![Natarajan Meghanathan et al. (Eds) : CCSEA, NCOM, AIFU, DKMP, EMSA, SIPRO, SEA - 2018
pp. 29– 34, 2018. © CS & IT-CSCP 2018 DOI : 10.5121/csit.2018.80303
ON THE DUALITY FEATURE OF P-CLASS
PROBLEMS AND NP COMPLETE
PROBLEMS
WenhongTian1, 2
1
University of Electronic Science and Technology of China,
Chengdu, China.
2
Chongqing Institute of Green and Intelligent Technology,
Chinese Academy of Science.
ABSTRACT
In term of computational complexity, P-class (abbreviated as P) problems are polynomial-time
solvable by deterministic Turing machine while NP complete (abbreviated as NPC) problems
are polynomial-time solvable by nondeterministic Turing machine. P and NPC problems are
regularly treated in different classes. Determining whether or not it is possible to solve NPC
problems quickly, is one of the principal unsolved problems in computer science today. In this
paper, a new perspective is provided: both P problems and NPC problems have the duality
feature in terms of computational complexity of asymptotic efficiency of algorithms.
KEYWORDS
P Problems; NP problems; NP Complete Problems; the P versus NP Problem; The Duality
Feature.
1. INTRODUCTION
In 1971, Cook [1] firstly established a theorem that a class of problems can be P-reducible
(polynomial time reducible) to each other and each of them can be P-reducible to Boolean
Satisfiability (SAT) problem, this class of problems is called NP (nondeterministic polynomial
time) problems. Karp [2] applied Cook’s 1971 theorem that the SAT problem is NP Complete
(also called the Cook-Levin theorem) to show that there is a polynomial time reduction from the
SAT problem to each of 21 combinatorial problems, thereby showing that they are all NP
complete (NPC). This was one of the first demonstrations that many natural computational
problems occurring throughout computer science are computationally intractable, and it drove
interest in the study of NP-completeness and the P versus NP problem [3]. The P versus NP
problem, determining whether or not it is possible to solve NP problems quickly, is one of the
principal unsolved problems in computer science today and listed as one of seven millennium
problems [3,4], challenging tens of thousands of researchers.
Simply speaking, P problems mean that the class of problems can be solved exactly in polynomial
time while NPC problems stands for a class of problems which might not be solvable in
polynomial time. NPC problems has far-reaching consequences to other problems in
mathematics, biology, philosophy and cryptography. More specifically, in Big O-notation (refer](https://image.slidesharecdn.com/csit88303-180306093251/85/ON-THE-DUALITY-FEATURE-OF-P-CLASS-PROBLEMS-AND-NP-COMPLETE-PROBLEMS-1-320.jpg)
![30 Computer Science & Information Technology (CS & IT)
to Definition 1 and 2) of computational complexity in asymptotic efficiency of algorithms, P
problems can be solved in polynomial time of O
input to the problem, while NPC problems may have computational complexity of O
including both exponential time and sub
than zero.
Karp [2] ever claimed that if any of NPC problems have efficient polynomial time algorithms,
then they all do. It is for this reason that research into the P versus NP problem centers on NPC
problems, i.e., looking for efficient polynomial time algorithms
decades’ efforts by many researchers, it is still an open question. There are quite many results but
none of them is commonly accepted yet by the research community.
The current author also classified NPC problems based on their
found that NPC problems are not equivalent in computational complexity.
In the following, we show a new perspective: both P problems and NPC problems have
features of each other in terms of computational complex
algorithms, especially by considering the representation of the input.
2. PROBLEM FORMULATIONS
The following definitions are based on the computational complexity of different problems in the
worst case.
Definition 1: The asymptotic efficiency of algorithms [6,7]: concerns with how the running time
of an algorithm increases with the size of the input
without bound.
Definition 2: The O-notation
bounds a function from above within a constant factor [6,7]. For a given function
by O(g(n)) :
O(g(n))={f(n): there exist positive constants
Fig.1 shows the intuition behind O
function f(n) is on or below g(n).
Computer Science & Information Technology (CS & IT)
to Definition 1 and 2) of computational complexity in asymptotic efficiency of algorithms, P
problems can be solved in polynomial time of O(nk
) for some constant k where n
input to the problem, while NPC problems may have computational complexity of O
including both exponential time and sub-exponential time, where c is a positive constant larger
Karp [2] ever claimed that if any of NPC problems have efficient polynomial time algorithms,
then they all do. It is for this reason that research into the P versus NP problem centers on NPC
problems, i.e., looking for efficient polynomial time algorithms for NPC problems. Through
decades’ efforts by many researchers, it is still an open question. There are quite many results but
none of them is commonly accepted yet by the research community.
The current author also classified NPC problems based on their natures and other methods [5] and
found that NPC problems are not equivalent in computational complexity.
In the following, we show a new perspective: both P problems and NPC problems have
in terms of computational complexity of asymptotic efficiency of
algorithms, especially by considering the representation of the input.
ORMULATIONS
The following definitions are based on the computational complexity of different problems in the
asymptotic efficiency of algorithms [6,7]: concerns with how the running time
of an algorithm increases with the size of the input in the limit, as the size of the input increases
of computational complexity of an algorithm: asymptotically
bounds a function from above within a constant factor [6,7]. For a given function g(n)
: there exist positive constants c and n0 such that 0≤f(n)≤cg(n) for n≥ n0
Fig.1 shows the intuition behind O-notation. For all values n to the right of n0, the value of
.
Figure 1.f(n)=O(g(n)) [6]
to Definition 1 and 2) of computational complexity in asymptotic efficiency of algorithms, P
is the size of
input to the problem, while NPC problems may have computational complexity of O(2cn
)
ve constant larger
Karp [2] ever claimed that if any of NPC problems have efficient polynomial time algorithms,
then they all do. It is for this reason that research into the P versus NP problem centers on NPC
for NPC problems. Through
decades’ efforts by many researchers, it is still an open question. There are quite many results but
natures and other methods [5] and
In the following, we show a new perspective: both P problems and NPC problems have duality
ity of asymptotic efficiency of
The following definitions are based on the computational complexity of different problems in the
asymptotic efficiency of algorithms [6,7]: concerns with how the running time
, as the size of the input increases
computational complexity of an algorithm: asymptotically
g(n), we denote
0}.
, the value of](https://image.slidesharecdn.com/csit88303-180306093251/85/ON-THE-DUALITY-FEATURE-OF-P-CLASS-PROBLEMS-AND-NP-COMPLETE-PROBLEMS-2-320.jpg)
![Computer Science & Information Technology (CS & IT) 31
Since O-notation describes an upper bound, when we use it to bound the worst-case running time
of an algorithm, we have a bound on the running time of the algorithm on every input. For
example, the doubly nested loop structure of the insertion sort algorithm has an O(n2
) upper
bound; equivalently, we mean that the worst-case running time is O(n2
).
Definition 3: P-class problems in term of O-notation of computational complexity: in the worst
case, their exact algorithms have computational complexity of O(nk
) for some constant k where n
is the size of input to the problem.
Definition 4: NPC problems: in the worst case, their exact solutions may have computational
complexity of O(2cn
) in O-notation, where n is the size of input to the problem and c is a
positive constant.
Table 1 provides a summary of variables used in this paper. Table 2 shows computational
complexities of exact solutions to some NPC problems (abstracted from [8]).
Table 1. The summary of variables
Variables Meaning
k, c, n0 a positive constant
n,e The size of inputs to a problem
s, m The number of bits (binary)
B0 min(2cn
, nk
, 2m
, 2s
)
W, C, b a positive number
Table 2. The complexities of exact solutions to some NPC problems from [8]
Problem Complexity Genre
MIS (maximum independent set) 1.1996n
Packing
Set covering problem (SCP) * 1.1996n
Covering
3CP (3-coloring of planar graph) 1.3446n
Partitioning
SSP (subset sum problem) [7] nW Numerical
3SAT 1.4802n
Satisfaction
TSP n2
2n
Sequencing
*: derived from MIS since SCP and MIS are complementary [7].
We also consider that the representation of the input to the problem by m bits (in binary) and the
space complexity of a problem by s bits (in binary) in memory.
Fact 1: A regular system (the computer hardware, called the system in the following) has
capability of handling B0 numbers in term of computational and space complexity, i.e., handling
the input efficiently within reasonable time (may be within a few hours or minutes depending on
the applications) without overflowing the system, where B0 may be related to the memory size or
whichever (CPU, memory, disk et.) is the bottleneck of the system.](https://image.slidesharecdn.com/csit88303-180306093251/85/ON-THE-DUALITY-FEATURE-OF-P-CLASS-PROBLEMS-AND-NP-COMPLETE-PROBLEMS-3-320.jpg)
![32 Computer Science & Information Technology (CS & IT)
Based on the definitions above, the capabilityB0can be represented by
B0=min(2cn
, nk
, 2m
, 2s
) (1)
3. RESULTS
Lemma 1. Dynamic programming algorithm for Subset Sum problem (SSP) is pseudo-
polynomial time.
Proof. removed■
Lemma 2. Any NPC problem can be reduced to SSP.
Theorem 1:The instances of SSP can have computational complexity of NPC problems and
P-class problems, this is called the duality feature in this paper.
Proof: removed■
Fact 2: The instances of other NPC problems also have the duality feature.
Proof: removed.■
In Table 3, records of optimum solutions to TSP problem are provided. One can see that the size
of the problem is increasing as year goes. Notice that the instance with 1904711 nodes is still not
yet solved exactly but just has a good lower bound [14]. Similar results are also observed for
other NPC problems. For real-life problems with small or moderate number of variables, they can
be solved to exactly very easily (LKH [15] and Concorde [16] are considered as two of the best
such solvers which can now find exact solutions for medium size TSP Problems). However, if the
problem size increases to very large, finding efficient solutions to NPC problems just become
intractable.
Theorem 2: The P-class Problems can have the duality feature in terms of computational
complexity.
Proof: removed. ■
Remarks: Actually SSP can be treated as polynomial time solvable problem (P-class problem)
when both n and W is not very large while it is of exponential complexity when both n and W are
very large.
Observation 1: Even one can find efficient solution (polynomial time algorithm) to one or
more NPC problems, some P problems may become exponential time solvable in
computational complexity when the representation of the input to them become very large.
4. DISCUSSIONS AND CONCLUSION
NPC problems have different natures, they can be classified into six basic genres [2, 7], i.e.,
Satisfaction, Packing, Covering, Partitioning, Sequencing, Numerical computing.
Originally, PRIMES and Graph Isomorphism were hard to determine to be in NPC or not in
Cook’s paper [1]. In 2004, PRIMES is found in P-class and accepted by the research community
[10]. And in 2015, Graph isomorphism is reported to have Quasipolynomial time solution [11,](https://image.slidesharecdn.com/csit88303-180306093251/85/ON-THE-DUALITY-FEATURE-OF-P-CLASS-PROBLEMS-AND-NP-COMPLETE-PROBLEMS-4-320.jpg)
![Computer Science & Information Technology (CS & IT) 33
12], though the results are still under verification. These show some new perspectives and trends
on NPC problems.
As another perspective, it is recently proved mathematically that memcomputing machines (a
novel non-Turing paradigm) have the same computational power of nondeterministic Turing
machines [13]. Therefore, they can solve NPC problems in polynomial time with resources that
only grow polynomially with the input size.
NPC problems and the P versus NP problem challenge many researchers to tackle them through
decades of efforts. In this paper, a new perspective is provided: both P problems and NPC
problems have the duality feature in terms of computational complexity of asymptotic efficiency
of algorithms.
Table 3. Records of optimum solutions to TSP problems [14] where n is the number of nodes in TSP. All
TSP problems in the table are solved to optimum except for the last one
n
Year
(solved) Node type
48 1954 USA cities
64 1971 random nodes
80 1975 random nodes
120 1977 Germany cities
318 1987 cities
532 1987 USA cities
666 1987 World cities
1002 1987 cities
2392 1987 cities
3038 1992 cities
13509 1998 USA cities
15112 2001 cities
24978 2004 Sweden cities
85900 2006 cities
100000 2009 Japan
1904711 2010* World TSP Challenge
ACKNOWLEDGMENTS
The abstract version of this manuscript is posted at my research gate.
REFERENCES
[1] S.A. Cook (1971). The Complexity of Theorem Proving Procedures, Proceedings of the third annual
ACM symposium on Theory of computing. pp. 151158, March of 1971.
[2] R. M. Karp (1972), Reducibility Among Combinatorial Problems, In R. E. Miller and J. W. Thatcher
(editors). Complexity of Computer Computations. New York: Plenum. pp. 85-103.
[3] Clay Mathematics Institute, http://www.claymath.org/millennium-problems/millennium-prize-
problems.](https://image.slidesharecdn.com/csit88303-180306093251/85/ON-THE-DUALITY-FEATURE-OF-P-CLASS-PROBLEMS-AND-NP-COMPLETE-PROBLEMS-5-320.jpg)
![34 Computer Science & Information Technology (CS & IT)
[4] Wikipedia, http://en.wikipedia.org/wiki/NP-complete.
[5] WenhongTian, NP Complete Problems Are not All Equivalent, under review, submitted 2017.
[6] T. Cormen,C. Leiserson, R.L. Rivest,C. Stein, Introduction to Algorithm, 2nd edition, MIT Press,
2004.
[7] J. Kleinberg, E. Tardos, Algorithm Design, 2006, Pearson Education Asia Limited, pp. 475-479.
[8] G. J. Woeginger, Exact algorithms for NP-hard Problems: A Survey, Combinatorial optimization -
Eureka, you shrink! Pages 185 - 207 , Springer-Verlag New York, Inc. New York, NY, USA ©2003
[9] WenhongTian, C. Huang, X. Wang, A Near Optimal Approach for Symmetric Traveling Salesman
Problem in Eclidean Space, In Proceedings of ICORES 2017, Portugal, also available at arxiv
https://arxiv.org/pdf/1502.00447.pdf
[10] M. Agrawal, N. Kayal, N. Saxena, PRIMES is in P, The Annals of Mathematics, Pages 781-793 from
Volume 160, Issue 2, 2004.
[11] L. Babai, Graph Isomorphism in Quasipolynomial Time, https://arxiv.org/abs/1512.03547, Dec 11,
2015.
[12] A. Cho , Science news, mathematician-claims-breakthrough-complexity-theory,
http://www.sciencemag.org/news/2015/11/mathematician-claims-breakthrough-complexity-theory,
Nov. 11, 2015.
[13] F. L. Traversa, C. Ramella, F. Bonani and M.D. Ventra, memcomputing NP-complete problems in
polynomial time using polynomial resources and collective states, Science, Vol. 1, no. 6, e1500031,
Nov. 2015.
[14] W. Cook, In Pursuit of the Traveling Salesman, Princeton University Press, 2012.
[15] LKH codes, http://www.akira.ruc.dk/~keld/research/LKH/
[16] Concorde codes, http://www.math.uwaterloo.ca/tsp/concorde.html](https://image.slidesharecdn.com/csit88303-180306093251/85/ON-THE-DUALITY-FEATURE-OF-P-CLASS-PROBLEMS-AND-NP-COMPLETE-PROBLEMS-6-320.jpg)
More Related Content
What's hot (19)
Similar to ON THE DUALITY FEATURE OF P-CLASS PROBLEMS AND NP COMPLETE PROBLEMS (20)
More from cscpconf (20)
Recently uploaded (20)
ON THE DUALITY FEATURE OF P-CLASS PROBLEMS AND NP COMPLETE PROBLEMS
- 1. Natarajan Meghanathan et al. (Eds) : CCSEA, NCOM, AIFU, DKMP, EMSA, SIPRO, SEA - 2018 pp. 29– 34, 2018. © CS & IT-CSCP 2018 DOI : 10.5121/csit.2018.80303 ON THE DUALITY FEATURE OF P-CLASS PROBLEMS AND NP COMPLETE PROBLEMS WenhongTian1, 2 1 University of Electronic Science and Technology of China, Chengdu, China. 2 Chongqing Institute of Green and Intelligent Technology, Chinese Academy of Science. ABSTRACT In term of computational complexity, P-class (abbreviated as P) problems are polynomial-time solvable by deterministic Turing machine while NP complete (abbreviated as NPC) problems are polynomial-time solvable by nondeterministic Turing machine. P and NPC problems are regularly treated in different classes. Determining whether or not it is possible to solve NPC problems quickly, is one of the principal unsolved problems in computer science today. In this paper, a new perspective is provided: both P problems and NPC problems have the duality feature in terms of computational complexity of asymptotic efficiency of algorithms. KEYWORDS P Problems; NP problems; NP Complete Problems; the P versus NP Problem; The Duality Feature. 1. INTRODUCTION In 1971, Cook [1] firstly established a theorem that a class of problems can be P-reducible (polynomial time reducible) to each other and each of them can be P-reducible to Boolean Satisfiability (SAT) problem, this class of problems is called NP (nondeterministic polynomial time) problems. Karp [2] applied Cook’s 1971 theorem that the SAT problem is NP Complete (also called the Cook-Levin theorem) to show that there is a polynomial time reduction from the SAT problem to each of 21 combinatorial problems, thereby showing that they are all NP complete (NPC). This was one of the first demonstrations that many natural computational problems occurring throughout computer science are computationally intractable, and it drove interest in the study of NP-completeness and the P versus NP problem [3]. The P versus NP problem, determining whether or not it is possible to solve NP problems quickly, is one of the principal unsolved problems in computer science today and listed as one of seven millennium problems [3,4], challenging tens of thousands of researchers. Simply speaking, P problems mean that the class of problems can be solved exactly in polynomial time while NPC problems stands for a class of problems which might not be solvable in polynomial time. NPC problems has far-reaching consequences to other problems in mathematics, biology, philosophy and cryptography. More specifically, in Big O-notation (refer
- 2. 30 Computer Science & Information Technology (CS & IT) to Definition 1 and 2) of computational complexity in asymptotic efficiency of algorithms, P problems can be solved in polynomial time of O input to the problem, while NPC problems may have computational complexity of O including both exponential time and sub than zero. Karp [2] ever claimed that if any of NPC problems have efficient polynomial time algorithms, then they all do. It is for this reason that research into the P versus NP problem centers on NPC problems, i.e., looking for efficient polynomial time algorithms decades’ efforts by many researchers, it is still an open question. There are quite many results but none of them is commonly accepted yet by the research community. The current author also classified NPC problems based on their found that NPC problems are not equivalent in computational complexity. In the following, we show a new perspective: both P problems and NPC problems have features of each other in terms of computational complex algorithms, especially by considering the representation of the input. 2. PROBLEM FORMULATIONS The following definitions are based on the computational complexity of different problems in the worst case. Definition 1: The asymptotic efficiency of algorithms [6,7]: concerns with how the running time of an algorithm increases with the size of the input without bound. Definition 2: The O-notation bounds a function from above within a constant factor [6,7]. For a given function by O(g(n)) : O(g(n))={f(n): there exist positive constants Fig.1 shows the intuition behind O function f(n) is on or below g(n). Computer Science & Information Technology (CS & IT) to Definition 1 and 2) of computational complexity in asymptotic efficiency of algorithms, P problems can be solved in polynomial time of O(nk ) for some constant k where n input to the problem, while NPC problems may have computational complexity of O including both exponential time and sub-exponential time, where c is a positive constant larger Karp [2] ever claimed that if any of NPC problems have efficient polynomial time algorithms, then they all do. It is for this reason that research into the P versus NP problem centers on NPC problems, i.e., looking for efficient polynomial time algorithms for NPC problems. Through decades’ efforts by many researchers, it is still an open question. There are quite many results but none of them is commonly accepted yet by the research community. The current author also classified NPC problems based on their natures and other methods [5] and found that NPC problems are not equivalent in computational complexity. In the following, we show a new perspective: both P problems and NPC problems have in terms of computational complexity of asymptotic efficiency of algorithms, especially by considering the representation of the input. ORMULATIONS The following definitions are based on the computational complexity of different problems in the asymptotic efficiency of algorithms [6,7]: concerns with how the running time of an algorithm increases with the size of the input in the limit, as the size of the input increases of computational complexity of an algorithm: asymptotically bounds a function from above within a constant factor [6,7]. For a given function g(n) : there exist positive constants c and n0 such that 0≤f(n)≤cg(n) for n≥ n0 Fig.1 shows the intuition behind O-notation. For all values n to the right of n0, the value of . Figure 1.f(n)=O(g(n)) [6] to Definition 1 and 2) of computational complexity in asymptotic efficiency of algorithms, P is the size of input to the problem, while NPC problems may have computational complexity of O(2cn ) ve constant larger Karp [2] ever claimed that if any of NPC problems have efficient polynomial time algorithms, then they all do. It is for this reason that research into the P versus NP problem centers on NPC for NPC problems. Through decades’ efforts by many researchers, it is still an open question. There are quite many results but natures and other methods [5] and In the following, we show a new perspective: both P problems and NPC problems have duality ity of asymptotic efficiency of The following definitions are based on the computational complexity of different problems in the asymptotic efficiency of algorithms [6,7]: concerns with how the running time , as the size of the input increases computational complexity of an algorithm: asymptotically g(n), we denote 0}. , the value of
- 3. Computer Science & Information Technology (CS & IT) 31 Since O-notation describes an upper bound, when we use it to bound the worst-case running time of an algorithm, we have a bound on the running time of the algorithm on every input. For example, the doubly nested loop structure of the insertion sort algorithm has an O(n2 ) upper bound; equivalently, we mean that the worst-case running time is O(n2 ). Definition 3: P-class problems in term of O-notation of computational complexity: in the worst case, their exact algorithms have computational complexity of O(nk ) for some constant k where n is the size of input to the problem. Definition 4: NPC problems: in the worst case, their exact solutions may have computational complexity of O(2cn ) in O-notation, where n is the size of input to the problem and c is a positive constant. Table 1 provides a summary of variables used in this paper. Table 2 shows computational complexities of exact solutions to some NPC problems (abstracted from [8]). Table 1. The summary of variables Variables Meaning k, c, n0 a positive constant n,e The size of inputs to a problem s, m The number of bits (binary) B0 min(2cn , nk , 2m , 2s ) W, C, b a positive number Table 2. The complexities of exact solutions to some NPC problems from [8] Problem Complexity Genre MIS (maximum independent set) 1.1996n Packing Set covering problem (SCP) * 1.1996n Covering 3CP (3-coloring of planar graph) 1.3446n Partitioning SSP (subset sum problem) [7] nW Numerical 3SAT 1.4802n Satisfaction TSP n2 2n Sequencing *: derived from MIS since SCP and MIS are complementary [7]. We also consider that the representation of the input to the problem by m bits (in binary) and the space complexity of a problem by s bits (in binary) in memory. Fact 1: A regular system (the computer hardware, called the system in the following) has capability of handling B0 numbers in term of computational and space complexity, i.e., handling the input efficiently within reasonable time (may be within a few hours or minutes depending on the applications) without overflowing the system, where B0 may be related to the memory size or whichever (CPU, memory, disk et.) is the bottleneck of the system.
- 4. 32 Computer Science & Information Technology (CS & IT) Based on the definitions above, the capabilityB0can be represented by B0=min(2cn , nk , 2m , 2s ) (1) 3. RESULTS Lemma 1. Dynamic programming algorithm for Subset Sum problem (SSP) is pseudo- polynomial time. Proof. removed■ Lemma 2. Any NPC problem can be reduced to SSP. Theorem 1:The instances of SSP can have computational complexity of NPC problems and P-class problems, this is called the duality feature in this paper. Proof: removed■ Fact 2: The instances of other NPC problems also have the duality feature. Proof: removed.■ In Table 3, records of optimum solutions to TSP problem are provided. One can see that the size of the problem is increasing as year goes. Notice that the instance with 1904711 nodes is still not yet solved exactly but just has a good lower bound [14]. Similar results are also observed for other NPC problems. For real-life problems with small or moderate number of variables, they can be solved to exactly very easily (LKH [15] and Concorde [16] are considered as two of the best such solvers which can now find exact solutions for medium size TSP Problems). However, if the problem size increases to very large, finding efficient solutions to NPC problems just become intractable. Theorem 2: The P-class Problems can have the duality feature in terms of computational complexity. Proof: removed. ■ Remarks: Actually SSP can be treated as polynomial time solvable problem (P-class problem) when both n and W is not very large while it is of exponential complexity when both n and W are very large. Observation 1: Even one can find efficient solution (polynomial time algorithm) to one or more NPC problems, some P problems may become exponential time solvable in computational complexity when the representation of the input to them become very large. 4. DISCUSSIONS AND CONCLUSION NPC problems have different natures, they can be classified into six basic genres [2, 7], i.e., Satisfaction, Packing, Covering, Partitioning, Sequencing, Numerical computing. Originally, PRIMES and Graph Isomorphism were hard to determine to be in NPC or not in Cook’s paper [1]. In 2004, PRIMES is found in P-class and accepted by the research community [10]. And in 2015, Graph isomorphism is reported to have Quasipolynomial time solution [11,
- 5. Computer Science & Information Technology (CS & IT) 33 12], though the results are still under verification. These show some new perspectives and trends on NPC problems. As another perspective, it is recently proved mathematically that memcomputing machines (a novel non-Turing paradigm) have the same computational power of nondeterministic Turing machines [13]. Therefore, they can solve NPC problems in polynomial time with resources that only grow polynomially with the input size. NPC problems and the P versus NP problem challenge many researchers to tackle them through decades of efforts. In this paper, a new perspective is provided: both P problems and NPC problems have the duality feature in terms of computational complexity of asymptotic efficiency of algorithms. Table 3. Records of optimum solutions to TSP problems [14] where n is the number of nodes in TSP. All TSP problems in the table are solved to optimum except for the last one n Year (solved) Node type 48 1954 USA cities 64 1971 random nodes 80 1975 random nodes 120 1977 Germany cities 318 1987 cities 532 1987 USA cities 666 1987 World cities 1002 1987 cities 2392 1987 cities 3038 1992 cities 13509 1998 USA cities 15112 2001 cities 24978 2004 Sweden cities 85900 2006 cities 100000 2009 Japan 1904711 2010* World TSP Challenge ACKNOWLEDGMENTS The abstract version of this manuscript is posted at my research gate. REFERENCES [1] S.A. Cook (1971). The Complexity of Theorem Proving Procedures, Proceedings of the third annual ACM symposium on Theory of computing. pp. 151158, March of 1971. [2] R. M. Karp (1972), Reducibility Among Combinatorial Problems, In R. E. Miller and J. W. Thatcher (editors). Complexity of Computer Computations. New York: Plenum. pp. 85-103. [3] Clay Mathematics Institute, http://www.claymath.org/millennium-problems/millennium-prize- problems.
- 6. 34 Computer Science & Information Technology (CS & IT) [4] Wikipedia, http://en.wikipedia.org/wiki/NP-complete. [5] WenhongTian, NP Complete Problems Are not All Equivalent, under review, submitted 2017. [6] T. Cormen,C. Leiserson, R.L. Rivest,C. Stein, Introduction to Algorithm, 2nd edition, MIT Press, 2004. [7] J. Kleinberg, E. Tardos, Algorithm Design, 2006, Pearson Education Asia Limited, pp. 475-479. [8] G. J. Woeginger, Exact algorithms for NP-hard Problems: A Survey, Combinatorial optimization - Eureka, you shrink! Pages 185 - 207 , Springer-Verlag New York, Inc. New York, NY, USA ©2003 [9] WenhongTian, C. Huang, X. Wang, A Near Optimal Approach for Symmetric Traveling Salesman Problem in Eclidean Space, In Proceedings of ICORES 2017, Portugal, also available at arxiv https://arxiv.org/pdf/1502.00447.pdf [10] M. Agrawal, N. Kayal, N. Saxena, PRIMES is in P, The Annals of Mathematics, Pages 781-793 from Volume 160, Issue 2, 2004. [11] L. Babai, Graph Isomorphism in Quasipolynomial Time, https://arxiv.org/abs/1512.03547, Dec 11, 2015. [12] A. Cho , Science news, mathematician-claims-breakthrough-complexity-theory, http://www.sciencemag.org/news/2015/11/mathematician-claims-breakthrough-complexity-theory, Nov. 11, 2015. [13] F. L. Traversa, C. Ramella, F. Bonani and M.D. Ventra, memcomputing NP-complete problems in polynomial time using polynomial resources and collective states, Science, Vol. 1, no. 6, e1500031, Nov. 2015. [14] W. Cook, In Pursuit of the Traveling Salesman, Princeton University Press, 2012. [15] LKH codes, http://www.akira.ruc.dk/~keld/research/LKH/ [16] Concorde codes, http://www.math.uwaterloo.ca/tsp/concorde.html