Query 117

#$&*

course Mth 277

11/29 2

11.5

*********************************************

Question: `q001. Consider f(x,y) = 3x^2 - 5xy + y^2 + 3. Find the critical points, and classify each point as either a relative maximum, relative minimum, or saddle point.

YYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYY

Your solution:

f_x is 6x - 5y

f_y is 5x + 2y

set = 0 and solve to get CP

both end up to be x , y = 0. so it would imply a maximum.

confidence rating #$&*:

^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

.............................................

Given Solution:

f_x = 6 x - 5 y and f_y = 5 x + 2 y. Our critical point therefore occurs when

6x - 5y = 0 and

5x + 2y = 0.

Solving simultaneously we get x = 0, y = 0 so our critical point is (0, 0).

The second derivatives are f_xx = 6, f_yy = 5 and f_xy = -5. f_xx and f_yy are both positive, indicating that if the critical point is not a saddle point, it is a minimum

f_xx * f_yy - f_xy^2 = 6 * 5 - (-5)^2 = 5, which is > 0 and indicates that the critical point is not a saddle point.

The one critical point for this function therefore corresponds to a relative minimum, which occurs at the point (0, 0, f(0, 0)) = (0, 0, 3).

&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&

Self-critique (if necessary):

I think f_yy is 2 not 5.

Also its hard for me to see that f_xy is -5. do you factor out the constant of -5 and fill in the other values x and y with 0 ?

Also I dont understand why you did that last equation.

------------------------------------------------

Self-critique rating:

@&

f_x = 6x - 5y, so

f_xy = (6x - 5y)_y = -5.

This is also equal to f_yx.

f_yy is 2, not 5.

The last step is necessary, since the saddle can be oriented so the 'drop-off' is at, say, the 45 degree angle and not along either the x or the y axis.

Corrected solution:

f_x = 6 x - 5 y and f_y = 5 x + 2 y. Our critical point therefore occurs when

6x - 5y = 0 and

-5x + 2y = 0.

Solving simultaneously we get x = 0, y = 0 so our critical point is (0, 0).

The second derivatives are f_xx = 6, f_yy = w and f_xy = -5. f_xx and f_yy are both positive, indicating that if the critical point is not a saddle point, it is a minimum.

f_xx * f_yy - f_xy^2 = 6 * 2 - (-5)^2 = -13, which is < 0 and indicates that the critical point is in fact a saddle point.

*@

*********************************************

Question: `q002. Consider f(x,y) = x^2 - y^2 + xy/16. Find the critical points, and classify each point as either a relative maximum, relative minimum, or saddle point.

YYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYY

Your solution:

f_x = 2x + y/16

f_y = -2y + x/16

set to zero and again CP is (0,0)

second derivatives is f_xx = 2

f_yy = -2

this implies a saddle point

confidence rating #$&*:

^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

.............................................

Given Solution: f_x = 2 x + y / 16 and f_y = -2 y + x / 16.

Setting the two equal to zero and solving we find that (0, 0) is our only critical point.

f_xx = 2 and f_yy = -2. This indicates that the y = 0 'slice' of the graph has a minimum while the x = 0 'slice' has a maximum. The critical point therefore yields a saddle point.

We could evaluate f_xy (which is 1/16) and test for a saddle point, but we've already seen that we do have a saddle point. So we normally wouldn't bother. However, just to illustrate that the test for a saddle point works in this case,

let's do the test.

f_xx * f_yy - f_xy ^ 2 = 2 * (-2) - (1/16)^2 = -4 - 1/16. This is clearly < 0, indicating what we already knew, that the point (0, 0) is a saddle point.

Note how f_xx * f_yy came out negative, since the signs of the two second derivatives differed. As soon as this happened the left-hand side was doomed to be negative, since the only other term - f_xy ^ 2 (being the negative of a square)

cannot be positive.

&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&

Self-critique (if necessary):

------------------------------------------------

Self-critique rating:

*********************************************

Question: `q003. Find the least squares regression line for the set of points {(4,-2), (3,-1), (0,0), (-1,3), (-2,1), (-3,2)}.

YYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYY

Your solution:

x values are ( 4, 3, 0, -1, -2, -3 )

y values are ( -2, -1, 0, 3, 1, 2 )

get to equations

4a + b + 2

3a + b + 1

b

-a + b -3

-2a + b - 1

-3a + b -2

square them all and sum together to get

39a^2 + 2ab + 44a + 6b^2 -6b + 24

take both partial derivatives and get a = 0.6 ; b = -.4

so y = 0.6x - 0.4

confidence rating #$&*:

^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

.............................................

Given Solution:

If y = a x + b, then the points on the line corresponding to x = 4, 3, 0, -1, -2, -3 have respective y coordinates 4 a + b, 3 a + b, b, -a + b, -2 a + b and -3 a + b.

These differ from the given y coordinates by respective amounts 4 a + b - (-2), 3 a + b - (-1), b, -a + b - 3, -2 a + b - 1 and -3a + b - 2.

The sum of the squares of these differences is thus

(4 a + b - (-2))^2 + (3 a + b - (-1))^2 + b^2 + (-a + b - 3)^2 + ( -2 a + b - 1 )^2 + (-3a + b - 2))^2 = 39 a^2 + 2 a b + 44 a + 6 b^2 - 6 b + 19

In order to find the regression line we will find the values of a and b that minimize this sum.

For the sake of convenient notation let

f(a, b) = 39 a^2 + 2 a b + 44 a + 6 b^2 - 6 b + 19.

f_a = 78 a + 2 b + 44

f_b = 12 b + 2 a - 6

Solving simultaneously we obtain approximate values a = -.58 and b = .60.

We conclude that the best-fit equation, with a and b accurate to 2 significant figures, is

y = -.58 x + .60.

&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&

Self-critique (if necessary):

for b i'm getting -0.4 and a = 0.6

solve for a with f_b. a = 3-6b

substitute into other equation to get 234 - 470b = -44

b = -0.4

fill back into small a equation

a = 0.6

would not have been able to do this without solution

------------------------------------------------

Self-critique rating:

*********************************************

Question: `q004. Consider these following functions, at each of which D = 0 at a critical point. Show whether each of the following is true or false:

f(x,y) = x^4 - y^4 has a saddle point at (0,0).

g(x,y) = x^2*y^2 has a relative minimum at (0,0).

h(x,y) = x^3 + y^3 has a relative maximum at (0,0).

f_xx is positive and f_yy is negative. The intersection of the graph of f(x, y) with the x-z plane is a curve with a relative min at (0, 0); the intersection with the x-y plane has a relative max at the same point.

Consider the line parameterized by x = t cos(theta), y = t sin(theta). Along this line we have z = t^4 cos^2(theta) sin^2(theta). For any value of theta except a multiple of 2 pi the function z vs. t has a relative minimum at t = 0.

YYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYY

Your solution:

f(x,y) True; has CP at point (0,0)

has saddle p because fxx*fyy-fxy^2 is negative. which is also at (0,0)

g(x,y) TRUE has CP (0,0)

and fxx*fyy-fxy^2 is pos. with all positive numbers so relative min.

h(x,y)

confidence rating #$&*:

^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

.............................................

Given Solution:

&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&

Self-critique (if necessary):

------------------------------------------------

Self-critique rating:

*********************************************

Question: `q005. We can use the exponential and logarithm to help us to linearize data that does not tend to change linearly. The following problems will demonstrate this.

Suppose we have y = kx^m. Show that by taking the natural logarithm of this equation we obtain a linear relationship Y = K + mX. Explain these new variables and the constant K.

The data that follows relates the periods of revolution, t (in days), of the six inner planets and to their semimajor axes a (in 10^6 km). Kepler conjectured that the relationship is t = ka^m for some k and m.

Transform the data as in the first part and find k and m.

t-data : (87.97, 224.7, 365.26, 686.98, 4332.59, 10759.2)

a-data : (58, 108, 149, 228, 778, 1426).

log(y) = log( k x^m) = log(k) + m log(x), which is of the form Y = K + m X for Y = log(y), K = log(k) and X = log(x).

It follows that the graph of Y vs. X is a straight line.

Logs of the a data are

1.76342799356294

2.03342375548695

2.17318626841227

2.35793484700045

2.89097959698969

3.15411952551585

Logs of the b data are

1.94433459197078

2.35160307241913

2.56260211477846

2.83694409365915

3.63674759298951

4.0317799805774

Both are given above to way too many significant figures.

We can find the best-fit line for these data.

The slope of the best-fit line will be our power m.

Since K = log(k), k = 10^K.

YYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYY

Your solution:

Since a slope is a change in Y over a change in X, we can use change in Y= log(y) which = log(k) + m log(x) / (change in X which = log(x)

confidence rating #$&*:

^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

.............................................

Given Solution:

&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&

Self-critique (if necessary):

------------------------------------------------

Self-critique rating:"

Self-critique (if necessary):

------------------------------------------------

Self-critique rating:

Self-critique (if necessary):

------------------------------------------------

Self-critique rating:

#*&!

&#Good responses. See my notes and let me know if you have questions. &#