Formal Methods of Software Design

an online course by Eric Hehner

"Formal methods of software design" means using mathematics to write error-free programs. The mathematics needed is not complicated; it's just basic logic. The word "formal" means the use of a formal language, so that the program logic can be machine checked. Our compilers already tell us if we make a syntax error, or a type error, and they tell us what and where the error is. Formal methods take the next step, telling us if we make a logic error, and they tell us what and where the error is. And they tell us this as we make the error, not after the program is finished. It is good to get any program correct while writing it, rather than waiting for bug reports from users. It is absolutely essential for programs that lives will depend on.

The course begins by introducing (or reviewing) the basic logic that will be used in the course as an aid to programming. Then we look at formal specifications, and how they are refined to become programs. At each refinement step, there is a small theorem to be proven (that the step is correct), so that at the end, the program is correct. Most of the course uses just those programming constructs that are common to most programming languages (assignment statement, if statement, array). We also look at parallel and interacting processes, at probabilistic programming, and functional programming. Along the way, we formally define the language features we use (both execution control and data structures). The emphasis is on program development to meet specifications, and on program modifications that preserve correctness, rather than on verification after a program is finished.

Prerequisites: some programming experience, and some mathematical ability.

The three best motivations for this course are written here, here, and here. Read comments on this course's formal methods.

Get the course textbook. Here are pages 223,..236 in case you would like to print the laws, precedence table, and distributing operators. Here are some study questions. And here is some tutorial material for Chapter 1, Chapter 3, and Chapter 4. Solutions to exercises are available to course instructors.

This course comes in two versions: for-credit, and not-for-credit.

For-Credit: You must be registered at a university that accepts this course. There is a start date and a finish date for the course. There are three one-hour tests during the term, each worth 20%, and a two-hour exam at the end, worth 40%. The times and places of the tests and exam are set by the university. The tests and exam will include the laws from the back of the textbook: pages 223,..236. In addition, you may bring into the tests and exam one letter-sized (or A4) piece of paper with anything you like written (or typed) on both sides. No other aids are allowed. For each question, a blank answer is worth approximately 33%; from that, marks will be awarded for relevant and correct information, and marks will be deducted for irrelevant or incorrect information. The graded tests will be given back; the graded exam will not be given back.

Not-For-Credit: You may start the course whenever you like, proceed at your own pace, and finish whenever you like. There are no tests or exams.

Whether you are taking the course for-credit or not-for-credit, it is very valuable for your understanding of the course material to do written exercises (assignments) as you go through the course. With each lecture segment, there is a suggested small exercise or two from the textbook, and there are hundreds more exercises in the textbook for you to choose from. You may work on them by yourself or in a group. In the for-credit version, there is an online chat room for students to discuss the exercises and course material with each other. In the not-for-credit version, you may email questions to the course instructor. If you choose to typeset your solutions in LaTeX, here is a helpful set of macros. When you have done an exercise, send your solution to the course instructor by email, and you will receive a solution by email to compare with your solution. Solutions to exercises will not be graded.

There are 34 lectures below for you to watch at your convenience. After each lecture segment listed below, there is a link to the lecture transcript, then the lecture time. Finally, there is a suggested exercise or two from the textbook. If you want to print the lecture visuals, here they are: visuals for Chapter 0, visuals for Chapter 1, visuals for Chapter 2, visuals for Chapter 3, visuals for Chapter 4, visuals for Chapter 5, visuals for Chapter 6, visuals for Chapter 7, visuals for Chapter 8, visuals for Chapter 9.

Lecture segment 0 and transcript 0 [11:56] Introduction; try Exercise 0
Lecture segment 1 and transcript 1 [29:29] Binary Theory; try Exercises 5 and 12
Lecture segment 2 and transcript 2 [11:16] Binary Theory continued, Number Theory, Character Theory; try Exercises 2(f,m,p,s,u) and 16
Lecture segment 3 and transcript 3 [14:49] Collections: Bunches and Sets; try Exercises 41 and 44
Lecture segment 4 and transcript 4 [27:06] Sequences: Strings and Lists; try Exercise 53
Lecture segment 5 and transcript 5 [10:31] Functions; try Exercise 55
Lecture segment 6 and transcript 6 [12:34] Quantifiers; try Exercise 63
Lecture segment 7 and transcript 7 [19:02] Function Fine Points; try Exercise 81
Lecture segment 8 and transcript 8 [30:26] Specification; try Exercise 99
Lecture segment 9 and transcript 9 [10:59] Refinement; try Exercises 116 and 120
Lecture segment 10 and transcript 10 [26:11] Program Development; try Exercise 125
Lecture segment 11 and transcript 11 [30:43] Time Calculation; try Exercise 133
Lecture segment 12 and transcript 12 [20:28] Searching; try Exercise 158
Lecture segment 13 and transcript 13 [22:27] Two Great Examples: Fast Exponentiation and Fibonacci; try Exercise 161
Lecture segment 14 and transcript 14 [15:15] Space Calculation; try Exercise 215
Lecture segment 15 and transcript 15 [17:52] Scope and Data Structures; try Exercise 263
Lecture segment 16 and transcript 16 [31:09] Control Structures; try Exercise 208(a)
Lecture segment 17 and transcript 17 [25:05] Mid-Course Review; try Exercise 202
Lecture segment 18 and transcript 18 [13:49] Time and Space Dependence and Assertions; try Exercise 276
Lecture segment 19 and transcript 19 [15:10] Subprograms and Aliasing; try Exercise 274
Lecture segment 20 and transcript 20 [26:54] Probabilistic Programming; try Exercises 279 and 280
Lecture segment 21 and transcript 21 [15:16] Functional Programming; try Exercise 144
Lecture segment 22 and transcript 22 [18:17] Recursive Data Definition; try Exercise 309
Lecture segment 23 and transcript 23 [8:50] Recursive Program Definition; try Exercise 321(a,b,c)
Lecture segment 24 and transcript 24 [29:29] Data Theory Design; try Exercise 338
Lecture segment 25 and transcript 25 [12:35] Program Theory Design; try Exercise 351
Lecture segment 26 and transcript 26 [22:33] Data Transformation; try Exercise 365
Lecture segment 27 and transcript 27 [12:03] Limited Queue; try Exercise 372
Lecture segment 28 and transcript 28 [14:45] Independent Composition; try Exercise 375
Lecture segment 29 and transcript 29 [21:00] Sequential to Parallel Transformation; try Exercise 381
Lecture segment 30 and transcript 30 [16:00] Interactive Variables; try Exercise 386
Lecture segment 31 and transcript 31 [20:06] Communication Channels; try Exercise 410
Lecture segment 32 and transcript 32 [15:33] Local Channel Declaration, Deadlock, Broadcast; try Exercise 414
Lecture segment 33 and transcript 33 [19:27] Final Review; try Exercise 226
total lecture time [10:49:07]