1. Normal Forms for Context-Free Grammars Definition: A symbol X in V  T is useless in a CFG G=(V, T, P, S) if there does not exist a derivation of the form S   wXy  * wxy where w, x, y are in T *. Definition: A symbol X in V  T is inaccessible in a CFG G = (V, T, P, S) if X does not appear in any sentential form. (inaccessible implies not reachable) Definition: A symbol X is generating if X   w for some w in T *.

2. Algorithm (NF1) (Is L(G) empty?) Input: A CFG G = (V, T, P, S). Output: “YES” if L(G) = , “NO” otherwise. Method: Construct sets N 0, N 1, …recursively as follows: 1.N 0 =  ; i = 1; 2.N i = {A | A  in P and  in (N i-1  T) * }  N i-1 ; 3. If N i  N i-1 then begin i = i +1; go to 2; end; 4.N e = N i ; 5.If S is in N e then output “NO” else output “YES”.

3. Example: S  AA | AB A  SA | AB B  BB | b

4. Algorithm (NF2): (Removal of inaccessible symbols) Input: A CFG G = (V, T, P, S). Output: CFG G = (V, T, P, S) such that (1) L(G) = L(G), (2) No X in V  T is unreachable. Method: Construct sets N 0, N 1, …recursively as follows: 1.N 0 = {S}; i = 1; 2.N i = {X | some A   X  is in P and A in N i-1 }  N i-1 ; 3.If N i  N i-1 then begin i=i+1; go to 2; end; 4.V = N i  V; T = N i  T; P = those productions of P with symbols from N i ; S = S;

5. Algorithm (NF3): (Useless Symbol Removal) Input: A CFG G = (V, T, P, S). Output: CFG G = (V, T, P, S) such that (1) L(G) = L(G), (2) No X in V  T is useless. Method: 1.Apply algorithm NF1 to G to get N e. 2.Let G 1 = (V  N e, T, P 1, S) where P 1 contains those productions of P involving only symbols from N e  T. 3.Apply algorithm NF2 to G 1 to obtain G = (V, T, P, S)

6. Example: S  a | A A  AB B  b

7. Definitions: 1.A   is called  -production 2.A   is called A-production 3.A  B where A and B are in V is called unit production (or single production). 4.A variable A is nullable if A   . 5.A CFG G = (V, T, P, S) is  -free if either (i) P has no  -productions, or (ii) there is exactly one  -production S   and S does not appear on the right side of any production in P.

8. Algorithm (NF4): Conversion to  -free grammar Input: A CFG G = (V, T, P, S). Output: Equivalent  -free CFG G = (V, T, P, S) Method: 1)Construct N e = {A | A in V and A  +  } 2)If A   0 B 1  1 B 2  2 … B k  k is in P, k  0 and for 1  i  k each B i in N e but no symbol in any  j is in N e, 0  j  k, add to P all productions of the form A   0 X 1  1 X 2  2 … X k  k where X i is either B i or , without adding A   to P 3)If S is in N e add to P the productions S   | S where S is a new symbol and set V = V  {S}. Otherwise set V = V and S =S. 4)G =(V, T, P, S)

9. Example: S  aSbS | bSaS | 

10. Algorithm (NF5): Removal of unit productions Input: An  -free CFG G = (V, T, P, S). Output: Equivalent  -free CFG G = (V, T, P, S) Method 1.Construct for each A in V the sets N A = {B | A  * B} as follows: (a) N 0 = {A}; i = 1; (b) N i = {C | B  C is in P and B in N i-1 }  N i-1 (c)If N i  N i-1 then i=i+1; go to (b); else N A = N i 2. Construct P as follows: If B   is in P and not a unit production, place A   in P for A such that B is in N A. 3. G = (V, T, P, S)

11. Example: E  E + T | T T  T * F | F E  ( E ) | a

12. Definition: A CFG G = (V, T, P, S) is said to be cycle-free if there is no derivation of the form A  * A for any A in V. G is said to be proper it it is cycle-free,  -free, and has no useless symbols.  -free and no unit productions imply cycle-free.

13. Chomsky Normal Form A CFG G = (V, T, P, S) is said to be in Chomsky Normal Form (CNF) if each production in P is one of the forms (1)A  BC with A, B, C in V, or (2)A  a with a in T, or (3)If  is in L(G), then S   is a production and S does not appear on the right side of any production

14. Algorithm (NF6) Conversion to CNF Input: A proper CFG G=(V, T, P, S) with no single production Output: A CFG G 1 =(V 1, T, P 1, S) such that L(G) = L(G 1 ) Method: From G, construct G 1 as follows: 1)Add each production of the form A → a in P to P 1 2)Add each production of the form A → BC in P to P 1 (A, B, C are non terminals)

15. 3) If S →  is in P, add S →  to P 1 4) For each production of the form A → X 1 …X k in P, k > 2, add to P 1 the following productions: A → Y 1 → Y 2 … → Y k-1 Y k where Y j stands for X i if X i is a nonterminal otherwise it is a new nonterminal added to V 1

16. 5) For each production of the form A → X 1 X 2 where either X 1 or X 2 or both are terminals add to P 1 the production A → Y 1 Y 2 6) For each terminal a replaced by a Y, add the production Y → a

17. Example: S  bA | aB A  bAA | aS | a B  aBB | bB | b

18. Greibach Normal Form Recursive, left-recursive, right-recursive: A nonterminal A in a CFG G = (V,T,P,S) is said to be recursive if A  +  A  for some  and . If  = , then A is left-recursive, if  = , then A is right-recursive. A grammar with at least one left-(right-) recursive nonterminal is said to be left- (right-) recursive

19. Lemma: Let G = (V,T,P,S) be a CFG in which A → A  1 | …|A  m |  1 | … |  n are all the A-productions in P and no  begins with A. Let G 1 = (V  {B}, T, P 1, S) where P 1 is P with the above productions are replaced by A →  1 | …|  n |  1 B| … |  n B B →  1 | … |  m |  1 B | …|  m B Then L(G) = L(G 1 )

20. Algorithm – elimination of left recursion. Input: A proper CFG G=(V,T,P,S) Output: A CFG G 1 with no left-recursion Method: Let V = {A 1,..., A n }. Transform G so that if A i →  is a production, then  begins with either a terminal of a symbol A j such that j > i. this can be accomplished by substitution for lower numbered productions and eliminating immediate left-recursion.

21. Example: A → BC | a B → CA | Ab C → AB | CC | a

22. Definition: A CFG G = (V,T,P,S) is said to be in Greibauch normal form (GNF) if G is  - free and each non  -production is of the form A → a  where a is a terminal and  is a string in V*.

23. Lemma: Let G=(V,T,P,S) be a non-left recursive grammar. Then there is linear order < on V such that A → B  is in P, the A < B. Proof: Let R be a relation A R B if and only if A  * B  for some . R is a partial order Extend R to a linear order

24. Algorithm: Conversion to GNF Input: G=(V,T,P,S), G non-left-recursive, proper CFG Output: A CFG G 1 in GNF Method: 1) Construct a linear order on V 2) Set i= n-1 3) If i=0 go to step 5. Otherwise replace each production of the form A i → A j , j>i, by A i →  1  | …|  m  where A j →  1 | …|  m are all the A j -productions (  i ’s begin with terminal)

25. 4) Set i=i-; go to step 2 5) For each production A → aX 1 …X k, if X j is a terminal replace it by a nonterminal Y j and add a production Y j → X j E → T | TE 1 E 1 → +T | +TE 1 T → F | FT 1 T 1 → *F | *FT 1 F → (E) | a