This page does not represent the most current semester of this course; it is present merely as an archive.

In any expression system with more than one operator, there is a need to clarify the meaning of a multi-operator system. For example, is 2-3+4 equal to 3 (2-3 plus 4) or -5 (2 minus 3+4)?

In arithmetic, an order of operations is sufficient to clarify meaning. However, as more complicated structures appear it no longer suffices. For example, in the code

Python
Java
for(int x : numbers) {
    sum += x;
}
for x in numbers:
    sum += x

which operator happens first: += or for? The answer is neither: they work together iteratively.

A versatile tool for understanding expressions, even complicated ones with components like loops, is the notion of a main operator. Any expression with an operator in it can be expressed as a single main operator with smaller expressions as its operands.

1 Scope and main operators

Every operator has some scope: a portion of the expression that it directly modifies. Logic has operators with three different ways of determining scope.

Binary operators

These have two operands, which comprise the operators scope. Examples include \land, \lor, \rightarrow, \oplus, etc.

Logic does not define an order of operations for binary operators: P \land Q \lor R is invalid notation; you must write either (P \land Q) \lor R or P \land (Q \lor R). The only time parentheses can be omitted is when the operators are associative1, as for example in P \oplus Q \oplus R.

Programming languages have parallels to many logical operators and do have operator precedence for them, so sometimes when logic is written by computer scientists they omit a few required parentheses and assume programming precedence rules apply. We won’t use that kind of omission in this class.

Not
The scope of \lnot is the term that immediately follows it. Thus \lnot A \lor B means (\lnot A) \lor B, not \lnot (A \lor B).
Quantifiers

The scope of a q quantifier is the entire expression that follows its terminating dot, though its scope cannot escape from parentheses. Thus, \forall x \;.\; A \land B means \forall x \;.\; (A \land B) not (\forall x \;.\; A) \land B.

There are actually two competing notations for quantifiers. MCS and most other computing texts use a dot after the quantifier and the from here until the end scope, as described above. ∀x and some other formal logic texts write the quantifier like a function instead, following it with parentheses that define its scope.

We only use the dot-notation quantifiers in this class.

The main operator of any expression is the operator whose scope is the entire expression.

2 Logic to English with main operators

When turning logic into English, the following process will always work:

  1. Identify the main operator
  2. Write the English for that operator, with place-holders for the operands
  3. Convert each operand to English using this process
  4. Simplify the resulting English, removing variables if possible.
Logic Example English template
\lnot x it is not the case that x
x \land y both x and y
x \lor y either x or y or both
x \oplus y either x or y but not both
x \rightarrow y if x then y
x \leftrightarrow y y if and only if x
\forall x \;.\; y for each x it is the case that y
\exists x \;.\; y there exists some x such that y
\nexists x \;.\; y there is no x such that y

Identify the main operators in each of the following:

  • (A \land B) \lor (C \land D) 2
  • (A \rightarrow B) \land (C \lor D) 3
  • \big((A \land B) \lor (C \land D)\big) 4
  • (A \land B) \lor \lnot(C \land D) 5
  • \lnot(A \land B) \lor \lnot(C \land D) 6
  • \lnot\big((A \land B) \lor (C \land D)\big) 7
  • \lnot A \rightarrow B 8
  • \lnot (A \rightarrow B) 9
  • \forall x \;.\; \lnot A(x) \rightarrow B(x) 10
  • \lnot \forall x \;.\; A(x) \rightarrow B(x) 11
  • \exists x \;.\; \lnot A(x) \rightarrow B(x) 12
  • \lnot \exists x \;.\; A(x) \rightarrow B(x) 13
  • \nexists x \;.\; A(x) \rightarrow B(x) 14
  • \forall x \;.\; \exists y \;.\; \lnot A(x) \rightarrow B(y) 15
  • \exists y \;.\; \forall x \;.\; \lnot A(x) \rightarrow B(y) 16

Convert the following to simple clear English

Predicate Meaning
P(x) x is a program
I(x,p) x is an input to program p
Q(p,x,y) program p solves input x as part of working on input y

\forall p \;.\; P(p) \rightarrow\Big(\exists i \;.\; I(i,p) \land\big(\forall j \;.\; I(j,p) \rightarrow Q(p,i,j)\big)\Big)

  1. The main operator is \forall p,

    For each p it is the case that …

    The remaining logic is P(p) \rightarrow \Big(\exists i \;.\; I(i,p) \land\big(\forall j \;.\; I(j,p) \rightarrow Q(p,i,j)\big)\Big)

  2. The main operator is \rightarrow

    For each p it is the case that if … then …

    The antecedent is P(p), which is just p is a program

    For each p it is the case that if p is a program then …

    We can simplify that English:

    For each program p it is the case that …

    The consequent is \Big(\exists i \;.\; I(i,p) \land\big(\forall j \;.\; I(j,p) \rightarrow Q(p,i,j)\big)\Big)

  3. The main operator is \exists i,

    For each program p it is the case that there exists some i such that …

    We can simplify that English:

    For each program p there exists some i such that …

    The remaining logic is I(i,p) \land\big(\forall j \;.\; I(j,p) \rightarrow Q(p,i,j)\big)

  4. The main operator is \land,

    For each program p there exists some i such that both … and …

    The first conjunct is I(i,p), which is just i is an input to p

    For each program p there exists some i such that both i is an input to p and …

    We can simplify that English:

    For each program p has some input i such that …

    The second conjunct is \big(\forall j \;.\; I(j,p) \rightarrow Q(p,i,j)\big)

  5. The main operator is \forall j,

    For each program p has some input i such that for each j it is the case that …

    The remaining logic is I(j,p) \rightarrow Q(p,i,j)

  6. The main operator is \rightarrow

    For each program p has some input i such that for each j it is the case that if … then …

    The antecedent is I(j,p), which is just j is an input to p;

    For each program p has some input i such that for each j it is the case that if j is an input to p then …

    We can simplify that English:

    For each program p has some input i such that for each input j it is the case that …

    The consequent is Q(p,i,j), which is program p solves input i as part of working on input j;

    For each program p has some input i such that for each input j it is the case that program p solves input i as part of working on input j.

    We can simplify that English:

    For each program p has some input i such that for each input j program p solves input i as part of working on input j.

  7. We have a full English sentence: For each program p has some input i such that for each input j program p solves input i as part of working on input j. But we’d rather not use variable in English.

    The variable p is the only program, so we can replace it with the program or the like: Every program has some input i such that for every input j, the program solves i as part of working on j.

    Since i and j are both inputs, the simple the input approach will not work, meaning we need more creativity in removing them from the phrase. A few examples:

    • Every program has an input that it solves along the way to solving every input.
    • For any program you care to pick, its has some base input: an input whose solution is part of every other inputs’ solution.
    • Every program has an input it solves during the solution of every other input.

3 English to logic with main operators

When turning English, the following process will always work:

  1. Identify the main operator implied by the English
  2. Write that operator as a symbol with its scope in parentheses
  3. Re-word the English to fit in the parentheses
  4. Convert each parenthesized English to logic

When doing this, be sure to look for implicit quantifiers. \forall is often omitted by making a general statement instead, which \exists may be as subtle as the user of a instead of the.

Also be careful about adding exists and if-then statements. I’ll marry someone who proposes means \exists x . I’ll marry x and x proposes not \exists x . I’ll marry x if x proposes; the latter is trivially true as long as there is someone somewhere who will never propose.

Identify the main operator of each of the following and re-write the sentence to use that operator as a symbol.

  • I’ll buy it if it works 17
  • I’ll buy anything that works 18
  • I’ll buy something that works 19
  • I’ll buy one that works 20
  • If a program passes all tests, it’s ready for release 21
  • If a program passes all tests, you are missing some tests 22
  • Every good program passes at least one test 23
  • There’s a test that every good program passes 24
  • You’ll need an IDE or debugger to solve this 25
  • A floating-point value is either normalized, denormalized, infinite, or NaN 26
  • This variable is either normalized, denormalized, infinite, or NaN 27
  • A floating-point value is a number 28
  • A floating-point value is missing 29
  • Using a floating-point value is a mistake 30

Convert the best program is never the cheapest program to logic

  1. This is a statement about all programs.

    \forall p\;. if p’s the cheapest, then p is not the best

  2. This is an if-then statement.

    \forall p\;. (p’s the cheapest) \rightarrow (p is not the best)

    The antecedent is p is the cheapest program.

    1. This is about the absence of something (nothing is cheaper than p).

      \nexists c\;. c is cheaper than p

    2. This is about cost. Which is not part of logic. So we turn it into a predicate:

      Predicate Meaning
      C(x,y) x is cheaper than y

      \nexists c \;.\; C(c,p).

    The consequent is p is the best program.

    1. This is about the absence of something (nothing is better than p).

      \nexists b\;. b is better than p

    2. This is about goodness. Which is not part of logic. So we turn it into a predicate:

      Predicate Meaning
      B(x,y) x is better than y

      \nexists b \;.\; B(b,p).

  3. Putting it all together:

    Predicate Meaning
    C(x,y) x is cheaper than y
    B(x,y) x is better than y

    \forall p\;.\; \big(\nexists c \;.\; C(c,p)\big) \rightarrow \big(\nexists b \;.\; B(b,p)\big)


  1. Some associative operator sequences are non-intuitive; for example, P \rightarrow (Q \lor R) and (P \rightarrow Q) \lor R are equivalent and hence can be written as P \rightarrow Q \lor R. You don’t need to knwo this, though, as it is always permitted to include parentheses even with associative operators.↩︎

  2. \lor↩︎

  3. \land↩︎

  4. \lor↩︎

  5. \lor↩︎

  6. \lor↩︎

  7. first \lnot↩︎

  8. \rightarrow↩︎

  9. \lnot↩︎

  10. \forall x↩︎

  11. \lnot↩︎

  12. \exists x↩︎

  13. \lnot↩︎

  14. \nexists x↩︎

  15. \forall x↩︎

  16. \exists y↩︎

  17. (it works) rightarrow (I’ll buy it)↩︎

  18. \forall x . (I’ll buy x if x works)↩︎

  19. \exists x . (I’ll buy x and x works)↩︎

  20. \exists x . (I’ll buy x and x works)↩︎

  21. \forall p . (if p passes all tests then p is ready for release↩︎

  22. (a program passes all tests) \rightarrow (you are missing some tests)↩︎

  23. \forall p . (if p is good then p passes at least one test)↩︎

  24. \exists t . (every good program passes t)↩︎

  25. (You’ll need an IDE to solve this) \lor (You’ll need a debugger to solve this)↩︎

  26. \forall n . (if n is a floating-point value then n is either normalized, denormalized, infinite, or NaN)↩︎

  27. (this variable is normalized) \oplus (this variable is denormalized) \oplus (this variable is infinite) \oplus (this variable is NaN)↩︎

  28. \forall n . (if n is a floating-point value then n is a number)↩︎

  29. \exists n . (if n is a floating-point value and n is missing)↩︎

  30. (using a float-point value) \rightarrow (a mistake)↩︎