In my early years (and ever since as well), at San Jose State University (SJSU) in San Jose California, I taught an introductory Management Science course called Quantitative Business Analysis. I adopted book after book searching for one that did not have some flaw in a topic that I covered. When I tired of the search, I finally realized that I would be happy only with a book of my own. So I put some connecting paragraphs into the lecture notes I had been using for a number of years and turned out a preliminary custom book that was published by Southwest Publishing, a division of Thompson Learning. Several semesters later, when the master discs for the book disappeared from their files and mine also at the same time, I realized that I must be on to something, and I decided to self publish the book from that point on.

Until recently, the book (see Table of Contents here) has been intended for my SJSU students only, but because some of the chapters contain material not available elsewhere, I have decided to offer the book online to the general public. It is designed for a one semester introductory course in Management Science, and hence does not cover all of the topics included in the usual sort of two semester books that are available in great number through the usual channels. But the topics that it does cover are given new content that makes my book different from other textbooks. In particular, note the following features:

• Introduction of the notion of "optimality condition" in the very first algorithm presented in the book, the simple Minimal Spanning Tree algorithm.
• Reinforcement of this fundamental concept in the context of the Transportation Method, which is presented after network models but prior to the general Linear Programming material. The central role of an optimal spanning tree is stressed, and the similarity between the optimality conditions for the two problem types is noted.
• Presentation of "stylistic guidelines" for the construction of linear programming models in the spreadsheet, using practices such as range names, parallel alignment, color coding, and border formatting. Use of such guidelines greatly reduced the likelihood of error in construction of the model, and facilitates documentation of the model for a co-worker.
• Gives an algorithm for doing project crashing of a CPM model by hand, and shows how linear programming with parametric programming can be used to obtain results on bigger project crashing models in the spreadsheet.
• Gives full particulars on the beta distribution in the context of PERT and PERT simulation. Enables the student to actually carry out PERT simulations in the spreadsheet, and enables them to find "best fits" of beta distributions to simulation results, or empirical data sets.

COMING SOON
Two additional chapters are under preparation which will incorporate some of my own research work. The first will be a separate chapter on Risk Averse Decision Making that will show how "certainty (i.e. cash) equivalence functions" for exponential utility can be used to facilitate the analysis of decision problems for risk-averse decision makers. Reasons for using CE-Value functions based on exponential utility will be explained, and the important parametric optimization technique called "Risk Tolerance Parametrics" will be presented, along with the associated chart and concept referred to as the "Maximal Value Frontier." The usual EVPI and EVSI concepts for the value of perfect and sample information are generalized to CEVPI and CEVSI (cash equivalent values) that are seen to be functions of the Risk Tolerance appropriate for the decision maker. An important result coming out of this generalization is the fact that information, especially CEVSI, can be much MORE for the risk-averse decision maker than the risk neutral one, and even in simple examples, it can be MUCH MORE. This is extremely important from the practical perspective, because it means that many worthwhile information alternatives may be erroneously passed up if one relies only on risk neutral EVSI evaluations.

The second new chapter pertains to a more thorough going use of the beta distribution in project duration analysis. If one assumes that ALL Early Start and Early Finish times developed in the forward pass are beta distributed, a set of reasonable approximations can be developed that make it possible to compute what they are. That is, a stochastic forward pass can be defined that develops the beta parameters for the ES and EF times going forward in an analytical way so that no simulation is necessary. The beta resulting for the last EF time in the project network is the distribution of project time, and it can be used in lieu of the histogram of a simulation result. The ability to develop the beta distribution for project duration analytically then leads to a stochastic project crashing formulation with a constraint on mean project duration. One can parametrically "tighten" this mean project duration constraint in analogy with the deterministic project crashing procedure to develop a trade-off curve relating mean project duration and total project cost (including crashing cost). The nonlinear programming Solver algorithm must be used in this case since the mean project duration is a (very complex) nonlinear function of the activity crashing fractions, but the Solver does just fine with this constraint on the example presented. Very interesting activity crashing plans are seen to result, that give insight into how to crash when the uncertainty in activity duration is taken into account.

I will be pursuing an aggressive publication effort to get the material in these two new chapters into the open literature, so that eventually, other texts will carry similar material. But until that happens, the only place to get this material is in my book. Those buying the current version will get the two new chapters as updates free of charge.