On algorithms preserving neighborhood, to file and retrieve information in a memory

Algorithms to file and retrieve information in a memory are used currently for many applications: compilers, systems, business, artificial intelligence, etc.... For example, binary search and Hash coding allow a compromise between memory size and number of access operations to be obtained. But, in many cases, distance or neighborhood has to be preserved. We propose two new methods having this quality. The algorithms of our methods are determined by properties of the set of information to be filed. The first is based on variation properties. Approximation methods are utilized. The second exploit the probability properties of the set. The resulting performances are comparable to Hash coding's and neighborhood is preserved.