What is the Directional Derivative at a Given Point?

  • Thread starter Thread starter jegues
  • Start date Start date
  • Tags Tags
    Derivatives
jegues
Messages
1,085
Reaction score
3

Homework Statement


See first figure.


Homework Equations


N/A


The Attempt at a Solution



See second figure. I defined direction of the line by which the two planes intersect as,

\vec{d}

and found that the point they are asking about is when,

t=1

and I'm stuck here. This is my first attempt at a question like this so I applogize in advance if I'm missing something obvious.

How do I figure out what direction is the direction of increasing x?

D_{u^{+}}f(x,y,z) = f_{x}(x,y,z)a + f_{y}(x,y,z)b + f_{z}(x,y,z)c

So,

D_{u^{+}}f(x,y,z) = ax^{2} + b2y -c2z

At the point (-1,1,3),

D_{u^{+}}f(-1,1,3) = a + 2b -6c

This is as far as I got.

Any suggestions?
 

Attachments

  • DIRDERIV.JPG
    DIRDERIV.JPG
    17.1 KB · Views: 402
  • DIRATT.JPG
    DIRATT.JPG
    39.7 KB · Views: 396
Physics news on Phys.org
To take a simpler example: If you have the function f(x,y)=x2+y2, and I told you to take the directional derivative along the line y=x...

The obvious vector to take is the vector (1,1) (scaled to be a unit vector). But that's not the only direction you could pick, (-1,-1) also is a vector pointing along the line y=x. If I addeed: take the direction of increasing x, then you would want to pick the vector that points along the direction in which x increases. This means that the vector pointing in the direction of the line has a positive x coordinate, so the right direction would be (1,1)
 
Office_Shredder said:
To take a simpler example: If you have the function f(x,y)=x2+y2, and I told you to take the directional derivative along the line y=x...

The obvious vector to take is the vector (1,1) (scaled to be a unit vector). But that's not the only direction you could pick, (-1,-1) also is a vector pointing along the line y=x. If I addeed: take the direction of increasing x, then you would want to pick the vector that points along the direction in which x increases. This means that the vector pointing in the direction of the line has a positive x coordinate, so the right direction would be (1,1)


Okay so the line I'm given is as follows,

\vec{l} = <0, -1, 2> + <-1,2,1>t

and for x to increase along the line l,

t \leq -1.

so if I simply compute t = -1, I'll get a vector with the direction,

\vec{u} = <1,-3,1>

Scaling this to be a unit vector,

\hat{u} = \frac{1}{\sqrt{11}}<1,-3,1>

Would this be a vector of magnitude 1 pointing in the direction of increasing x?
 
Bump, still looking for some help on this one.
 
Bumpity bump!
 
jegues said:
Okay so the line I'm given is as follows,

\vec{l} = <0, -1, 2> + <-1,2,1>t

and for x to increase along the line l,

t \leq -1.

so if I simply compute t = -1, I'll get a vector with the direction,

\vec{u} = <1,-3,1>

That's not right. Your \vec{u} is the vector from the origin to the point (1, -3, 1) on line l where t = -1. You want the direction of your line itself for which x is increasing. So you should just have \vec{u} = <1, -2, -1>. Make that into a unit vector and finish up.
 
Raskolnikov said:
That's not right. Your \vec{u} is the vector from the origin to the point (1, -3, 1) on line l where t = -1. You want the direction of your line itself for which x is increasing. So you should just have \vec{u} = <1, -2, -1>. Make that into a unit vector and finish up.


Okay so scaling it as a unit vector,

\hat{u} = \frac{1}{\sqrt{6}} < 1, -2, -1>

Now how can I use this to solve for a, b and c? That's what I want to do next right?
 
No...that's what you just found. \vec{u} = <a, b, c>
 
Raskolnikov said:
No...that's what you just found. \vec{u} = <a, b, c>

Oh okay! I think I get it now.

Thanks again.
 
  • #10
Sorry for ressurecting this thread, but I've got another quick question.

When someone asks me for the directional derivative of something at a given point, is my final answer a scalar?

In this case would it simply be the scalar provided by,

\nabla f(-1,1,3)\cdot \hat{u}

?

Thanks again!
 
  • #11
Yes, it's the scalar product of the gradient and a unit vector.
<br /> D_{u}f= \nabla f\cdot\frac{\bold{u}}{\left|\bold{u}\right|}<br />
 
Back
Top