Question about Using Noise Inputs in model

Use this forum to post Vensim related questions.
Post Reply
sudeepdesai
Junior Member
Posts: 2
Joined: Tue May 20, 2008 4:40 pm

Question about Using Noise Inputs in model

Post by sudeepdesai »

Hi All,

I am making a model which involves 2 process and a stock in between them. I need to introduce some noise in the system to show the variations of the process and see the effect on the stock inbetween.

I use the input generator to introduce noise in the model. I let the system start in equlibrium, introduce noise at some point of time and observe the behaviour of the system for the simulation. The stock always increases due to introduction of noise. I have observed different behaviour of the system based on these things:

1) When I use the same noise input for both the processes, the stock in model tends to faltten out at a particular level. (there are variations but around a specific level)

2) When I use different noise inpus for the two proceses. the stock never reaches an equilibrium.

In reality the two process will always have different variation that is why I made a new noise generater for the second process. But what I want to understand is why the system reached equilibrium in the first case?

I have taken care that the stock does not go below zero.
sudeepdesai
Junior Member
Posts: 2
Joined: Tue May 20, 2008 4:40 pm

Post by sudeepdesai »

I think I should add a bit more details...

THe two process are balanced. The noise in both processes is to show their instability. Average throughput of both the process is same.

The Nose start time and the Noise standard deviation for both noise inputs is the same. Only the noise seeds are different.

THanks!
bob@vensim.com
Senior Member
Posts: 1107
Joined: Wed Mar 12, 2003 2:46 pm

Post by bob@vensim.com »

If you really just have

stock = INTEG(noisy inflow - noisy outflow,init stock)

and both inflow and outflow are iid then the behavior is exactly what you would expect (actually the just need to same mean value, not even identical distributions). As TIME STEP tends toward 0 you will get closer and closer to a straight line for the level (assuming the power of the noise is constant so that the standard deviation decreases as TIME STEP decreases).
Post Reply