I asked a little while ago on Twitter about whether there’s a given (searchable) name for a class of problem, and I realize given the answers that there isn’t enough information in 140 characters.
So here’s an attempt to state the problem more clearly:
Suppose you are told there arepoints in
, but not their positions, and are also given a list of Euclidean distances between some (or all) pairs of those points. You are asked to determine whether the distances are feasible or not.
For example, suppose I say there are four points in a plane, ,
,
and
, and that the distance
,
,
and
; if I’ve done the sketching right, those distances are feasible for points in a plane. If however
, the distances are infeasible.
Another way to frame an example, I guess, is to simply ask is it possible that there are four points in ,
,
,
and
, and that the distance
,
,
,
and
?
What I’m wondering is whether this class of geometric constraint satisfaction problems has a particular name that one might, you know, look up. Like the Boolean satisfiability problem and the knapsack problem have special names all their own.
Bill -
This is not the answer, but it does have some set of the same words that you are looking for, and chasing references to and fro might work:
http://www.jstor.org/pss/2699760
and it looks like the answer might be somewhere in this piece of math
http://www.ams.org/mathscinet/msc/msc2010.html?t=&s=28A12&btn=Search&ls=s
This looks promising as well, at least for defining some terms:
http://en.wikipedia.org/wiki/Measure_(mathematics)
This makes *much* more sense than it did on Twitter
. Assuming you’re in a “standard” Euclidean space, you’re essentially asking if the given distances satisfy the triangle inequality. You can generalize this to other metrics & spaces, which takes you into measure theory as mentioned above.
As far as a class name, something like TriangleInequalityChecker? (I’m crap at names, I’m afraid.)
Thanks Nic. The triangle inequality thing should be a lead, as well. Ed’s metric space links also seem to be leading me to the correct mathematicians’ terms.
You and Ron Jeffries think alike. I never really meant “what should I name the class in a program?”; I meant class as in “type” or “category”!
It looks promising that questions of this type might generally be answered by restricting your google search to ams.org, e.g.
site:ams.org Euclidean distances between pairs
as a search string corresponding to the URL
http://www.google.com/search?client=safari&rls=en&q=site:ams.org+Euclidean+distances+between+pairs&ie=UTF-8&oe=UTF-8
has a page of stuff which includes some AMS feature articles which look like surveys worth noting.