Molecular Solutions For The Set-Partition Problem On Dna-Based Computing

International Journal of Computer Science & Information Technology (IJCSIT) Vol 5, No 6, December 2013

Molecular Solutions For The Set-Partition Problem
On Dna-Based Computing
Sientang Tsai and Wei-Yeh Chen
Department of Information Management, Southern Taiwan University of Science and
Technology, Yuan Kung District, Tainan City, Taiwan, R.O.C.

ABSTRACT
Consider that the every element in a finite set S having q elements is a positive integer. The set-partition
problem is to determine whether there is a subset T ⊆ S such that

∑x = ∑x, where T
x ∈T

= {x| x ∈ S and

x ∈T

x ∉ T}. This research demonstrates that molecular operations can be applied to solve the set-partition
problem. In order to perform this goal, we offer two DNA-based algorithms, an unsigned parallel adder
and a parallel Exclusive-OR (XOR) operation, that formally demonstrate our designed molecular solutions
for solving the set-partition problem.

KEYWORDS
DNA-based Computing, the Set-Partition Problem, the NP-Complete Problems, the NP-Hard Problems.

1. Introduction
Feynman first offered bio-molecular computation in 1961, but his idea was not implemented by
experiments for a few decades [16]. After almost thirty years later, Adleman finally achieved his
experiment of the Hamiltonian path problem by manipulating DNA strands in a test tube [1].
From [19], that an optimal solution of every NP-complete or NP-hard problem is determined from
its characteristics is manifested. DNA-based algorithms had been proposed to solve many
computational problems, and those consisted of the satisfiability problem [21], the maximal
clique [23], three-vertex-coloring [5], the subset-sum problem [11], the maximum cut problem
[29], and the binary integer programming problem [30]. One potentially significant area of
application for DNA algorithms is the breaking of encryption schemes [9]. From [18], DNAbased arithmetic algorithms are proposed, and from [25] DNA-based algorithms for constructing
DNA databases are also offered. Here we use the molecular operations in the Adleman-Lipton
filtering model to develop the DNA-based algorithms for solving the set-partition problem. We
also construct an unsigned parallel adder and a parallel Exclusive-OR (XOR) operation in the
procedure of demonstration.

2. DNA Computation
In this section we describe the available techniques for dealing with DNA strands that will be
used to solve the set-partition problem.

DOI : 10.5121/ijcsit.2013.5602

21


2.1. Descriptions of Molecular Operations in DNA Manipulations
There have been revolutionary advances in the field of biomedical engineering particularly in
recombinant DNA and RNA manipulating in the last decade. Due to the industrialization of the
biotechnology field, laboratory techniques for recombinant DNA and RNA manipulation are
becoming highly standardized. Basic principles about recombinant DNA can be found in [6, 7, 15,
28]. In this subsection we describe eight biological operations useful for solving the set-partition
problem. The method of constructing DNA solution space for the set-partition problem is based
on the proposed method in [3, 10].
The filtering model developed by Adleman is memory less in the sense that the strings do not
change during a computation. A computation composes of a sequence of operations on finite
multi-set of strings containing the alphabet {A, C, G, T}. A test tube is a finite multi-set of strings.
This work is based on the following operations:
1. Extract: Consider a test tube P and a short DNA substring x, create two new tubes + (P, x) and
− (P, x), where + (P, x) consists of all strands in P containing x as a sequence, while − (P, x)
produces all strands in P not containing x as a sequence.
2. Merge: Take two test tubes P1 and P2, produce their union P1 ∪ P2, and place the result into the
tube ∪(P1, P2), where ∪(P1, P2) = P1 ∪ P2.
3. Detect: Pick a test tube P and output ‘yes’ if P contains at least one DNA molecule; otherwise
output ‘no’.
4. Amplify: Start with a test tube P, this operation, Amplify (P, P1, P2), will duplicate two new
copies P1 and P2 of P and P becomes an empty tube.
5. Append: Take a test tube P and a short strand x, this operation will create a tube that contains
all strands having string x at the end of every strand in P.
6. Append-head: Consider a test tube P and a short strand x, and generate a test tube that
comprises all strands having string x at the beginning of every strand in P.
7. Read: this operation describes each of the resulting solutions contained in tube P. If P is empty,
then no answer is found.

3. DNA Algorithms for Solving the Set-Partition Problem
3.1. Definition of the Set-Partition Problem
Assume that a finite set S is {s1, s2,…,sq}, where sm is the mth element for 1≤ m ≤ q. Also consider
every element in S is a positive integer. The set-partition problem is to decide if there is a subset T
⊆ S such that

∑ x = ∑ x, where T = {x| x ∈ S and x ∉ T}. The set-partition problem has been
x ∈T

x ∈T

verified to be the NP-complete problem [13, 17].
Consider that a finite set S = {1, 2, 3}. The complete subsets for S are ∅, {1}, {2}, {3}, {1, 2},
{2, 3}, {1, 3} and {1, 2, 3} respectively. According to the definition above of the set-partition
−

−

−

problem, two subsets T and subset T are disjoint, i.e. T ∩ T = ∅, and they form S, i.e. T ∪ T = S.
−

−

Therefore, the set S has four partitions: (1) T ={1,3} and T ={2}; (2) T ={2,3} and T ={1}; (3) T
−

−

−

={3} and T ={1, 2}; (4) T = {1, 2, 3} and T = ∅. Subsequently, the sum for each pair (T, T ) is

22


(4, 2), (5, 1), (3, 3) and (6, 0). According to the definition of Set-Partition Problem, the solution
−

for a finite set S is T = {3} and

T

= {1, 2}.

3.2. Creating the Solution Space of DNA Strands
Suppose that xq xq −1… x2 x1 is a q-bit binary number applied to represent q elements in a finite set
S, where the value of each bit xm is either 1 or 0 for 1 ≤ m ≤ q. From [3, 10], for every bit xm
representing the mth element in S for 1≤ m ≤ q, two distinct 15-base value sequences are designed.
One represents the value “0” for xm and the other represents the value “1” for xm. For sake of
convenience in our representation, assume that x1 , which corresponding element sm belongs to
m
0
the subset of S, denotes the value of xm to be 1 and x m , which corresponding element sm does not
belong to the subset of S, denotes the value of xm to be 0. The following DNA-based algorithm is
used to construct solution the space of DNA strands for solving the set-partition problem of a qelement set S.

Procedure Init (T0, q)
(1) Append(T1, x1 −1 )
q
0
(2) Append(T2, xq −1 )

(3) T0 = ∪(T1, T2)
(4) For m = q − 2 downto 1
(4a) Amplify(T0, T1, T2)
(4b) Append(T1, x1 )
m
0
(4c) Append(T2, x m )

(4d) T0 = ∪(T1, T2)
EndFor
(5) Append-head(T0, x1 )
q
EndProcedure
Consider that a finite set S is equal to {001, 010, 011}. The number of elements in the finite set S,
q is three. When the algorithm, Init (T0, q), is called from step (1) of Algorithm 1 in Subsection
3.8, it is applied to construct the solution space of DNA strands. Tube T0 is an empty tube. We
think of it as an input tube for Init (T0, q) and the number of elements in the finite set S, q, as the
second parameter for Init (T0, q). After the execution of Step(1) and step (2) are performed, tube
0
0
T1= { x1 } and tube T2 = { x2 }. Then, the merge operation makes tube T0 = { x1 , x2 }, tube T1= ∅,
2
2
and tube T2 = ∅ after step (3).
Because the value of q is three, step (4a) through (4d) will be executed one time. We perform
0
0
Amplify operation in step (4a), then tube T0 = ∅, tube T1= { x1 , x2 }, and tube T2 = { x1 , x2 }.
2
2
Next, after the execution of step (4b) and the execution of step (4c) are performed, tube T1 =
1
0 1
0
{ x1 x1 , x2 x1 } and tube T2 = { x1 x10 , x2 x10 }. Then, after having finished step (4d), the merge
2
2
1
0
1
0
operation makes tube T0 = { x1 x 1 , x2 x 1 , x1 x 10 , x2 x10 }, tube T1= ∅ and tube T2 = ∅. Finally,
2
2

1
1 0
1
1 0
1
1
after the execution of step (5) is done, tube T0 = { x 3 x1 x 1 , x 3 x2 x 1 , x 3 x1 x 10 , x 3 x2 x10 }.
2
2

23


The result is shown in Table 1. Lemma 1 is used to demonstrate the correctness of the algorithm
Init (T0, q).
Table 1. Init(T0, q) procedure produces the following results.

Tube
T0

The result generated by Init(T0, q)
1
1 0
1
1 0
1
1
{ x 3 x1 x 1 , x 3 x2 x 1 , x 3 x1 x 10 , x 3 x2 x10 }
2
2

T1

∅

T2

∅

Lemma 1: The Init (T0, q) procedure creates the solution space of DNA strands for a q-element
set S.
Proof: The Init (T0, q) procedure is implemented by means of amplify, append, append-head and
merge operations. Step (1) and (2) append the DNA sequences representing value “1” for xq − 1 and
value “0” for xq − 1, respectively, onto the end of every strand in tube T1 and T2. Hence subsets
including the (q − 1)th element appear in T1 and subsets not including the (q − 1)th element appear
in T2. After step (3), tube T0 is merged by tube T1 and T2. This means that DNA strands in T0
include sequences of xq − 1 = 1 and xq − 1 = 0.
Then, each time step (4a) is performed, it uses the amplify operation to copy the content of tube T0
into two new tubes, T1 and T2, which are copies of T0. Tube T0 becomes empty. Step(4b) and step
(4c) are used to subsequently append DNA sequences, respectively, representing the value “1” or
“0” for xm, onto the end of every strand in tube T1 and tube T2. This implies that subsets containing
the mth element appear in tube T1 and subsets not containing the mth element appear in tube T2.
Finally, tube T0 is merged by tube T1 and T2 in step (4d). This indicates that DNA strands in tube
T0 include DNA sequences of xm = 1 and xm = 0. After For loop in step 4 is performed completely,
T0 is comprised of 2q-1 DNA sequences. After step (5) is finished, a DNA sequence, representing
the value “1” for xq, is appended onto the head of every strand in tube T0. We conclude that this
procedure can create 2q-1 partitions of a q-element set S with DNA strands.

3.3. Solution Space of the Value for Every Element of Each Subset for Solving the
Set-Partition Problem of a Finite Set
For the sake of designing a better and simpler DNA-based algorithm for solving the set-partition
problem of a q-element finite set S, suppose that for an element, sm ∈ T, its value is represented as
−

a binary number of n bits, sm,n,, sm,n-1, …, sm,2, sm,1 and for an element, rm ∈ T , its value is
represented as a binary number of n bits, rm, n, rm, n-1, …, rm,2, rm,1. Also suppose that sm, n and rm, n
are the most significant bit, and sm, 1 and rm, 1 are the least significant bit. For every bit sm, k and rm,
k, 1≤ m ≤ q and 1 ≤ k ≤ n, from [3, 10] two distinct DNA sequences are designed. One
corresponds to the value “0” for sm, k and rm, k and the other corresponds to the value “1” for sm, k
1
and rm, k. For the sake of convenience in our representation, assume that s 1 , k and rm ,k denote the
m
0
0
value of sm, k and rm, k to be 1 and sm, k and rm , k define the value of sm, k and rm, k to be 0. The

following algorithm is employed to construct the value of each corresponding element in 2q-1
partitions of a q-element set S.
24


Procedure Value(T0, q, n)
(1) For m = 1 to q
1

(1a) T1 = + (T0, x1 ) and T2= − (T0, xm)
m
(1b) For k = n downto 1
(1c) Append (T1, sm, k)
0
(1d) Append (T2, sm,k )

End For
(1e) T0 = ∪ (T1, T2)
EndFor
(2) For m = 1 to q
1

(2a) T1= + (T0, x1 ) and T2= − (T0, xm)
m
(2b) For k = n downto 1
0
(2c) Append (T1, rm ,k )

(2d) Append (T2, rm k)
End For
(2e) T0 = ∪ (T1, T2)
EndFor
EndProcedure
When the algorithm, Value(T0, q, n), is called from step (2) of Algorithm 1 in Subsection 3.8, it
is used to encode the value of each element in 2q-1 partitions of a q-element set S. We think of
tube T0 with the result shown in Table 1 as an input tube for the algorithm, Value(T0, q, n), the
number of elements in S, q is regarded as the second parameter, and the number of bits for each
element, n, is regarded as the third parameter. Step(1) is the first nested loop and is applied to
−

construct the value of each element in T in each pair of (T, T ). After we perform step (1a), then
tube T1 = {x31x21x11, x31 x20 x11} and tube T2 = {x31 x21 x10, x31 x20 x10}. Then, after the first execution
of step (1c) and (1d) is performed, tube T1 = {x31x21x11s1, 30, x31x20x11s1, 30} and tube T2 = {x31x21x10
s1, 30, x31x20x10s1, 30}. After every execution for step (1c) and (1d) is finished, tube T1 =
{x31x21x11s1,30s1,20s1,11,
x31x20x11s1,30s1,20s1,11}
and
tube
T2 =
{x31x21x10s1,30s1,20s1,10,
1 0 0
0
0
0
x3 x2 x1 s1,3 s1,2 s1,1 }. Then, after the first execution of step (1e) is finished, tube T0 =
{x31x21x11s1,30s1,20s1,11, x31x20x11s1,30s1,20s1,11, x31 x21 x10s1,30s1,20s1,10, x31x20x10s1,30s1,20s1,10}, tube T1 = ∅
and tube T2 = ∅. Next, after each execution of step (1a) through (1e) in the first nested loop is
completed, tube T0 = {x31 x21 x11 s1,30 s1,20 s1,11 s2,30 s2,21 s2,10 s3,30 s3,21 s3,11, x31 x21 x10 s1,30 s1,20 s1,10
s2,30 s2,21 s2,10 s3,30 s3,21 s3,11, x31 x20 x11 s1,30 s1,20 s1,11 s2,30 s2,20 s2,10 s3,30 s3,21 s3,11, x31 x20 x10s1,30 s1,20
s1,10 s2,30 s2,20 s2,10 s3,30 s3,21 s3,11}, tube T1 = ∅ and tube T2 = ∅. Step(2) is the second nested loop
−

−

and is used to construct the value of each element in T in each pair of (T, T ). The contents of
tube T0 are shown as above. When the first execution of step (2a) is performed, tube T1 = {x31 x21
x11 s1,30s1,20s1,11 s2,30 s2,21 s2,10 s3,30 s3,21 s3,11, x31 x20 x11 s1,30 s1,20 s1,11 s2,30 s2,20 s2,10 s3,30 s3,21 s3,11} and
tube T2 = {x31 x21 x10 s1,30 s1,20 s1,10s2,30 s2,21 s2,10 s3,30 s3,21 s3,11, x31 x20 x10s1,30 s1,20 s1,10 s2,30 s2,20 s2,10
s3,30 s3,21 s3,11 }. Then, after the first execution, the second execution and the third execution for
step (2c) and (2d) are finished, “r1,30 r1,20 r1,10” is appended onto the tail of each bit pattern in tube
T1, and “r1,30 r1,20 r1,11” is appended onto the tail of each bit pattern in tube T2. Next, after the first
execution of step (2e) is performed, tube T0 = {x31 x21 x11 s1,30 s1,20 s1,11 s2,30 s2,21 s2,10 s3,30 s3,21 s3,11
r1,30 r1,20 r1,10, x31 x21 x10 s1,30 s1,20 s1,10 s2,30 s2,21 s2,10 s3,30 s3,21 s3,11, r1,30 r1,20 r1,11, x31 x20 x11 s1,30 s1,20
25


s1,11 s2,30 s2,20 s2,10 s3,30 s3,21 s3,11, r1,30 r1,20 r1,10, x31 x20 x10s1,30 s1,20 s1,10 s2,30 s2,20 s2,10 s3,30 s3,21 s3,11 r1,30
r1,20 r1,11}, tube T1 = ∅ and tube T2 = ∅. Then, after the second For loop is done, tube T1 = ∅, tube
T2 = ∅ and the contents of tube T0 are shown in Table 2. Lemma 2 is applied to prove the
correctness of the algorithm, Value (T0, q, n).
Table 2. The result generated by Value (T0, q, n).

The result generated by Value(T0, q, n)

Tube
T0

1

1

{x3 x2 x11 s1,30 s1,20 s1,11 s2,30 s2,21 s2,10 s3,30 s3,21 s3,11 r1,30 r1,20 r1,10 r2,30 r2,20
r3,20 r3,10,
x31 x21 x10 s1,30 s1,20 s1,10 s2,30 s2,21 s2,10 s3,30 s3,21 s3,11, r1,30 r1,20 r1,11 r2,30 r2,20
r3,20 r3,10,
x31 x20 x11 s1,30 s1,20 s1,11 s2,30 s2,20 s2,10 s3,30 s3,21 s3,11, r1,30 r1,20 r1,10 r2,30 r2,21
r3,20 r3,10,
x31 x20 x10s1,30 s1,20 s1,10 s2,30 s2,20 s2,10 s3,30 s3,21 s3,11 r1,30 r1,20 r1,11 r2,30 r2,21
r3,20 r3,10}

r2,10 r3,30
r2,10 r3,30
r2,10 r3,30
r2,10 r3,30

−

Lemma 2: The value of each element in each pair of (T, T ) for a q-element finite set, S, can be
constructed from the algorithm, Value (T0, q, n).
Proof: Refer to Lemma 1.

3.4. The Implementation of a Parallel One-bit Adder
A one-bit adder has three inputs: the two data input and a carry input. It forms the arithmetic sum
of these inputs. Two of the input bits represent two significant bits to be added. The third input
represents the carry from the previous lower significant position. The least significant bit of the
sum for augend, addend and previous carry comes from first output. The output carry transferred
into the input carry of the next one-bit adder comes from second output. The truth table of a onebit adder is as follows:
Table 3. The truth table of a one-bit adder.

Augend
bit
0
0
0
0
1
1
1
1

Addend
bit
0
0
1
1
0
0
1
1

Previous
carry bit
0
1
0
1
0
1
0
1

Sum
bit
0
1
1
0
1
0
0
1

Carry
bit
0
0
0
1
0
1
1
1

Suppose that two one-bit binary numbers, αm-1, k and αm, k, represent the first input (augend) and
the first output (sum) of a one-bit adder for 1≤ m ≤ q and 1≤ k ≤ n, respectively. A one-bit binary
number, βm, k, is applied to represent the second input (addend) of a one-bit adder. γm, k-1 is used to
represent the third input (previous carry) and γm, k represent the second output (current carry) of a
26


one-bit adder. From [3, 10], two distinct DNA sequences are designed to represent the value “0”
1
and “1” for every corresponding bit. Also suppose that β m , k contains the value of βm, k to be 1 and
0
1
β m ,k contains the value of βm, k to be 0. Also suppose that α m −1, k denotes the value of αm-1, k to be

0
1
1 and α m −1, k defines the value of αm-1, k to be 0. Similarly, suppose that α m , k contains the value of

0
1
αm, k to be 1 and α m,k denotes the value of αm, k to be 0, γ m,k −1 denotes the value of γm, k -1 to be 1
0

and γ m,k −1 contains the value of γm, k-1 to be 0. γm, k1 defines the value of γm, k to be 1 and γm, k0
contains the value of γm, k to be 0. The following algorithm is offered to perform the Boolean
function of a parallel one-bit adder.
Procedure ParallelOneBitAdder(T0, αm-1, k, βm, k, γm, k-1, m, k)
1
1
(1) T1 = +(T0, α m−1,k ) and T2 = ( T0, α m−1,k )

－

1
1
(2) T3 = +(T1, β m, k ) and T4 = −(T1, β m ,k )

1

1

(3) T5 = +(T2, β m ,k ) and T6 = −(T2, β m ,k )
1

1

(4) T7 = +(T3, γ m,k −1 ) and T8 = −(T3, γ m,k −1 )
1

1

(5) T9 = +(T4, γ m,k −1 ) and T10 = −(T4, γ m,k −1 )
1

1

1

1

(6) T11 = +(T5, γ m,k −1 ) and T12 = −(T5, γ m,k −1 )
(7) T13 = +(T6, γ m,k −1 ) and T14 = −(T6, γ m,k −1 )
(8) If (Detect (T7) = = ”yes”) then
1

1
Append-head (T7, α m,k ) and Append-head (T7, γ m,k )

EndIf
0

1

Append-head (T8, α m ,k ) and Append-head (T8, γ m,k )
EndIf
0

1

EndIf
1

0

EndIf
0

1

EndIf
(13) If (Detect (T12) = =”yes”) then
1

0

EndIf
27


1

0

EndIf
0

0

EndIf
(16) T0 = ∪(T7, T8, T9, T10, T11, T12, T13, T14)
EndProcedure
Lemma 3: The ParallelOneBitAdder(T0, αm-1, k, βm, k, γm, k-1, m, k) procedure can be applied to
implement a parallel one-bit adder.
Proof: The algorithm ParallelOneBitAdder(T0, αm-1, k, βm, k, γm, k-1, m, k) is implemented via the
extract, append-head and merge operations. Steps from (1) to (7) employ the extract operation to
form some different test tubes including different strands (T1 to T14). That is T1 includes all of the
strands that have αm-1, k = 1, T2 includes all of the strands that have αm-1, k = 0. T3 includes all of the
strands that have αm-1, k = 1 and βm, k= 1. T4 includes those that have αm-1, k = 1 and βm, k= 0. T5
includes those that have αm-1, k = 0 and βm, k= 1. T6 includes those that have αm-1,k =0 and βm,k=0.
T7 includes those that have αm-1,k =1, βm,k=1, and γm,k-1=1. T8 includes those that have αm-1, k = 1,
βm,k= 1, and γm, k-1= 0. T9 includes those that have αm-1, k = 1, βm, k= 0, and γm, k-1= 1. T10 includes
those that have αm-1, k = 1, βm, k= 0, and γm, k-1= 0. T11 includes those that have αm-1, k = 0, βm, k= 1,
and γm , k-1= 1. T12 includes those that have αm-1, k = 0, βm, k = 1, and γm, k-1 = 0. T13 includes those that
have αm-1, k = 0, βm, k = 0, and γm, k-1 = 1. Finally T14 consists of those that have αm-1, k = 0, βm, k= 0,
and γm, k-1 = 0. After step (1) through step (7) are performed, those eight corresponding results of a
one-bit adder as shown in Table 3 are poured into tube T7 through T14 respectively.
Next, we use step (8) through step (15) to check whether it contains any DNA strand for tubes T7,
T8, T9, T10, T11, T12, T13, and T14 or not. If any a “yes” returned for those steps, then append-head
1
0
0
1
operations will be performed correspondingly. The results are α m ,k or α m ,k , and γ m , k or γ m , k are
appended onto the head of every strand in the corresponding tubes. After performing step (8)
through step (15), we find that one of eight different outputs of a one-bit adder in Table 3 is
correspondingly appended into tubes T7 through T14.The last execution of step (16) applies the
merge operation to pour tubes T7 through T14 into tube T0. T0 contains the strands performing the
addition of three input bits.

3.5. The Implementation of a Parallel N-bit Adder
The parallel one-bit adder introduced in Section 3.4 figures out the arithmetic sum of two bits and
a previous carry. A binary parallel n-bit adder also performs the arithmetic sum for the two input
operands of n-bit and the input carry through performing this one-bit adder n times. The following
algorithm is offered to perform the arithmetic sum for a parallel n-bit adder.
Procedure ParallelAdder(T0, α, β, γ, q, n)
(1) For k = n downto 1
0
(1a) Append (T0, α 0 ,k )

EndFor
28


(2) For m = 1 to q
0
(2a) Append (T0, γ m , 0 )

(2b) For k = 1 to n
(2c) ParallelOneBitAdder(T0, αm-1, k, βm, k, γm, k-1, m, k)
EndFor
EndFor
EndProcedure
When the first execution of ParallelAdder(T0, α, β, γ, q, n) is called from step (3) of Algorithm 1
in Section 3.8, tube T0 with the result in Table 2 in Section 3.3 is regarded as the input tube. From
the second parameter to the fourth parameter are replaced by the actual arguments A, s and z,
respectively. The values for the fifth and sixth parameters, q and n, are both three. Because the
value of n is equal to three, three bits A 00, 3 , A 00, 2 and A 00,1 are appended onto the tail of each bit
pattern in tube T0 after the execution of step (1a). Since the values for n and q are both three, step
(2a) will be executed three times and step (2c) will be executed nine times. The bit, z10, 0 , from the
first execution of step (2a) is appended onto the tail of each bit pattern in tube T0. Next, after the
first execution, the second execution and the third execution for step (2c), that call the algorithm,
ParallelOneBitAdder(T0, αm-1, k, βm, k, γm, k-1, m, k), are performed, tube T0 = {z1,30 A1,30 z1,20A1,20
z1,10 A1,11 x31 x21 x11 s1,30 s1,20 s1,11 s2,30 s2,21 s2,10 s3,30 s3,21 s3,11 r1,30 r1,20 r1,10 r2,30 r2,20 r2,10 r3,30 r3,20 r3,10
A0,30 A0,20 A0,10 z1,00, z1,30 A1,30 z1,20A1,20 z1,10 A1,10 x31 x21 x10 s1,30 s1,20 s1,10 s2,30 s2,21 s2,10 s3,30 s3,21 s3,11,
r1,30 r1,20 r1,11 r2,30 r2,20 r2,10 r3,30 r3,20 r3,10 A0,30 A0,20 A0,10 z1,00, z1,30 A1,30 z1,20A1,20 z1,10 A1,11 x31 x20 x11
s1,30 s1,20 s1,11 s2,30 s2,20 s2,10 s3,30 s3,21 s3,11, r1,30 r1,20 r1,10 r2,30 r2,21 r2,10 r3,30 r3,20 r3,10, A0,30 A0,20 A0,10
z1,00, z1,30 A1,30 z1,20A1,20 z1,10 A1,10 x31 x20 x10 s1,30 s1,20 s1,10 s2,30 s2,20 s2,10 s3,30 s3,21 s3,11 r1,30 r1,20 r1,11
r2,30 r2,21 r2,10 r3,30 r3,20 r3,10 A0,30 A0,20 A0,10 z1,00}. Similarly, after each operation in step (2) is
performed, the result for tube T0 is shown in Table 4.
Table 4. The first execution result generated by ParallelAdder(T0, α, β, γ, q, n).

Tube
T0

The first execution result generated by ParallelAdder(T0, α, β, γ, q, n)
{z3,30 A3,31 z3,21A3,21 z3,11 A3,10 z2,30 A2,30 z2,20A2,21 z2,10 A2,11 z1,30 A1,30 z1,20A1,20 z1,10 A1,11
x31 x21 x11 s1,30 s1,20 s1,11 s2,30 s2,21 s2,10 s3,30 s3,21 s3,11 r1,30 r1,20 r1,10 r2,30 r2,20 r2,10 r3,30 r3,20
r3,10 A0,30 A0,20 A0,10 z1,00 z2,00 z3,00,
z3,30 A3,31 z3,21A3,20 z3,11 A3,11 z2,30 A2,30 z2,20A2,21 z2,10 A2,10 z1,30 A1,30 z1,20A1,20 z1,10 A1,10
x31 x21 x10 s1,30 s1,20 s1,10 s2,30 s2,21 s2,10 s3,30 s3,21 s3,11, r1,30 r1,20 r1,11 r2,30 r2,20 r2,10 r3,30 r3,20
r3,10 A0,30 A0,20 A0,10 z1,00 z2,00 z3,00 ,
x31 x20 x11 s1,30 s1,20 s1,11 s2,30 s2,20 s2,10 s3,30 s3,21 s3,11, r1,30 r1,20 r1,10 r2,30 r2,21 r2,10 r3,30 r3,20
r3,10, A0,30 A0,20 A0,10 z1,00 z2,00 z3,00 ,
x31 x20 x10 s1,30 s1,20 s1,10 s2,30 s2,20 s2,10 s3,30 s3,21 s3,11 r1,30 r1,20 r1,11 r2,30 r2,21 r2,10 r3,30 r3,20
r3,10 A0,30 A0,20 A0,10 z1,00 z2,00 z3,00}

When the second execution of ParallelAdder(T0, α, β, γ, q, n) is called from step (4) of
Algorithm 1 in Section 3.8, tube T0 with the result in Table 4 is regarded as the input tube. From
the second parameter to the fourth parameter are replaced by the actual arguments B, r and y
respectively. The values for the fifth and sixth parameters, q and n, are both three. Because the
value of n is equal to three, three bits B 00, 3 , B 00 , 2 and B 00,1 are appended onto the tail of each bit
pattern in tube T0 after the execution of step (1a). Since the values for n and q are both three, step
29

0
(2a) will be executed three times and step (2c) will be executed nine times. The bit, y1, 0 , from the

first execution of step (2a) is appended onto the tail of each bit pattern in tube T0. Next, after the
first execution, the second execution and the third execution for step (2c), that call the algorithm,
ParallelOneBitAdder(T0, αm-1, k, βm, k, γm, k-1, m, k), are performed, tube T0 = {y1,30 B1,30 y1,20B1,20
y1,10 B1,11 z3,30 A3,31 z3,21A3,21 z3,11 A3,10 z2,30 A2,30 z2,20A2,21 z2,10 A2,11 z1,30 A1,30 z1,20A1,20 z1,10 A1,11 x31 x21
x11 s1,30 s1,20 s1,11 s2,30 s2,21 s2,10 s3,30 s3,21 s3,11 r1,30 r1,20 r1,10 r2,30 r2,20 r2,10 r3,30 r3,20 r3,10 A0,30 A0,20 A0,10
z1,00 z2,00 z3,00 B0,30 B0,20 B0,10 y1,00, y1,30 B1,30 y1,20B1,20 y1,10 B1,11z3,30 A3,31 z3,21A3,20 z3,11 A3,11 z2,30 A2,30
z2,20A2,21 z2,10 A2,10 z1,30 A1,30 z1,20A1,20 z1,10 A1,10 x31 x21 x10 s1,30 s1,20 s1,10 s2,30 s2,21 s2,10 s3,30 s3,21 s3,11,
r1,30 r1,20 r1,11 r2,30 r2,20 r2,10 r3,30 r3,20 r3,10 A0,30 A0,20 A0,10 z1,00 z2,00 z3,00 B0,30 B0,20 B0,10 y1,00, y1,30 B1,30
y1,20B1,20 y1,10 B1,10z3,30 A3,31 z3,21A3,20 z3,11 A3,10 z2,30 A2,30 z2,20A2,20 z2,10 A2,11 z1,30 A1,30 z1,20A1,20 z1,10
A1,11 x31 x20 x11 s1,30 s1,20 s1,11 s2,30 s2,20 s2,10 s3,30 s3,21 s3,11, r1,30 r1,20 r1,10 r2,30 r2,21 r2,10 r3,30 r3,20 r3,10
A0,30 A0,20 A0,10 z1,00 z2,00 z3,00 B0,30 B0,20 B0,10 y1,00, y1,30 B1,30 y1,20B1,20 y1,10 B1,11z3,30 A3,30 z3,20A3,21 z3,10
A3,11 z2,30 A2,30 z2,20A2,20 z2,10 A2,10 z1,30 A1,30 z1,20A1,20 z1,10 A1,10 x31 x20 x10 s1,30 s1,20 s1,10 s2,30 s2,20 s2,10
s3,30 s3,21 s3,11 r1,30 r1,20 r1,11 r2,30 r2,21 r2,10 r3,30 r3,20 r3,10 A0,30 A0,20 A0,10 z1,00 z2,00 z3,00 B0,30 B0,20 B0,10
y1,00}. Similarly, after each operation in step (2) is performed, the result for tube T0 is shown in
Table 5.
Lemma 4: The ParallelAdder(T0 , α, β, γ, q, n) procedure can be applied to perform a binary
parallel adder of n bits.
Proof: Refer to Lemma 1.
Table 5. The second execution result generated by ParallelAdder(T0, α, β, γ, q, n).

Tube
T0

The second execution result generated by ParallelAdder(T0, α, β, γ, q, n)
{y3,30 B3,30 y3,20B3,20 y3,10 B3,10 y2,30 B2,30 y2,20B2,20 y2,10 B2,10 y1,30 B1,30 y1,20B1,20 y1,10 B1,11
z3,30 A3,31 z3,21A3,21 z3,11 A3,10 z2,30 A2,30 z2,20A2,21 z2,10 A2,11 z1,30 A1,30 z1,20A1,20 z1,10 A1,11 x31
x21 x11 s1,30 s1,20 s1,11 s2,30 s2,21 s2,10 s3,30 s3,21 s3,11 r1,30 r1,20 r1,10 r2,30 r2,20 r2,10 r3,30 r3,20 r3,10
A0,30 A0,20 A0,10 z1,00 z2,00 z3,00 B0,30 B0,20 B0,10 y1,00 y2,00 y3,00,
y3,30 B3,30 y3,20B3,20 y3,10 B3,11 y2,30 B2,30 y2,20B2,20 y2,10 B2,11 y1,30 B1,30 y1,20B1,20 y1,10 B1,11z3,30
A3,31 z3,21A3,20 z3,11 A3,11 z2,30 A2,30 z2,20A2,21 z2,10 A2,10 z1,30 A1,30 z1,20A1,20 z1,10 A1,10 x31 x21
x10 s1,30 s1,20 s1,10 s2,30 s2,21 s2,10 s3,30 s3,21 s3,11, r1,30 r1,20 r1,11 r2,30 r2,20 r2,10 r3,30 r3,20 r3,10
A0,30 A0,20 A0,10 z1,00 z2,00 z3,00 B0,30 B0,20 B0,10 y1,00 y2,00 y3,00,
x11 s1,30 s1,20 s1,11 s2,30 s2,20 s2,10 s3,30 s3,21 s3,11, r1,30 r1,20 r1,10 r2,30 r2,21 r2,10 r3,30 r3,20 r3,10
A0,30 A0,20 A0,10 z1,00 z2,00 z3,00 B0,30 B0,20 B0,10 y1,00 y2,00 y3,00,
x10 s1,30 s1,20 s1,10 s2,30 s2,20 s2,10 s3,30 s3,21 s3,11 r1,30 r1,20 r1,11 r2,30 r2,21 r2,10 r3,30 r3,20 r3,10 A0,30
A0,20 A0,10 z1,00 z2,00 z3,00 B0,30 B0,20 B0,10 y1,00 y2,00 y3,00}

3.6. Constructing the Parallel One-bit XOR Operation on Bio-molecular Computing
The Exclusive-OR (XOR) operation of a bit for Boolean variables A and B generates an output of
1 if both A and B have different values and 0 if they are equal. The ⊕ symbol represents the XOR
operation. The four possible combinations for the XOR operation are 0 ⊕ 0 = 0, 0 ⊕ 1 = 1, 1 ⊕ 0
= 1, 1 ⊕ 1 = 0. A truth table is usually used with logic operation to represent all possible
30


combinations of inputs and the corresponding outputs. The truth table for the XOR operation is
shown in Table 6.
Table 6. Truth table for the XOR operation of a bit.

Input
A
0
0
1
1

B
0
1
0
1

output
C=A⊕B
0
1
1
0

Assume that two one-bit binary numbers, Aq, k and Bq, k for 1 ≤ k ≤ n, are used to respectively
represent the first input and the second input for the XOR operation of a bit. Also assume that Cq,
k for 1 ≤ k ≤ n, is used to represent the output for the XOR operation of a bit. For the sake of
convenience, assume that Aq, k1 denotes that the value of Aq, k is 1 and Aq, k0 denotes that the value
of Aq, k is 0. Similarly, suppose that Bq ,k1 denotes that the value of Bq, k is 1 and Bq, k0 denotes that
the value of Bq, k is 0. Assume that Cq, k1 denotes that the value of Cq, k is 1 and Cq, k 0 denotes that
the value of Cq, k is 0. The following algorithm is used to perform the parallel one-bit XOR
operation.
Procedure ParalleOneBitXOR(T0, Aq, k, Bq, k, q, k)
(1) T1 = +(T0, Aq, k1) and T2 = −(T0, Aq, k 1)
(2) T3 = +(T1, Bq, k 1) and T4 = −(T1, Bq, k 1)
(3) T5 = +(T2, Bq, k 1) and T6 = −(T2, Bq, k 1)
(4) If (Detect(T3) = = “yes”) then
(4a) Append-head(T3, Cq, k0)
EndIf
(5a) Append-head (T4, Cq, k1)
EndIf
(6a) Append-head(T5, Cq, k1)
EndIf
(7a) Append-head(T6, Cq, k 0)
EndIf
(8) T0 = ∪(T3, T4, T5, T6)
EndProcedure
Lemma 5: The ParallelOneBitXOR(T0, Aq, k, Bq, k, q, k) procedure can be used to implement the
parallel XOR operation of one-bit.
Proof: The algorithm, ParallelOneBitXOR(T0, Aq, k, Bq, k, q, k), is implemented by the extract,
detect, merge and append-head operations. Steps from (1) to (3) employ the extract operations to
yield different tubes consisting of different inputs (T1 to T6). This implies, T1 includes all of the
inputs that have Aq, k = 1, T2 contains all of the inputs that have Aq, k = 0, T3 includes those inputs
that have Aq, k = 1 and Bq, k = 1, T4 consists of those inputs that have Aq, k = 1 and Bq,k = 0, T5
comprises of those inputs that have Aq, k = 0 and Bq, k = 1, and T6 includes those that have Aq,k = 0
and Bq, k = 0. After having performed separation operation from step (1) to (3), the results of
XOR operation shown in Table 6 are poured into tubes T3 through T6 respectively.
31


Step (4a) through (7a) are applied to check whether it contains any input for tubes T3, T4, T5, and
T6 or not respectively. If any a “yes” returned for those steps, then append-head operations will be
performed correspondingly. Because tubes T3, T4, T5 and T6, subsequently, contains the input for
the fourth row, the third row, the second row and the first row in Table 6. Cq, k0 is appended onto
the head of every input in tubes T3 and T6 if the detection result from step (4) or step (7) is yes. Cq,
1
k is subsequently appended onto the head of every input in tubes T4 and T5 when the detection
result from step (5) or step (6) is yes. After performing step (4) through (7), we discover that one
of four different outputs for the XOR operation of a bit as shown in Table 6 is correspondingly
appended into tubes T3 through T6. The last execution of step (8) uses the merge operation to pour
tubes T3 through T6 into tube T0. Tube T0 contains the result finishing the XOR operation of a bit
as shown in Table 6.

3.7. Constructing the Parallel N-bit XOR Operation on Bio-molecular Computing
Simultaneously, the parallel XOR operation of n bits generates the corresponding n-bit
outputs for XOR operation with two n-bit Boolean variables A, represented by Aq,nAq,n-1…Aq,2Aq,1,
and B, represented by Bq,nBq,n-1…Bq,2Bq,1. The following algorithm is proposed to perform the
parallel n-bit XOR operation for 2q-1 partitions of a q-element set S.
Procedure ParallelXOR(T0, A, B, q, n)
(1) For k = 1 to n
(1a) ParallelOneBitXOR(T0, Aq, k, Bq, k, q, k).
EndFor
EndProcedure
When the algorithm, ParallelXOR(T0, A, B, q, n), is invoked from step (5) in Algorithm 1, tube
T0 with the result in Table 5 is regarded as the input tube. The second parameter is replaced by the
total sum, A, for the value of each element in each T and the third parameter is replaced by the
−
total sum, B, for the value of each element in each T . The fourth parameter q and the fifth
parameter n are both three. Because the value of n is equal to three, step (1a) will be executed
three times. After the first execution of step (1a), that calls the algorithm, ParallelOneBitXOR(T0,
Aq, k, Bq, k, q, k), is finished, tube T0 = {C3,10 … B3,10 … A3,10… x31x21 x11 …, C3,10… B3,11 … A3,11 …
x31 x21 x10 …, C3,10 … B3,10… A3,10 … x31 x20x11 …, C3,10 … B3,11… A3,11 … x31 x20 x10 …}. Then,
after performing each execution of step (1a), the result for tube T0 is shown in Table 7. Lemma 6
is applied to prove the correctness of the algorithm, ParallelXOR(T0, A, B, q, n).
Lemma 6: The procedure, ParallelXOR(T0, A, B, q, n), can be used to finish the parallel XOR
operation of n bits.
Proof: Tube T0 is generated from the algorithm, ParallelAdder(T0, α, β, γ, q, n) and contains
−

those DNA strands representing the individual sum of two disjoint subsets, T and T in each pair
−

of (T and T ) in 2q-1 partitions of a q-element set S. Step (1) is the single loop and is mainly
applied to perform the function of parallel XOR operation of n bits for 2q-1 partitions of a qelement set S. Each execution of step (1a) calls ParallelOneBitXOR(T0, Aq, k, Bq, k, q, k) to finish
the XOR operation for the kth bit of two operands, A and B, in tube T0 . After repeating the
execution of step (1a) until the most significant bit of each operand, A and B, is processed, tube T0
includes the result of performing the parallel XOR operation of n bits. In other words, after
finishing the execution of single loop, tube T0 contains the strands representing the results of
parallel XOR operations for 2q-1 partitions of a q-element set S.
32


Table 7. The result generated by ParallelXOR(T0, A, B, q, n).

Tube

The result generated by ParallelXOR(T0, A, B, q, n)

T0

{C3,31 C3,21C3,10 y3,30 B3,30 y3,20B3,20 y3,10 B3,10 y2,30 B2,30 y2,20B2,20 y2,10 B2,10 y1,30 B1,30
y1,20B1,20 y1,10 B1,11 z3,30 A3,31 z3,21A3,21 z3,11 A3,10 z2,30 A2,30 z2,20A2,21 z2,10 A2,11 z1,30 A1,30
z1,20A1,20 z1,10 A1,11 x31 x21 x11 s1,30 s1,20 s1,11 s2,30 s2,21 s2,10 s3,30 s3,21 s3,11 r1,30 r1,20 r1,10
r2,30 r2,20 r2,10 r3,30 r3,20 r3,10 A0,30 A0,20 A0,10 z1,00 z2,00 z3,00 B0,30 B0,20 B0,10 y1,00 y2,00 y3,00,
C3,31 C3,20C3,10 y3,30 B3,30 y3,20B3,20 y3,10 B3,11 y2,30 B2,30 y2,20B2,20 y2,10 B2,11 y1,30 B1,30
y1,20B1,20 y1,10 B1,11z3,30 A3,31 z3,21A3,20 z3,11 A3,11 z2,30 A2,30 z2,20A2,21 z2,10 A2,10 z1,30 A1,30
z1,20A1,20 z1,10 A1,10 x31 x21 x10 s1,30 s1,20 s1,10 s2,30 s2,21 s2,10 s3,30 s3,21 s3,11, r1,30 r1,20 r1,11
r2,30 r2,20 r2,10 r3,30 r3,20 r3,10 A0,30 A0,20 A0,10 z1,00 z2,00 z3,00 B0,30 B0,20 B0,10 y1,00 y2,00
y3,00,
z1,20A1,20 z1,10 A1,11 x31 x20 x11 s1,30 s1,20 s1,11 s2,30 s2,20 s2,10 s3,30 s3,21 s3,11, r1,30 r1,20 r1,10
r2,30 r2,21 r2,10 r3,30 r3,20 r3,10, A0,30 A0,20 A0,10 z1,00 z2,00 z3,00 B0,30 B0,20 B0,10 y1,00 y2,00
y3,00,
y3,00 }

3.8. A DNA Algorithm for Solving the Set Partition Problem
The following algorithm is used to solve the set-partition problem of a q-element set S. Notations
used in the following algorithm are denoted in the previous sections.
Algorithm 1: Solving the set-partition problem.
(1) Init (T0, q)
(2) Value (T0, q, n)
(3) ParallelAdder(T0, A, s, z, q, n)
(4) ParallelAdder(T0, B, r, y, q, n)
(5) ParallelXOR(T0, A, B, q, n)
(6) For k = 1 to n
(6a) T1 = + (T0, Cq, k0) and T2= −(T0, Cq, k0)
(6b) T0 = ∪(T1, T0)
(6c) Discard (T2)
EndFor
(7) If (Dectect (T0) = = ”yes”) then
(7a) Read(T0)
EndIf
EndAlgorithm
Consider that Algorithm 1 is used to solve the set-partition problem of a set S with {001, 010,
011}. When step (1) is performed, it calls the algorithm, Init (T0, q). The first actual argument,
tube T0, is an empty tube and the second actual argument, q is equal to three. After all operations
in Init (T0, q) is performed, the result for tube T0 is shown in Table 1. Then, step (2) calls the
33


algorithm, Value (T0, q, n). Tube T0 with the result in Table 1 is regarded as the first actual
argument, and the second actual argument and the third actual argument are equal to three. After
each operation in Value (T0, q, n) is performed, the result for tube T0 is shown in Table 2.
Next, step (3) calls the algorithm, ParallelAdder (T0, A, s, z, q, n). Tube T0 with the result in
Table 2 is regarded as the first actual argument. The second actual argument A is used to represent
the total sum for the value of each element in T, which is in the four partitions of a three-element
set S. The third actual argument s is used to represent the value of each element in T. The fourth
actual argument z is used to represent the carry for performing addition of each element in T. The
fifth actual argument q and the sixth actual argument n are equal to three. After each operation in
ParallelAdder(T0, A, s, z, q, n) is performed, the result for tube T0 is shown in Table 4.
Then, step (4) calls the algorithm, ParallelAdder(T0, B, r, y, q, n). Tube T0 with the result in
Table 4 is regarded as the first actual argument. The second actual argument B is used to represent
−
the total sum for the value of each element in T , which is in the four partitions of a three-element
−
set S. The third actual argument r is used to represent the value of each element in T The fourth
−

actual argument y is used to represent the carry for performing addition of each element in T The
fifth actual argument q and the sixth actual argument n are equal to three. After each operation in
ParallelAdder (T0, B, r, y, q, n) is finished, the result for tube T0 is shown in Table 5.
Then, step (5) calls the algorithm, ParallelXOR(T0, A, B, q, n). Tube T0 with the result in Table 5
is regarded as the first actual argument. The second actual argument A is used to represent the
total sum for the value of each element in T. The third actual argument B is used to represent the
−

total sum for the value of each element in T .The fourth actual argument q and the fifth actual
argument n are equal to three. After each operation in ParallelXOR(T0, A, B, q, n) is finished, the
result for tube T0 is shown in Table 7.
step (6) is a single loop and is used to search the answer of the set-partition problem. Because
C3,10 appears in each bit pattern in tube T0 in Table 7, after the first execution of step (6a) is
performed, tube T1 obtains the same result in Table 7 and tube T0 = ∅, tube T2 = ∅. Then, after
the first execution of step (6b) is finished, tube T1 = ∅, and tube T0 obtains the same result in
Table 7. After the first execution of step (6c) is finished, tube T2 = ∅. Similarly, after each
operation in step (6) is finished, the result for tube T0 is shown in Table 8. Finally, after the
−

execution of step (7) is performed, the answer is that T = {3} and T = {1, 2}. Theorem 1 is used
to prove the correctness of Algorithm 1.
Table 8. The final result generated by Algorithm 1.

Tube
T0

The final result generated by Algorithm 1
{C3,30 C3,20C3,10 y3,30 B3,30 y3,20B3,21 y3,10 B3,11 y2,30 B2,30 y2,20B2,21 y2,10 B2,11 y1,30 B1,30
y3,00}

Theorem 1: From those steps in Algorithm 1, the set partition problem for 2q-1 partitions of a qelement set S can be solved.
Proof: On the execution of step (1), it calls Init (T0, q) to construct solution space for 2q-1
34


partitions of a q-element set S. This means that the tube T0 includes strands encoding 2q-1
partitions. Next the execution of step (2) calls Value (T0, q, n) to encode the value of each
element in 2q-1 partitions. On the execution of step (3), it calls ParallelAdder(T0, A, s, z, q, n) to
perform the function of a parallel adder for computing the sum to the value of each element in T
−

in each pair of (T and T ). Similarly, on the execution of step (4), it calls ParallelAdder(T0, B, r,
y, q, n) to perform the function of a parallel adder for computing the sum to the value of each
−

−

element in T in each pair of (T , T ). Then, after step (5) is executed, it calls ParallelXOR(T0, A,
B, q, n) to perform the function of parallel XOR operation of n bits for the total sum to each pair
−

of (T , T ).
Then, step (6) is a single loop and is employed individually to retrieve those DNA strands
encoded zero value as the result of parallel XOR operation of n bits. Each time step (6a) uses the
extract operation to form two test tubes: T1 and T2 .The first tube T1 includes all of the strands that
have Cq,k=0. The second tube T2 contains all of the strands that have Cq,k=1. Then, on each
execution of step (6b), it employs the merge operation to pour tube T1 into T0 and each execution
of step (6c) uses the discard operation to discard tube T2 containing the result of parallel XOR
operation is not equal to zero. After each step in the single loop is performed n times, tube T0
includes DNA strands that encodes the sum of subset T to be equal to that of its exclusive
−
subset T , the answer for the set-partition is found from tube T0. Finally, the execution of step (7)
uses the detect operation to check if there is any DNA strand in tube T0. If it returns a “yes”, then
the execution of step (7a) applies the read operation to read the answer. Otherwise, no solution
exists. Therefore, the set-partition problem for 2q-1 partitions of a q-element set S can be computed
from those steps in Algorithm 1.

3.9 The Complexity of Algorithm 1
Theorem 2: Suppose that a finite set S is {s1, s2,…, sq}. The set partition problem for S can be
solved with O (q × n) biological operations, O (2n) library strands, O (1) tubes and the longest
library strands, O (q × n), where n is the number of bits for representing the value of each element
in S.
Proof: refer to Algorithm 1.

4. Conclusions
In this paper, based on biological operations, we propose DNA-based algorithms for solving the
set partition problem of a q-element set. The computational complexity is clearly reduced to O
(q*n), where q is the numbers of element in set S we concerned, n is the number of bits for
representing the value of each element in S.
Although the future of molecular computers is vague at present, it is likely that in the future DNA
computers will be the better choice for performing massively parallel computations. However,
there are still severe challenges to face and many technical difficulties to overcome before this
becomes a reality. We hope that this paper helps to manifest that molecular computing is a
technology worth pursuing.

Acknowledgements
The author is very grateful to Dr. Amos who is the author of the 5th reference and the 6th reference
for proposing valuable information about implementation of biological operations.
35


References
[1]
[2]

[3]

[4]
[5]
[6]
[7]
[8]

[9]

[10]

[11]
[12]
[13]
[14]

[15]
[16]
[17]
[18]
[19]

[20]
[21]
[22]
[23]
[24]

Adleman, L. M. (1994) “Molecular computation of solutions to combinatorial problems”, Science,
266, November 11, 1021-1024.
Adleman, L.M., Rothemund, P. W. K., Roweis S., and Winfree, E. (1999) “On applying molecular
computation to the Data Encryption Standard”, The 2nd annual workshop on DNA Computing,
Princeton University, DIMACS: series in Discrete Mathematics and Theoretical Computer Science,
American Mathematical Society, 31-44.
Adleman,L.M., Braich, R.S., Johnson, C., Rothemund, P.W.K., Hwang, D. and Chelyapov, N. (2002)
“Solution of a 20-Variable 3-SAT Problem on a DNA Computer”, Science, Volume 296, Issue 5567,
499-502, 19 April.
Ahrabian, H. and Nowzari-Dalini,A. (2004) “DNA simulation of nand Boolean circuits”, Advanced
Modeling and Optimization, Volume 6, Number 2, 33-41.
Amos, M. (1997) DNA Computation, Ph.D. Thesis. department of computer science, the University
of Warwick.
Amos,M.(2005) Theoretical and Experimental DNA Computation. Springer, ISBN3540657738.
Blackburn, G. M. and Gait, M. J. (1990) Nucleic Acids in Chemistry and Biology. IRL Press.
Boneh, D., Dunworth, C., Lipton, R. J. and Sgall J., 1996. “On the Computational Power of DNA”,
Discrete Applied Mathematics, Special Issue on Computational Molecular Biology, Volume 71, 7994.
Boneh, D., Dunworth, C., and Lipton, R. J. (1996) “Breaking DES using a molecular computer”, In
Proceedings of the 1st DIMACS Workshop on DNA Based Computers, 1995, American
Mathematical Society. In DIMACS Series in Discrete Mathematics and Theoretical Computer
Science, Volume 27, 37-66.
Braich, R. S., Johnson, C., Rothemund, P.W.K., Hwang, D., Chelyapov, N. and Adleman, L. M.
(2001) “Solution of a satisfiability problem on a gel-based DNA computer”, Proceedings of the 6th
International Conference on DNA Computation in the Springer-Verlag Lecture Notes in Computer
Science series, 1-27.
Chang, W.L., Ho, M. and Guo, M. (2004) “Molecular solutions for the subset-sum problem on DNAbased supercomputing”, BioSystems, Volume 73, No. 2, 117-130.
Chang, W.L., Ho, M. and Guo, M. (2005) “Fast parallel molecular algorithms for DNA-based
computation: factoring integers”, IEEE Transactions on Nanobioscience, Volume 4, No. 2, 149-163.
Cormen, T. H., Leiserson, C. E. and Rivest, R. L. (2001) Introduction to algorithms (the second
edition). MIT Press.
Cukras, A. R., Faulhammer,D., Lipton, R.J. and Landweber, L. F. (1998) “Chess games: A model for
RNA-based computation”, In Proceedings of the 4th DIMACS Meeting on DNA Based Computers,
held at the University of Pennsylvania, June 16-19, 27-37.
Eckstein F. (1991) Oligonucleotides and Anologues. Oxford University Press.
Feynman, R. P. (1961) “In miniaturization”, D.H. Gilbert, Ed., Reinhold Publishing Corporation, New
York, 282-296.
Garey, M. R. and Johnson, D. S. (1979) Computer and intractability. Freeman, San Fransico, CA.
Guarnieri,F., Fliss, M. and Bancroft, C. (1996) “Making DNA add”, Science, Vol. 273, 220–223.
Guo, M., Chang, W. L., Ho, M., Lu, J. and Cao, J. (2005) “Is optimal solution of every NP-complete
or NP-hard problem determined from its characteristic for DNA-based computing”, Biosystems, Vol.
80, No. 1. 71-82.
Ho, Michael (2005) “Fast parallel molecular solutions for DNA-based supercomputing: the subsetproduct problem”, BioSystems, Volume 80, 233–250.
Lipton, R. J. (1995) “DNA solution of hard computational problems”, Science, 268, 542:545.
Paun,G., Rozenberg, G. and Salomaa, A. (1998) DNA Computing: New Computing Paradigms.
Springer-Verlag, New York.
Quyang, Q. P., Kaplan, Liu, D. S. and Libchaber A. (1997) “DNA solution of the maximal clique
problem”, Science, 278:446-449.
Roweis, S., Winfree, E., Burgoyne, R., Chelyapov, N. V., Goodman, M. F., Rothemund, Paul W.K.
and Adleman, L. M. (1999) “A sticker based model for DNA computation”, 1999, 2nd annual
workshop on DNA Computing, Princeton University. Eds. L. Landweber and E. Baum, DIMACS:
36


[25]
[26]
[27]
[28]
[29]
[30]

series in Discrete Mathematics and Theoretical Computer Science, American Mathematical Society,
1-29.
Schuster, A. (2005) DNA databases. BioSystems, Volume 81, 234–246.
Sinden, R. R. (1994) DNA Structure and Function. Academic Press.
Watson,J., Gilman, M., Witkowski,J. and Zoller, M. (1992) Recombinant DNA (2nd edition).
Scientific American Books, W. H. Freeman and Co..
Watson, J., Hoplins, N., Roberts, J. and et al. (1987) Molecular Biology of the Gene.
Benjamin/Cummings Menlo Park CA.
Xiao, D., Li, W., Zhang, Z. and He, L. (2005) ”Solving the maximum cut problems in the Adleman–
Lipton model”, BioSystems, Volume 82, 203-207.
Yeh, C.W., Chu, C. P. and Wu, K.R. (2006) “Molecular solutions to the binary integer programming
problem based on DNA computation”, Biosystems, Volume: 83, Issue: 1, January, 56-66.

Authors
Sientang Tsai received his B.S. from department of physics, National Taiwan Normal
University in 1974 and got M.S. degree in computer science from University of
Georgia, Athens, U.S.A. in 1983. He is currently an associate professor at the
department of information management in Southern Taiwan University of Science and
Technology. His researching interests include parallel computing, quantum computing
and DNA computing.
Wei-Yeh Chen received his B.S. degree from National Cheng Kung University in
1986, M.S. degree from National Chiao Tung University in 1988, and Ph. D. degree in
information management from National Taiwan University of Science and Technology
in 2004. From 1990 to 2004, he was a lecturer in the Department of Electronic
Engineering at Northern Taiwan Institute of Science and Technology. From 2004 to
2005, he was an associate professor of the Institute. Since August 2005, he has been an
associate professor in the Department of Information Management, Southern Taiwan
University of Science and Technology, Tainan, Taiwan. His current research interests
are resource allocation of mobile communications systems, protocol design of wireless
mesh networks, and spectrum management of cognitive radio networks.

37

Molecular Solutions For The Set-Partition Problem On Dna-Based Computing

More Related Content

What's hot (20)

Viewers also liked (17)

Similar to Molecular Solutions For The Set-Partition Problem On Dna-Based Computing (20)

Recently uploaded (20)

Molecular Solutions For The Set-Partition Problem On Dna-Based Computing