A cinética da transformação de fase UAl3 + Al → UAl4 foi estudada em duas ligas Al-U, com 31,4 % e 33,4 % U em peso respectivamente, através de metalografia quantitativa. Os resultados mostraram que esta transformação é um processo de nucleação e crescimento termicamente ativado, com a nucleação ocorrendo heterogeneamente nas interfaces UAl3/AI(∞) e o crescimento sendo controlado por difusão em volume. A energia de ativação empírica do processo foi determinada, cujo valor médio é da ordem de 54,8 kcal/mol. Foi verificado que a cinética de crescimento da fase UAl4 obedece uma lei parabólica. As interfaces UAl4/ UAl3 e UAl4/Al(∞) migram em direções opostas, sendo que a velocidade da interface UAl4/ UAl3 é aproximadamente cinco vezes maior que aquela da interface UAl4/Al(∞). O coeficiente de difusão química do Al e do U na fase UAl4 foi avaliado ser da ordem de 10-9 cm 2/s a 600°C. / The kinetics of phase transformation UAl3 + Al → UAl4 of two Al-U alloys, with 31.4 and 33.4 wt % U respectively, was studied by quantitative microscopy. The results have shown that this transformation is a nucleation and thermally activated growth process. The nucleation occurs heterogeneously at the UAl3/AI(∞) interfaces and growth is controlled by volume diffusion. The empirical activation energy of the process was determined, whose mean value is of about 54.8 kcal/mol. It was verified that a parabolic growth law is obeyed. The UAl4/ UAl3 and UAl4/Al(∞) interfaces migrates in opposite directions, with the UAl4/ UAl3 interface velocity was approximately 5 times greater than that of UAl4/Al(∞) interface. The chemical diffusion coefficient of Al and U in the UAl4 phase were evaluated to be of the order of 10-9 cm 2/s at 600°C.

This thesis focuses on the phase transitions, including ferro-magnetic, anti-ferromagnetic, metal to "Mott" insulator and structural transitions in perovskite and Ruddlesden-Popper ruthenates. The thesis is mainly composed of two parts. The first half presents Density Functional Theory (DFT)+Dynamical Mean Field Theory (DMFT) studies of the electronically driven phase transitions in various ruthenate materials. We study cubic perovskite BaRuO$_3$ via DFT add DMFT method using interaction parameters which were found in previous studies to be appropriate for the related materials, CaRuO$_3$ and SrRuO$_3$. The calculated variation in transition temperature between the Ba and Sr compounds is consistent with experiment, confirming the assignment of the compounds to the Hund's metal family of materials, the appropriateness of the single-site dynamical mean field approximation for these materials as well as confirming the appropriateness of the values for the interaction parameters. The results provide insights into the origin of magnetism and the role of the van Hove singularity in the physics of Hund's metals. We also study the metal-insulator transition (MIT) and magnetic transitions in Ca$_2$RuO$_4$. The Ru-O bonds lengths are found to be the most important control parameters for the metal-insulator transitions and rotations are found to be less important. The calculation successfully captures the important features of the para-magnetic (PM) "Mott" insulating state, including the orbital occupancy disproportionation and the orbitally resolved electron spectral function. It shows the advantage of single set DFT+DMFT in dealing with strongly correlated multi-orbital systems without the assumption of spin symmetry breaking. In the second half, we present a Landau free energy model that incorporates the electronic energetics, the coupling of the electronic state to local distortions and the coupling of local distortions to long-wavelength strains. The model is used to elucidate important experimental features in thermal and current-induced MIT in Ca$_2$RuO$_4$ and Ca$_3$Ru$_{2-x}$Ti$_x$O$_7$ materials. The investigation of lattice and electronic energetics and determination of parameters using DFT+DMFT methods is explained. The change in lattice energy across the metal-insulator transition is shown to be comparable to the change in electronic energy. Important consequences are a strongly first order transition, a sensitive dependence of the phase boundary on pressure and that the geometrical constraints on in-plane lattice parameter associated with epitaxial growth on a substrate typically change the lattice energetics enough to eliminate the metal-insulator transition entirely. The change in elasto-resistance across the MIT is determined. The DFT+U relaxation study shows the octahedron relaxation with respect to uniaxial strain on a and b axes are very different. This sensitive a and b axes dependence is observed in calculations on both Ca$_2$RuO$_4$ and Ca$_3$Ru$_2$O$_7$. The theory model is also generalized to investigate spatially non-homogeneous solutions. Important features of the stripe patterns at the domain boundaries of metallic and insulating phases are discussed and compared with experiments.

University of Macau / Faculty of Science and Technology. / Department of Computer and Information Science

Este trabalho discute a permanência dos espaços públicos centrais frente às sucessivas transformações metropolitanas e seu papel na construção da cidade. A partir de uma praça signifi cativa da ocupação territorial de São Paulo, a Praça do Patriarca, procuramos apontar, no espaço construído, as imposições e as possibilidades defi nidas ao longo do processo histórico e dos diversos ideais urbanos responsáveis pela sua confi- guração. A Praça e seu entorno foram observados quanto à conformação espacial, ao desenho formal de seus edifícios e às possibilidades de uso e apropriação que permitia em cada período estudado. A esse levantamento, foram sobrepostas referências históricas sobre a construção da cidade e seus distintos propósitos, procurando esclarecer as relações entre a conjuntura sócio-política e a consequente construção do espaço. / This work discusses the permanence of central public spaces in between the successive urban transformations and their role in the construction of the city. As off the Praça do Patriarca, an important square for the territorial occupation of São Paulo, we will show the impositions and possibilities defi ned during historic process in the constructed space and the various urban concepts responsible for its confi guration. The relation of the square and its environment was researched concerning their spatial confi guration, the formal design of their buildings and the possibilities of use and appropriation permitted during each of the observed periods. Added to this survey were historic references about the city construction and their distinct purposes, with the idea to show the relationship between a current social-politic situation and the resulting spatial construction.

利率模型廣泛應用於利率衍生品的定價。為了吻合實證利率的分佈和隱含波動率，一種可能的辦法是用Lévy過程替換Hull- White模型中的布朗隨機變量的利率模型，但是這種方法很難實施。本文建立了一種有效的網絡數值方法對利率進行估測。利用Lévy過程的馬爾可夫性質， FFT網絡實質上是多項樹模型的擴展。這種數值方法的優勢在於一直固定不變的狀態點，對現時利率期限結構的超級校準以及基於對Lévy過程的特徵方程的快速傅裡葉變換(FFT) 去恢復概率密度函數以實現轉移概率的計算過程。這種網絡數值方法對利率衍生品的定價與利率樹類似。對利率上限期權和交換期權的解析解和數值解的比較表明網絡數值方法是準確和有效的。FFT網絡還可以對百慕達式利率交換期權以及美式期權進行定價。最後， FFT網絡被擴展去適應路徑依賴變量，因此，能對利率依賴的結構性票據進行定價，比如目標贖回票據和範團積息結構票據。 / Short rate models are widely used in valuing interest rate derivatives. To fit empirical distribution of interest rates and implied volatility, a possible way is to replace Brownian motion by a Lévy process in short rate models. However, this approach is difficult to implement. This thesis establishes an efficient network approach for interest rate valuation. The FFT-network is essentially an extension of multinomial tree model, taking advantage of the Markov property of Lévy processes. Its fixed and unchanged states at all time, super-calibration ability to the current term structure, and elegant computation procedure for transition probabilities using the fast Fourier transform (FFT) from the characteristic function of Lévy processes make it attractive and distinct from other numerical methods. The interest rate derivatives value is determined in a way similar to that of the tree approach. The comparison between the closed-form solution of interest rate caplets and swaptions and the numerical results under the network demonstrates that the proposed network is accurate and efficient. In addition, the FFT-network can also be used to pricing the Bermudan swaption and American-style option. Finally, the FFT-network is expanded to accommodate path-dependent variables, and hence can be used for pricing some path-dependent structured notes, such as the target redemption notes and range of accrual notes. / Detailed summary in vernacular field only. / Xu, Zhuolu. / Thesis (M.Phil.)--Chinese University of Hong Kong, 2012. / Includes bibliographical references (leaves 91-93). / Abstracts also in Chinese. / Chapter 1 --- Introduction --- p.1 / Chapter 2 --- Preliminaries --- p.4 / Chapter 2.1 --- Elementary techniques --- p.4 / Chapter 2.1.1 --- Characteristic function --- p.4 / Chapter 2.1.2 --- Cumulant generating function --- p.5 / Chapter 2.1.3 --- Fourier Transform --- p.6 / Chapter 2.1.4 --- Fast Fourier Transform (FFT) --- p.8 / Chapter 2.2 --- Lévy Processes --- p.10 / Chapter 2.2.1 --- Definition --- p.10 / Chapter 2.2.2 --- Lévy-Khintchine --- p.11 / Chapter 2.2.3 --- Lévy Processes in Interest Rate --- p.13 / Chapter 2.3 --- Hull-White Model --- p.13 / Chapter 2.3.1 --- Model setup --- p.14 / Chapter 2.3.2 --- Interest rate caps --- p.15 / Chapter 2.3.3 --- European Swaptions --- p.16 / Chapter 2.3.4 --- A Tree-building procedure --- p.19 / Chapter 3 --- HW-Lévy Model --- p.20 / Chapter 3.1 --- Model Setup --- p.20 / Chapter 3.2 --- The Characteristic Function --- p.22 / Chapter 3.3 --- Analytic result on interest rate derivatives --- p.26 / Chapter 4 --- Valuation: FFT Network Model --- p.35 / Chapter 4.1 --- Drawbacks of Tree Approach --- p.35 / Chapter 4.2 --- FFT Network Setup --- p.37 / Chapter 4.3 --- Transition Probability Matrix --- p.38 / Chapter 4.4 --- Yield Curve Fitting --- p.42 / Chapter 4.5 --- Pricing Algorithm under FFT Network --- p.45 / Chapter 4.5.1 --- European Interest Rate Derivatives Pricing --- p.45 / Chapter 4.5.2 --- Bermudan Interest Rate Derivatives Pricing --- p.49 / Chapter 5 --- Extended FFT Network for Path-dependent Structured Notes --- p.55 / Chapter 5.1 --- Extended FFT-netwok --- p.55 / Chapter 5.2 --- Target Redemption Notes (TARN) --- p.61 / Chapter 6 --- Numerical Study --- p.69 / Chapter 6.1 --- Numerical Scheme --- p.69 / Chapter 6.2 --- Numerical Examples --- p.74 / Chapter 7 --- Conclusion --- p.89 / Bibliography --- p.91

Shik, Hoi Yin = 自旋模型中量子相變與量子糾纏的研究 / 石海燕. / Thesis (Ph.D.)--Chinese University of Hong Kong, 2011. / Includes bibliographical references (leaves 90-96). / Electronic reproduction. Hong Kong : Chinese University of Hong Kong, [2012] System requirements: Adobe Acrobat Reader. Available via World Wide Web. / Abstract also in Chinese. / Shik, Hoi Yin = Zi xuan mo xing zhong liang zi xiang bian yu liang zi jiu chan de yan jiu / Shi Haiyan.

Shum Yu Hing. / Thesis (M.Phil.)--Chinese University of Hong Kong, 1994. / Includes bibliographical references (leaves 140-143). / ABSTRACT --- p.1 / Chapter 1. --- INTRODUCTION / Chapter 1.1. --- Fourier Analysis --- p.3 / Chapter 1.2. --- Wavelet Analysis --- p.6 / Chapter 1.3. --- Parallelization --- p.10 / Chapter 1.3.1. --- Data Dependency Analysis / Chapter 2. --- LITERATURE SURVEY / Chapter 2.1. --- One Dimensional Fast Wavelet Transform (Discrete) --- p.13 / Chapter 2.2. --- Shared Memory Architecture : Parallel Virtual Machine (PVM) / Chapter 2.3. --- Distributed Memory Architecture : Massively Parallel Machine (DECmpp) --- p.21 / Chapter 3. --- THEORY / Chapter 3.1. --- Parallel Processing / Chapter 3.1.1. --- Amdahl ´ةs Law --- p.25 / Chapter 3.1.2. --- Quality Factor --- p.31 / Chapter 3.2. --- Parallel Architecture / Chapter 3.2.1. --- Pipelining --- p.32 / Chapter 3.2.2. --- Vector Processors --- p.34 / Chapter 3.2.3. --- Multiprocessor --- p.34 / Chapter 3.2.4 --- Array Processors --- p.36 / Chapter 3.2.5. --- Systolic Array Processing --- p.37 / Chapter 3.2.6. --- Granularity --- p.40 / Chapter 3.2.7. --- Load Balancing & Throughput --- p.42 / Chapter 3.3. --- Parallel Programming --- p.43 / Chapter 3.4. --- Parallel Numerical Algorithm / Chapter 3.4.1. --- Parallelism Within a Statement --- p.44 / Chapter 3.4.2. --- Parallelism Between Statements --- p.47 / Chapter 4. --- IMPLEMENTATION / Chapter 4.1. --- Sequential Version --- p.49 / Chapter 4.2. --- Parallel Version / Chapter 4.2.1. --- Matrix Representation of Wavelet Transform / Chapter 4.2.1.1. --- Decomposition --- p.52 / Chapter 4.2.1.2. --- Reconstruct ion --- p.55 / Chapter 4.2.2. --- Parallel Virtual Machine (PVM) / Chapter 4.2.2.1. --- Parallel Algorithm / Chapter (a) --- HOST --- p.56 / Chapter (b) --- NODE --- p.57 / Chapter 4.2.2.2. --- Flowcharts --- p.59 / Chapter 4.2.2.3. --- Timing Model Analysis --- p.65 / Chapter 4.2.2.4 --- Quality Factor / Chapter (a) --- Decomposition --- p.71 / Chapter (b) --- Reconstruction --- p.72 / Chapter 4.2.3. --- Massively Parallel Machine - DECmpp --- p.73 / Chapter 4.2.3.1. --- Parallel Algorithm for ACU & PEs / Chapter 4.2.3.2. --- Flowcharts --- p.75 / Chapter 4.2.3.3. --- Timing Model Analysis / Chapter (a) --- Communication Strategy --- p.77 / Chapter (b) --- Decomposition --- p.78 / Chapter (c) --- Reconstruct ion --- p.87 / Chapter 4.2.3.4. --- Quality Factor / Chapter (a) --- Decomposition --- p.89 / Chapter (b) --- Reconstruction --- p.89 / Chapter 4.2.3.5. --- Mapping --- p.92 / Chapter 5. --- RESULT / Chapter 5.1. --- Parallel Virtual Machine (PVM) / Chapter 5.1.1. --- Sequential Version --- p.93 / Chapter 5.1.2. --- Parallel Version --- p.103 / Chapter 5.2. --- Massively Parallel Machine - DECmpp / Chapter 5.2.1. --- Sequential Vers ion --- p.104 / Chapter 5.2.2. --- Parallel Version --- p.110 / Chapter 5.3. --- Output File Generated from both machines --- p.118 / Chapter 6. --- DISCUSSION / Chapter 6.1. --- Application on real time situation --- p.123 / Chapter 6.2. --- Two dimensional or Multidimensional case --- p.123 / Chapter 6.3. --- Block Algorithm Approach / Chapter 6.3.1. --- Blocked --- p.124 / Chapter 6.3.2. --- Row Wrapped --- p.126 / Chapter 6.4. --- Memory Requirement --- p.127 / Chapter 6.5. --- Signal Size Prediction / Chapter 6.5.1. --- Method A --- p.131 / Chapter 6.5.2. --- Method B --- p.133 / Chapter 7. --- CONCLUSION --- p.134 / Chapter 8. --- FUTURE MODIFICATION --- p.138 / REFERENCE --- p.140 / LISTING --- p.144 / APPENDIX I - Technical Information of PVM --- p.145 / Chapter II - --- Technical Information of DECmpp --- p.152 / Chapter III - --- Some Tips/Guide --- p.165

by Shui-Qin Zhou. / Publication date from spine. / Thesis (Ph.D.)--Chinese University of Hong Kong, 1995. / Includes bibliographical references (leaves 152-157). / Acknowledgments --- p.i / Abstract --- p.ii / Contents --- p.vi / Abbreviation --- p.viii / List of Figures --- p.xi / List of Tables --- p.xvi / Chapter 1. --- Introduction --- p.1 / Chapter 1.1. --- Project's Significance --- p.1 / Chapter 1.2. --- Research background.....................................: --- p.1 / Chapter 1.2.1. --- Linear PNIPAM chains in aqueous solution --- p.2 / Chapter 1.2.1.1. --- Origins of the LCST --- p.2 / Chapter 1.2.1.2. --- Experimental studies --- p.4 / Chapter 1.2.2. --- PNIPAM gels --- p.9 / Chapter 1.2.2.1. --- Origins of the volume phase transition --- p.10 / Chapter 1.2.2.2. --- Experimental studies --- p.11 / Chapter 1.2.3. --- Surfactant effects --- p.13 / Chapter 1.3. --- The objective of the project --- p.15 / Chapter 1.4. --- Experimental difficulties --- p.17 / Chapter 2. --- Basic Theory --- p.20 / Chapter 2.1. --- Laser light scattering (LLS) --- p.20 / Chapter 2.2. --- Polymer chains in solution --- p.23 / Chapter 2.3. --- Swollen polymer gels --- p.27 / Chapter 3. --- Results and Discussion --- p.32 / Chapter 3.1. --- Linear PNIPAM chains --- p.32 / Chapter 3.1.1. --- D = kDM-αD for PNIPAM in THF and water --- p.32 / Chapter 3.1.2. --- Coil-to-globule transition of single PNIPAM chains in water --- p.46 / Chapter 3.1.3. --- Additional chain-conformation broadening of the line-width distribution in dynamic LLS --- p.64 / Chapter 3.1.4. --- Internal motions of long linear PNIPAM chains --- p.72 / Chapter 3.1.5. --- Structural model of spherical water/AOT/n-hexane microemulsion --- p.80 / Chapter 3.2. --- Spherical PNIPAM microgels --- p.94 / Chapter 3.2.1. --- Volume phase transition of PNIPAM microgels --- p.94 / Chapter 3.2.2. --- Internal motions of short PNIPAM subchains inside the gel networks --- p.107 / Chapter 3.2.3. --- Surfactant effects on the volume phase transition --- p.113 / Chapter 3.2.4. --- Swelling and drying kinetics of a very thin PNIPAM gel film --- p.129 / Chapter 4. --- Conclusions --- p.141 / Chapter 5. --- Experimental --- p.145 / Chapter 5.1 --- Sample preparation --- p.145 / Chapter 5.2 --- Physical measurement --- p.149 / Chapter 6. --- References --- p.152

La superélasticité (SE) est la capacité d'un alliage à recouvrer sa forme initiale après une importante déformation. Cet effet, dû à une transformation martensitique réversible, est très convoité pour de nombreuses applications biomédicales. Les alliages nickel-titane sont les alliages les plus utilisés dans les applications fonctionnelles nécessitant de la SE. Cependant, l'utilisation de cet alliage dans les dispositifs biomédicaux est controversée par la présence du nickel, considéré comme élément cytotoxique et allergène. De ce fait, ces dernières années, une attention accrue a été portée aux alliages de titane [3-métastable pouvant constituer une alternative pour des applications biomédicales. En effet, ils ont l'avantage d'être élaborés à partir d'éléments biocompatibles et de présenter un comportement SE. L'objectif de cette thèse réside ainsi dans l'étude de la réponse SE des alliages de titane [3-métastable à différentes échelles dans leur forme massive et sous forme de revêtements. Deux alliages [3-métastable ont été élaborés: le Ti-27Nb {% at) et le Ti-24Nb-4Zr-4Sn (% mass). Les propriétés SE de ces alliages à l'état massif ont été caractérisées à l'échelle macroscopique et à l'échelle submicrométrique et comparées à celles du NiTi superélastique et du CP-Ti élasto-plastique. La nanoindentation a d'abord été utilisée pour sonder l'effet SE dans ces alliages massifs à l'échelle locale, et dans un deuxième temps, pour étudier l'effet de l'orientation cristallographique sur la réponse SE et mécanique à l'échelle du grain. Enfin, les propriétés mécaniques et SE de ces alliages sous forme de revêtements ont été évaluées avec ce même procédé de nanoindentation. / The superelasticity is the ability of an alloy to recover its original shape after significant deformation. This effect, due to a reversible stress-induced martensitic transformation, is highly sought after for many biomedical applications. Nickel-titanium alloys that have a very large strain recovery (in bulk and thin film state), are the alloys currently used for functional applications requiring superelasticity. However, the use of this alloy in biomedical devices is controversial by the presence of nickel, considered as allergen and presumed cytotoxic for the body. As a result, in recent years, increased attention has been given to metastable f3 titanium alloys, which may be an alternative for biomedical applications. Indeed, they have the advantage of being elaborated from biocompatible elements and exhibit a superelastic behavior. The objective of this thesis lies in the study of the superelastic response of metastable f3 titanium alloys at different scales in bulk and thin film state. Two metastable f3 titanium alloys were elaborated: Ti-27Nb (at%) alloy and Ti-24Nb-4Zr-8Sn (wt %) alloy. The superelastic properties of these two bulk alloys were characterized at macroscopic and sub-micrometric scale and compared to superelastic NiTi and elastoplastic CP-Ti. The nanoindentation was first used to evaluate the superelastic effect of various bulk titanium alloys at local scale, and in a second time, to study the effect of crystallographic orientations on the superelastic and mechanical responses at the grain scale. Finally, the mechanical and superelastic properties of metastable f3 titanium coatings were evaluated with this same nanoindentation process.

針對Lévy過程下的二維期權定價問題，本文提出了一種基於快速傅利葉變換(FFT)的解決方案，稱之為二維快速傅利葉變換網絡。不論是時間從屬還是線性組合，此方法適用於所有能取得聯合特徵函數的二維Lévy構建。快速傅利葉變換的種種優點使得比數值方法在不影響結果精確性的前提下，大大降低了所需計算時間。理論上，更高維的Lévy期權定價問題也可以通過擴展數值網絡解決。除此之外，我們還探究了資產波動性亦服從Lévy過程的單資產期權定價。這種資產價值和波動性由一組相關Lévy過程驅動的模型被稱為時間轉換Lévy過程。最後，關於美式及奇異期權定價的數值算例驗證了文中方法的準確性和有效性。 / We propose a two-dimensional network to retrieve the price of two-asset option under Lévy processes by using the fast Fourier transform (FFT). It can be applied to different multivariate Lévy constructions such as subordination and linear combination provided that the joint characteristic function is obtainable. With the prevalent implementation of FFT, the network approach results in significant computational time reduction while maintaining satisfactory accuracy. In general, multi-dimensional option pricing problems are also solvable by extending this network. Furthermore, we investigate option pricing on a single asset where the asset return and its volatility are driven by a pair of dependent Lévy processes. Such a model is also called the random time-changed Lévy process. Numerical examples are given to demonstrate the efficiency and accuracy of FFT-network applied to exotic and American-style options. / Detailed summary in vernacular field only. / Wang, Weiyin. / Thesis (M.Phil.)--Chinese University of Hong Kong, 2013. / Includes bibliographical references (leaves 41-43). / Electronic reproduction. Hong Kong : Chinese University of Hong Kong, [2012] System requirements: Adobe Acrobat Reader. Available via World Wide Web. / Abstracts also in Chinese. / List of Tables --- p.ii / List of Figures --- p.iii / Chapter 1 --- Introduction --- p.1 / Chapter 2 --- Literature Review --- p.4 / Chapter 2.1 --- Lévy Process --- p.4 / Chapter 2.1.1 --- Definition and Properties --- p.4 / Chapter 2.1.2 --- Multivariate Lévy Construction --- p.6 / Chapter 2.2 --- Fast Fourier Transform (FFT) in Option Pricing --- p.9 / Chapter 2.2.1 --- European Option on One Asset --- p.9 / Chapter 2.2.2 --- European Option on Two Assets --- p.11 / Chapter 3 --- Two-dimensional FFT-network Model --- p.13 / Chapter 3.1 --- Two-dimensional FFT-network --- p.15 / Chapter 3.2 --- Two-asset Option Pricing --- p.22 / Chapter 3.2.1 --- General Model --- p.22 / Chapter 3.2.2 --- Specific Models --- p.23 / Chapter 3.3 --- Random Time-changed Lévy Process --- p.25 / Chapter 3.3.1 --- Model --- p.26 / Chapter 3.3.2 --- Correlation Adjustment --- p.28 / Chapter 4 --- Numerical Examples --- p.31 / Chapter 4.1 --- Two-asset Option --- p.31 / Chapter 4.1.1 --- Spread Option Pricing --- p.31 / Chapter 4.1.2 --- Pricing under Diffierent Multivariate Lévy Constructions --- p.36 / Chapter 4.2 --- One-asset Option under Random Time-changed Lévy Process --- p.37 / Chapter 5 --- Conclusion --- p.40 / Bibliography --- p.41