Thread: Storing position/velocity of agents -- opinions wanted

Hi,

I have to do some 2D and 3D agent-based-models, need to track position/velocity of agents. In the past I had used some vector and agent classes I had written myself. Lately, I find the vector class cumbersome and have become more acquainted with std::vector.

I figure I could do the simulations I have to do by just adjusting my agent class to use std::vector -- I don't think I can outdo STL code. Any thoughts? Is there any hypothetical benefit to keeping custom vector classes anymore? I need them only to track floating-point co-ordinates in 2 or 3-space.

EDIT: The reason I made vector classes in the first place was during my first learning stages with c++. I haven't done agent-based simulations in a while, and so I am looking at my old code and would like to improve it. If anyone with graphics programming experience can weigh in on the most efficient way to store position, velocity of agents which you have to animate in openGL.

Keep using your custom vector class. `std::vector` is a storage container, not a mathematical object.

Oh, do you mean that I can't think of them as Euclidean vectors and do +, -, * with them? Yeah, if I have to write new math functions to for the std::vectors it does seem to defeat the point.

Out of curiosity, what is the motivation for calling these STL containers "vectors"?

Originally Posted by Ocifer

Oh, do you mean that I can't think of them as Euclidean vectors and do +, -, * with them? Yeah, if I have to write new math functions to for the std::vectors it does seem to defeat the point.

That's exactly what I mean.

Originally Posted by Ocifer

Out of curiosity, what is the motivation for calling these STL containers "vectors"?

I honestly have no idea.

Originally Posted by Ocifer

Out of curiosity, what is the motivation for calling these STL containers "vectors"?

I imagine the name std::array was considered, but rejected because "array" implies certain things, like linear arrangement in memory, which are not actually guaranteed by the container. Actually, I think it was a stupid choice.

Also, your attempts to simultaneously store position and velocity will be defeated... By Werner Heisenberg.

Originally Posted by brewbuck

I imagine the name std::array was considered, but rejected because "array" implies certain things, like linear arrangement in memory, which are not actually guaranteed by the container.

I had the impression that the lack of a C++98 guarantee that std::vector's elements will be contiguous in memory was a defect from the start rather than something deliberate that was changed in C++03. Maybe if they had come up with the new std::array earlier, they would have chosen std::dynamic_array instead, heh.

Yeah, I guess if you think about it REALLY in the abstract, a vector is just an ordered set of data (thinking back to Linear Algebra when my professor would often switch from using the vernacular "list of basis elements" to "basis vector" when he mathematized his vernacular English). Weird, that word still has a lot of baggage. I just ASSumed it meant Euclidean vector. Anyway, thanks, guys.

Originally Posted by manasij7479

Well, a group of similar things, that can change in size, does look like a mathematical vector.
(I am referring the "can change in size" from my limited knowledge in linear algebra.. I may be somewhat wrong.)

o_O
A mathematical vector can't change size. It is always a fixed size, but you can naturally "make" an infinitely long vector.

Originally Posted by Ocifer

Out of curiosity, what is the motivation for calling these STL containers "vectors"?

Historically, arrays were used in Algol to represent vectors, and the terms were used interchangeably - albeit imprecisely. Algol is one of those languages that influenced the nomenclature of computer science, and is also a precursor of most modern programming languages including C and C++.

Mathematically, a vector is defined as a "collection of similar objects that may be manipulated and characterised collectively using a set of defined rules".

The "Euclidean vector" you are using is a special case of that. For example, the x, y, z in a 3D euclidean vector are the "similar objects", there are defined rules for addition, dot product, cross product, etc, and the vector has characteristics (properties like length) that are not owned by its elements.

A (mathematical) vector is just something that obeys certain linearity conditions which are expressed in terms of addition and scalar multiplication. Anything that adheres to those rules can be a vector. That includes scalar values, matrices, infinite arrays of things, anything that has a well-defined addition and scalar multiplication along with the aforementioned linearity conditions.

You're thinking of the concept from linear algebra called a "vector space" (or sometimes "linear space"), brewbuck.

Vectors are also used in non-linear algebra which, by definition, is not concerned with linear spaces (except maybe as a special case). For example, the coefficients of a polynomial are often represented using vectors, but operations on polynomials are nonlinear (they don't satisfy properties of additivity or homogeneity).