- #1
evilcman
- 41
- 2
I have a scalar function f dependent on a few variables $x_i$, and I would like to change variables, so that [tex]y_i = \sum_j {M_{ij} x_j},[/tex] where M is an invertible matrix independent of the x_i-s, and compute:
[tex]
\frac{\partial f}{\partial x_i} = \frac{\partial f}{\partial \left( \sum_j {M^{-1}_{ij} y_j} \right)} = \sum_j Q_{ij} \frac{\partial f}{\partial y_j}
[/tex]
I suggest that the last identity is true for some matrix Q. Is there a general formula for Q?
[tex]
\frac{\partial f}{\partial x_i} = \frac{\partial f}{\partial \left( \sum_j {M^{-1}_{ij} y_j} \right)} = \sum_j Q_{ij} \frac{\partial f}{\partial y_j}
[/tex]
I suggest that the last identity is true for some matrix Q. Is there a general formula for Q?