# Set configuration definition

Tags:
1. Mar 11, 2015

### zrek

Please help me to define correctly, in the language of mathematics, the configuration of sets shown on the picture.

1. The problem statement, all variables and given/known data
I'd like to define the following rules:
U is a set with infinite members.
L is a list or set of properties. Every property (Ls1, Ls2 ... ) have a value ( 0,1,2,...P )
Every member of U is defined by its unique property list. Every member is paired with one and only one value from each of the Ls1,Ls2... properties.
The U contains i+1 subsets.
F is a list of functions, or it is the set of subsets of U.
Every subset of U is paired with a function in F, except one, which contains all of the members which are not in any subsets defined by F.

The subsets of U are determined the following way:
An Xk member of U is in subset Fj, ( 1<=j<=i ) if the function Fj outputs "true" after processing the Ls1, Ls2 ... values of Xk.

For example:
Properties of X0 are:
Ls1: 3
Ls2: 2
Ls3: 5
...
Since only the function F1 returns "true" for this input, the X0 is solely in the F1 subset of U.

3. The attempt at a solution
I have difficulties even to correctly define the L set and the members of U.

#### Attached Files:

• ###### subsets2.png
File size:
33.1 KB
Views:
78
2. Mar 11, 2015

### Staff: Mentor

What do you mean with "correctly define the L set"?

You can let each element in U be a function uj: M -> {1,2,...,P}
Where M is the set of properties.

3. Mar 12, 2015

### haruspex

zrek, I find your notation very confusing. Do you mean that the Ls are functions from U to {1 ... P}, and that given x in U the vector (Ls1(x), Ls2(x)...) is unique?

4. Mar 12, 2015

### zrek

I mean I don't know how to describe it in the language of mathemathics, with clear formulas.
Is L a set at all? Or maybe a list? (The index of the elements of the L is an important data, I think)
L is only used to define the requirements of the elements of U.
The requirement of the elements of U is that to have infinite values, one from each member of the L.
For example u0 is a zero element, if all of its values are 0. u0={0,0,0, ...} Is this a good notation for it?
Let's say umx is a "max element", if if all of its values are P. umx={P,P,P, ...}

I don't want to say that the elements of U are functions. The functions describe the subsets of U.
There are elements in U which are can not be accepted by any of the functions in F.

Thank you for your help, mfb.

Last edited: Mar 12, 2015
5. Mar 12, 2015

### zrek

Sorry, you are right, it is confusing, I'm not experienced describing sets, I'll try to learn it.

I'm uncertain of the meaning of "vector". I know that v(x,y) is a vector, if x and y are real numbers. In our case the index values are discrete. Can we say that u(a,b) is a vector, if "a" and "b" can have only discrete values from [0,1,2...P] ? If this is ok, then it makes easier, we can say that U is a set of vectors.

I found a good analogy. Please consider the followings as an example:

U is a set of infinite many marbles.
The marbles have properties. All of the property types are listed in the list L. (there are infinite many of them)
Every property value is "normalized" between 0 and P.
The first property type (Ls1) is the weight. The weight of a marble can be between 0 and P.
The second property type (Ls2) is the roundness. The roundness also a value between 0 and P.
The marbels are colorful. There are 3 color component: R,G,B.
The third property type (Ls3) is the R component ....
(...and so on, there are infinite many property types.)
Every marble have to have a value for every property type.

Now we can create subsets, there are i+1 subsets.
For example we can create a function that determines whether or not a marble is heavy.
F1 function outputs true for those marbles with the Ls1 property is more than P/2. F1 function describes the "heavy marbles" subset of marbles.
F2 function outputs true if the Ls3 value of a marble is P, and the Ls4 and Ls5 (G,B components) are 0. The F2 determines the "pure red" subset of marbles.
(...)
There are finite number of functions.

There are marbles which will be accepted by one function.
There are marbles which will be accepted by several functions.
There are marbles which will be not accepted by any of the functions.

I hope that this analogy makes it clear.

6. Mar 12, 2015

### Staff: Mentor

They do what functions are doing. The description will become easier if you call them functions.
Those are different (types of) functions.

With mathematical vectors this is a bit tricky (it can be tricky to find a corresponding vector space), with computer science vectors it is no problem.

7. Mar 12, 2015

### haruspex

Not quite. It's the (x, y) that's the vector. v(x,y) looks like a function of two variables (or, equivalently, a function of the vector (x, y)).
The correct term is a 'tuple' (http://en.wikipedia.org/wiki/Tuple). Some allow tuple as a species of vector; others, as you say, require vectors to live in a vector space.
The notation for that is $Ls_j:U\rightarrow \{0,...,P\}$. (In words, "The function Lsj is a map from U to the set {0,..,P}")
No, U is not a set of vectors. U is your plain set of things x. Given an x in U, you can construct the (infinite) tuple L(x) = (Ls1(x), Ls2(x), ... ). Given distinct x, y in U, $L(x) \neq L(y)$, right?
I'll call the set of all such tuples T. T = {(t1, t2, t3, ...}:ti ∈{0, ... , P}}. So $L:U\rightarrow T$.
U is infinite, so has infinitely many subsets. I assume you mean there is some particular collection of subsets A of interest. |A|=i+1.
Every member of A is paired... etc.
You have a choice of how to define Fj. For clarity I'll write F' here for one of the choices.
You can have them as functions of x in U
$F'_j:U \rightarrow \{ true, false \}$
or as functions of the tuples.
$F_j:T\rightarrow\{true, false\}$
These are connected by $F'_j(x) = F_j(L(x)) \forall x \in U$. I.e. F'j is the composition of Fj on L.
For each F'j, you can define a subset Aj of U by $A_j = \{x \in U: F_j(x) = true\}$. (Or, more succinctly, $A_j = \{x \in U: F_j(x)\}$.)
This is sometimes written as an inverse function: $A_j = {F'}_j^{-1}(true)$, but it's a slight abuse of notation (http://people.clas.ufl.edu/groisser/files/inverse_images.pdf).
Note that the Aj are not necessarily distinct.
Then $A = \{A_j\} \cup \{U- \cup A_j \}$.

Hope this helps.

8. Mar 13, 2015

### zrek

Mfb, haruspex, thank you for the answers. I got so many information, I'd like to thinking about it (few days) before my next question.

9. Mar 18, 2015

### zrek

Thank you for your help Mfb and haruspex, finally I think I figured out what to do.