Content deleted Content added
{{Copied |from=Index (computer science) |from_oldid=470513254 |to=Array data structure |to_diff=470647949 |to_oldid=470644178 |date=17:50, 10 January 2012}} |
→Arrays not analogous to vectors or matrices: new section |
||
Line 282:
The implementation of the iliffe array presented here makes the assumption that each row is allocated separately. However, Numerical Recipes in C provide a version of such vector with better locality by allocating the data as a contiguous memory block and have the indexing array points to beginning of rows in this block. Tests show few to no overhead between this and a traditional Dope vector. Would it be worthwhile to speak of this distinction ?
[[User:Joel falcou|Joel falcou]] ([[User talk:Joel falcou|talk]]) 18:27, 4 January 2012 (UTC)
== Arrays not analogous to vectors or matrices ==
(Referring to the 3rd paragraph) Arrays are not analogous to mathematical vectors, matrices or tensors, but rather to the objects called tuples in mathematical terminology. I appreciate that they have often been named this way, and there may even be cases where it is helpful to explain them like this, but the statement as it stands is not correct. I believe at the least there should be a mention of tuples. I am not aware of the mathematical term for multidimensional tuples.
Example:
Vectors and the related objects are linear objects. An array of customer details is valid in computer science, but it has no meaning to act on it using coordinate transformations, linear functions, inner products or any of the other machinery associated with linear mathematics. Conversely Maxwell's equations contain vectors but do not contain any arrays (or tuples). When a vector can be represented as a tuple/array, it must have elements that are members of a field, this is not the case for a general array.
I see that the paragraph doesn't have a reference, so unless there is some disagreement or discussion I will delete it, or possibly edit it.
David Drakard
| |||