4 ms·
Where's the vector Jacobian matrix?
by melony 4y ago
Where's the vector Jacobian matrix?
- omnicognate 4y agoYou don't need to explicitly form a Jacobian matrix to do AD. The key insight is that the matrix is just a representation of a linear function. In forward mode AD you evaluate the linear function the Jacobian represents (i.e. the derivative) at the same time as evaluating the function you are differentiating, usually without ever explicitly building the matrix. The derivative is a local linear approximation of the original function, so it composes the same way. This allows you to compute the parts of it as you compute the parts of the original function, and pass on the intermediate results alongside the intermediate results of the original calculation.