Feynman Lectures on Computation

This series of lectures, Like Feynmans physics lectures, start from the very beginning and proceed quickly. Read each chapter several times before moving on to the next. This is not a quasi coffee table "physics for poets" text. Feyman assumes you will actually work out the problems he presents, follow the logical flow of how a computer circuit works, etc. However, if you do work through each chapter, the insights are astounding. The subject matter of this books touches on information theory (Shannon et al), quantum computing, infophysics, etc. If you have a passing interest in these subjects, read this book. It will make all of these subjects much more clear.

Yes, I think you can teach the theory of computation from this book. And you can learn it from this book. Some of the material isn't all that recent, but much of it doesn't need to be. 35 years ago, if one were teaching a course on the theory of computation, I'd have recommended Minsky's book (it came out in 1967). That was a great text. Nowadays, there are numerous choices. But one could still use books that originally came out well before Feynman's notes, such as Lewis & Papadimitriou or Hopcroft, Motwani, and Ullman. The question boils down to the quality of what is in the book, as well as what material it has that other books do not, and what material it is missing that most other texts have. This book is quite readable and preserves much of Feynman's teaching style. So let's look at what it is missing. First, it doesn't talk much about real neurons. Of course, even Minsky doesn't dwell much on that, and other computation books avoid that topic too. But now, there's a more serious omission. Feynman spends something like two pages on grammars! If you were using Lewis and Papadimitriou (first edition) there would be a chapter of over 70 pages on context-free languages alone. As a teacher or a student, would you really want to miss all that? No, as a student, you would have to read up on all that material elsewhere. And as a teacher, you would have to use another book or write your own notes. That material is too much a part of most required curricula. But that doesn't take away from the value of the book when it comes to the rest of the material. And the final four chapters, which discuss coding and information theory, reversible computation and the thermodynamics of computing, quantum mechanical computers, and some physical aspects of computation, are all useful material that you often won't see in other computation texts. As a student, I'd read the book. As a teacher, I'd recommend it to my students. But as either, I wouldn't expect to use it as the only textbook.

The book starts out at such a leisurely pace that one is fooled into thinking that it will be finshed in a few days read, but Feynman soon plunges into the much deeper aspects of computation. Some chapters are material that are covered by others much more extensively (such as theory of computation) but they are often treated in his unique approach, other topics (such as Quantum mechanical computers) are such rare gems that they alone would be worth getting the book for.

This book is not easy, but like his physics lecture, the effort in following his lectures and working out the questions and problems that he poses make this, in my opinion, one of the most beautiful, albeit difficult and terse, books on computation I have come across in a long time. Certainly belongs in the library of anyone who is serious about the theoretical aspects of computation.

This reference is derived from Feynman's lectures at Caltech between 1983-1986 for the course 'Potentialities and Limitations of Computing Machines'. This small volume introduces computers as a file clerk performing his tasks, moves on to show how the 'file clerk' can be built out of simple gates, how the gates can be built out actual transistors, discusses essential issues in computation theory such as computability and Turing machines, and then discusses essential issues in information theory such as data compression. The physics of computing from a thermodynamics context is then considered. If the general reader ignores the gas equations, this chapter is fairly easy to read and enlightening. The next chapter continues with a discussion of quantum mechanical computers. The final chapter discusses how real transistors function at the atomic level and fabrication techniques for real integrated circuits. Lectures given by invited experts on computer science topics such as vision, robots, expert systems, etc, are not included. Although this reference does not discuss alternative architectures for computation, such as the ones found in the brains of animals, this reference is ideal to introduce the motivated general reader to the concept of computation and the techniques used in commercial computers.