Generate Random Numbers

AIMMS has a number of distribution functions included, in order for you to be able to generate random numbers very efficiently. This capability might be extremely useful when you want to quickly populate a multidimensional parameter. This functionality was used to randomize input values for the Knapsack example, example that we will be using as an example here.

See also

You can find the complete list of functions available. A few of them are: Normal Distribution, Log Normal Distribution and Uniform.

Example

On Knapsack example all input parameters are over s_items set. And this set is defined by p_itemQuantity as shown:

Set s_items {
        Index: i_item;
        Parameter: ep_selectedItem;
        Definition: ElementRange(1, p_itemQuantity, prefix:"item-", fill:1);
}

p_itemQuantity can be changed to any integer value from the example, starting as 14 default. For this model, we would like to generate numbers for 3 parameters, so it would be from 3 generated values (for 1 item) to up to 42 values (for 14 items) or even more.

To populate this data, the procedure pr_randomizeData was created:

1empty p_itemValue, p_itemWeight, p_itemBound;
2
3p_itemValue(i_item) := uniform(0, 200)*1[$];
4p_itemWeight(i_item) := uniform(0[lb], p_maxWeightKnapsack/3);
5p_itemBound(i_item) := ceil(uniform(0, 10));

  • where p_maxWeightKnapsack is the maximum weight of the knapsack.

If we would plot our parameter distribution we would end up with a uniform distribution.