Discrepancy between implied growth rate and dilution rate

Software
1

Hi developers,
I find the implied growth rate does not reflect the growth rate estimated from dilution rate in turbidostat runs. It seems to change more slowly than the actual dilution rate.

As shown below, the bottom curve is dilution rate calculated by 1/(timestamp of nth dilution - timestamp of (n-1)th dilution), and the upper curve is the implied dilution rate copied from the data csv. The two curves roughly agree with each other but substantial discrepancy is seen towards the end.

Any suggestions on how to tune the parameters of growth rate calculation to make it match better the dilution rate?

Thanks!

image
image834×442 124 KB

2

@CamDavidsonPilon would implementing dilution rate as a “control input” into the kalman filter not be the way to solve this. This way it knows what to expect instead of just reacting to it.

3

Thanks! Looking forward to the update.