Non-Informative Prior
A prior distribution for a parameter vector which is intended to express ignorance about This can be tricky, and often leads to the use of densities which have an infinite integral (known as improper priors), such as a uniform density or
 
 
     
 
  Normal Distribution
In our usage, this includes both univariate and multivariate distributions. The density is
 
 
     
 
  NP-Complete
It is desirable that algorithms terminate in a time which is polynomial in the size of the problem and the accuracy required. Let P denote all problems for which there is such an algorithm. If we allow nondeterministic algorithms (which are allowed to choose the best option whenever there is a choice) the class of problems is called NP. Equivalently, NP is the class of problems for which a solution could be verified in polynomial time. It is widely believed that NP is strictly larger than P, but this remains an open research problem. A problem in NP is called NP-complete if proving if proving it was in P would establish P = NP, which should be regarded as strong evidence that no polynomial algorithm will ever be found. (Cormen et al., 1990; Sedgewick, 1990; Garey & Johnson, 1979.)
 
 
     
 
  NP-Hard
A NP-hard problem is one that implies a solution to every problem in NP (see NP-complete) but is not known to be in NP. Thus NP-hardness is strong evidence that no polynomial algorithm for the problem will ever be found.