QA08

#$&*

course MTH 279

2/25 11:48 am

q_a_08110216

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

Question:  `q001.  Consider the equation

y ' = 3 y^2 * sqrt(t)

This is a separable equation and can be solved.  However a very similar equation could be written down which is not solvable.  So let's pretend for the moment that this one isn't either.

Suppose we want to calculate an approximate solution curve for which y(1) = .5.  Our solution curve will therefore pass through the point (1, .5).

For reference, plot a direction field for this equation on the interval 1 <= t <= 1.2, .5 <= y <= .7.

Now construct an approximate solution by answering the following questions:

What is y ' at the point (1, .5)?

If we follow this slope as we move .1 unit to the right, in the y vs. t plane, ending up at a new point, what will be our rise?

What will then be the coordinates of our new point?

What is y ' at this new point?

If we follow this slope as we move .1 unit to the right, in the y vs. t plane, ending up at a new point, what will be our rise?

What will then be the coordinates of our new point?

We have found two new points by this method.  It should be clear that we could continue the process to find as many more new points as we might wish.

Plot these points on your sketch of the direction field.  Do they make sense in the context of that field?

The new points lie close to, but not exactly on, the solution curve through (1, .5). 

Why do we expect the points to lie close to the solution curve? 

Why don't we expect them to lie exactly on the solution curve?

Do our points lie above or below the solution curve?  Explain how you know.

YYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYY

Your solution:

 y'(1,0.5) = 0.75

rise = 0.075

(1.1, 0.575)

y' = 1.04

rise = 1.04*0.1= 0.104

(1.2, 0.679)

They make sense as they fit well in the space that are very close to the solution curve. They lie close because we have made small incremental steps in our process of approximating solution values but they do not lie along the curve since they are not found using the exact solution. This is an approximation using known values to find new values and then new values based on those values and as we continue this the error actually gets progressively larger.

confidence rating #$&*:

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

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

Given Solution:

Self-critique:

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

Self-critique Rating:

 

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

Question: `q002.  Solve the equation

y ' = 3 y^2 * sqrt(t)

for the initial condition y(1) = .5.

Evaluate your solution for t = 1.1 and t = 1.2, and plot the resulting two points on your graph of the direction field.

How close was the first new point obtained in the preceding problem to the actual solution?

How close was the second new point obtained in the preceding problem to the actual solution?

YYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYY

Your solution:

1/y^2= 3 sqrt t dt

-1/y= 2t^1.5+C

y= C-1/2t^-1.5

@& The negative reciprocal of the right-hand side is

y = - 1 / (2 t^(3/2) + c)

and the initial condition yields c = -4 so that the solution with initial condition is

y = -1 / (2 t^(3/2) - 4)

Evaluated at t = 1.1 and 1.2 this yields y values 0.591 and 0.729.

The error in our approximation grows progressively from a discrepancy of .016 at t = 1.1 to .05 at t = 1.2.

The graph below depicts our approximation and the solution curve y = -1 / (2 t^(3/2) - 4).

*@

y(1) = 0.5 --> C= 1

y(1.1) = 0.567

y(1.2)= 0.619

0.567- 0.575= -0.008

0.619-0.679= -0.06

confidence rating #$&*:

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

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

Given Solution:

Self-critique:

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

Self-critique Rating:

 

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

Question:  `q003.  We could get the t = 1.1 approximation by using the same process, with two intervals, each of .05.

What is your approximation to the y value for t = 1.05, based on the slope at (1, .5)?

What is your approximation to the y value for t = 1.10, based on the slope at the point you just found?

How far from the known solution is your new t = 1.10 approximation?

Originally we got to t = 1.1 by a single step.  We have now used two steps to get a new approximation. 

By what factor did our approximation error change?

Did doubling the number of steps reduce the error in the approximation by a factor of two, less than two or more than two?

Why did the approximation error change as it did?

YYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYY

Your solution:

y = 0.5375

y= 0.5666

0.567-0.5666= 0.0004

So much closer to and actually over the actual value.

Our error decreased by 1/20 or 0.05

It was reduced significantly greater than 2.

It changed in relation to the value of the interval, that is with a 0.1 interval we had a 0.1 error and with at 0.05 interval we had a 0.05 error. So we divided our by 1/2 but our error changed by 1/20.

@& I like the way your accuracy improved. My solution doesn't improve quite that much. Let me know if you find a flaw in my solution:

The t = 1.05 approximation is y = .5 + .75 * .05 = .5375.

This yields y ' = 3 * .5375^2 * sqrt(1.05) = .88, approximately, so that the new rise is .88 * .05 = .44 and the t = 1.10 approximation is .5815.

This approximation differs from the accurate t = 1.1 solution .591 by about .0095.

The direction field and approximate solution are depicted at

http://vhcc2.vhcc.edu/dsmith/geninfo/qa_query_etc/differential_equations/q_a_083.gif

This solution along with the approximate solution and direction field are depicted at

http://vhcc2.vhcc.edu/dsmith/geninfo/qa_query_etc/differential_equations/q_a_083.gif

*@

confidence rating #$&*:

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

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

Given Solution:

Self-critique:

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

Self-critique Rating:

 

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

Question:  `q004.  BB data:  Shot at the same trajectory the BB follows a path similar to that of a good sand wedge shot in golf, but I believe it starts out with greater velocity and loses speed more quickly.  In the sunlight against a blue sky you can track the thing for just about its entire flight.  Pretty neat.  Shot at the Moon, without correcting for gravity, and didn't really miss it by all that much.  Don't want to steal NASA's thunder to I'll leave it at that.  Shot into water at a low enough trajectory it will skip without losing too much speed.  Shot at a nearly vertical trajectory the water stops it within a few centimeters.  All this appears to be consistent with the gun's rating of 0.3 Joules and the BB's .12 gram mass (which is in turn consistent with the 250 ft/sec muzzle velocity).  At that speed the BB should climb for about 1000 feet, if shot in a vacuum, which is what leads me to believe it's losing velocity.

How could differential equations be used to provide models for this discussion?

What tests are suggested by the solutions of those equations?

YYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYY

Your solution:

We have the Fnet Eqs that involve drag changing to the square of the velocity. Of course a drag coefficient would be involved to distinguish from the water and the air. Seems that coefficient would be related to density, maybe. We could use energy considerations since we have an energy rating and that would involve KE, PE and Work by NonCons Forces, namely the fluid resistance. Actually a Bernoulli's eq comes to mind for some reason, I guess just thinking about an object moving through a fluid but that's not the same as the motion of the fluid.

Maybe something like:

1/2mv^2 - mgy - W_NC

where v = dy/dt = y' and W_NC = F_frict*dy

Or

m dv/dt - kv

Sub in dv/dy for the vertical shots and k is drag coefficient.

@& Good.

I would tend to think it would be

F_net = - m g - k v

or

F_net = -m g - gamma v^2

yielding either

y '' + k/m y ' = - g

which could be written

v ' + k/m v = -g

or

y '' = - g - gamma (y ')^2 / m , which could be written

v ' = - g - gamma v^2 / m.

Either equation could be solved. The two conditions v(0) = 250 ft / sec and v = 0 when y = 120 ft could be used to evaluate the integration constant and k or gamma, depending on the equation.*@