Book Reviews

Blackett’s War: The Men Who Defeated the Nazi U-Boats and Brought Science to the Art of Warfare

Budiansky, Stephen. 2013. Blackett’s War: The Men Who Defeated the Nazi U-Boats and Brought Science to the Art of Warfare. Knopf Publishing Group. 306 pp. $27.95.

Practitioners and researchers of operations research (OR) in sports may be intrigued to learn that as a result of a tennis match, OR may not have developed. Stephen Budiansky recalls how Professor Frederick Lindemann, Oxford’s professor of experimental philosophy and director of its Clarendon laboratory, who was to become Churchill’s confidante and scientific adviser on scientific aspects of modern warfare, first met Churchill when he partnered Mrs. Churchill at a charity tennis tournament in 1932. Budiansky also reports that almost all the good work that Blackett, a physicist, and his colleagues, including Henry Tizard, a chemist, and A.V. Hill, a biologist, carried out before and during World War II was done despite Lindemann’s advice to Churchill. Blackett and Lindemann did not see eye to eye! The first reference to OR was by A.P. Rowe at Bawdsey Research Station, where radar was being developed, as a result of pressure from Henry Tizard. Blackett and A.V. Hill both resigned from the Tizard committee because of disagreements with Lindemann, who wrote to the secretary of state for air that with those two obstacles out of the way, he was confident “this extremely important work will now be pursued more effectively and with due energy” (p. 78).

Fortunately, they were not out of the way, and were able to play a significant role for the United Kingdom and the Allies during World War II. Budiansky details how the work of physicists, biologists, and mathematicians, led by Blackett, Tizard, and Hill, in applying scientific thinking to military problems eventually turned the tide against the Nazi enemy. They were able to influence the military to make decisions based on data rather than prejudice. However, Budiansky does more than that. In this extremely well-written book, he goes back to World War I, Blackett’s experience in it, the influence of submarines during this war, and the buildup of the German U-boat fleet in the 1930s, to help the reader see how the situation in the late 1930s, relevant to the story he is telling, came about.

In his preface, Budiansky states that “it is no exaggeration to say that few men did more to win the war against Nazi Germany than Patrick Blackett…. As director of the antisubmarine analysis effort for the Royal Air Force and Royal Navy during World War II, Blackett not only helped win that battle, and the war, but in doing so founded the new science of operational research” (p. x). Blackett is regarded as the father of OR. Budiansky does an excellent job in telling the legendary stories about the start of OR; examples include determining the size of convoys, painting planes white rather than black, and creating a new policy at which to set depth charges. He describes how scientists became intimately involved with what had once been the distinct province of military commanders—convincing the disbelieving military to place trust in OR solutions. They did this by getting their hands dirty in undertaking the tasks that they were analyzing.

So, who was Blackett? He was one of the most distinguished experimental physicists of the 20th century. He was awarded the 1948 Nobel Prize in acknowledgment of his fundamental contribution to the knowledge and understanding of cosmic radiation, and was apparently thought unlucky not to have shared the 1936 Nobel Prize for the discovery of the positron. His first engagement with defense research was when he was appointed to the Aeronautical Research Committee in 1934.

Of course, Blackett’s war did not just involve Blackett. Key characters in Budiansky’s history include Karl Donitz, who rose from being a watch officer on a U-boat in 1917 to succeeding Hitler as Führer in the last days of the war, and Philip Morse, an MIT physicist, who was appointed head of the U.S. Navy’s Antisubmarine Warfare Operations Research Group in 1942.

Although Morse’s involvement in the start of the Operations Research Society of America in 1952 is documented, a surprising omission is any reference to Blackett’s subsequent involvement in British OR. It may be true, as has been asserted, that he viewed with some suspicion the development of OR as a new profession; nevertheless, in 1947, he was one of four diners at a meeting at the exclusive London club, the Athenaeum, which led to the creation of the Operational Research Club in 1948; this became the Operational Research Society in 1953. In 1950, Blackett wrote the opening article in the first issue of the Operational Research Quarterly; in this article, he set out his vision for civilian operational researchers. From 1965 to 1970, he was the president of the Royal Society. His portrait as president faces those who deliver the Operational Research Society’s prestigious annual Blackett lecture in a room of the Royal Society. Budiansky reports that when a colleague commented that this portrait made him look serious, Blackett’s response was “But I am a very serious man!” (p. 45). He was made a Companion of Honour in 1965, decorated with the award of the Order of Merit in 1967, and became Baron Blackett of Chelsea in 1969.

Blackett and his colleagues made a huge contribution to the creation of OR and to the defeat of the Third Reich. Without both, our lives would be very different today. Their story is well worth reading.

Graham Rand

Department of Management Science, Lancaster University, United Kingdom, [email protected]

Handbook of Newsvendor Problems: Models, Extensions and Applications

Choi, Tsan-Ming, ed. 2012. Handbook of Newsvendor Problems: Models, Extensions and Applications. Springer. 384 pp. $179.00.

Most operations researchers are exposed to the classical newsvendor, or newsboy, problem during their studies. This problem was first stated within a bank context as far back as the 18th century when a bank needed to determine the level of cash reserves it required to cover the demand from its customers. In this problem, the normal distribution was used to estimate the probability that a given amount of cash is sufficient to cover total cash demand. Another, possibly the most well-known, example addresses a newsboy’s need to determine the number of newspapers to order to meet future uncertain demand, given that the newspaper becomes obsolete (in other examples, it is perishable) at the end of the day. Given a stationary demand distribution and a known cost-revenue structure, the newsboy can determine the optimal order quantity by either maximizing the expected profit or minimizing the expected cost. Today, a huge amount of research literature is available on this topic and related problem areas; application areas are extensive and include many publications. However, as the editor of this handbook states: “what is absent is a comprehensive reference source of state-of-the-art findings on both theoretical and applied research on the newsvendor problem” (p. v). The goal of this handbook is therefore “to consolidate many latest research findings and applications of the newsvendor problem into one volume” (p. v).

This book has two parts: Part 1 contains the results of research on basic models and extensions; Part 2 focuses on a variety of practical applications of the extended newsvendor problem. Each part has eight chapters, and each chapter is dedicated to a specific topic in the field. Research related to this area has greatly increased over the past few years, and this handbook presents some of the most current research.

Chapter 1 discusses the multiproduct newsvendor problem (MPNP). In the MPNP, a number of products must be obtained and the demand for each product is assumed to be stochastic. This chapter presents a review of current MPNP research with specific focus on a variety of buyer stocking constraints, product substitution through one- and two-way substitution, and complementary products; it also discusses ideas for future research. Chapter 2 considers a different aspect of the MPNP. The objective is to replace the traditional risk-neutral performance measure by risk-aversion measures. Although risk-neutral problems can be decomposed, decomposition is not possible in the case this chapter considers. The entire portfolio of products must be considered as a whole; in this way, results are derived that show the impact of risk aversion and demand dependencies on optimal order quantities. In addition, this chapter shows that for general law-invariant coherent measures of risk, fairly accurate approximations of the optimal order quantities for a large number of products with independent demand can be obtained.

When a number of multiple-demand sources are available, decision makers can pool their inventories. Chapter 3 and a few other chapters cover the pooling problem, thus indicating the attractiveness of this mechanism. Inventory can be held at one or more physical locations to serve multiple demand points with fast transshipments to these demand points. This chapter focuses on the inventory pooling problem within the newsvendor framework; its objectives are to study the change in inventory levels when inventories are pooled and develop models to determine optimal inventory levels under such circumstances. It also introduces the copula framework to model various dependency structures among pooled demands.

To improve service offerings, reduce costs, and increase profit margins, many companies are employing collaboration and cooperation strategies. For these companies, the main drivers are the economies of scale they can achieve and risk pooling, such as virtual pooling of inventories in which stocks are kept locally but can be transferred to satisfy demand. Chapter 4 addresses a repeated newsvendor game with transshipments. Retailers act as they do in the classical newsvendor problem; however, after demand has been satisfied, the individual retailers must decide how much leftover stock to share with other retailers. When they do decide to share stock, residual inventories are transshipped to meet residual demands and residual profits are determined based on dual allocations. Chapter 5 presents a review of newsvendor games. It discusses a simple newsvendor game and its extensions, such as large games and multiple warehouses. The ultimate aim is to determine whether the anticipated benefits can allow a stable collaboration to be formed.

Chapters 6 and 7 consider newsvendor risk preferences. Newsvendor models typically assume risk neutrality. However, we are increasingly seeing attempts to introduce and use other risk preferences to describe newsvendor decision-making behavior. Chapter 7 focuses on risk preferences within the expected utility theory and prospect theory frameworks and directions for further research in the context of these frameworks. Chapter 8 discusses a comparative analysis of the value-of-risk (VaR) measure, the conditional value at risk (CVaR), and the mean-CVaR as objectives and constraints, respectively, in newsvendor models.

Chapters 9 and 10 address a fairly recent phenomenon in which (1) end-of-life products are returned to the original producer, and (2) the producer can, depending on the product’s condition, decide to remanufacture the used product. Mobile phones are one of the numerous examples of such products. In one case discussed, a two-period newsvendor-type approach is presented to model the optimal production and remanufacture strategies for a closed-loop supply chain. In such a situation, the link between production and the sales of new products and subsequent supply of used products is important. The second application is the development of a remanufacturing model that assists the newsvendor in making inspection decisions that yield maximum expected profits in a situation with a fixed number of end-of-life items.

Risk pooling, or statistical economies of scale, is a well-known managerial mechanism in different industries. In Chapter 11, using the newsvendor setting, a centralized inventory model is developed whereby inventory is pooled. The inventory allocation to the various retailers in the system is based on each retailer’s contribution, which is defined as the number of units the retailer produces and deposits at the central warehouse. A model of a pooled inventory system with priorities shows its advantages over a nonpooled inventory system.

The uncertain reliability of machines in metal stamping plants is a huge concern. Chapter 12 discusses an enhancement to the classical newsvendor problem to address this issue; the model incorporates multiple items, machine setup times, and the machine’s lack of reliability.

An additional, relevant application is the introduction of carbon emission policies, because high levels of carbon emissions are a major concern in today’s world. Chapter 13 addresses this topic by analyzing the classic single-period newsvendor problem under three carbon emission policies: the mandatory carbon emission capacity, the carbon emission tax, and the cap-and-trade system.

Chapter 14 discusses a practical example in which a fresh product could decay or lose freshness as a result of being transported over long distances. Decisions about the order quantity and selling price are made in sequence; however, the ordering decision is made before the product is ready to be transported, but the selling price depends on the freshness of the product when it arrives at the destination and the timing of the market. The model developed is a variation of the traditional newsvendor problem.

Chapter 15 attempts to address the profit target-setting approach for multiple divisions within a firm. The firm has a number of independent, autonomous divisions, each of which has full responsibility and accountability for its own operations and decisions. Each division is given a profit target and must determine both a production level and a selling price to maximize the probability of achieving the divisional profit targets. It also examines how to set the profit targets.

Finally, Chapter 16 investigates a portfolio approach to the MPNP with budget constraints. The procurement strategy for each newsvendor product is designed as a portfolio contract, which consists of a fixed-price contract and an option contract. With a fixed-price contract, the retailer pays a unit fixed cost for procuring each product; with an option contract, the retailer pays a unit reservation cost up front and receives a commitment from the supplier. For each unit procured under the specified commitment level, the retailer pays a unit execution cost. The objective is to find the optimal quantities for the portfolio contracts to maximize the total expected profit.

The Handbook of Newsvendor Problems: Models, Extensions and Applications addresses a wide variety of problems illustrating the breadth and depth of the current research related to the newsvendor problem—research on different models, model enhancements and extensions, and real-world applications. Most of the papers presented clearly have a direct link to real-life situations, thus making this handbook particularly relevant in the supply chain management environment. This publication includes many applications and suggestions for future research, which will be valuable to researchers, practitioners, and postgraduate students who are interested in working on the newsvendor problem.

Hans W. Ittmann

University of Johannesburg, Johannesburg, South Africa, [email protected]

Measuring the Performance of Public Services: Principles and Practice

Pidd, Michael. 2012. Measuring the Performance of Public Services: Principles and Practice. Cambridge University Press. 332 pp. $75.00.

In this lucid book, Professor Pidd demonstrates that quantitative performance measurement can greatly improve the quality of public services. However, he also demonstrates that ill-conceived measurement programs can cause misallocations, policy errors, and other distortions. As he explains, performance measurement is like medication: at its best, it greatly reduces the patient’s suffering; at its worst, its side effects are so adverse that the cure is worse than the disease.

Professor Pidd proceeds calmly and clearly to discuss the principles and techniques of sound performance measurement, illustrating his points with stimulating examples (mostly from the United Kingdom, although their broader relevance is obvious). The book contains no formal exercises; however, it has the effect of stimulating readers to create their own exercises and attempting to complete them. For example, he describes a formula that was devised in the United Kingdom to “measure the standards of healthcare against public expectations” (p. 233). Two thousand British citizens were chosen at random and asked to allocate 60 chips among various health risks to indicate their relative importance to the respondent. From this exercise, the performance P of a healthcare system was estimated using the formula:

where C = cancer death rate (per 100,000 people), H = heart disease death rate, A = death rate from avoidable diseases, Q = hip-replacement rate, and W and L relate to the times that patients slated for admission to a hospital must wait to enter.

In this formulation, the higher the value of P, the worse the assessment of the system. (The negative coefficient of Q reflects the perception that more hip replacements imply greater system responsiveness.)

When presenting this model, Professor Pidd does not remind us that P values could vary in different settings for reasons unrelated to the healthcare system per se. For example, C could be lower in one setting than in another because far fewer people smoke cigarettes in the first setting. However, he does not need to remind us because, elsewhere in the book, he discusses at length how differences in input factors across public systems can make it highly inappropriate to compare their performances based on raw output measures. Because different sections of the book mutually reinforce each other, readers can find themselves as active participants in learning from the examples. For this reason, the cumulative effect of Professor Pidd’s discussions is consistent with the principle that “the whole is greater than the sum of its parts.”

If any question exists about this book, it concerns the intended audience. Although the book is not dense with equations, it does assume some mathematical sophistication on the part of its readers. This point is perhaps most clearly implied when Professor Pidd introduces a projection formula for the number of patients in a hospital emergency department (ED):

He then states that “this is a little clumsy and we can make it easier to read by developing some algebra” (p. 72). Once he has done so, he reaches:

In my experience, many intelligent policy professionals who would readily understand the previous verbal formulation would be utterly flustered by the more mathematical one.

Yet, Professor Pidd does not assume that his readers are highly mathematical, as evidenced by his inclusion of a primer on linear programming as part of his treatment of data envelopment analysis. He also discusses regression analyses and time-series smoothing in ways that assume little prior familiarity with the techniques. These mathematical treatments, however, are too brief to substitute for more formal introductions that would yield deeper understanding by readers.

Perhaps this book is most valuable as an accompaniment to university courses in statistics and optimization that are geared to operations research professionals. The textbooks on these subjects devote much effort to discussing why various methods work and how they do so, but their examples are often unimaginative. Indeed, these examples routinely include the kind of simplistic assumptions that Professor Pidd properly criticizes in public sector policy analyses. Operations researchers who read this book will understand the importance of devoting as much intellectual energy to developing a mathematical model as to analyzing it.

That this last statement is true makes clear that Professor Pidd’s book constitutes a major contribution. It can benefit all operations researchers, including those with interests in the private rather than the public sector.

Arnold I. Barnett

MIT Sloan School of Management, Cambridge, Massachusetts 02142, [email protected]

Portfolio Decision Analysis: Improved Methods for Resource Allocation

Salo, Ahti, Jeffrey Keisler, Alec Morton, eds. 2011. Portfolio Decision Analysis: Improved Methods for Resource Allocation. Springer. 424 pp. $189.00.

This edited volume on portfolio decision analysis is a successful effort to capture many perspectives on an important problem in the industry: the process for selecting the right project investments to pursue. One could easily argue that portfolio decisions represent a key part of the strategy implementation process in any organization. When does strategy become reality within an organization? One answer is, at the moment that vision statements and business case reports transform into a resource budget, which is accompanied by a somewhat imperfect plan and a somewhat fluid team of functional experts or, said differently, at the moment that words transform into actions. In that regard, I fully echo and applaud the spirit of this book, which tries to elevate the importance and centrality of the portfolio decision process within organizations.

The chapters form a diversified set of perspectives that can aid both academics and practitioners. In particular, Part II, which addresses methodology, should be useful to anyone who holds the position of a portfolio decision analyst within an organization. Academics could use the examples described in this part and integrate them into their MBA and higher-level courses. I would not advise using this material in undergraduate programs, because the problems discussed are too complex to be understood by an audience that does not have some experience with project funding and execution.

I found Chapter 6 to be particularly useful because it attempts to empirically analyze the portfolio decision process. We need more efforts like this and on a larger scale. The lack of clean and useful data is a major issue; however, reaching out to practitioners with books such as this volume is a good starting point to argue for the value of data that can improve portfolio decisions. Once practitioners accept the need for more data, more empirical efforts should follow. Chapter 7, Behavioural Issues in Portfolio Decision Analysis, shows promise. I have performed several in-class exercises that use a novel simulation game (Loch and Mihm 2012) to address portfolio decisions. Clearly, aspects that relate to behavioural biases are an important determinant of portfolio decisions. Portfolio decisions require judgments of uncertain outcomes; as such, they are naturally subject to all biases observed in the behavioural economics literature. I would add to the perspective of Chapter 7 more specificity on the biases that are present in portfolio decision contexts and the respective (potential) remedies that we should consider.

Relative to future enhancements to this book, I think it has neglected two important discussion topics. The first is the process of translating strategy into portfolio criteria. This process is not self-explanatory, though it is extremely important because it reflects the definition of the criteria that govern the portfolio decision process. In addition, it is during this translation of strategic vision to portfolio criteria that the majority of what I call “endogenous diversity” emerges. That is, individuals who judge uncertain investments tend to use different lenses for the softer dimensions of evaluation (e.g., whether a project fits into the company strategy). As such, they infuse diversity into the process despite the seemingly common perspective (i.e., one common strategy, one common set of opportunities). Second, Chapter 8, A Framework for Innovative Management, needs improvement. I read this chapter with the expectation that I would read about an overall framework on innovation; the way I interpreted this was a positioning of the portfolio decision within innovation processes and, above it perhaps, a linkage between the portfolio decision and the different phases of innovation. Although this chapter has some discussion on the phases of ideation and selection, it is more of an attempt to introduce innovation tournaments. I found this an overly restrictive perspective on innovation and innovation processes in particular. Many other types of processes take place within organizations (e.g., stage-gate processes, internal agile-based processes) and this chapter includes little discussion about them. Moreover, it provides little discussion on whether diverse portfolios require different processes, which I believe is an essential message.

In summary, I believe that this volume is a necessary effort in the process of consolidating our knowledge about an important decision process within organizations. In its final section, it offers a variety of practitioner-based instances that make it useful reading for portfolio analysts. It also has some valuable information and future directions for academics who are interested in exploring the challenges of portfolio decisions. Hopefully, in the foreseeable future, a new version will address some of its weaker points.

Stelios Kavadias

Judge Business School, University of Cambridge, Cambridge, United Kingdom, [email protected]

Managing Supply Chain Risk

Sodhi, ManMohan, Christopher S. Tang. 2012. Managing Supply Chain Risk. Springer. 351 pp. $179.00.

In the foreword to Managing Supply Chain Risk, the authors explain honestly and intently that the book’s purpose is as a conceptual exploration of supply chain risk. They describe the structure of the book as layer upon layer of concepts drawn from the surrounding business world to try to understand supply chain risk in its entirety. Supply chain risk is a budding new field of research, they say, and has yet to establish its position as a vital and integral part of policy in daily business. Accordingly, their intended audience is mainly scholars who have the ability to digest the academic approach to this pragmatic problem; they hope that these scholars will continue to pave the way forward in this unfamiliar territory. The authors also believe that the world of academia, if removed from the realities of daily business, loses all value. However, because supply chain risk is a problem in real-life business, its exploration should also be targeted at managers. If practitioners and senior managers could not find value in what they have discovered, maybe their findings were not originally about supply chain risk.

This humble, balanced, and structured book is a pleasure to read. It is divided into four parts: Part I: Introduction–Identifying, Assessing, Mitigating and Responding to Supply Chain Risk; Part II: Mitigation Approaches and Applications; Part III: Review of Quantitative Models for Managing Supply Chain Risk; and Part IV: Perspectives and Topics for Future Research. In Chapter 1 of Part I, the authors lay the foundation for their research, taking the time to explain the background, context, and need for supply chain risk management. The subsequent chapters in this first part explore the basic concepts of risk management, starting with existing identification methodologies, such as the butterfly depiction, walking through the steps to risk management via mitigation strategies and the time-based risk management framework, and ending with a case study from Samsung Electronics to illustrate how a global manufacturing company manages its supply chain risk.

Part II deals with broad strategic and tactical approaches for mitigating supply chain risk and applies these, piecemeal, to specific domains: long-term demand uncertainty, outsourcing, product development, and product recall. Part III lays the foundations for mathematical modeling through three detailed reviews of the research literature. First, it introduces mathematical risk models, and then enhances them to include value flexibility to increase the robustness and stochastic programming for planning under demand uncertainty. In the final part, Part IV, the authors describe their perspectives and topics for future research. In doing so, they compare existing research literature against the perceived need for research, as discussed in Chapter 16, Researchers’ Perspectives on Supply-Chain Risk Research. In the final two chapters, they call upon the practitioners to engage with them, by means of “prescriptive research questions that are important for managers and for which empirical and quantitative modeling research could provide answers as managerial implications” (p. 303).

Although the chapters in Part III are primarily for researchers and technically trained practitioners, the chapters in Parts I, II, and IV are aimed at, and can benefit, both practitioners and researchers. The authors have structured this book as clearly as they would an encyclopedia, which enables all types of readers who are interested in supply chain risk to use it. They have attentively laid out their road map, the onion layers, from the daily dogma and routine acceptance to the mathematical core. At the beginning of each level of the diagnosis, they cautiously and clearly explain their intentions; conversely, at the end of each layer, they reflect upon what they have described. Anyone who wishes to travel along with their rationale, from any origin to any destination along their path, is welcome to do so. The signposts are clear, thus making the navigation childishly easy.

The authors clearly explain both the outline and the concepts stated within that outline. They accompany each methodology by an illustration, a practical example, and a discussion. Accordingly, the reader hops from one perspective to another, through the layers of diagnosis, with a continual thread of self-reflection, from theory to practice and back again. Using this approach, the authors never stray from their aim and their ambition, which they so clearly stated in the book’s preface. Modestly, but truthfully, they state that they cannot claim that this book is a valuable gem; however, I experienced it as a dedicated and unwavering testament to the importance of supply chain risk management and its value in business. I would advise each supply chain manager and researcher to add this book to his (her) bookshelf.

Tessa Möller

TNT Express, Duiven, the Netherlands, [email protected]

Public Policy Analysis: New Developments

Thissen, Wil A.H., Warren E. Walker, eds. 2013. Public Policy Analysis: New Developments. Springer. 278 pp. $129.00.

Perhaps the best known phrase in public policy analysis is speaking truth to power, which was most likely coined by Quakers during the 1950s, and became the title of a highly influential book on policy analysis (Wildavsky 1979). Thissen, Walker, and their colleagues aim to present a way of incorporating rational analysis in political policy and decision making, where power play is common. They do so, recognizing that over-simplified notions of rational decision making, which so often dominate papers published in operations research and management science (OR/MS) journals, serve only as ideal types in policy analysis, at best.

Thus, this book takes a broad view of public policy analysis and, thankfully, does not over-simplify the often messy world of public policy. In particular, its chapters stress the need for appropriate methods and processes to support stakeholder involvement and democratic accountability. The writing is thorough, if rather cold and clinical and, unlike many multiple-author books, the chapters fit together well and present a coherent argument. In this respect, the editors have done an unusually good job. The book occasionally includes vignettes of particular policy analyses based on the authors’ experiences in the Netherlands.

The book has two parts: Part I, which is the longest and most detailed, consists of six interlinked chapters that discuss policy analysis in a multiple-actor context. The three chapters in Part II focus on modelling for policy analysis. Some readers may think this balance is wrong because it suggests that modelling plays only a minor role in public policy analysis; however, I think the authors have made a sound decision. In my limited experience with public policy development and analysis, I have seen more effort and time go into boundary management, stakeholder management, and negotiation than into modelling. Reading the book makes the reason for this clear.

Chapter 1 provides a short introduction to policy analysis; the substantive content begins in Chapter 2, which is devoted to policy analysis in a multiple-actor context. It sets the scene for the rest of the book by offering five different models of policymaking: as a rational decision-making process, as a political game, as discourse, as a garbage can, and as an institutional process. None of these are new; however, comparing them serves as the backdrop for later material.

This multifaceted approach sensibly dominates the rest of Part I. Chapter 3 argues that no single correct or best view of policy analysis exists, and offers six approaches. Chapter 4 uses the five models of Chapter 2 to suggest how policy issues might be diagnosed and understood, stressing the need to consider the substantive and objective aspects, the characteristics of the actors and stakeholders, and the nature of the institutions and actor networks that exert power and influence. Chapter 5 argues that much of the policy analysis literature takes a naïve approach to the design and implementation of the processes that underpin policy analysis, which are important in democratic states. Chapter 6 proposes principles that should guide the policy analysis process, arguing strongly for approaches that support participation by encouraging and managing stakeholders, but prevent descent into chaos and near anarchy.

Part II consists of Chapters 7, 8, and 9. Chapter 7 is recognizable to most in the OR/MS field, and focuses on system models that have an inevitable trade-off between rigour and relevance. Chapter 8 describes actor models in policy analysis; however, I found this chapter weak, with a naïve presentation of agent-based approaches and gaming. However, it does recognize that the aim of such modelling, with its focus on actors and their interactions, may be to aid in understanding implementation issues, rather than to decide between policy options. Chapter 9 concludes the book with a discussion of approaches to uncertainty in policy analysis. This topic is helpful and should have been a more important element of Part II.

When I started to read this book, I was expecting to find more references to what has come to be called soft OR, as summarized in Rosenhead and Mingers (2001). Soft OR focuses on process management, stakeholder analysis, and the handling of uncertainty and offers much to public policy analysis. In particular, I was surprised to find no mention of the strategic choice approach (Friend and Hickling 2004), which is aimed at public policy and offers, I think, a more sophisticated view of uncertainty and its management.

Overall, this book is a welcome addition to the literature. OR/MS practitioners working in the private sector might learn something by reading it; those working in public policy will certainly do so, as will students. If they read it with books on soft OR, they will gain even more.

Michael Pidd

Department of Management Science, Lancaster University, Lancaster, United Kingdom, [email protected]

Global Optimization: A Stochastic Approach

Schäffler, Stefan. 2012. Global Optimization: A Stochastic Approach. Springer, 164 pp. $99.00.

Improvements in computer technologies have allowed more advanced exploration of global optimization. In particular, numerical solution methods for global optimization problems, which incorporate stochastic aspects, have been analyzed and used commonly in applied and engineering sciences. Global Optimization: A Stochastic Approach focuses on these methods, while providing background and theory about them.

This book includes a well-written and structured state-of-the-art survey, which gives the interested reader, both practitioner and researcher, essential information on what is necessary for global optimization. The book also provides information on recent and ongoing scientific investigations worldwide; thus, it invites readers to do their own scientific studies.

Global Optimization: A Stochastic Approach is structured into six chapters. Chapter 1 gives an introduction to the stochastic approach. It begins with a discussion and basic theoretical background of random search algorithms. An example, supported by several figures, shows the importance of selecting probability distributions within the algorithm. This chapter then proceeds to give a concise overview of adaptive search, Markovian, and population algorithms.

Chapter 2 covers the basics of unconstrained local optimization. It highlights the theory of the curve of steepest descent for a local optimization problem, describes the initial condition problem of steepest descent, and explores the background of numerical approximation. In this context, implicit and semi-implicit Euler methods are highlighted to solve the initial condition problem. Several examples also demonstrate the steepest descent curve and its initial value problem.

After emphasizing local optimization problems in Chapter 2, Chapter 3 addresses unconstrained global optimization. In particular, this chapter focuses on finding the starting point, which is close enough to the global optimum, because local optimization methods can also be used efficiently in global problems. It analyzes a randomized curve of steepest descent with an n-dimensional Brownian motion, and gives a concise overview of the numerical analysis of the descent curves. It also develops the semi-implicit Euler algorithm to obtain a numerical approximation for the global optimization problem, applies the algorithm in eight global optimization problems, and discusses the algorithm’s efficiency using these examples. Chapter 3 concludes with a discussion of the Euler method for gradient approximations. Here, approximate gradients are calculated using centered difference. Chapter 4 shows an application of the stochastic-unconstrained global optimization problem. In this demonstration, the digital communication channel coding problem is solved.

Chapter 5 highlights the constrained global optimization problem. Like Chapter 3, this chapter emphasizes finding an appropriate starting point, and not developing a method from scratch. Then, it uses local optimization approaches to solve the problems. After providing a concise summary of the sequential order quadratic method and Karush-Kuhn-Tucker conditions, it explores the penalty approach and global optimization problems with equality constraints. These approaches are also supported by sample problems, which are useful in helping the reader understand the underlying logic. Finally, this chapter addresses a general case of the global optimization problem.

Chapter 6 addresses vector optimization problems using a stochastic approach. Steepest descent methods are generalized for vector optimization and a randomized curve of the dominated points’ method is developed.

The most beneficial aspect of this book is the link between stochastic and global optimization problems and numerical solution techniques. The theoretical background and exercises that support this link are also valuable. However, the reader should have some knowledge of optimization and numerical analysis or have sufficient interest to gain that knowledge.

We believe that both today’s and future generations of students, teachers, researchers, and industry representatives could benefit from this book.

Miray Hanım (Aslan) Yıldırım, Gerhard-Wilhelm Weber

Institute of Applied Mathematics, METU, Ankara, Turkey {[email protected], [email protected]}