Institute of Informatics TU München Prof. Carle. Network Security. Chapter 2 Basics

Chair for Network Architectures and Services Institute of Informatics TU München Prof. Carle. Network Security. Chapter 2 Basics Chair for Network Architectures ad Services Istitute of Iformatics TU Müche Prof. Carle Network Security Chapter 2 Basics 2.4 Radom Number Geeratio for Cryptographic Protocols Motivatio It is crucial to

More information

5 Boolean Decision Trees (February 11)

5 Boolean Decision Trees (February 11) 5 Boolea Decisio Trees (February 11) 5.1 Graph Coectivity Suppose we are give a udirected graph G, represeted as a boolea adjacecy matrix = (a ij ), where a ij = 1 if ad oly if vertices i ad j are coected

More information

INVESTMENT PERFORMANCE COUNCIL (IPC) Guidance Statement on Calculation Methodology

INVESTMENT PERFORMANCE COUNCIL (IPC) Guidance Statement on Calculation Methodology Adoptio Date: 4 March 2004 Effective Date: 1 Jue 2004 Retroactive Applicatio: No Public Commet Period: Aug Nov 2002 INVESTMENT PERFORMANCE COUNCIL (IPC) Preface Guidace Statemet o Calculatio Methodology

More information

QUADRO tech. FSA Migrator 2.6. File Server Migrations - Made Easy

QUADRO tech. FSA Migrator 2.6. File Server Migrations - Made Easy QUADRO tech FSA Migrator 2.6 File Server Migratios - Made Easy FSA Migrator Cosolidate your archived ad o-archived File Server data - with ease! May orgaisatios struggle with the cotiuous growth of their

More information

PUBLIC RELATIONS PROJECT 2016

PUBLIC RELATIONS PROJECT 2016 PUBLIC RELATIONS PROJECT 2016 The purpose of the Public Relatios Project is to provide a opportuity for the chapter members to demostrate the kowledge ad skills eeded i plaig, orgaizig, implemetig ad evaluatig

More information

Conceptualization with Incremental Bron- Kerbosch Algorithm in Big Data Architecture

Conceptualization with Incremental Bron- Kerbosch Algorithm in Big Data Architecture Acta Polytechica Hugarica Vol. 13, No. 2, 2016 Coceptualizatio with Icremetal Bro- Kerbosch Algorithm i Big Data Architecture László Kovács 1, Gábor Szabó 2 1 Uiversity of Miskolc, Istitute of Iformatio

More information

Domain 1 Components of the Cisco Unified Communications Architecture

Domain 1 Components of the Cisco Unified Communications Architecture Maual CCNA Domai 1 Compoets of the Cisco Uified Commuicatios Architecture Uified Commuicatios (UC) Eviromet Cisco has itroduced what they call the Uified Commuicatios Eviromet which is used to separate

More information

Making State Explicit for Imperative Big Data Processing

Making State Explicit for Imperative Big Data Processing Makig State Explicit for Imperative Big Data Processig Raul Castro Feradez, Imperial College Lodo; Matteo Migliavacca, Uiversity of Ket; Evagelia Kalyviaaki, City Uiversity Lodo; Peter Pietzuch, Imperial

More information

A probabilistic proof of a binomial identity

A probabilistic proof of a binomial identity A probabilistic proof of a biomial idetity Joatho Peterso Abstract We give a elemetary probabilistic proof of a biomial idetity. The proof is obtaied by computig the probability of a certai evet i two

More information

ADAPTIVE NETWORKS SAFETY CONTROL ON FUZZY LOGIC

ADAPTIVE NETWORKS SAFETY CONTROL ON FUZZY LOGIC 8 th Iteratioal Coferece o DEVELOPMENT AND APPLICATION SYSTEMS S u c e a v a, R o m a i a, M a y 25 27, 2 6 ADAPTIVE NETWORKS SAFETY CONTROL ON FUZZY LOGIC Vadim MUKHIN 1, Elea PAVLENKO 2 Natioal Techical

More information

STUDENTS PARTICIPATION IN ONLINE LEARNING IN BUSINESS COURSES AT UNIVERSITAS TERBUKA, INDONESIA. Maya Maria, Universitas Terbuka, Indonesia

STUDENTS PARTICIPATION IN ONLINE LEARNING IN BUSINESS COURSES AT UNIVERSITAS TERBUKA, INDONESIA. Maya Maria, Universitas Terbuka, Indonesia STUDENTS PARTICIPATION IN ONLINE LEARNING IN BUSINESS COURSES AT UNIVERSITAS TERBUKA, INDONESIA Maya Maria, Uiversitas Terbuka, Idoesia Co-author: Amiuddi Zuhairi, Uiversitas Terbuka, Idoesia Kuria Edah

More information

Automatic Tuning for FOREX Trading System Using Fuzzy Time Series

Automatic Tuning for FOREX Trading System Using Fuzzy Time Series utomatic Tuig for FOREX Tradig System Usig Fuzzy Time Series Kraimo Maeesilp ad Pitihate Soorasa bstract Efficiecy of the automatic currecy tradig system is time depedet due to usig fixed parameters which

More information

On the Capacity of Hybrid Wireless Networks

On the Capacity of Hybrid Wireless Networks O the Capacity of Hybrid ireless Networks Beyua Liu,ZheLiu +,DoTowsley Departmet of Computer Sciece Uiversity of Massachusetts Amherst, MA 0002 + IBM T.J. atso Research Ceter P.O. Box 704 Yorktow Heights,

More information

Floating Codes for Joint Information Storage in Write Asymmetric Memories

Floating Codes for Joint Information Storage in Write Asymmetric Memories Floatig Codes for Joit Iformatio Storage i Write Asymmetric Memories Axiao (Adrew Jiag Computer Sciece Departmet Texas A&M Uiversity College Statio, TX 77843-311 [email protected] Vaske Bohossia Electrical

More information

PROCEEDINGS OF THE YEREVAN STATE UNIVERSITY AN ALTERNATIVE MODEL FOR BONUS-MALUS SYSTEM

PROCEEDINGS OF THE YEREVAN STATE UNIVERSITY AN ALTERNATIVE MODEL FOR BONUS-MALUS SYSTEM PROCEEDINGS OF THE YEREVAN STATE UNIVERSITY Physical ad Mathematical Scieces 2015, 1, p. 15 19 M a t h e m a t i c s AN ALTERNATIVE MODEL FOR BONUS-MALUS SYSTEM A. G. GULYAN Chair of Actuarial Mathematics

More information

Optimal Adaptive Bandwidth Monitoring for QoS Based Retrieval

Optimal Adaptive Bandwidth Monitoring for QoS Based Retrieval 1 Optimal Adaptive Badwidth Moitorig for QoS Based Retrieval Yizhe Yu, Iree Cheg ad Aup Basu (Seior Member) Departmet of Computig Sciece Uiversity of Alberta Edmoto, AB, T6G E8, CAADA {yizhe, aup, li}@cs.ualberta.ca

More information

A Secure Implementation of Java Inner Classes

A Secure Implementation of Java Inner Classes A Secure Implemetatio of Java Ier Classes By Aasua Bhowmik ad William Pugh Departmet of Computer Sciece Uiversity of Marylad More ifo at: http://www.cs.umd.edu/~pugh/java Motivatio ad Overview Preset implemetatio

More information

Case Study. Normal and t Distributions. Density Plot. Normal Distributions

Case Study. Normal and t Distributions. Density Plot. Normal Distributions Case Study Normal ad t Distributios Bret Halo ad Bret Larget Departmet of Statistics Uiversity of Wiscosi Madiso October 11 13, 2011 Case Study Body temperature varies withi idividuals over time (it ca

More information

A Balanced Scorecard

A Balanced Scorecard A Balaced Scorecard with VISION A Visio Iteratioal White Paper Visio Iteratioal A/S Aarhusgade 88, DK-2100 Copehage, Demark Phoe +45 35430086 Fax +45 35434646 www.balaced-scorecard.com 1 1. Itroductio

More information

Your organization has a Class B IP address of 166.144.0.0 Before you implement subnetting, the Network ID and Host ID are divided as follows:

Your organization has a Class B IP address of 166.144.0.0 Before you implement subnetting, the Network ID and Host ID are divided as follows: Subettig Subettig is used to subdivide a sigle class of etwork i to multiple smaller etworks. Example: Your orgaizatio has a Class B IP address of 166.144.0.0 Before you implemet subettig, the Network

More information

Virtual Machine Scheduling Management on Cloud Computing Using Artificial Bee Colony

Virtual Machine Scheduling Management on Cloud Computing Using Artificial Bee Colony , March 12-14, 2014, Hog Kog Virtual Machie Schedulig Maagemet o Cloud Computig Usig Artificial Bee Coloy B. Kruekaew ad W. Kimpa Abstract Resource schedulig maagemet desig o Cloud computig is a importat

More information

CREATIVE MARKETING PROJECT 2016

CREATIVE MARKETING PROJECT 2016 CREATIVE MARKETING PROJECT 2016 The Creative Marketig Project is a chapter project that develops i chapter members a aalytical ad creative approach to the marketig process, actively egages chapter members

More information

Systems Design Project: Indoor Location of Wireless Devices

Systems Design Project: Indoor Location of Wireless Devices Systems Desig Project: Idoor Locatio of Wireless Devices Prepared By: Bria Murphy Seior Systems Sciece ad Egieerig Washigto Uiversity i St. Louis Phoe: (805) 698-5295 Email: [email protected] Supervised

More information

A Modified Key Partitioning for BigData Using MapReduce in Hadoop

A Modified Key Partitioning for BigData Using MapReduce in Hadoop Joural of Computer Sciece Origial Research Paper A Modified Key Partitioig for BigData Usig MapReduce i Hadoop Gothai Ekambaram ad Balasubramaie Palaisamy Departmet of CSE, Kogu Egieerig College, Erode-638052,

More information

Clustering Algorithm Analysis of Web Users with Dissimilarity and SOM Neural Networks

Clustering Algorithm Analysis of Web Users with Dissimilarity and SOM Neural Networks JONAL OF SOFTWARE, VOL. 7, NO., NOVEMBER 533 Clusterig Algorithm Aalysis of Web Users with Dissimilarity ad SOM Neal Networks Xiao Qiag School of Ecoomics ad maagemet, Lazhou Jiaotog Uiversity, Lazhou;

More information

iprox sensors iprox inductive sensors iprox programming tools ProxView programming software iprox the world s most versatile proximity sensor

iprox sensors iprox inductive sensors iprox programming tools ProxView programming software iprox the world s most versatile proximity sensor iprox sesors iprox iductive sesors iprox programmig tools ProxView programmig software iprox the world s most versatile proximity sesor The world s most versatile proximity sesor Eato s iproxe is syoymous

More information

Chatpun Khamyat Department of Industrial Engineering, Kasetsart University, Bangkok, Thailand [email protected]

Chatpun Khamyat Department of Industrial Engineering, Kasetsart University, Bangkok, Thailand ocpky@hotmail.com SOLVING THE OIL DELIVERY TRUCKS ROUTING PROBLEM WITH MODIFY MULTI-TRAVELING SALESMAN PROBLEM APPROACH CASE STUDY: THE SME'S OIL LOGISTIC COMPANY IN BANGKOK THAILAND Chatpu Khamyat Departmet of Idustrial

More information

Convention Paper 6764

Convention Paper 6764 Audio Egieerig Society Covetio Paper 6764 Preseted at the 10th Covetio 006 May 0 3 Paris, Frace This covetio paper has bee reproduced from the author's advace mauscript, without editig, correctios, or

More information

MapReduce Based Implementation of Aggregate Functions on Cassandra

MapReduce Based Implementation of Aggregate Functions on Cassandra Iteratioal Joural of Electroics Commuicatio ad Couter Techology (IJECCT) MapReduce Based Ilemetatio of Aggregate Fuctios o Cassadra Aseh Daesh Arasteh Departmet of Couter ad IT Islamic Azad Uiversity of

More information

Enhancing Oracle Business Intelligence with cubus EV How users of Oracle BI on Essbase cubes can benefit from cubus outperform EV Analytics (cubus EV)

Enhancing Oracle Business Intelligence with cubus EV How users of Oracle BI on Essbase cubes can benefit from cubus outperform EV Analytics (cubus EV) Ehacig Oracle Busiess Itelligece with cubus EV How users of Oracle BI o Essbase cubes ca beefit from cubus outperform EV Aalytics (cubus EV) CONTENT 01 cubus EV as a ehacemet to Oracle BI o Essbase 02

More information

Optimize your Network. In the Courier, Express and Parcel market ADDING CREDIBILITY

Optimize your Network. In the Courier, Express and Parcel market ADDING CREDIBILITY Optimize your Network I the Courier, Express ad Parcel market ADDING CREDIBILITY Meetig today s challeges ad tomorrow s demads Aswers to your key etwork challeges ORTEC kows the highly competitive Courier,

More information

Research Method (I) --Knowledge on Sampling (Simple Random Sampling)

Research Method (I) --Knowledge on Sampling (Simple Random Sampling) Research Method (I) --Kowledge o Samplig (Simple Radom Samplig) 1. Itroductio to samplig 1.1 Defiitio of samplig Samplig ca be defied as selectig part of the elemets i a populatio. It results i the fact

More information

Lesson 15 ANOVA (analysis of variance)

Lesson 15 ANOVA (analysis of variance) Outlie Variability -betwee group variability -withi group variability -total variability -F-ratio Computatio -sums of squares (betwee/withi/total -degrees of freedom (betwee/withi/total -mea square (betwee/withi

More information

Vladimir N. Burkov, Dmitri A. Novikov MODELS AND METHODS OF MULTIPROJECTS MANAGEMENT

Vladimir N. Burkov, Dmitri A. Novikov MODELS AND METHODS OF MULTIPROJECTS MANAGEMENT Keywords: project maagemet, resource allocatio, etwork plaig Vladimir N Burkov, Dmitri A Novikov MODELS AND METHODS OF MULTIPROJECTS MANAGEMENT The paper deals with the problems of resource allocatio betwee

More information

Major Coefficients Recovery: a Compressed Data Gathering Scheme for Wireless Sensor Network

Major Coefficients Recovery: a Compressed Data Gathering Scheme for Wireless Sensor Network This full text paper was peer reviewed at the directio of IEEE Commuicatios Society subject matter experts for publicatio i the IEEE Globecom proceedigs. Major Coefficiets Recovery: a Compressed Data Gatherig

More information

Entropy of bi-capacities

Entropy of bi-capacities Etropy of bi-capacities Iva Kojadiovic LINA CNRS FRE 2729 Site école polytechique de l uiv. de Nates Rue Christia Pauc 44306 Nates, Frace [email protected] Jea-Luc Marichal Applied Mathematics

More information

Non-life insurance mathematics. Nils F. Haavardsson, University of Oslo and DNB Skadeforsikring

Non-life insurance mathematics. Nils F. Haavardsson, University of Oslo and DNB Skadeforsikring No-life isurace mathematics Nils F. Haavardsso, Uiversity of Oslo ad DNB Skadeforsikrig Mai issues so far Why does isurace work? How is risk premium defied ad why is it importat? How ca claim frequecy

More information

E-Plex Enterprise Access Control System

E-Plex Enterprise Access Control System Eterprise Access Cotrol System Egieered for Flexibility Modular Solutio The Eterprise Access Cotrol System is a modular solutio for maagig access poits. Employig a variety of hardware optios, system maagemet

More information

Overview of some probability distributions.

Overview of some probability distributions. Lecture Overview of some probability distributios. I this lecture we will review several commo distributios that will be used ofte throughtout the class. Each distributio is usually described by its probability

More information

Authentication - Access Control Default Security Active Directory Trusted Authentication Guest User or Anonymous (un-authenticated) Logging Out

Authentication - Access Control Default Security Active Directory Trusted Authentication Guest User or Anonymous (un-authenticated) Logging Out FME Server Security Table of Cotets FME Server Autheticatio - Access Cotrol Default Security Active Directory Trusted Autheticatio Guest User or Aoymous (u-autheticated) Loggig Out Authorizatio - Roles

More information

Patentability of Computer Software and Business Methods

Patentability of Computer Software and Business Methods WIPO-MOST Itermediate Traiig Course o Practical Itellectual Property Issues i Busiess November 10 to 14, 2003 Patetability of Computer Software ad Busiess Methods Tomoko Miyamoto Patet Law Sectio Patet

More information

.04. This means $1000 is multiplied by 1.02 five times, once for each of the remaining sixmonth

.04. This means $1000 is multiplied by 1.02 five times, once for each of the remaining sixmonth Questio 1: What is a ordiary auity? Let s look at a ordiary auity that is certai ad simple. By this, we mea a auity over a fixed term whose paymet period matches the iterest coversio period. Additioally,

More information

Firewall Modules and Modular Firewalls

Firewall Modules and Modular Firewalls Firewall Modules ad Modular Firewalls H. B. Acharya Uiversity of Texas at Austi [email protected] Aditya Joshi Uiversity of Texas at Austi [email protected] M. G. Gouda Natioal Sciece Foudatio

More information

An Efficient Polynomial Approximation of the Normal Distribution Function & Its Inverse Function

An Efficient Polynomial Approximation of the Normal Distribution Function & Its Inverse Function A Efficiet Polyomial Approximatio of the Normal Distributio Fuctio & Its Iverse Fuctio Wisto A. Richards, 1 Robi Atoie, * 1 Asho Sahai, ad 3 M. Raghuadh Acharya 1 Departmet of Mathematics & Computer Sciece;

More information

Soving Recurrence Relations

Soving Recurrence Relations Sovig Recurrece Relatios Part 1. Homogeeous liear 2d degree relatios with costat coefficiets. Cosider the recurrece relatio ( ) T () + at ( 1) + bt ( 2) = 0 This is called a homogeeous liear 2d degree

More information

The Power of Both Choices: Practical Load Balancing for Distributed Stream Processing Engines

The Power of Both Choices: Practical Load Balancing for Distributed Stream Processing Engines The Power of Both Choices: Practical Load Balacig for Distributed Stream Processig Egies Muhammad Ais Uddi Nasir #1, Giamarco De Fracisci Morales 2, David García-Soriao 3 Nicolas Kourtellis 4, Marco Serafii

More information

Annuities Under Random Rates of Interest II By Abraham Zaks. Technion I.I.T. Haifa ISRAEL and Haifa University Haifa ISRAEL.

Annuities Under Random Rates of Interest II By Abraham Zaks. Technion I.I.T. Haifa ISRAEL and Haifa University Haifa ISRAEL. Auities Uder Radom Rates of Iterest II By Abraham Zas Techio I.I.T. Haifa ISRAEL ad Haifa Uiversity Haifa ISRAEL Departmet of Mathematics, Techio - Israel Istitute of Techology, 3000, Haifa, Israel I memory

More information

Development of 3D-CFD code for Heat Conduction Process using CUDA. Dissertation

Development of 3D-CFD code for Heat Conduction Process using CUDA. Dissertation Developmet of 3D-CFD code for Heat Coductio Process usig CUDA Dissertatio Submitted i partial fulfilmet of Evaluatio of Master of Techology Computer Egieerig By Yogesh Bhadke MIS No: 121222007 Uder the

More information

Plug-in martingales for testing exchangeability on-line

Plug-in martingales for testing exchangeability on-line Plug-i martigales for testig exchageability o-lie Valetia Fedorova, Alex Gammerma, Ilia Nouretdiov, ad Vladimir Vovk Computer Learig Research Cetre Royal Holloway, Uiversity of Lodo, UK {valetia,ilia,alex,vovk}@cs.rhul.ac.uk

More information

Determining the sample size

Determining the sample size Determiig the sample size Oe of the most commo questios ay statisticia gets asked is How large a sample size do I eed? Researchers are ofte surprised to fid out that the aswer depeds o a umber of factors

More information

Optimization of Large Data in Cloud computing using Replication Methods

Optimization of Large Data in Cloud computing using Replication Methods Optimizatio of Large Data i Cloud computig usig Replicatio Methods Vijaya -Kumar-C, Dr. G.A. Ramachadhra Computer Sciece ad Techology, Sri Krishadevaraya Uiversity Aatapuramu, AdhraPradesh, Idia Abstract-Cloud

More information

The Stable Marriage Problem

The Stable Marriage Problem The Stable Marriage Problem William Hut Lae Departmet of Computer Sciece ad Electrical Egieerig, West Virgiia Uiversity, Morgatow, WV [email protected] 1 Itroductio Imagie you are a matchmaker,

More information