For the number of weed
f, the rate at which it changes over time
t is given by
dtdf=af−a1cfwhere
c is the number of grass carp which predate upon this grass. The constant
a>0 denotes the rate at which the weed grows. More weed means more growth. The second term
−a1cf describes the rate of 'death' for this weed. It depends on the constant
a1, and effectively the number of predators
c, as well as the number of weed
f.
Note that if
c increases, it means more predators about hence a decrease in weed growth. Also, if
f increases, weeds compete against each other for nutrients. So the entire negative effect
cf is bunched up into one term.
For the carps, the ODE is
dtdc=bcf−b1cwhere
b,b1 are constants of growth and death respectively.
Currently, these models are dimensional. Is the number of weed / carps in hundreds? Thousands? Millions??? Are they densities per some volume of space?
We can bring these models down via a process called non-dimensionalisation whereby we simply scale the variables (in this case f,c,t) in such a way as to eliminate as many constants as possible.
In your link, time is brought down via
u=at. This means that we can simply divide our equations through by the constant
a and obtain
dudf=f−aa1cfdudc=abcf−ab1cWe can also scale
f=Az and
c=By where
A,B are dimensional constants (for us to rewrite in terms of the known constants), and
z,y are dimensionless variables.
Substituting these in yields
dudz=z−aa1Byzdudy=abAyz−ab1yWe can bring the coefficient of
yz in the first eqn down to
−1 if we impose that
B=a1a.
We can also choose
A=ba in order to bring the coefficient of
yz in the second eqn down to
1.
This leaves us with
dudz=z−yzdudy=yz−ab1ywhereby we can combine the ratio
ab1 into a single constant
β.
Thus, via nondimensionalisation, we have reduced the number of constants in our model down from 4 to only 1. The leftover variables z,y,u do not have any dimensions and can therefore be treated on level ground without having to worry about dimensions and their scaling factors.