Vector Form of the Chain Rule
The chain rule is the sum of 2 products, and thus it can be
written as the dot product of 2 vectors. That is,
|
|
dw
dt
|
= |
¶f
¶x
|
|
dx
dt
|
+ |
¶f
¶y
|
|
dy
dt
|
= |
|
|
¶f
¶x
|
, |
¶f
¶y
|
|
· |
|
|
dx
dt
|
, |
dy
dt
|
|
|
| (3) |
Let's introduce some new notation to exploit this interpretation of the
inner product.
If r( t) =
á x( t) ,y(t)
ñ is a curve in the xy-plane, then we let w = f( r) denote the function w( t) = f( x( t) ,y( t) ) . Moreover, the last
vector in (3) is the velocity v of r( t) , which we denote as
|
v = |
dr
dt
|
= |
|
|
dx
dt
|
, |
dy
dt
|
|
|
|
Next, we define the gradient of a function f( x,y) to
be
|
Ñf = |
|
|
¶f
¶x
|
, |
¶f
¶y
|
|
|
|
where ``Ñ" is pronounced ``del'' and is the gradient operator. In
terms of this new notation, (3) can be written as
Other notations for the gradient of a function include grad(f) and df/dr. This last notation is especially nice
because it allows us to write the chain rule (4) as
which is reminiscent of the chain rule for functions of a single variable.
EXAMPLE 3 Evaluate df/dt using (4) given
that f( x,y) = 2x2y+y2 and r( t) =
á cos( t) ,sin2( t)
ñ
Solution: Since Ñf =
á 4xy,2x2+2y
ñ
, substitution of r( t) =
á cos(t) ,sin2( t)
ñ yields
|
|
|
|
á 4cos( t) sin2( t) ,2cos2( t) +2sin2( t)
ñ |
| |
|
|
á 2sin( t) 2sin( t) cos(t) ,2
ñ |
| |
|
|
|
Since v =
á -sin( t) ,2sin( t)cos( t)
ñ =
á -sin( t) ,sin( 2t)
ñ , the chain rule (4) implies that
|
|
df
dt
|
= Ñf·v =
á 2sin( t)sin( 2t) ,2
ñ ·
á -sin(t) ,sin( 2t)
ñ |
|
Evaluating the inner product then leads to
|
|
|
|
-2sin2( t) sin( 2t) +2sin( 2t) |
| |
|
| |
|
|
|
EXAMPLE 4 Evaluate df/dt using (4) given
that f( x,y) = sin( pxy) and r(t) =
á et,e-t
ñ
Solution: Since Ñf =
á pycos( pxy) ,pxcos( pxy)
ñ , substitution of r( t) =
á et,e-t
ñ yields
|
|
|
|
á pe-tcos( pete-t) ,petcos( pete-t)
ñ |
| |
|
|
á pe-tcos( p) ,petcos( p)
ñ |
| |
|
|
|
Since v =
á et,-e-t
ñ , the chain rule (4) implies that
|
|
|
| |
|
|
á -pe-t,-pet
ñ ·
áet,-e-t
ñ |
| |
|
| |
|
|
|
Check your Reading: How many trig identities did we use in
example 3?