Skip to content

Latest commit

 

History

History
56 lines (46 loc) · 2.39 KB

Complexity.md

File metadata and controls

56 lines (46 loc) · 2.39 KB
Raw
f(n) = O(g(n))  ⇔  |f(n)| ≤ |g(n)|·k           "grows less than"
f(n) = Ω(g(n))  ⇔  f(n) ≥ g(n)·k               "grows more than"
f(n) = Θ(g(n))  ⇔  g(n)·k1 ≤ f(n) ≤ g(n)·k2    "bounded by"

(Insert "∃k>0: ∃n0: ∀n>n0" where needed.)

Master theorem

An algorithm has complexity T(n) = a T(n/b) + f(n).

  1. f(n) = Θ(n^(<logb(a))) → T(n) = Θ(n^(logb(a)))
  2. f(n) = Θ(n^(logb(a)) log(n)^k) → T(n) = Θ(n^(logb(a)) log(n)^(k+1))
  3. f(n) = Θ(n^(>logb(a))) → T(n) = Θ(f(n))

Typical complexities

Complexity Description
O(1) Getting an element from an array
O(m α(m,n)) Best minimum spanning tree (α = inverse Ackermann)
O(log n) Binary search
O(n) Maximum of an unsorted array
O(n log n) Best comparison sort
O(n^2) Naive vector cross product
O(n^3) Naive matrix multiplication
2^O(log n) P; Karmarkar (Linear programming); AKS (Primes)
2^o(n) Integer factorization
2^O(n) E; TSP, 3-SAT, Graph-coloring
  • SAT: can a given expression with variables, AND, OR, NOT and nesting be true?
  • TSP: minimum distance for a traveling saleswoman which must go through a set of cities.
  • Graph coloring: give each vertex a different color than its neighbors, with a fixed number of colors (2 is O(n), 3 is O(1.3289^n), k is O(2.445^n)).
  • Knapsack: keep a maximal value out of a set of elements with a value and a mass, given a limit to how much mass you can keep. It has a known O(n Mass) dynamic programming solution, but is NP-complete.
  • Exact cover: given a bunch of subsets (tetris piece locations) of a set (board), select the subsets that cover the whole set with no overlap. Pentomino tilings, Sudoku and N queens are of that form. Donald Knuth implements it using "Dancing Links".

Complexity classes

  • P: solved on a deterministic Turing machine in polynomial time.
  • NP: solved on a non-deterministic Turing machine in polynomial time (has overlapping parallel universes). A solution is verified by a deterministic Turing machine in polynomial time. (The non-deterministic solution generates all candidates and checks them in parallel.)
  • NP-hard: if that problem was O(1), all problems in NP would be in P.
  • NP-complete: NP-hard and NP.

P could be equal to NP, but for all practical purposes, it is not (which may be proved in the future).