This article includes a list of references, related reading or external links, but its sources remain unclear because it lacks inline citations. (December 2017) (Learn how and when to remove this template message)
In computer science, the computational complexity, or simply complexity of an algorithm is the amount of resources required for running it. The computational complexity of a problem is the minimum of the complexities of all possible algorithms for this problem (including the unknown algorithms).
As the amount of needed resources varies with the input, the complexity is generally expressed as a function n → f(n), where n is the size of the input, and f(n) is either the worst-case complexity, that is the maximum of the amount of resources that are needed for all input of size n, or the average-case complexity, that is average of the amount of resources over all input of size n.
When the nature of the resources is not explicitly given, this is usually the time needed for running the algorithm, and one talks of time complexity. However, this depends on the computer that is used, and the time is generally expressed as the number of needed elementary operations, which are supposed to take a constant time on a given computer, and to change by a constant factor when one changes of computer.
Otherwise, the resource that is considered is often the size of the memory that is needed, and one talks of space complexity.
The study of the complexity of explicitly given algorithms is called analysis of algorithms, while the study of the complexity of problems is called computational complexity theory. Clearly, both areas are strongly related, as the complexity of an algorithm is always an upper bound of the complexity of the problem solved by this algorithm.
- 1 Resources
- 2 Complexity as a function of input size
- 3 Asymptotic complexity
- 4 Models of computation
- 5 Problem complexity (lower bounds)
- 6 Use in algorithm design
- 7 See also
- 8 References
The resource that is most commonly considered is the time, and one talks of time complexity. When "complexity" is used without being qualified, this generally means time complexity.
The usual units of time are not used in complexity theory, because they are too dependent on the choice of a specific computer and of the evolution of the technology. Therefore, instead of the real time, one generally consider the elementary operations that are done during the computation. These operations are supposed to take a constant time on a given machine, and are often called steps.
Another important resource is the size of computer memory that is needed for running algorithms.
The number of arithmetic operations in another resource that is commonly used. In this case, one talks of arithmetic complexity. If one knows an upper bound on the size of the binary representation of the numbers that occur during a computation, the time complexity is generally the product of the arithmetic complexity by a constant factor.
For many algorithms the size of the integers that are used during a computation is not bounded, and it is not realistic to consider that arithmetic operations take a constant time. Therefore the time complexity, generally called bit complexity in this context, may be much larger than the arithmetic complexity. For example, the arithmetic complexity of the computation of the determinant of a n×n integer matrix is for the usual algorithms (Gaussian elimination). The bit complexity of the same algorithms is exponential in n, because the size of the coefficients may grow exponentially during the computation. On the other hand, if these algorithms are coupled with multi-modular arithmetic, the bit complexity may be reduced to O~(n4).
Complexity as a function of input size
- For clarity, only time complexity is considered in this section, but everything applies (with slight modifications) to the complexity with respect to other resources.
It is impossible to count the number of steps of an algorithm on all possible inputs. As the complexity increases generally with the size of the input, the complexity is generally expressed as a function of the size n (in bits) of the input, and therefore, the complexity is a function of n. However, the complexity of an algorithm may vary dramatically for different inputs of the same size. Therefore several complexity functions are commonly used.
The worst-case complexity is the maximum of the complexity over all inputs of size n, and the average-case complexity is the average of the complexity over all inputs of size n (this makes sense, as the number of possible inputs of a given size is finite). Generally, when "complexity" is used without being further specified, this is the worst-case time complexity that is considered.
It is generally very difficult to compute exactly the worst-case and the average-case complexity. Moreover this makes not really sense, as any change of computer or of model of computation would change the complexity. Moreover, the resource use is not critical for small values of n, and this makes that, for small n, the ease of implementation is generally more interesting than a good complexity.
For these reasons, one generally focuses on the behavior of the complexity for large n, that is on its asymptotic behavior when n tends to the infinity. Therefore, the complexity is generally expressed by using big O notation.
For example, the usual algorithm for integer multiplication has a complexity of this means that there is a constant such that the multiplication two integers of at most n digits may be done in a time less than This bound is sharp in the sense that the worst-case complexity and the average-case complexity are which means that there is a constant such that these complexities are larger than The radix does not appear in these complexity, as changing of radix changes only the constants and
Models of computation
The evaluation of the complexity relies on the choice of a model of computation, which consist in defining the basic operations that are done in a unit of time. When the model of computation is not explicitly specified, this is generally meant as being multitape Turing machine.
A deterministic model of computation is a model of computation such that the successive states of the machine and the operations to be performed are completely determined by the preceding state. Historically, the first deterministic models were recursive functions, lambda calculus, and Turing machines. The model of Random access machines (also called RAM-machines) is also widely used, as closer to real computers.
When the model of computation is not specified, it is generally the model of multitape Turing machines. For most algorithms, the complexity is the same on multitape Turing machines and on RAM-machines, although, some care may be needed in storing data for getting this equivalence.
In a non-deterministic model of computation, such as non-deterministic Turing machines, some choices may be done at some steps of the computation. In complexity theory, one considers all possible choices simultaneously, and the non-deterministic time complexity is the time needed, when the best choices are always done. In other words, one considers that the computation is done simultaneously on as many processors as needed, and the non-deterministic computation time time is the time spent by the first computer that finishes the computation.
Such a computation model is not realistic, but it has some practical applications, mostly related with P = NP problem. A problem is in the complexity class NP, if is may solved in polynomial time over a non-deterministic machine. A problem is NP-complete if, roughly speaking, it is in NP and is not easier than any other NP problem. Many combinatorics problems, such as Knapsack problem, travelling salesman problem, and Boolean satisfiability problem are NP-complete. For all these problems, the best known algorithm has an exponential complexity. If any of these problem could be solved in polynomial time on a deterministic machine, then all NP problems could also be solved in polynomial time, and one would have P = NP. It generally conjectured that P ≠ NP, and this has the practical implication that NP problems are intrinsically difficult (in the worst case).
Parallel and distributed computation
Parallel and distributed computing consist of splitting computation on several processors, which work simultaneously. The difference between the different model lies mainly in the way of transmitting information between processors. Typically, in parallel computing the data transmission between processors is very fast, while, in distributed computating, the data transmission is done through a network and is therefore much slower.
The time needed for a computation on N processors is at least the quotient by N of the time needed by a single processor. In fact this theoretically optimal bound can never be reached, because some subtasks cannot be parallelized, and some processors may have to wait a result from another processor.
The main complexity problem is thus to design algorithms such that the product of the computation time by the number of processors is as close as possible to the time needed for the same computation on a single processor.
A quantum computer is a computer whose model of computation is base on quantum mechanics. The Church–Turing thesis applies to quantum computers, that is, every problem that can be solved by a quantum computer may be also solved by a Turing machine. However, some problems may theoretically be solved with a much lower complexity over a quantum computer than over classical computer. This is, for the moment, purely theoretical, as one does not know how building an efficient quantum computer.
Quantum complexity theory has been developed for studying computational complexity of quantum computing. It is used in post-quantum cryptography, which consists of designing cryptographic protocols that will resist to attacks with quantum computers, when such computers will really exist.
Problem complexity (lower bounds)
The complexity of a problem is the infimum of the complexities of the algorithms that may solve the problem, including the unknowns algorithm. Thus the complexity of a problem is not greater than the complexity of any algorithm that solves the problems.
It follows that every complexity that is expressed with big O notation is a complexity of the algorithm as well as of the corresponding problem.
On the other hand, it is generally hard to obtain nontrivial lower bounds for problem complexity, and there are few methods for obtaining such lower bounds.
For solving most problems, it is required to read all input data, which, normally, needs a time proportional to the size of the data. Thus, such problems have a complexity that is at least linear, that is, using big omega notation, a complexity
The solution of some problems, typically in computer algebra and computational algebraic geometry, may be very large. In such a case, the complexity is lower bounded by the maximal size of the output, since the output must be written. For example, a system of n polynomial equations of degree d in n indeterminates may have up to complex solutions, if the number of solutions is finite (this is Bézout's theorem). As these solutions must be written down, the complexity of this problem is For this problem, an algorithm of complexity is known, which may thus be considered as asymptotically quasi-optimal.
A nonlinear lower bound of is known for the number comparisons needed for a sorting algorithm. Thus the best sorting algorithms are optimal, as their complexity is This lower bound results from the fact that there are n! ways of ordering n objects. As each comparison splits in two parts this set of n! orders, the number of N of comparisons that are needed for distinguishing all orders must verify which implies by Stirling's formula.
A standard method for getting lower bounds of complexity consists of reducing a problem to another problem. More precisely, suppose that one may encode a problem A of size n into a subproblem of size f(n) of a problem B, and that the complexity of A is Without loss of generality, one may suppose that the function f increases with n and has an inverse function h. Then the complexity of the problem B is This is this method that is used for proving that, if P ≠ NP (a unsolved conjecture), the complexity of every NP-complete problem is for every positive integer k.
Use in algorithm design
Evaluating the complexity of an algorithm is an important part of algorithm design, as this gives useful information on the performance that may be expected.
It is a common misconception that the evaluation of the complexity of algorithms becomes less important, because of the exponential growth (Moore's law) of the power of modern computers. This is wrong because, this power increase allows working with large input data (big data). For example, when one want to sort alphabetically a list of a few hundreds of entries, such as the bibliography of a book, any algorithm should work well in less than a second. On the other hand, for a list of a million of entries (the phone numbers of a large town, for example), the elementary algorithms that require comparisons would have to do a trillion of comparisons, which would need more than one day at the speed of a million of comparisons per second. On the other hand the quicksort and merge sort require only comparisons (as average-case complexity for the former, as worst-case complexity for the latter). For n = 1,000,000, this gives approximately 30,000,000 comparisons, which may be done in a few seconds on a powerful computer.
Thus the evaluation of the complexity may allow eliminating many inefficient algorithms before any implementation. This may also be used for tuning complex algorithms without testing all variants. By determining the most costly steps of a complex algorithms, the study of complexity allows also focusing on these steps the effort for improving the efficiency of an implementation.
- Arora, Sanjeev; Barak, Boaz (2009), Computational Complexity: A Modern Approach, Cambridge, ISBN 978-0-521-42426-4, Zbl 1193.68112
- Du, Ding-Zhu; Ko, Ker-I (2000), Theory of Computational Complexity, John Wiley & Sons, ISBN 978-0-471-34506-0
- Garey, Michael R.; Johnson, David S. (1979), Computers and Intractability: A Guide to the Theory of NP-Completeness, W. H. Freeman, ISBN 0-7167-1045-5
- Goldreich, Oded (2008), Computational Complexity: A Conceptual Perspective, Cambridge University Press
- van Leeuwen, Jan, ed. (1990), Handbook of theoretical computer science (vol. A): algorithms and complexity, MIT Press, ISBN 978-0-444-88071-0
- Papadimitriou, Christos (1994), Computational Complexity (1st ed.), Addison Wesley, ISBN 0-201-53082-1
- Sipser, Michael (2006), Introduction to the Theory of Computation (2nd ed.), USA: Thomson Course Technology, ISBN 0-534-95097-3