how to define a subset of a population which give the max values for a set of functions

90 Views Asked by At

I am a computer scientist, struggling with writing a mathematical definition for a subset of a population which contains individuals that are the best scoring in the population for a set of functions. I have a population, $P$. Each individual in the population, $p \in P$, is assessed over a number of objectives, given by the functions $a_1(p),\dotsc,a_n(p)$. I wish to define a subset of $P$, which is the set of individuals in $P$ that return the maximal values for each assessment.

Currently, I have the following definition:

Let the set $P$ denote the current population, with an individual solution defined as $p \in P$. Next, for a given domain with $n$ objectives, defined by the functions $a_1(p),\dotsc,a_n(p)$, where $a: P \rightarrow \mathbb{R}$, let the set $A = \{a_1(p),a_2(p),\dotsc, a_{n}(p)\}$. Let the subset $P_{max} \subset P$ contain the maximal solutions for each assessment, where $P_{max} : \underset{a \in A, p\in P}{max} a(p)$.

I think that (if it makes any sense at all) in the way I have written it, $\underset{a \in A, p\in P}{max} a(p)$ refers to set of maximal values for each assessment, not the individuals in $P$ which return the maximal value for each assessment. I'm not sure how to express the subset correctly

1

There are 1 best solutions below

2
On BEST ANSWER

The terminology Arg max is used to denote the set of points that maximize a real-valued function.

Aside: An Arg max is well defined for each of the objectives $a_1,\ldots, a_n$, but wouldn't the maximizing arguments (individuals) differ for each of the $n$ objectives? It's unclear how you would define a subset of individuals who maximize the ensemble of objectives, unless you find a way to combine the assessments over all $n$ objectives into a global score of some sort, and find the individuals who maximize that global score.