# Propositional formula

For example: The assertion: "This cow is blue. That horse is orange but this horse here is purple." is actually a compound proposition linked by "AND"s: ( ("This cow is blue" AND "that horse is orange") AND "this horse here is purple" ) .
Example: "That purple dog is running", "This cow is blue", "Switch M31 is closed", "This cap is off", "Tomorrow is Friday".

For the purposes of the propositional calculus a compound proposition can usually be reworded into a series of simple sentences, although the result will probably sound stilted.

In particular, simple sentences that employ notions of "all", "some", "a few", "one of", etc. called logical quantifiers are treated by the predicate calculus. Along with the new function symbolism "F(x)" two new symbols are introduced: ∀ (For all), and ∃ (There exists ..., At least one of ... exists, etc.). The predicate calculus, but not the propositional calculus, can establish the formal validity of the following statement:

"All blue pigs have wings but some pigs have no wings, hence some pigs are not blue".

While some of the familiar rules of arithmetic algebra continue to hold in the algebra of propositions (e.g. the commutative and associative laws for AND and OR), some do not (e.g. the distributive laws for AND, OR and NOT).

Engineers analyze the logic circuits they have designed using synthesis techniques and then apply various reduction and minimization techniques to simplify their designs.

Arbitrary propositional formulas are built from propositional variables and other propositional formulas using propositional connectives. Examples of connectives include:

The following are the connectives common to rhetoric, philosophy and mathematics together with their truth tables. The symbols used will vary from author to author and between fields of endeavor. In general the abbreviations "T" and "F" stand for the evaluations TRUTH and FALSITY applied to the variables in the propositional formula (e.g. the assertion: "That cow is blue" will have the truth-value "T" for Truth or "F" for Falsity, as the case may be.).

The connectives go by a number of different word-usages, e.g. "a IMPLIES b" is also said "IF a THEN b". Some of these are shown in the table.

Engineering symbols have varied over the years, but these are commonplace. Sometimes they appear simply as boxes with symbols in them. "a" and "b" are called "the inputs" and "c" is called "the output".

The following three propositions are equivalent (as indicated by the logical equivalence sign ≡ ):

Example: The proposition " IF 'Winston Churchill was Chinese' THEN 'The sun rises in the east' " evaluates as a TRUTH given that 'Winston Churchill was Chinese' is a FALSEHOOD and 'The sun rises in the east' evaluates as a TRUTH.

As shown above, the CASE (IF c THEN b ELSE a ) connective is constructed either from the 2-argument connectives IF ... THEN ... and AND or from OR and AND and the 1-argument NOT. Connectives such as the n-argument AND (a & b & c & ... & n), OR (a ∨ b ∨ c ∨ ... ∨ n) are constructed from strings of two-argument AND and OR and written in abbreviated form without the parentheses. These, and other connectives as well, can then used as building blocks for yet further connectives. Rhetoricians, philosophers, and mathematicians use truth tables and the various theorems to analyze and simplify their formulas.

Substitution: The variable or sub-formula to be substituted with another variable, constant, or sub-formula must be replaced in all instances throughout the overall formula.

This inductive definition can be easily extended to cover additional connectives.

If used in an axiomatic system, the symbols 1 and 0 (or T and F) are considered to be well-formed formulas and thus obey all the same rules as the variables. Thus the laws listed below are actually axiom schemas, that is, they stand in place of an infinite number of instances. Thus ( x ∨ y ) ≡ ( y ∨ x ) might be used in one instance, ( p ∨ 0 ) ≡ ( 0 ∨ p ) and in another instance ( 1 ∨ q ) ≡ ( q ∨ 1 ), etc.

Thus the formula can be parsed—but because NOT does not obey the distributive law, the parentheses around the inner formula (~c & ~d) is mandatory:

d & c ∨ p & ~(c & ~d) ≡ c & d ∨ p & c ∨ p & ~d rewritten is ( ( (d & c) ∨ ( p & ~((c & ~(d)) ) ) ) ≡ ( (c & d) ∨ (p & c) ∨ (p & ~(d)) ) )

Omitting parentheses in strings of AND and OR: The connectives are considered to be unary (one-variable, e.g. NOT) and binary (i.e. two-variable AND, OR, IMPLIES). For example:

( (c & d) ∨ (p & c) ∨ (p & ~d) ) above should be written ( ((c & d) ∨ (p & c)) ∨ (p & ~(d) ) ) or possibly ( (c & d) ∨ ( (p & c) ∨ (p & ~(d)) ) )

However, a truth-table demonstration shows that the form without the extra parentheses is perfectly adequate.

OR distributes over AND and AND distributes over OR. NOT does not distribute over AND or OR. See below about De Morgan's law:

NOT, when distributed over OR or AND, does something peculiar (again, these can be verified with a truth-table):

Absorption, in particular the first one, causes the "laws" of logic to differ from the "laws" of arithmetic:

The engineering symbol for the NAND connective (the 'stroke') can be used to build any propositional formula. The notion that truth (1) and falsity (0) can be defined in terms of this connective is shown in the sequence of NANDs on the left, and the derivations of the four evaluations of a NAND b are shown along the bottom. The more common method is to use the definition of the NAND from the truth table.

Example: The following shows how a theorem-based proof of "(c, b, 1) ≡ (c → b)" would proceed, below the proof is its truth-table verification. ( Note: (c → b) is defined to be (~c ∨ b) ):

In the following truth table the column labelled "taut" for tautology evaluates logical equivalence (symbolized here by ≡) between the two columns labelled d. Because all four rows under "taut" are 1's, the equivalence indeed represents a tautology.

Any propositional formula can be reduced to the "logical sum" (OR) of the active (i.e. "1"- or "T"-valued) minterms. When in this form the formula is said to be in disjunctive normal form. But even though it is in this form, it is not necessarily minimized with respect to either the number of terms or the number of literals.

In the following table, observe the peculiar numbering of the rows: (0, 1, 3, 2, 6, 7, 5, 4, 0). The first column is the decimal equivalent of the binary equivalent of the digits "cba", in other words:

When working with Karnaugh maps one must always keep in mind that the top edge "wrap arounds" to the bottom edge, and the left edge wraps around to the right edge—the Karnaugh diagram is really a three- or four- or n-dimensional flattened object.

Produce the formula's truth table. Number its rows using the binary-equivalents of the variables (usually just sequentially 0 through n-1) for n variables.

Technically, the propositional function has been reduced to its (unminimized) conjunctive normal form: each row has its minterm expression and these can be OR'd to produce the formula in its (unminimized) conjunctive normal form.

Example: ((c & d) ∨ (p & ~(c & (~d)))) = q in conjunctive normal form is:

However, this formula be reduced both in the number of terms (from 4 to 3) and in the total count of its literals (12 to 6).

Steps in the reduction using a Karnaugh map. The final result is the OR (logical "sum") of the three reduced terms.

Use the values of the formula (e.g. "p") found by the truth-table method and place them in their into their respective (associated) Karnaugh squares (these are numbered per the Gray code convention). If values of "d" for "don't care" appear in the table, this adds flexibility during the reduction phase.

Minterms of adjacent (abutting) 1-squares (T-squares) can be reduced with respect to the number of their literals, and the number terms also will be reduced in the process. Two abutting squares (2 x 1 horizontal or 1 x 2 vertical, even the edges represent abutting squares) lose one literal, four squares in a 4 x 1 rectangle (horizontal or vertical) or 2 x 2 square (even the four corners represent abutting squares) lose two literals, eight squares in a rectangle lose 3 literals, etc. (One seeks out the largest square or rectangles and ignores the smaller squares or rectangles contained totally within it. ) This process continues until all abutting squares are accounted for, at which point the propositional formula is minimized.

For example, squares #3 and #7 abut. These two abutting squares can lose one literal (e.g. "p" from squares #3 and #7), four squares in a rectangle or square lose two literals, eight squares in a rectangle lose 3 literals, etc. (One seeks out the largest square or rectangles.) This process continues until all abutting squares are accounted for, at which point the propositional formula is said to be minimized.

Example: The map method usually is done by inspection. The following example expands the algebraic method to show the "trick" behind the combining of terms on a Karnaugh map:

Minterms #3 and #7 abut, #7 and #6 abut, and #4 and #6 abut (because the table's edges wrap around). So each of these pairs can be reduced.

Observe that by the Idempotency law (A ∨ A) = A, we can create more terms. Then by association and distributive laws the variables to disappear can be paired, and then "disappeared" with the Law of contradiction (x & ~x)=0. The following uses brackets [ and ] only to keep track of the terms; they have no special significance:

q = ( (~p & d & c ) ∨ (p & d & c) ∨ (p & d & ~c) ∨ (p & ~d & ~c) ) = ( #3 ∨ #7 ∨ #6 ∨ #4 )
( [ (d & c) ∨ (~p & p) ] ∨ [ (p & d) ∨ (~c & c) ] ∨ [ (p & ~c) ∨ (c & ~c) ] )

Given the following examples-as-definitions, what does one make of the subsequent reasoning:

(1) "This sentence is simple." (2) "This sentence is complex, and it is conjoined by AND."

Then assign the variable "s" to the left-most sentence "This sentence is simple". Define "compound" c = "not simple" ~s, and assign c = ~s to "This sentence is compound"; assign "j" to "It [this sentence] is conjoined by AND". The second sentence can be expressed as:

If truth values are to be placed on the sentences c = ~s and j, then all are clearly FALSEHOODS: e.g. "This sentence is complex" is a FALSEHOOD (it is simple, by definition). So their conjunction (AND) is a falsehood. But when taken in its assembled form, the sentence a TRUTH.

In the abstract (ideal) case the simplest oscillating formula is a NOT fed back to itself: ~(~(p=q)) = q. Analysis of an abstract (ideal) propositional formula in a truth-table reveals an inconsistency for both p=1 and p=0 cases: When p=1, q=0, this cannot be because p=q; ditto for when p=0 and q=1.

Analysis requires a delay to be inserted and then the loop cut between the delay and the input "p". The delay must be viewed as a kind of proposition that has "qd" (q-delayed) as output for "q" as input. This new proposition adds another column to the truth table. The inconsistency is now between "qd" and "p" as shown in red; two stable states resulting:

A "clocked flip-flop" memory ("c" is the "clock" and "d" is the "data"). The data can change at any time when clock c=0; when clock c=1 the output q "tracks" the value of data d. When c goes from 1 to 0 it "traps" d = q's value and this continues to appear at q no matter what d does (as long as c remains 0).

Without delay, inconsistencies must be eliminated from a truth table analysis. With the notion of "delay", this condition presents itself as a momentary inconsistency between the fed-back output variable q and p = qdelayed.

About the simplest memory results when the output of an OR feeds back to one of its inputs, in this case output "q" feeds back into "p". Given that the formula is first evaluated (initialized) with p=0 & q=0, it will "flip" once when "set" by s=1. Thereafter, output "q" will sustain "q" in the "flipped" condition (state q=1). This behavior, now time-dependent, is shown by the state diagram to the right of the once-flip.

The formula known as "clocked flip-flop" memory ("c" is the "clock" and "d" is the "data") is given below. It works as follows: When c = 0 the data d (either 0 or 1) cannot "get through" to affect output q. When c = 1 the data d "gets through" and output q "follows" d's value. When c goes from 1 to 0 the last value of the data remains "trapped" at output "q". As long as c=0, d can change value without causing q to change.

The state diagram is similar in shape to the flip-flop's state diagram, but with different labelling on the transitions.

The use of the word "everything" in the law of excluded middle renders Russell's expression of this law open to debate. If restricted to an expression about BEING or QUALITY with reference to a finite collection of objects (a finite "universe of discourse") -- the members of which can be investigated one after another for the presence or absence of the assertion—then the law is considered intuitionistically appropriate. Thus an assertion such as: "This object must either BE or NOT BE (in the collection)", or "This object must either have this QUALITY or NOT have this QUALITY (relative to the objects in the collection)" is acceptable. See more at Venn diagram.