What is Optimizer

​

Optimizer refers to the optimization function of the feed design model.

When proceeding from feed evaluation to feed design, it is required that each nutrient such as roughage, carbohydrates, protein, vitamins and minerals meet the required amount and fall within the physiological and theoretical range. In addition to that, improving production volume (sales) and gross profit could also be a goal.

By increasing the amount of each ingredient in the feed by 500g and checking the nutrient concentration, production response, and benefits one by one through trial and error, one day we will arrive at the "optimized feed content." However, no amount of time is enough.

Therefore, an automatic calculation program, Optima isor has been devised and introduced to CNCPS.

There are two types of optimizers:

​

• ​linear optimizer

This is a calculation method that seeks the optimum solution based on linear relationships. A well-known example is the minimum cost calculation, and NDS has introduced a program to find the minimum cost recipe solution within the range of minimum/maximum amount of feed and within the limits of each nutrient.

​

• nonlinear optimizer

It is a method applied to factors that are in a curve relationship. NDS employs a design system that derives the best possible solution from a range of possibilities for each factor, given arbitrarily chosen objectives and constraints.

​

​

​

​

​

 

 

 

 

 

 

The figure above shows the linear relationship on the left and the non-linear relationship on the right. In this chart, there are only two relevant factors, the vertical and horizontal axes. If more than one factor is added to the non-linear horizontal axis, it will look like the figure below. By changing not only the number of factors on the horizontal axis, but also the number of factors on the vertical axis, and by changing the limit range and priority of each factor, the number of "peaks" will also increase.

​

​

​

​

​

​

​

​

​

​

The NDS nonlinear optimizer incorporates several nonlinear algorithms (calculation methods) and is programmed to quickly and efficiently converge candidate solutions from multiple "mountains" to find the optimal solution. (Figure below).

​

​

​

​

​

​

​

On the other hand, CNCPS is inherently classified as a non-linear model.

The linear model assumes constant availability of nutrients in each feed ingredient. For example, if the intake of CP16% feed is increased to 10kg, 20kg, and 30kg, the optimal solution will be calculated on the assumption that the protein available to the animal will always be 16% of the intake. ​

In the non-linear model, calculations are based on actual physiology, such as an increase in intake, an increase in transit velocity and a decrease in decomposition rate. An example of a non-linear approach is to increase the supply of a certain nutrient and look at the production response to it, and set the optimal supply using the point at which the reaction rate changes as an "inflection point."

For nonlinear models, a large number of complexly intertwinedIt can be said that nonlinear optimizers calculated from factors are more suitable for field applications.