Information Technology Reference
In-Depth Information
τ 1
τ 2
τ 3
τ j
τ n
1
1
1
1
x 1
x 2
x i = 1
x n
Figure 9.9 When x i = 1thereisapath P to τ 1 such that each gate on P has value 1.
We now derive a stronger result: we show that every monotone circuit for binary merging
has a size that is Ω( n log n ) . Binary merging is realized by a function f ( n )
n
n , n =
merge :
B
→B
2 k , defined as follows: given two sorted binary k -tuples x and y , the value of f ( n )
merge ( x , y )
is the n -tuple that results from sorting the n -tuple formed by concatenating x and y .Thus,
a binary merging circuit can be obtained from one for sorting simply by restricting the values
assumed by inputs to the sorting circuit. (Binary merging is a subfunction of binary sorting.)
It follows that a lower bound on C Ω mon
f ( n )
merge is a lower bound on C Ω mon
f ( n )
sort .
THEOREM 9.6.2 Let n be even. Then the monotone circuit size for f ( n )
n
n
merge :
B
→B
satisfies
the following bounds:
C Ω mon f ( n )
merge = O ( n log n )
( n/ 2 )log 2 n
O ( n )
merge follows from the construction given in The-
orem 6.8.2 after max and min comparison operators are replaced by AND sand OR s, respec-
tively.
Let k = n/ 2. The function f ( n )
f ( n )
Proof The upper bound on C Ω mon
merge operates on two k -tuples x and y to produce the
merged result f ( n )
merge ( x , y ) ,where x and y are in descending order; that is, x 1
x 2
···≥
x k and y 1
y 2
≥···≥
y k . As stated above for binary sorting, the output functions
are τ 1 , τ 2 , ... , τ n .
Let x 1 = x 2 =
···
= x r− 1 = 1, x r + 1 =
···
= x k = 0, y 1 = y 2 =
···
= y s = 1,
and y s + 1 =
= y k = 0. Let x r be unspecified. Since the circuit is monotone, the value
computed by each gate circuit is 0, 1, or x r .Also,
···
t<r + s
1
τ t ( x , y )=
x r
t = r + s
t>r + s
0
It follows that there must be a path P ( r + s r of gates from the input labeled x r to the
output labeled τ r + s such that each gate output is x r .If x r = 0, since the components of x
 
Search WWH ::




Custom Search