3

For all $i \in I,j\in J$ and $k\in K$, define variables $x_{ij}, z_{ijk}\in\{0,1\}$, $y_{ij}\geq 0$ and constants $c_j, e_{ijk}, d_j, f_j >0$.

We have the following constraint $$\sum_{j\in J_1}c_j\frac{x_{ij}}{\sum_{k\in K}e_{ijk}z_{ijk}} + \sum_{j\in J_2}c_jf_j x_{ij} = \sum_{j\in J_1}d_j\frac{y_{ij}}{\sum_{k\in K}e_{ijk}z_{ijk}} + \sum_{j\in J_2}d_jf_j y_{ij} \quad \forall i\in I,$$

where $J_1, J_2 \subset J$.

We also know that $$\sum_{k\in K}e_{ijk}z_{ijk}=1 \quad \forall i\in I,j\in J.$$

Now is there a way to linearize these two components? $$\frac{x_{ij}}{\sum_{k\in K}e_{ijk}z_{ijk}}, \quad \frac{y_{ij}}{\sum_{k\in K}e_{ijk}z_{ijk}}$$

I saw some interesting answers here, here and here but I'm having some problems using them for our problem.

Vitamin Z
  • 103
  • 5

1 Answers1

6

The two denominators are equal to $1$, so just omit the denominators, yielding a linear constraint.

RobPratt
  • 32,006
  • 1
  • 44
  • 84
  • Thank you. I didn't notice that! Is there a way in general to linearize division of two variables? – Vitamin Z Jan 01 '22 at 15:36
  • Here is a Taylor series approximation for a function of two variables https://math.stackexchange.com/questions/69610/taylor-series-in-two-variables – Daniel Jan 02 '22 at 03:25