Homework: 11/6

14.7
6, 16. Find the local maximum and minimum values and saddle point(s) of the function. If you have three-dimensional graphing software, graph the function with a domain and viewpoint that reveal all the important aspects of the function. 6. f(x, y) = xy 2x 2y x2 y2. Theorem 2 in the chapter says that if a function f has a local maximum or minimum at (a, b) and the first-order partial derivatives of f exist there, then fx(a, b) = 0 and fy(a, b) = 0. An obvious first step is to find where fx = fy = 0 then. First, lets get their general forms: fx(x, y) = y 2 2x fy(x, y) = x 2 2y Setting fx(x, y) = 0 0 = y 2 2x 2 + 2x = y Substituting into fy(x, y) = 0 x y 2y = 0 x (2 + 2x) 2(2 + 2x) = 0 -3x 6 = 0 x = -2 y = -2 (a, b) = (-2, -2) This implies theres only one point of interest (i.e. critical point). We might be tempted at this point to say that the function reaches a local maximum or minimum here, but close attention to Theorem 2 will reveal that we dont know if this is a local maximum, or a local minimum, or a saddle point. All we know is that if it were one, itd behave this way. We have to apply the second derivative test, which says that D > 0 and fxx(a, b) > 0, D > 0 and fxx(a, b) < 0, and D < 0, means (a, b) is a minimum point, maximum point, or neither respectively, where D = fxx(a, b) fyy(a, b) [fxy(a, b)]2 which means we have to find fxx, fyy, and fxy if we want to apply the test: fxx = -2 fyy = -2 fxy = 1 D = (-2)(-2) (1)2 = 3 > 0

Because fxx = -2 < 0, by the second derivative test we see f(-2, -2) = 4, is a local maximum. Notice, D, which is theoretically a function of x and y, doesnt depend on either here.

You might be concerned that we did not pick up on all the possible critical points. After all, our equations only yielded one. If you work through the logic of Theorem 2 though, youll find that whenever there are any points of interests, theyll show up on our radar if we set the two partials equal to 0. Concern dealt with.

16. f(x, y) = ey(y2 x2). Following the same logic as in the previous problem, we go through the three steps: 1) Find critical points (i.e. find where first partials equal 0 and solve system of equations). 2) Find fxx, fyy, and fxy in anticipation of evaluating D. 3) Check for what kind of critical points each from part 1) are. So: 1) fx = -2xey, fy = (2y + y2 x2)ey

fx = -2xey 0 = -2xey x = 0 (no matter what value y takes on, ey 0; so x must be the 0). fy = (2y + y2 x2)ey 0 = (2y + y2 x2)ey 0 = (2y + y2)ey 0 = y(2 + y) (since, as above, no matter what value y takes on, ey 0; so (2y + y2) must be the 0) y = 0 or -2 So, our two critical points are (0, 0) and (0, -2) 2) fxx = -2ey, fyy = 2ey + 2yey + 2yey + y2ey x2 = ey(2 + 4y + y2 x2), fxy = -2xey

fxx(0, 0) = -2, fyy(0, 0) = 2, fxy(0, 0) = 0 D(0, 0) = (-2)(2) (0)2 = -4 < 0. Since D < 0, (0, 0) is neither a minimum nor a maximum; we conclude that it is a saddle point. fxx(0, -2) = -2e-2, fyy(0, -2) = -2e-2, fxy(0, -2) = 0 -2 -2 2 -4 D(0, -2) = (-2e )(-2e ) (0) = 4e > 0. fxx(0, -2) = -2e-2 < 0. Since D > 0, and fxx < 0, (0, -2) is a local maximum.

(Sorry the picture is so bad; my Wolfram Alpha wasnt working).

34. Find the absolute maximum and minimum values of f on the set D: f(x, y) = xy2, D = {(x, y) | x 0, y 0, x2 + y2 3}. According to the Extreme Value Theorem for functions of two variables, there will always be absolute minimum and absolute maximum values on the domain D if f is continuous and D is closed and bounded. Box 9 in the textbook states that to find the absolute values, we first find the critical values in D, then find values on the boundary of D, and then choose the largest/smallest out of those. Lets get started: fx = y2, fy = 2xy, 2 fx = 0 0 = y y = 0. At this point we need look no further to realize a key fact: there are no critical values on the interior of D. Notice, all of the interior points would require y 0, since y = 0 implies the point is on the boundary. So, we just have to check the boundary of D.

Theres D for you. As you can see, there are three parts to the boundary, so well take them one at a time: L1 is on the y-axis, so its constraint is that x = 0. Looking at the function f(x, y), we now make the appropriate adjustment f(0, y) = 0. L2 is on the x-axis, so its constraint is that y = 0. Then f(x, y) f(x, 0) = 0 L3 is the only tricky one. However, we know that on that line, the constraint is x2 + y2 = 3 (since its on the boundary; if it wasnt then wed only know x2 + y2 3). Then we have a substitution. Let y = (3 x2), so f(x, y) f(x, (3 x2)) = 3x x3 for 0 x 3. The restriction on x is extremely important, since, we might be tempted to let x vary freely, and then we could find very strange values for f. Applying a simple derivative to 3x x3 will yield any single-variable critical points: (3x x3) = 3 3x2 3 3x2 = 0 x = 1. Ah but wait! Remember we put a restriction on x (0 x 3), so we conclude that the only point of interest is x = 1. If x = 1, then we solve our substitution (y = (3 x2)) to find y = 2. Plugging into the original function f(1, 2) = 2 yields that (1, 2) is a maximum. As for a minimum, since x is restricted, we must check the endpoints: x = 0 y = 3 f(0, 3) = 0; x = 3 y = 0 f(3, 0) = 0. Remember that we checked all the interior points and nothing was of interest and all of the points along L1 and L2 were equal to 0 (this includes the two points we just found). Therefore, we say that the absolute minimum is 0, along lines L1 and L2.

50. The base of an aquarium with given volume V is made of slate and the sides are made of glass. If slate costs five times as much (per unit area) as glass, find the dimensions of the aquarium that minimize the cost of the materials. Were given that the volume of a presumably rectangular solid (without a top) is a fixed value V. What varies is the cost. Then cost equals [price of base] + [price of walls] = [price of slate]x[surface area of slate] + [price of glass]x[surface area of walls] = 5x[price of glass]x[area of slate] + [price of glass]x[area of walls] (since slate = 5[glass price]) = [price of glass](5[area of slate] + [area of walls]) (1) Now, lets establish x and y as the dimensions of the base, and z as the height. Then the surface area of the base is xy, and the surface area of one wall is either xz or yz. There are two xz walls, and two yz walls. So then equation (1) becomes: [price of glass](5xy + 2xz + 2yz). Lets let c be the constant to represent price of glass: = c(5xy + 2xz + 2yz) = C(x, y) As fun as it would be to work with three variables, we dont need to. We can substitute out one of the variables with use a constant. Recall that volume is fixed: V = xyz. Then, making the appropriate substitution: C(x, y) = 5xy + 2V(x + y)/xy = 5xy + 2V(1/x + 1/y).

Now that we have our function of two variables which we wish to minimize, this has become a simple find the extremum problem. As usual then, we start with partials: Cx = 5y + (-1)2V(1/x2) = 5y 2Vx-2, Cy = 5x 2Vy-2 Cx = 0 5y 2Vx-2 = 0 y = 2V/(5x2) Cy = 0 5x 2V(2V/5x2)-2 = 0 5x = 2V(5x2)2 * (2V)-2 2V = 5x3 x = 3(2V/5) Plugging this back into one of our partials equations will yield that x = y (its a little messy; just keep track of everything). Then we plug back in to V = xyz to find that z = V1/3(5/2)2/3 Then the dimensions are x = (2V/5)1/3, y = (2V/5)1/3, z = V1/3(5/2)2/3.

54. Three alleles (alternative versions of a gene) A, B, and O determine the four blood types A (AA or AO), B (BB or BO), O (OO) and AB. The Hardy-Weinberg Law states that the proportion of individuals in a population who carry two different alleles is P = 2pq + 2pr + 2rq where p, q, and r are the proportions of A, B, and O in the population. Use the fact that p + q + r = 1 to show that P is at most 2/3. Another way to interpret this problem is to find the maximum for this function P and then prove that it equals 2/3 at that point. The problem, as with the one above, is that even though we have an equation for all three of our variables, were only really good at looking at two variables at a time. So, we need a way to remove one variable. No problem; they give us p + q + r = 1. Lets substitute for r: r = 1 p q P = 2pq + 2(1 p q)(p + q) = -2pq 2p2 2q2 + 2p + 2q. So far so good. Lets find the partials: Pp = -2q 4p + 2, Pq = -2p 4q + 2 Pp = 0 -2q 4p + 2 = 0 q = -2p + 1, Pq = 0 -2p 4q + 2 = 0 -2p 4(-2p + 1) + 2 = 0 6p = 2 p = 1/3 q = 1/3. That means we have one critical point: (p, q) = (1/3, 1/3). P(1/3, 1/3) = 2/3. Now, you might be tempted to say were done at this point. However, all we did was find a critical point; we did not verify that it was a maximum, or a minimum, or otherwise. On top of that, it should concern you (well if it doesnt no biggie) that P(p, q, r) and P(p, q) seem like a very open expressions (i.e. do they even have bounds?), and there seems to exist a restrictive relationship. We could take second derivatives (Ppp = -4, Pqq = -4, Ppq = -2 D > 0, Ppp < 0, therefore this is a local maximum), but this only gives local maximums, so who knows if it goes to infinite. The only way to check this (at our level) is to set up boundaries.

We know for a fact that p 0, and q 0. Since r 0, and 1 p q = r, p + q 1. Then our domain is the space enclosed by the p-axis, the q-axis, and the line 1 = p + q. So: Along the p-axis, q = 0 P(p, q) P(p, 0) = 2p 2p2, 0 p 1. The greatest value that P could attain here is when p = (you could check this with single-variable maximization or properties of parabolas). Similarly, along the q-axis, P(0, ) = is the maximum. On the segment p + q = 1, P(p, q) P(p, 1 p) = 2p 2p2 (do the algebra yourself) 0 p 1. Then we find that on all three segments along the boundary, the maximum value is . Therefore, we see that the point we found on the interior, (1/3, 1/3), is the maximum = 2/3.

14.8
4, 6, 9. Use Lagrange multipliers to find the maximum and minimum values of the function subject to the given constraint. 4. f(x, y) = 3x + y; x2 + y2 = 10 The basic setup of a Lagrange multiplier is a function you want to maximize, minus a scalared function that equals a constant value, is equal to 0. In particular, it becomes fx = gx, fy = gy, (and fz = gz). Then lets get to it (let g(x, y) = x2 + y2 = 10): fx = 3, gx = 2x, 3 = 2x (1) fy = 1, gy = 2y, 1 = 2y (2) 2 2 x + y = 10 (3) Using all three, we can solve our system of equations (first equate (1) and (2) to eliminate ) to find that x = 3y, and y2 = 1 y = 1. Then our critical points are (3, 1), and (-3, -1). Compute: f(3, 1) = 10, f(-3, -1) = -10. Then we know that the maximum is 10, and the minimum is -10. The actual theory behind Lagrange multipliers is explained in the book. If anyone asks, Ill send a more detailed (i.e. easier to understand) explanation.

6. f(x, y) = exy; x3 + y3 = 16 Nothing to it: fx = yexy, gx = 3x2 yexy = 3x2 fy = xexy, gy = 3y2 xexy = 3y2 x3 + y3 = 16

(1) (2) (3)

While solving your system, keep in mind that in general we try and use equations (1) and (2) to eliminate . Doing so in this situation, you must take a little precaution. In order to solve for

in either equation, we require that x2 or y2 in the denominator, and for this to happen neither may be equal to 0. But thats okay. Consider if one of them was 0. Then by (1) or (2), we can see that the other must be 0 as well. If both of them were 0, then wed contradict (3). Therefore, we know that neither of them will be 0. Solving the system yields x = y, and that x = 2 = y. Checking at (2, 2), we find f(2, 2) = e4. Now, there are two ways to decide if this is a minimum or a maximum. One is to plug in any point you can think of and then realize that itll be less than e4. The other is to look really hard at equation (3). If you pick a super negative value for, say, x, then we must pick an slightly superer positive value for y. If we try and plug these into f(x, y), we can get a super super-er negative number in the exponent of e. This means well get very close to 0. But, we can pick a super-er negative value for x, and a slightly super-er-er positive value for y, and get a super-er super-er-er negative value in the exponent of e. So we get even closer to 0. Well never get to 0, but we can always get closer, so there is no minimum here. Thus, f(2, 2) = e4 is a maximum.

9. f(x, y, z) = xyz; x2 + 2y2 + 3z2 = 6 And again (except with three variables): fx = yz, gx = 2x yz = 2x fy = xz, gy = 4y xz = 4y fz = xy, gz = 6z xy = 6z 2 2 2 x + 2y + 3z = 6

(1) (2) (3) (4)

We find the same potential pitfall here as in the previous question. What happens if x, y, or z is 0? In the same way, consider if any of them in 0. Then (1) (2) and (3) imply all three of them are 0, and if all three of them are 0, then we contradict (4). Thus, we conclude none of them is 0. Working through the three questions, we arrive at x2 = 2y2 and z2 = 2/3 y2. Thus (4) implies that 6y2 = 6, or y = 1. Then the critical points are (2, 1, 2/3), (2, 1, -2/3), (-2, 1, 2/3), and (-2, 1, -2/3). Testing these out, we see that since the magnitudes in all cases are equal, we only need to see if f will be positive or negative. Its positive when exactly two are negative, or all are positive, and f = 2/3. In all other cases, we have a minimum, and f = -2/3

42. Find the maximum and minimum volumes of a rectangular box whose surface area is 1500 cm2 and whose total edge length is 200 cm. Let us set up the dimensions of the box as x, y, and z. Then cursory geometry will tell us that the total surface area will be S(x, y, z) = 2(xy + xz + yz) = 1500, and E(x, y, z) = 4(x + y + z) = 200. Alternatively, xy + xz + yz = 750, and x + y + z = 50.

We want to consider the extrema of V(x, y, z) = xyz. Lucky for us, we have two constant equations to help us. The theory is basically the same as with the previous problems, but instead were going to subject our function to the constraint of two functions. Its easier to just show: Vx = yz, Sx = (y + z) Ex = yz = (y + z) + (1) Vy = xz, Sy = (x + z) Ey = xz = (x + z) + (2) Vz = xy, Sz = (x + y) Ez = xy = (x + y) + (3) xy + xz + yz = 750 (4) x + y + z = 50 (5) Subtracting (1) and (2) implies that z(y x) = (y x), or z = as long as y x. Subtracting (1) and (3) implies that y(z x) = (z x), or = y, as long as z x. So then y = z = (assume x is distinct), and from (5) we get x = 50 2. From (4) we get x(2) + 2 = 750. Then, we can solve for : 50 - 2 = (750 2)/(2) 32 100 + 750 = 0 = (50 510)/3. Plugging this in, we get the point ((1/3)(50 1010), (1/3)(50 + 510), (1/3)(50 + 510)), and the point ((1/3)(50 + 1010), (1/3)(50 510), (1/3)(50 510)). Plugging these values into V yields that the minimum is at the first point, equal to (1/27)(87,500 250010), and so the maximum is at the second point, equal to (1/27)(87,500 + 250010).

43. The plane x + y + 2z = 2 intersects the paraboloid z = x2 + y2 in an ellipse. Find the points on this ellipse that are nearest to and farthest from the origin. Another way to interpret this problem is to find the extrema of the function f(x, y, z) = x2 + y2 + z2 (distance to origin) subject to the constraints g(x, y, z) = x + y + 2z = 2, and h(x, y, z) = x2 + y2 z = 0. So, this is identical to the last problem in many ways: fx = 2x gx = hx = 2x 2x = + 2x (1) fy = 2y gy = hx = 2y 2y = + 2y (2) fz = 2z gz = 2 hz = - 2z = 2 (3) x + y + 2z = 2 (4) 2 2 x +y z=0 (5) Cancelling in (1) and (2) yields 2(x y) = 2(x y), so if x y, = 1. Using this for equation (3) yields 2z = 2 1 = z + . However, letting = 1 in (1) says that = 0. Then we combine these two and get that z + = 0 z = -. Plugging this value into (4) and (5) gives us x + y 3 = 0 and x2 + y2 + = 0 respectively. However, look at the last equation. The sum of two squared numbers and a positive constant is equal to zero? Impossible. Therefore, we conclude that our original condition (if x y) must have been false (or at least, it yields none of the points were looking for). So, x = y.

Using this fact in (4) and (5) yields 2x + 2z = 2 x + z + 1, and 2x2 z = 0, respectively, which implies z = 1 x, and z = 2x2, respectively. Equating the two, we get 2x2 + x 1, which factors out to (2x 1)(x + 1) = 0. So x = or -1. Thus, plugging x back in, our two critical points to check are (1/2, 1/2, 1/2) and (-1, -1, 2): f(1/2, 1/2, 1/2) = 3/4, f(-1, -1, 2) = 6. Therefore, we conclude that (1/2, 1/2, 1/2) is a minimum, and (-1, -1, 2) a maximum. Or to translate back into the language of the original problem, (1/2, 1/2, 1/2) is the point closest to the origin, and then the point (-1, -1, 2) is furthest from the origin.

(Disclaimer: this solution set is subject to typos or careless errors. I apologize in advance for any of those.)