Question.. (1 Viewer)

+Po1ntDeXt3r+

Active Member
Joined
Oct 10, 2003
Messages
3,527
Gender
Undisclosed
HSC
2003
wats a directional derivative?

if g(x,y) = 2xy - y^3 at (4,2) in the direction , u =(2,1) [a vector]... how is the directional derivative found? and wat is it?
 

turtle_2468

Member
Joined
Dec 19, 2002
Messages
408
Location
North Shore, Sydney
Gender
Male
HSC
2002
There's some formula involving the partial derivative of g wrt x and pd of g wrt y. It should be in any decent multivariate textbook...
 

+Po1ntDeXt3r+

Active Member
Joined
Oct 10, 2003
Messages
3,527
Gender
Undisclosed
HSC
2003
thanks!
yer i kno the formula.. but i just wanted to kno how to visualise it .. yey!
 

+Po1ntDeXt3r+

Active Member
Joined
Oct 10, 2003
Messages
3,527
Gender
Undisclosed
HSC
2003
i kno im not suppose to.. but i need it cos i keep forgettting sumfing.. :S

i got up to a) but b) im forgettin sumthing :s its for practice but im stumped
 

wogboy

Terminator
Joined
Sep 2, 2002
Messages
653
Location
Sydney
Gender
Male
HSC
2002
The subspace W is equal to span{a,b,c,d}, where:
a = [1, 1, 1, 1], b = [0, -1, -1, -1], c = [0, 0, 1, -1]

To produce an orthonormal basis for W using the Gram Schmidt method, you do the following:

1. Take one of the vectors, say a, and divide by its norm to get q1. q1 = a/|a|

2. Take another of vectors, say b, and subtract it's own projection onto q1, then divide this new vector you get by its norm to get q2. q2 = (b - (b.q1)q1 )/|b - (b.q1)q1|

3. Take the final vector, c, and subtract it's own projection on span{q1, q2}. Since q1 and q2 are orthogonal, it suffices to simply subtract the projection of each vector from c (otherwise you'd have to use the subspace projection formula), and then divide this resulting vector by it's norm to get q3. q3 = (c - (c.q2)q2 - (c.q1)q1)/|c - (c.q2)q2 - (c.q1)q1|

Now q1, q2, and q3 form an orthonormal basis for W. This procedure can be extended to more than three vectors if required. And keep in mind the answer to this question is not unique.
 

Users Who Are Viewing This Thread (Users: 0, Guests: 1)

Top