Binary variable to count appearances

Question

Let $x \in \mathbb{R}^n$ be an optimization variable. Now, at a constraint, I would like to count how many times a value, say $2$, appears in $x$ decision.

I think we can have a binary variable $y_i$ indicating whether $x_i =2$. So, $x_i - 2 = 0$ should imply $y_i = 1$. But, anything except $0$ should imply $y_i = 0$. What is the easiest way for this?

Note: since we can subtract $2$ from each element of $x$, we are interested in the number of zeros in $x-2$. So, 'the number of zeros in a decision vector' constraint will also make it.

We may assume $x$ consists of elements $x_i< M$ for some constant $M$

You can't really check for exact inequality. If you're willing to allow a small tolerance around 2 (say), then the approach described here might work to set your $y_i$ variables. — LarrySnyder610, Sep 12 '19 at 02:07
I think there should be something with demeaning $x$ and then taking the absolute values... But, not sure. — independentvariable, Sep 12 '19 at 02:10
$x_i - 2$ for all $i$. Now I tried to use Yalmip's iff command. Apparently, this works. But idk how — independentvariable, Sep 12 '19 at 02:23
Just to clarify, what you want is to define a variable $y$ associated with an $x$ such that $y=1$ iff $x=2$ (otherwise $y=0$)? — Nikos Kazazakis, Sep 12 '19 at 11:09

score 8 · Answer 1 · answered Sep 12 '19 at 17:43

8

As LarrySnyder610 said, you cannot do exactly what you want when $x_i$ is continuous. (You can if it is an integer variable.) I discussed how to model this particular issue here: Flagging a Specific Variable Value.

answered Sep 12 '19 at 17:43

prubin

39,078
3
37
104

Binary variable to count appearances

1 Answers1