39
1 Programming Languages (CS 550) Lecture 9 Summary Introduction to Formal Semantics Jeremy R. Johnson

1 Programming Languages (CS 550) Lecture 9 Summary Introduction to Formal Semantics Jeremy R. Johnson TexPoint fonts used in EMF. Read the TexPoint manual

Embed Size (px)

Citation preview

Page 1: 1 Programming Languages (CS 550) Lecture 9 Summary Introduction to Formal Semantics Jeremy R. Johnson TexPoint fonts used in EMF. Read the TexPoint manual

1

Programming Languages (CS 550)

Lecture 9 SummaryIntroduction to Formal Semantics

Jeremy R. Johnson

Page 2: 1 Programming Languages (CS 550) Lecture 9 Summary Introduction to Formal Semantics Jeremy R. Johnson TexPoint fonts used in EMF. Read the TexPoint manual

2

Theme

This lecture introduces three techniques for formally specifying the semantics of programming languages: operational semantics (formal machine model), denotational semantics, and axiomatic semantics.

So far we have extensively used operational semantics (meta-circular interpreter and interpreters built using other languages), the approaches outlined are related but more mathematical.

Page 3: 1 Programming Languages (CS 550) Lecture 9 Summary Introduction to Formal Semantics Jeremy R. Johnson TexPoint fonts used in EMF. Read the TexPoint manual

3

Outline

Operational SemanticsReduction machineProlog implementation

Denotational SemanticsTranslating programs to mathematical functionsScheme implementation

Axiomatic SemanticsSpecificationsPredicate transformersCorrectness proofs

Page 4: 1 Programming Languages (CS 550) Lecture 9 Summary Introduction to Formal Semantics Jeremy R. Johnson TexPoint fonts used in EMF. Read the TexPoint manual

4

Mini Language Syntax

1. < program > → < stmt-list>

2. < stmt-list> → < stmt > ; < stmt-list > | < stmt >

3. < stmt > → < assign-stmt > | < if-stmt > | < while-stmt >

4. < assign-stmt > → < identifier > := < expr >

5. < if-stmt > → if < expr > then < stmt-list > else < stmt-list > fi

6. < while-stmt > → while < expr > do < stmt-list > od

7. < expr > → < expr > + < term > | < expr > - < term > | < term >

8. < term > → < term > * < factor > | < factor >

9. < factor > → ( < expr > ) | < number > | < identifier >

10. < number > → < number > < digit > | < digit >

11. < digit > → 0 | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9

12. < identifier > → < identifier > < letter > | < letter >

13. < letter > → a | b | c | ... | z

Page 5: 1 Programming Languages (CS 550) Lecture 9 Summary Introduction to Formal Semantics Jeremy R. Johnson TexPoint fonts used in EMF. Read the TexPoint manual

5

Environments

Let an Environment be a map from indentifiers to values = integers undefined

Mini language programs can be thought of as a map from an initial Environment to a final Environment (assuming it terminates)

The initial environment maps all identifiers to an undefined

Each statement is defined in terms of what it does to the current environment (another mapping)

Page 6: 1 Programming Languages (CS 550) Lecture 9 Summary Introduction to Formal Semantics Jeremy R. Johnson TexPoint fonts used in EMF. Read the TexPoint manual

6

Semantics of Mini Language Statements

1. Env: Identifier → Integer Union {undef}

2. (Env and {I = n})(J) = n if J=I, Env(J) otherwise

3. Env_0 = undef for all I

4. for if-stmt, if expr evaluates to value greater than 0, then evaluate stmt-list after then, else evaluate stmt-list after else

5. for while-stmt, as long as expr evaluates to a value greater than 0, stmt-list is repeatedly executed and expr evaluated.

Page 7: 1 Programming Languages (CS 550) Lecture 9 Summary Introduction to Formal Semantics Jeremy R. Johnson TexPoint fonts used in EMF. Read the TexPoint manual

7

Example Mini Language Program

1. n := 0 - 5;

2. if n then i := n else i := 0 - n fi;

3. fact := 1;

4. while i do fact := fact * i; i := i - 1 od

What is the final environment?

Page 8: 1 Programming Languages (CS 550) Lecture 9 Summary Introduction to Formal Semantics Jeremy R. Johnson TexPoint fonts used in EMF. Read the TexPoint manual

8

Operational Semantics

Define language by describing its actions in terms of operations of an actual or hypothetical machine. Need precise description of machineProgram Control Store

Reduction machineReduce program to a semantic “value”Reduction rules (logical inference rules)

Page 9: 1 Programming Languages (CS 550) Lecture 9 Summary Introduction to Formal Semantics Jeremy R. Johnson TexPoint fonts used in EMF. Read the TexPoint manual

9

Operational Semantics of Mini Language Expressions

(1) ‘0’ 0,…, ‘9’ 9(2) V’0’ 10*V,…,V’9’ 10*V+9

(3) V1 ‘+’ V2 V1 + V2

(4) V1 ‘+’ V2 V1 + V2

(5) V1 ‘*’ V2 V1 * V2

Page 10: 1 Programming Languages (CS 550) Lecture 9 Summary Introduction to Formal Semantics Jeremy R. Johnson TexPoint fonts used in EMF. Read the TexPoint manual

10

Mini Language Expressions

(7) E E1 _____________________________________________________________________

E ‘+’ E2 E1 ‘+’ E2

(8) E E1 _____________________________________________________________________

E ‘-’ E2 E1 ‘-’ E2

(9) E E1 _____________________________________________________________________

E ‘*’ E2 E1 ‘*’ E2

Page 11: 1 Programming Languages (CS 550) Lecture 9 Summary Introduction to Formal Semantics Jeremy R. Johnson TexPoint fonts used in EMF. Read the TexPoint manual

11

Mini Language Expressions

(10) E E1 _____________________________________________________________________

V ‘+’ E V ‘+’ E1

(11) E E1 ____________________________________________________________________

V ‘-’ E V ‘-’ E1

(12) E E1 ____________________________________________________________________

V ‘*’ E V ‘*’ E1

(14) E E1, E1 E2 [transitive closure] _____________________________________________________________________

E E2

Page 12: 1 Programming Languages (CS 550) Lecture 9 Summary Introduction to Formal Semantics Jeremy R. Johnson TexPoint fonts used in EMF. Read the TexPoint manual

12

Implementation in Prolog% reduce_all(times(plus(2,3),minus(5,1)),V).

% V = 20 ?

reduce(plus(E,E2),plus(E1,E2)) :- reduce(E,E1).

reduce(minus(E,E2),minus(E1,E2)) :- reduce(E,E1).

reduce(times(E,E2),times(E1,E2)) :- reduce(E,E1).

reduce(plus(V,E),plus(V,E1)) :- reduce(E,E1).

reduce(minus(V,E),minus(V,E1)) :- reduce(E,E1).

reduce(times(V,E),times(V,E1)) :- reduce(E,E1).

reduce(plus(V1,V2),R) :- integer(V1), integer(V2), !, R is V1+V2.

reduce(minus(V1,V2),R) :- integer(V1), integer(V2), !, R is V1-V2.

reduce(times(V1,V2),R) :- integer(V1), integer(V2), !, R is V1*V2.

reduce_all(V,V) :- integer(V), !.

reduce_all(E,E2) :- reduce(E,E1), reduce_all(E1,E2).

Page 13: 1 Programming Languages (CS 550) Lecture 9 Summary Introduction to Formal Semantics Jeremy R. Johnson TexPoint fonts used in EMF. Read the TexPoint manual

13

Environments and Assignment

(7) <E | Env> <E1| Env> ______________________________________________________________________________________________________________________________

<E ‘+’ E2 | Env> < E1 ‘+’ E2 | Env>

(15) Env(I) = V ____________________________________________________________________________

<I | Env> <V | Env>

Page 14: 1 Programming Languages (CS 550) Lecture 9 Summary Introduction to Formal Semantics Jeremy R. Johnson TexPoint fonts used in EMF. Read the TexPoint manual

14

Environments and Assignment

(16) <I ‘:=’ V | Env> Env & {I = V}

(17) <E | Env> <E1 | Env> ______________________________________________________________________________________________________________________

<I ‘:=’ E | Env> <I ‘:=’ E1 | Env>

(18) <S | Env> Env1 ______________________________________________________________________________________________

<S ‘;’ L | Env> <L | Env1>

(19) L < L | Env0>

Page 15: 1 Programming Languages (CS 550) Lecture 9 Summary Introduction to Formal Semantics Jeremy R. Johnson TexPoint fonts used in EMF. Read the TexPoint manual

15

Implementation in Prolog

Configurations<E | Env> Config(E,Env)

Environments[value(I1,v1),...,value(In,vn)]

Predicate to lookup values Lookup(Env,I,V)

Page 16: 1 Programming Languages (CS 550) Lecture 9 Summary Introduction to Formal Semantics Jeremy R. Johnson TexPoint fonts used in EMF. Read the TexPoint manual

16

Implementation in Prolog

Configurations<E | Env> Config(E,Env)

Environments[value(I1,v1),...,value(In,vn)]

Predicate to lookup values Lookup(Env,I,V)

lookup([value(I,V)|_],I,V).

lookup([_|Es],I,V) :- lookup(Es,I,V), !.

Page 17: 1 Programming Languages (CS 550) Lecture 9 Summary Introduction to Formal Semantics Jeremy R. Johnson TexPoint fonts used in EMF. Read the TexPoint manual

17

Implementation in Prolog

% reduce_value(config(times(plus(x,3),minus(5,y)),[value(x,2),value(y,1)]),V).

% V = config(20,[value(x,2),value(y,1)]) ?

reduce(config(plus(E,E2),Env),config(plus(E1,E2),Env)) :-

reduce(config(E,Env),config(E1,Env)).

reduce(config(I,Env),config(V,Env)) :- atom(I), lookup(Env,I,V).

reduce_all(config(V,Env),config(V,Env)) :- integer(V), !.

reduce_all(config(E,Env),config(E2,Env)) :-

reduce(config(E,Env),config(E1,Env)), reduce_all(config(E1,Env),config(E2,Env)).

reduce_value(config(E,Env),V) :- reduce_all(config(E,Env),config(V,Env)).

Page 18: 1 Programming Languages (CS 550) Lecture 9 Summary Introduction to Formal Semantics Jeremy R. Johnson TexPoint fonts used in EMF. Read the TexPoint manual

18

If Statements

(20) <E | Env> <E1| Env> __________________________________________________________________________________________________________________________________

<‘if’ E ‘then’ L1 ‘else’ L2 ‘fi’ | Env> <‘if’ E1 ‘then’ L1 ‘else’ L2 ‘fi’ | Env>

(21) V > 0 ______________________________________________________________________________________________________________________________

<‘if’ V ‘then’ L1 ‘else’ L2 ‘fi’ | Env> < L1|Env>

(22) V 0 _____________________________________________________________________

<‘if’ V ‘then’ L1 ‘else’ L2 ‘fi’ | Env> < L2|Env>

Page 19: 1 Programming Languages (CS 550) Lecture 9 Summary Introduction to Formal Semantics Jeremy R. Johnson TexPoint fonts used in EMF. Read the TexPoint manual

19

While Statements

(23) <E | Env> <V| Env>, V 0 ________________________________________________________________________________________________________________

<‘while’ E ‘do’ L ‘od’|Env> Env

(24) <E | Env> <V| Env>, V > 0 _____________________________________________________________________________________________________________

<‘while’ E ‘do’ L ‘od’|Env> <L;‘while’ E ‘do’ L ‘od’|Env>

Page 20: 1 Programming Languages (CS 550) Lecture 9 Summary Introduction to Formal Semantics Jeremy R. Johnson TexPoint fonts used in EMF. Read the TexPoint manual

20

Implementation in Prolog% Test cases:

% reduce_exp_all(config(plus(times(2,5),minus(2,5)),[]),V).

% V = config(7,[])

% reduce_exp_all(config(plus(times(x,5),minus(2,y)),[value(x,2),value(y,5)]),V).

% V = config(7,[value(x,2),value(y,5)])

% reduce_all(config(seq(assign(x,3),assign(y,4)),[]),Env).

% Env = [value(x,3),value(y,4)]

% reduce(config(if(3,assign(x,3),assign(x,4)),[]),Env).

% Env = [value(x,3)]

% reduce(config(if(0,assign(x,3),assign(x,4)),[]),Env).

% Env = [value(x,4)]

% reduce_all(config(if(n,assign(i,0),assign(i,1)),[value(n,3)]),Env).

% Env = [value(n,3),value(i,0)]

Page 21: 1 Programming Languages (CS 550) Lecture 9 Summary Introduction to Formal Semantics Jeremy R. Johnson TexPoint fonts used in EMF. Read the TexPoint manual

21

Implementation in Prolog% reduce_all(config(while(x,assign(x,minus(x,1))),[value(x,3)]),Env).

% Env = [value(x,0)]

% reduce_all(config(

% seq(assign(n,minus(0,3)),

% seq(if(n,assign(i,n),assign(i,minus(0,n))),

% seq(assign(fact,1),

% while(i,seq(assign(fact,times(fact,i)),assign(i,minus(i,1)))))))

% ,[]),Env).

% Env = [value(n,-3),value(i,0),value(fact,6)]

Page 22: 1 Programming Languages (CS 550) Lecture 9 Summary Introduction to Formal Semantics Jeremy R. Johnson TexPoint fonts used in EMF. Read the TexPoint manual

22

Denotational Semantics

Use functions to describe semantics of a programming language. Associate semantic value to syntactically

correct constructMap syntactic domain to semantic domainVal: Expression IntegerVal(2 + 3*4) = 1414 “denotes” the value of the expression 2+3*4

P: Program (Input Output)Program Input Output [right associate]

Page 23: 1 Programming Languages (CS 550) Lecture 9 Summary Introduction to Formal Semantics Jeremy R. Johnson TexPoint fonts used in EMF. Read the TexPoint manual

23

Denotational Semantics of Mini Language (Expressions)

E: Expression Environment Integer

E[[E1 ‘+’ E2]](Env) = E[[E1]](Env) + E[[E2]](Env)

E[[E1 ‘-’ E2]](Env) = E[[E1]](Env) - E[[E2]](Env)

E[[E1 ‘*’ E2]](Env) = E[[E1]](Env) * E[[E2]](Env)

E[[I]](Env) = Env(I)E[[N]](Env) = N

Page 24: 1 Programming Languages (CS 550) Lecture 9 Summary Introduction to Formal Semantics Jeremy R. Johnson TexPoint fonts used in EMF. Read the TexPoint manual

24

Implementation in Scheme

(define (exprE expr)

(cond ((number? expr) (numE expr))

((ident? expr) (identE expr))

((plus? expr) (plusE expr))

((minus? expr) (minusE expr))

((times? expr) (timesE expr))

(else (error "illegal expression"))))

(define (env exp)

(if (eq? exp 'x) 3 'undef))

((exprE '(+ 2 x)) env)

;Value: 5

(define (numE expr)

(lambda (env) expr))

(define (identE expr)

(lambda (env) (env expr)))

(define (plusE expr)

(lambda (env)

(let ((expr1 (cadr expr)) (expr2 (caddr expr)))

(+ ((exprE expr1) env) ((exprE expr2) env)))))

Page 25: 1 Programming Languages (CS 550) Lecture 9 Summary Introduction to Formal Semantics Jeremy R. Johnson TexPoint fonts used in EMF. Read the TexPoint manual

25

Denotational Semantics of Mini Language (Expressions)

P: Program EnvironmentP[[L]] = L[[L]](Env0)

L: Statement-list Environment EnvironmentL[[L1 ‘;’ L2]] = L[[L1]] L[[L2]]

L[[S]] = S[[S]]

Page 26: 1 Programming Languages (CS 550) Lecture 9 Summary Introduction to Formal Semantics Jeremy R. Johnson TexPoint fonts used in EMF. Read the TexPoint manual

26

Implementation in Scheme(define (progP prog)

(if (stmt-list? prog)

(stmt-listL prog)

(error "illegal program")))

(define (stmt-listL stmt-list)

(let ((first-stmt (car stmt-list)) (remaining-stmts (cdr stmt-list)))

(if (null? remaining-stmts)

(stmtS first-stmt)

(compose (stmt-listL remaining-stmts) (stmtS first-stmt)))))

(define (compose f g)

(lambda (x)

(f (g x))))

Page 27: 1 Programming Languages (CS 550) Lecture 9 Summary Introduction to Formal Semantics Jeremy R. Johnson TexPoint fonts used in EMF. Read the TexPoint manual

27

Denotational Semantics of Mini Language (Statements)

S: Statement Environment EnvironmentS[[I ‘:=’ E]](Env) = Env & {I = E[[E]](Env)} S[[‘if’ E ‘then’ L1 ‘else’ L2]](Env) =

if E[[E]](Env) > 0 then L[[L1]](Env) else L[[L2]](Env)

S[[‘while’ E ‘do’ L od’]](Env) = if E[[E]](Env) 0 then Env else S[[‘while’ E ‘do’ L od’]](L[[L]](Env))

Page 28: 1 Programming Languages (CS 550) Lecture 9 Summary Introduction to Formal Semantics Jeremy R. Johnson TexPoint fonts used in EMF. Read the TexPoint manual

28

Implementation in Scheme(define (stmtS stmt)

(cond ((assign-stmt? stmt) (assignS stmt))

((if-stmt? stmt) (ifS stmt))

((while-stmt? stmt) (whileS stmt))

(else (error "illegal statement"))))

(define (assignS stmt)

(let ((ident (cadr stmt)) (expr (caddr stmt)))

(lambda (env)

(lambda (var)

(if (eq? var ident) ((exprE expr) env) (env var))))))

Page 29: 1 Programming Languages (CS 550) Lecture 9 Summary Introduction to Formal Semantics Jeremy R. Johnson TexPoint fonts used in EMF. Read the TexPoint manual

29

Implementation in Scheme(define (ifS stmt)

(let ((expr (cadr stmt)) (S1 (caddr stmt)) (S2 (cadddr stmt)))

(lambda (env)

(if (> ((exprE expr) env) 0)

((stmt-listL S1) env)

((stmt-listL S2) env)))))

(define (whileS stmt)

(let ((expr (cadr stmt)) (S (caddr stmt)))

(lambda (env)

(if (<= ((exprE expr) env) 0)

env

((whileS stmt) ((stmt-listL S) env))))))

Page 30: 1 Programming Languages (CS 550) Lecture 9 Summary Introduction to Formal Semantics Jeremy R. Johnson TexPoint fonts used in EMF. Read the TexPoint manual

30

Implementation in Scheme(define prog

'((assign n (- 0 5))

(if n

((assign i n))

((assign i (- 0 n))))

(assign fact 1)

(while i

((assign fact (* fact i))

(assign i (- i 1))))))

(define env0 (lambda (ident) 'undef))

(env0 'x);Value: undef

(define envf ((progP prog) env0));Value: envf

(envf 'n);Value: -5

(envf 'i);Value: 0

(envf 'fact);Value: 120

Page 31: 1 Programming Languages (CS 550) Lecture 9 Summary Introduction to Formal Semantics Jeremy R. Johnson TexPoint fonts used in EMF. Read the TexPoint manual

31

Axiomatic Semantics

Describe semantics of language constructs by their effect on assertions about the data manipulated by the program. Pre and post conditions

{x = A} x := x + 1 {x = A+1} {y 0} x := 1/y {x = 1/y}

Program specifications { n ≥ 0, 1 i n, a[i] = A[i]} sort-program {sorted(a)

and permutation(a,A)}

Correctness Proofs

Page 32: 1 Programming Languages (CS 550) Lecture 9 Summary Introduction to Formal Semantics Jeremy R. Johnson TexPoint fonts used in EMF. Read the TexPoint manual

32

Weakest Precondition

Given Q, Lots of preconditions P such that{P}C{Q}{x = 3}x := x + 1{x > 0}{x ≥ 3}x := x + 1{x > 0}…{x > -1}x := x + 1{x > 0}

Weakest (or most general) preconditionwp(C,Q){P}C{Q} iff P wp(C,Q)

Page 33: 1 Programming Languages (CS 550) Lecture 9 Summary Introduction to Formal Semantics Jeremy R. Johnson TexPoint fonts used in EMF. Read the TexPoint manual

33

Properties of wp

Law of the Excluded miraclewp(C,false) = false

Distributivity of Conjunctionwp(C,P and Q) = wp(C,P) and wp(C,Q)

Law of MonotonicityIf Q R then wp(C,Q) wp(C,R)

Distributivity of Disjunctionwp(C,P) or wp(C,Q) wp(C,P or Q)Equal if C is deterministic

Page 34: 1 Programming Languages (CS 550) Lecture 9 Summary Introduction to Formal Semantics Jeremy R. Johnson TexPoint fonts used in EMF. Read the TexPoint manual

34

Axiomatic Semantics of Mini Language

Semantics of C is the functionwp(C,_) from assertions to assertionsPredicate transformer

Statement-listwp(L1;L2,Q) = wp(L1,wp(L2,Q))

Assignment Statementswp(I := E,Q) = Q[E/I]wp(x:=x+1,x>0) = (x+1 > 0) = (x > -1)wp(x:=x+1,x=A) = (x+1 = A_ = (x = A-1)

Page 35: 1 Programming Languages (CS 550) Lecture 9 Summary Introduction to Formal Semantics Jeremy R. Johnson TexPoint fonts used in EMF. Read the TexPoint manual

35

Axiomatic Semantics of Mini Language

If Statementswp(if E the L1 else L2 fi,Q) = (E > 0

wp(L1,Q)) and (E 0 wp(L2,Q))

wp(if x then x := 1 else x := -1,x=1)(x > 0 1=1) and (x 0 -1=1) true and (x > 0) x > 0

Page 36: 1 Programming Languages (CS 550) Lecture 9 Summary Introduction to Formal Semantics Jeremy R. Johnson TexPoint fonts used in EMF. Read the TexPoint manual

36

Axiomatic Semantics of Mini Language

While StatementsHi(while E do L od,Q), while executes i

iterations and terminates satisfying Q

H0(while E do L od,Q) = (E 0 Q)

Hi+1(while E do L od,Q) = (E > 0 wp(L,Hi(while E do L od,Q)

wp(while E do L od,Q) = i, Hi(while E do L od,Q)

Page 37: 1 Programming Languages (CS 550) Lecture 9 Summary Introduction to Formal Semantics Jeremy R. Johnson TexPoint fonts used in EMF. Read the TexPoint manual

37

Correctness Proofs

Loop Invariantswhile E do L odFind W such that W wp(while…,Q)

W and (E > 0) wp(L,W) W and (E 0) Q P W

If while loop terminates , W wp(while…,Q)Proves {P}while E do L od{Q}

Page 38: 1 Programming Languages (CS 550) Lecture 9 Summary Introduction to Formal Semantics Jeremy R. Johnson TexPoint fonts used in EMF. Read the TexPoint manual

38

Correctness Proofs

Loop invariant (fact = (i+1)n, i≥0){n > 0}i := n;fact := 1;while i do

fact := fact*i; i := i-1;

od

Page 39: 1 Programming Languages (CS 550) Lecture 9 Summary Introduction to Formal Semantics Jeremy R. Johnson TexPoint fonts used in EMF. Read the TexPoint manual

39

Correctness Proofs

W = (fact = (i+1)n, i≥0)wp(fact := fact*i,i:=i-1,W)= wp(fact := fact*i,wp(i:=i-1,W))= wp(fact := fact*i,fact=((i-1)+1)n, i-1≥0)= wp(fact := fact*i,fact=in, i-1≥0)= (fact*i=in, i-1≥0) = (fact = (i+1)n, i-1≥0)

W and i > 0 wp(L,W)W and i 0 fact = n!n > 0 wp(i:=n,fact := 1,W) = (n ≥ 0)