PDA

View Full Version : Chain rule for implicit differentiation


Tuomo
Jan21-12, 12:00 PM
I have derivative dx/dt = y(u(t)) * z(u(t)) + u(t)
Now, what is dx/du ? I know the chain rule should help, but I am stuck :-(

fluidistic
Jan22-12, 11:44 AM
I have derivative dx/dt = y(u(t)) * z(u(t)) + u(t)
Now, what is dx/du ? I know the chain rule should help, but I am stuck :-(

Maybe \frac{dx}{du}=\frac{dx}{dt} \cdot \frac{dt}{du}.

Tuomo
Jan23-12, 05:01 AM
Maybe \frac{dx}{du}=\frac{dx}{dt} \cdot \frac{dt}{du}.

Thanks fluidistics.

Ok, so, what I get is:

\frac{dx}{du}=\frac{dx}{dt} \cdot \frac{dt}{du}=\left[y(u(t)) \cdot z(u(t)) + u(t)\right] \cdot \frac{dt}{du(t)}

Now, what the heck is this \frac{dt}{du(t)} ? Any idea how should I treat it?
Is it just a reciprocal of \frac{du(t)}{dt}? In other words, can I just divide the original function with the time derivative of u(t)?

\frac{dx}{du}=\frac{dx}{dt} \cdot \frac{dt}{du}=\left[y(u(t)) \cdot z(u(t)) + u(t)\right] \cdot \left[\frac{du(t)}{dt}\right]^{-1}

Thanks for help and comments! Its is almost... uhhhh... 15 years since I studied this stuff in university.