Scribd
Upload
81 views1 page

Jacob Ian

The document introduces Jacobians by discussing how derivatives and gradients can be used to calculate small changes in output given small changes in input for differentiable functions. It then defines the Jacobian matrix J as a linear mapping between input and output vectors for a function from Rm to Rn, such that a change in the output vector dy is equal to J multiplied by the change in the input vector dx. The Jacobian is an m×n matrix with entries given by the partial derivatives of each output variable with respect to each input variable.

Uploaded by

Li Xuanji
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
81 views1 page

Jacob Ian

The document introduces Jacobians by discussing how derivatives and gradients can be used to calculate small changes in output given small changes in input for differentiable functions. It then defines the Jacobian matrix J as a linear mapping between input and output vectors for a function from Rm to Rn, such that a change in the output vector dy is equal to J multiplied by the change in the input vector dx. The Jacobian is an m×n matrix with entries given by the partial derivatives of each output variable with respect to each input variable.

Uploaded by

Li Xuanji
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 1

Jacobians

1 Introduction
To motivate the study of the Jacobian matrix and the Jacobian we begin by considering two operators
that you should have learnt as a little child, the derivative and the gradient, and consider them as a way
of answering this question: for a ’nice’ function, how do you calculate a small change in the output given
a small change in the input?

1.1 R→R
dy
Consider a function y(x) : R → R. If the function is differentiable and continuous we have dx = ( dx )dx.
dy
At any given point dx is a constant; hence small changes in y are approximately linearly related to small
changes in y.

1.2 Rn → R
Consider a function f (~r) : Rn → R, for instance, the scalar electric potential as a function of position.
We wish to know df as a linear function of d~r. We generalize the scalar derivitive to give the gradient
~ · d~r. In 3-D cartesian coordinates ∇f
such that df = ∇f ~ = ( df , df , df ).
dx dy dz

1.3 Rm → Rn
Consider a function ~y (~x) : Rm → Rn Notice that if x ⊂ Rn then dx ⊂ Rn . to find dy as a function of dx
we need a linear mapping between Rn and Rm . When m = n = 1 this mapping was multiplication by a
constant; when n = 1 the mapping was dot product with a constant vector. In the general case we would
probably need the mapping multiplication by a constant m by n matrix.

2 Construction
We now explicitly write out ~y as (y1 , ...yn ). yi is a function of ~x, so can use our previous result to write
~ i · dx. Continuing we construct the Jacobian matrix J, a m by n matrix such that d~x = Jd~y
dyi = ∇y
and
∂y1 ∂y1
    
dy1 ··· dx1
 ..   ∂x1 ∂xn   .. 
 .   . .. ..   . 
= . . (1)
 .   . .  . 
   
.
 .   ∂y  . .
m ∂y m
dym ··· dxm
∂x1 ∂xn

You might also like