18.02A Topic 26: Tangent plane approximation, directional derivatives.Author: Jeremy OrloffRead: TB: 19.1, 19.2 SN: TAFrom last time we have the tangent plane approximation.∆x = (x − x0), ∆y = (y − y0), ∆w = f(x, y) − f(x0, y0)∆w ≈∂w∂x0∆x +∂w∂y0∆yThe supplemental notes §TA give an analytic argument for thisand its genaralization to 3 dimensions:Suppose w = f(x, y, z) and∆x = (x − x0), ∆y = (y − y0), ∆z = (z − z0) then∆w = f(x, y) − f(x0, y0) ≈∂w∂x0∆x +∂w∂y0∆y +∂w∂z0∆zxyz....................................................................•Vector fieldsExample: F =xx2+ y2i +yx2+ y2j.This can be visualized as a bunch of vectors in the plane.Gradient∂w∂x0= rate of change inbi direction,∂w∂y0= rate of change inbj direction.∂w∂x0,∂w∂y0= gradient of w = ∇wTo evaluate at P0we write ∇w(P0). (Will use this in a moment.)Example: Let f(x, y) = x2y + y2x + y3⇒ ∇f = (2xy + y2)i + (x2+ 2xy + 3y2)j.Note: If f(x, y) is a scalar function then ∇f is a vector field.Note the graph if z = f(x, y) is a surface in 3 dimensional space, but the gradient∇f is a vector field in the plane.Directional derivativeFix a directionbu and a point P0in the plane.The directional derivative of w at P0in the directionbu isdefined asdwdsP0,bu= lim∆s→0∆w∆s.Here ∆w is the change in w caused by a step of length ∆s inthe direction ofbu (all in the xy-plane). This is illustrated inthe figure at right.Theorem:dwdsP0,bu= ∇w(P0) ·bu.xyw.......................))••P0∆w∆s∆sbu(continued)118.02A topic 26 2Proof: The tangent plane approximation and the picture atright show∆w∆s≈∂w∂x0∆x∆s+∂w∂y0∆y∆s≈∂w∂x0cos φ +∂w∂y0sin φButbu = hcos φ, sin φi since it is a unit vector. Thus, the lastformula is just ∇w ·bu. In the limit the approximations becomeexact and we get the boxed equation. QEDxy•P0??bu=hcos φ,sin φi∆s∆x∆yφExample: (Algebraic example) Let w = x3+ 3y2.Computedwdsat P0= (1, 2) in the direction of v = 3i + 4k.i) ∇w = h3x2+ 3y2i ⇒ ∇w|(1,2)= h15, 12i = 15 i + 12 j.ii)bu =v|v|=35i +45j.iii)dwdsP0,bu= ∇w|(1,2)·bu = (15 i + 12 j) · (34i +45j) =935.Example: (Geometric example) Letbu be the direction of h1, −1i.Using the picture at right estimate∂w∂xP,∂w∂yp, anddwdsP,bu.By measuring from P to the next in level curve in thex direction we see that ∆x ≈ −.5.⇒∂w∂xP≈∆w∆x≈10−.5= −20.Similarly, we get∂w∂yP≈ 20..Measuring in the u direction we get ∆s ≈ −.3⇒dwdsP,bu≈∆w∆s≈10.3= −33.3.xyw = 25w = 15w = 5P•bu∆xOO∆yoo11Direction of maximum change:bu =∇w|∇w|Proof: angle = 0. That is,dwds= ∇w · u = |∇w| cos θ. This has a max when θ = 0.Another proof: ∆w ≈ fx∆x + fy∆y = (fxi + fyj) · (∆x i + ∆y j)= gradient·displacement. So ∆w is maximized when diplacement isparallel to the gradient.The Gradient is perpendicular to level curves∇w ⊥ level curve (surface) w = c..................................•P0??bu−−→∇w(continued)18.02A topic 26 3Example: Consider the graph of y = ex. Find a vector perpindicular to the tangentto at the point (1, e).Old method: Find the slope take the negative reciprocal and make the vector.New method: This graph is the level curve of w = y − exwith w = 0.∇w = h−ex, 1i ⇒ normal = ∇w(1, e) = h−e, 1i.proof: Ifbu is tangent to the level curve at P0thendwdsP0,bu= 0since w is constant along the level curve, i.e.,dwdsP0,bu= ∇w(P0) ·bu = 0. QEDExample: Show the level curves and gradient field for z = (x2+ y2)/3 and z =−(x2+ y2)/3.xy//ooOO??__//ooOO??__//ooOO??__z = x2+ y2xyoo//OO__??oo//OO__??oo//OO__??z = −(x2+ y2)Higher dimensionsFor w = f(x, y, z) we get level surfaces. The gradient is normal to level surfaces.Example: Find the tangent plane to the surface x2+ 2y2+ 3z2= 6 at the pointP = (1, 1, 1).Introduce a new variable w = x2+ 2y2+ 3z2. Our surface is the level surface w = 6⇒ normal to surface is ∇w = h2x, 4y, 6zi. At the point P we have ∇w|P= h2, 4, 6i.Using point normal form the equation of the tangent plane is 2(x − 1) + 4(y − 1) +6(z − 1) = 0 ⇔ 2x + 4y + 6z = 12.Example: (Abstract version of previous example) Find the normal to the tangentplane of the graph of z = f(x, y) at (x, y, z).answer: Let w = f(x, y) − z. The graph of z = f(x, y) is just the level surfacew = 0 ⇒ ∇w = hfx, fy, −1i is the normal.Example: (Not done in class) Let θ = tan−1(y/x).⇒ ∇θ = −yx2+ y2i +xx2+ y2j = −yr2i +xr2j. Letbu = dir∇θ⇒dθdsbu= |∇θ| =1r⇒dsdθ= r ⇒ s = rθ (around the
View Full Document