Context-Free Grammars
4 Sets
V – Variables
Σ – Terminals
P or R – Production or Rules
S – Start Variable/Symbol
A -> x
Comment:
A is a variable
X is any combo of terminals and variables.
Problem 1:
Want: 0011
G = {(S), (0,1), (S -> 0S1 | ɛ)}
S => 0S1 => 00S11 -> 0011
Note:
The => symbol called yields.
The underlined called derivations, S => … => x ∈ Σ* this means from S we apply a bunch
of rules and then we get something in the form x where this is in sigma star.
Σ* means set of all possible strings including empty strings.
x ∈ Σ* means that x is a string composed of zero or more symbols from the
alphabet/terminal Σ.
Problem 2:
G = {(S,A,B), (a,b), (S -> AB | ɛ, A -> aSB, B -> b | SA)}
S => AB
Note:
Leftmost derivation always replace leftmost variable.
Every derivation has an equivalent left most derivation.
Derivation in CFG means replacing variable with non-terminals/terminals.
Problem 3:
G = {(S), (), (S - > SS | ɛ)}
S => ɛ
S => SS => S => ɛ
Note:
Both of them are different leftmost derivations, we called it ambiguous grammar
because the string that grammar makes can be achieved with two ways of derivations.
S -> ɛ this grammar is unambiguous.
Questions:
1) Can we “remove” ambiguity from a CFG?
Answer: No in general.
2) Are there “inherently ambiguous” CFLs? Every CFG is ambiguous. Are there languages
that every grammar is ambiguous?
Answer: Yes
Problem 4:
L = (0 U 1)*
Solution:
G = {(S), (), (S -> ɛ | S0 | S1)}
Answer: The created CFG can be empty, starts with 0 or 1, and ends with 0 or 1.
Problem 5:
L = { 0n1m or 0*1*: n, m >= 0 }
Note:
All leftside should be 0, then rightside must be 1.
Solution:
G = {(S,P), (0,1), (S -> 0S | P, P -> 1P | ɛ)}
Want = 001
S => 0S => 00S => 001P => 001
Problem 6:
L = { palindromes over {0, 1} }
Note:
If you see 1 at the beginning then the end must be 1, same with 0.
G = {(S), (0,1), (S -> 0S0 | S -> 1S1 | 1 | 0 | ɛ)}
Problem 7:
La = L 1 U L 2
Lb = L 1 • L2
Lc = L 1*
Note:
L1, L2 are CFLs
G1, G2 are CFGs
S1,S2 are start variables
Answer:
La = S -> S1 | S2
Lb = S -> S1S2
Lc = S -> ɛ | S1S
Problem 8:
L = { aibjck : i != j }
G = {(S, Ai<j, Ai>j B, C, X), (a,b,c), (S -> Ai<jC | Ai>jC, Ai<j -> aAi<jb | bB, B -> bB | ɛ, C -> cC | ɛ, Ai>j ->
aAi>jb | aX, X -> aX | ɛ)}
S -> Ai<jC | Ai>jC
Ai<j -> aAi<jb | bB
B -> bB | ɛ
C -> cC | ɛ
Ai>j -> aAi>jb | aX
X -> aX | ɛ
Problem 9:
L = { 0n1m2m3n : n, m > 0 }
S -> 0S3 | X
X -> 1x2 | ɛ
Problem 10:
L = { 0n1n2m3m : n, m >= 0 }
S -> X1X2
X1 -> 0X11 | ɛ
X2 -> 2X23 | ɛ
Problem 11:
L = { non-palindromes over { 0, 1} }
S -> 0S0 | 1S1 | 0X1 | 1X0
X -> 0X | 1X | ɛ
Chomsky Normal Form
Important steps:
1. Ensure start variable not on RHS.
2. Eliminate ɛ.
3. Eliminate unit rules (A -> B) one variable on RHS.
4. Ensure RHS is all variables or single terminal (S0 - > ɛ).
5. Reduce “long” RHS’s (>= 3 variables) to 2.
