In this post you will see how to linearize max functions, min functions, and absolute value functions. These can be expressed mathematically as:

Let’s start with the **modulus**, as the other two depend on it.

Here you simply replace that one constraint by the following two:

The **min function** can be linearized as:

First, we decompose the min function into a max and a modulus.

Then we get rid of the max.

Finally, the answer is using the following five constraints:

Remember to penalize and such that only one of them will be different from zero in any solution.

The **max function** can be linearized as follows:

From here on things are very similar to the min function…

And finally, the five necessary constraints are:

And again, remember to penalize and .

*Thanks to Lieke van der Heide.*

This was useful? Buy me a cup of coffee to help me keep this website running!