L A T E X

An Introduction to L A T E X Thesis Template of Tsinghua University Dissertation Submitted to Tsinghua University in partial fulfillment of the requirement for the degree of Doctor of Engineering by Xue Ruini ( Computer Science and Technology ) Dissertation Supervisor : Associate Supervisor : Professor Zheng Weimin Chen Wenguang February, 2009

1 2 3

1 2 ThuThesis 5 TEX L A TEX CJK I

Abstract Abstract An abstract of a dissertation is a summary and extraction of research work and contributions. Included in an abstract should be description of research topic and research objective, brief introduction to methodology and research process, and summarization of conclusion and contributions of the research. An abstract should be characterized by independence and clarity and carry identical information with the dissertation. It should be such that the general idea and major contributions of the dissertation are conveyed without reading the dissertation. An abstract should be concise and to the point. It is a misunderstanding to make an abstract an outline of the dissertation and words the first chapter, the second chapter and the like should be avoided in the abstract. Key words are terms used in a dissertation for indexing, reflecting core information of the dissertation. An abstract may contain a maximum of 5 key words, with semicolons used in between to separate one another. Key words: TEX; L A TEX; CJK; template; thesis II

1 English...1 1.1...1 1.2...1 1.3...3 1.3.1...3 1.3.2...3 1.3.3...7 1.4...8 1.5... 12 1.6... 12 1.7... 13 2... 14 2.1... 14 2.1.1... 14 2.1.2... 14... 17... 19 A... 20 A.1 Single-Objective Programming... 20 A.1.1 Linear Programming... 21 A.1.2 Nonlinear Programming... 22 A.1.3 Integer Programming... 23 B... 25 B.1... 25 B.1.1... 25 B.1.2... 26 B.1.3... 26 C... 27 III

... 28 IV

HPC cluster Itanium SMP API PI MPI PBI MPBI PY PMDA-BDA G χ E m c P T v (High Performance Computing) N- N- (Activation Free Energy) (Transmission Coefficient) V

VI

1 English 1 English ThuThesis ThuThesis 1 1.1 cover.tex denation.tex appendix01.tex resume.tex 2 1.2 1037-1101 1 2 768-824 1

1 English 2

1 English 1.3 1.3.1 booktabs array longtabular \hlinewd booktabs \toprule \midrule \bottomrule longtable hlinewd{xpt} 1.1 chapter thuthesis.ins thuthesis.dtx thuthesis.cls thuthesis.cfg thubib.bst thutils.sty L A TEX docstrip 1 2 cls cfg Bibtex 1 2 1.1 L A TEX \footnote 1.3.2 1.2 3

1 English 1.2 1 y First Half Second Half x 1st Qtr 2nd Qtr 3rd Qtr 4th Qtr East 20.4 27.4 90 20.4 West 30.6 38.6 34.6 31.6 ThuThesis * ** 1.2 1 tabularx X 2 \backslashbox 1 2 1.3 1.4 1.5 2.1.2.2 1.3 1.4 111 222 222 333 111 222 222 333 4

1 English 1.5 (a) 111 222 222 333 (b) 111 222 222 333 L A TEX 1.6 1.6 2 Network Topology # of nodes # of clients Server Waxman GT-ITM 600 Transit-Stub 2% 10% 50% Max. Connectivity Inet-2.1 6000 Rui Ni Xue ThuThesis ABCDEF 5

1 English TEX/L A TEX Word L A TEX longtable supertabular longtable 1.7 longtable 1.7 (s) (s) (s) (s) (s) KB CG.A.2 23.05 0.002 0.116 0.035 0.589 32491 CG.A.4 15.06 0.003 0.067 0.021 0.351 18211 CG.A.8 13.38 0.004 0.072 0.023 0.210 9890 CG.B.2 867.45 0.002 0.864 0.232 3.256 228562 CG.B.4 501.61 0.003 0.438 0.136 2.075 123862 CG.B.8 384.65 0.004 0.457 0.108 1.235 63777 6

1 English 1.7 (s) (s) (s) (s) (s) KB MG.A.2 112.27 0.002 0.846 0.237 3.930 236473 MG.A.4 59.84 0.003 0.442 0.128 2.070 123875 MG.A.8 31.38 0.003 0.476 0.114 1.041 60627 MG.B.2 526.28 0.002 0.821 0.238 4.176 236635 MG.B.4 280.11 0.003 0.432 0.130 1.706 123793 MG.B.8 148.29 0.003 0.442 0.116 0.893 60600 LU.A.2 2116.54 0.002 0.110 0.030 0.532 28754 LU.A.4 1102.50 0.002 0.069 0.017 0.255 14915 LU.A.8 574.47 0.003 0.067 0.016 0.192 8655 LU.B.2 9712.87 0.002 0.357 0.104 1.734 101975 LU.B.4 4757.80 0.003 0.190 0.056 0.808 53522 LU.B.8 2444.05 0.004 0.222 0.057 0.548 30134 EP.A.2 123.81 0.002 0.010 0.003 0.074 1834 EP.A.4 61.92 0.003 0.011 0.004 0.073 1743 EP.A.8 31.06 0.004 0.017 0.005 0.073 1661 EP.B.2 495.49 0.001 0.009 0.003 0.196 2011 EP.B.4 247.69 0.002 0.012 0.004 0.122 1663 EP.B.8 126.74 0.003 0.017 0.005 0.083 1656 1.3.3 \caption*{} XX XX L A TEX \caption* 1.111 1.111 7

1 English 1.111 ThuThesis Figure 1.111 773-819 1 2 1 2 1.4 1.1 c = a 2 b 2 (1-1) = (a + b)(a b) (1-2) 1.1 8

1 English 1.1 1.1 φ(x, z) = z γ 10 x γ mn x m z n = z Mr 1 x Mr (m+n) x m z n ζ 0 = (ξ 0 ) 2, (1-3) ζ 1 = ξ 0 ξ 1, (1-4) ζ 2 = (ξ 1 ) 2, (1-5) 1.1 x y + 1 (mod m 2 ) (1-6) x y + 1 mod m 2 (1-7) x y + 1 (m 2 ) (1-8) 9

1 English 1.1 b { b } [ f (x) 2 g(y) 2 + f (y) 2 g(x) 2 ] 2 f (x)g(x) f (y)g(y) dx dy a a b b b b } = {g(y) 2 f 2 + f (y) 2 g 2 2 f (y)g(y) f g dy a a a a 1.1 y = 1 y = 0 (1-9a) (1-9b) 10

1 English 1.1 V i = v i q i v j, X i = x i q i x j, U i = u i, for i j; (1-10) V j = v j, X j = x j, U j u j + q i u i. (1-11) i j 1.1 f (x ) p λ j g j (x ) = 0 j=1 λ j g j (x ) = 0, j = 1, 2,, p λ j 0, j = 1, 2,, p. (1-12) 1.1 Andrew S. Tanenbaum W. Richard Stevens 1.1 Poincare Conjecture If in a closed three-dimensional space, any closed curves can shrink to a point continuously, this space can be deformed to a sphere. 1.1 1.1 label ref 11

1 English 1.5 bibitem BIBTEX thubib.bst [1 3] [4 6] [7 9] [10,11] [12,13] [14] [15,16] [17] [18] lang= zh bst [19] bbl [12] 1.6 (1-13) p(y x) p(x) p(x) p(y x) = p(x, y) p(x) = p(x y)p(y) p(x) (1-13) TEX happy amsmath det K(t = 1, t 1,..., t n ) = I n ( 1) I i I t i (D j + λ j t j ) det A (λ) (I I) = 0. (1-14) j I amsmath b a { b a } [ f (x) 2 g(y) 2 + f (y) 2 g(x) 2 ] 2 f (x)g(x) f (y)g(y) dx dy b b b b } = {g(y) 2 f 2 + f (y) 2 g 2 2 f (y)g(y) f g dy a a a a 12

1 English max F(x, y x 1, y 2,, y m) subject to: G(x) 0 (y 1, y 2,, y m) solves problems (i = 1, 2,, m) max f i (x, y 1, y 2,, y m ) y i subject to: g i (x, y 1, y 2,, y m ) 0. (1-15) 1.7 \pozhehao (1) (2) paralist \itemsep paralist 13

2 2 2.1 1 (1-13) p(y x) = p(x, y) p(x) = p(x y)p(y) p(x) (2-1) 2.1.1 pstricks pgf pgf Postscript L A TEX GUI XFig(jFig), WinFig, Tpx, Ipe, Dia, Inkscape, LaTeXPiX, jpicedt, jaxdraw 2.1.2 L A TEX 2ε subfig 2.1.2.1 1 L A TEX float [H] 2.1 Hello, Xfig! LittleLeo 2.1 Xfig 1 This is not a bug, but a feature of L A TEX! 14

2 (a) (b) caption 2.2 2.1.2.2 minipage parbox 2.2 2.2(a) 2.2(b) \subfloat \subfigure \subtable 15

2 2.3 2.4 16

[1] Knuth D E. The TEX Book. 15th ed., Reading, MA: Addison-Wesley Publishing Company, 1989. [2] Goosens M, Mittelbach F, Samarin A. The L A TEX Companion. Reading, MA: Addison- Wesley Publishing Company, 1994: 112 125. [3] Gröning P, Nilsson L, Ruffieux P, et al. Encyclopedia of Nanoscience and Nanotechnology, volume 1. American Scientific Publishers, 2004: 547 579. [4] Krasnogor N. Towards robust memetic algorithms. In: Hart W, Krasnogor N, Smith J, (eds.). Proceedings of Recent Advances in Memetic Algorithms. New York: Springer Berlin Heidelberg, 2004: 185 207. [5].., 2001: 185 207. [6].. :,,,.. :, 1998: 65 69. [7] Chafik El Idrissi M, Roney A, Frigon C, et al. Measurements of total kinetic-energy released to the N = 2 dissociation limit of H 2 evidence of the dissociation of very high vibrational Rydberg states of H 2 by doubly-excited states. Chemical Physics Letters, 1994, 224(10):260 266. [8] Mellinger A, Vidal C R, Jungen C. Laser reduced fluorescence study of the carbon-monoxide nd triplet Rydberg series-experimental results and multichannel quantum-defect analysis. J. Chem. Phys., 1996, 104(5):8913 8921. [9] Shell M. How to Use the IEEEtran L A TEX Class. Journal of L A TEX Class Files, 2002, 12(4):100 120. [10]. [ ]. :, 2005. [11] Jeyakumar A R. Metamori: A library for Incremental File Checkpointing[M]. Blacksburg: Virgina Tech, June 21, 2004. [12]. [ ]. :, 2005. [13] Zadok E. FiST: A System for Stackable File System Code Generation[D]. USA: Computer Science Department, Columbia University, May, 2001. [14] IEEE Std 1363-2000. IEEE Standard Specifications for Public-Key Cryptography. New York: IEEE, 2000. [15] Kim S, Woo N, Yeom H Y, et al. Design and Implementation of Dynamic Process Management for Grid-enabled MPICH. Proceedings of the 10th European PVM/MPI Users Group Conference, Venice, Italy, 2003. 17

[16] Kocher C, Jaffe J, Jun B. Differential Power Analysis. In: Wiener M, (eds.). Proceedings of Advances in Cryptology (CRYPTO 99), volume 1666 of Lecture Notes in Computer Science. Springer-Verlag, 1999. 388 397. [17] Woo A, Bailey D, Yarrow M, et al. The NAS Parallel Benchmarks 2.0. Technical report, The Pennsylvania State University CiteSeer Archives, December 05, 1995. http://www.nasa.org/. [18],,,.., 1800, 224:260 266. [19],,,.. N,,, 2006. 18

xxx xxx xxx xx xx ThuThesis 19

A A As one of the most widely used techniques in operations research, mathematical programming is defined as a means of maximizing a quantity known as objective function, subject to a set of constraints represented by equations and inequalities. Some known subtopics of mathematical programming are linear programming, nonlinear programming, multiobjective programming, goal programming, dynamic programming, and multilevel programming [1]. It is impossible to cover in a single chapter every concept of mathematical programming. This chapter introduces only the basic concepts and techniques of mathematical programming such that readers gain an understanding of them throughout the book [2,3]. A.1 Single-Objective Programming The general form of single-objective programming (SOP) is written as follows, max f (x) subject to: (123) g j (x) 0, j = 1, 2,, p which maximizes a real-valued function f of x = (x 1, x 2,, x n ) subject to a set of constraints. Definition A.1 In SOP, we call x a decision vector, and x 1, x 2,, x n decision variables. The function f is called the objective function. The set S = { x R n g j (x) 0, j = 1, 2,, p } (456) is called the feasible set. An element x in S is called a feasible solution. 20

A Definition A.2 A feasible solution x is called the optimal solution of SOP if and only if f (x ) f (x) (A-1) for any feasible solution x. One of the outstanding contributions to mathematical programming was known as the Kuhn-Tucker conditionsa-2. In order to introduce them, let us give some definitions. An inequality constraint g j (x) 0 is said to be active at a point x if g j (x ) = 0. A point x satisfying g j (x ) 0 is said to be regular if the gradient vectors g j (x) of all active constraints are linearly independent. Let x be a regular point of the constraints of SOP and assume that all the functions f (x) and g j (x), j = 1, 2,, p are differentiable. If x is a local optimal solution, then there exist Lagrange multipliers λ j, j = 1, 2,, p such that the following Kuhn- Tucker conditions hold, f (x ) p λ j g j (x ) = 0 j=1 λ j g j (x ) = 0, j = 1, 2,, p (A-2) λ j 0, j = 1, 2,, p. If all the functions f (x) and g j (x), j = 1, 2,, p are convex and differentiable, and the point x satisfies the Kuhn-Tucker conditions (A-2), then it has been proved that the point x is a global optimal solution of SOP. A.1.1 Linear Programming If the functions f (x), g j (x), j = 1, 2,, p are all linear, then SOP is called a linear programming. The feasible set of linear is always convex. A point x is called an extreme point of convex set S if x S and x cannot be expressed as a convex combination of two points in S. It has been shown that the optimal solution to linear programming corresponds to an extreme point of its feasible set provided that the feasible set S is bounded. This 21

A fact is the basis of the simplex algorithm which was developed by Dantzig as a very efficient method for solving linear programming. Table 1 of tables This is an example for manually numbered table, which would not appear in the list Network Topology # of nodes # of clients Server Waxman GT-ITM 600 Transit-Stub 2% 10% 50% Max. Connectivity Inet-2.1 6000 Rui Ni Xue ThuThesis ABCDEF Roughly speaking, the simplex algorithm examines only the extreme points of the feasible set, rather than all feasible points. At first, the simplex algorithm selects an extreme point as the initial point. The successive extreme point is selected so as to improve the objective function value. The procedure is repeated until no improvement in objective function value can be made. The last extreme point is the optimal solution. A.1.2 Nonlinear Programming If at least one of the functions f (x), g j (x), j = 1, 2,, p is nonlinear, then SOP is called a nonlinear programming. A large number of classical optimization methods have been developed to treat special-structural nonlinear programming based on the mathematical theory concerned with analyzing the structure of problems. Figure 1 list of figures This is an example for manually numbered figure, which would not appear in the Now we consider a nonlinear programming which is confronted solely with maximizing a real-valued function with domain R n. Whether derivatives are available or 22

A not, the usual strategy is first to select a point in R n which is thought to be the most likely place where the maximum exists. If there is no information available on which to base such a selection, a point is chosen at random. From this first point an attempt is made to construct a sequence of points, each of which yields an improved objective function value over its predecessor. The next point to be added to the sequence is chosen by analyzing the behavior of the function at the previous points. This construction continues until some termination criterion is met. Methods based upon this strategy are called ascent methods, which can be classified as direct methods, gradient methods, and Hessian methods according to the information about the behavior of objective function f. Direct methods require only that the function can be evaluated at each point. Gradient methods require the evaluation of first derivatives of f. Hessian methods require the evaluation of second derivatives. In fact, there is no superior method for all problems. The efficiency of a method is very much dependent upon the objective function. A.1.3 Integer Programming Integer programming is a special mathematical programming in which all of the variables are assumed to be only integer values. When there are not only integer variables but also conventional continuous variables, we call it mixed integer programming. If all the variables are assumed either 0 or 1, then the problem is termed a zero-one programming. Although integer programming can be solved by an exhaustive enumeration theoretically, it is impractical to solve realistically sized integer programming problems. The most successful algorithm so far found to solve integer programming is called the branch-and-bound enumeration developed by Balas (1965) and Dakin (1965). The other technique to integer programming is the cutting plane method developed by Gomory (1959). Uncertain Programming (BaoDing Liu, 2006.2) 23

A References NOTE: these references are only for demonstration, they are not real citations in the original text. [1] Donald E. Knuth. The TEXbook. Addison-Wesley, 1984. ISBN: 0-201-13448-9 [2] Paul W. Abrahams, Karl Berry and Kathryn A. Hargreaves. TEX for the Impatient. Addison-Wesley, 1990. ISBN: 0-201-51375-7 [3] David Salomon. The advanced TEXbook. New York : Springer, 1995. ISBN:0-387-94556-3 24

B B B.1 p(y x) = p(x, y) p(x) = p(x y)p(y) p(x) (123) B.1.1 1 Network Topology # of nodes # of clients Server Waxman GT-ITM 600 Transit-Stub 2% 10% 50% Max. Connectivity Inet-2.1 6000 Rui Ni Xue ThuThesis ABCDEF 25

B B.1.2 Hello, Xfig! LittleLeo 1 B.1.3 26

C C \input 27

xxxx xx xx xx xx xxxx 9 xx xx xx xxxx 7 xx xxxx 9 xx xx xx [1] Yang Y, Ren T L, Zhang L T, et al. Miniature microphone with silicon- based ferroelectric thin films. Integrated Ferroelectrics, 2003, 52:229-235. (SCI, :758FZ.) [2],,,.., 2005, 16(14):1289-1291. (EI, :0534931 2907.) [3],,,.., 2003, 24(S4):192-193. (EI.) [4] Yang Y, Ren T L, Zhu Y P, et al. PMUTs for handwriting recognition. In press. ( Integrated Ferroelectrics. SCI.) [5] Wu X M, Yang Y, Cai J, et al. Measurements of ferroelectric MEMS microphones. Integrated Ferroelectrics, 2005, 69:417-429. (SCI, :896KM.) [6],,,.., 2006, 28(1):117-119. (EI, :06129773469.) [7],,,. MEMS., 2003, 53:59-61. 28

[1],,,. :, CN1602118A. (.) [2] Ren T L, Yang Y, Zhu Y P, et al. Piezoelectric micro acoustic sensor based on ferroelectric materials: USA, No.11/215, 102. (.) 29