Numerical versus Analytical Solutions

We’ve started working on the physics of motion in my programming class, and really it boils down to solving differential equations using numerical methods. Since the class has a calculus co-requisite I thought a good way to approach teaching this would be to first have the solve the basic equations for motion (velocity and acceleration) analytically–using calculus–before we took the numerical approach.

Constant velocity

  • Question 1. A ball starts at the origin and moves horizontally at a speed of 0.5 m/s. Print out a table of the ball’s position (in x) with time (t) (every second) for the first 20 seconds.

Analytical Solution:
Well, we know that speed is the change in position (in the x direction in this case) with time, so a constant velocity of 0.5 m/s can be written as the differential equation:

 \frac{dx}{dt} = 0.5

To get the ball’s position at a given time we need to integrate this differential equation. It turns out that my calculus students had not gotten to integration yet. So I gave them the 5 minute version, which they were able to pick up pretty quickly since integration’s just the reverse of differentiation, and we were able to move on.

Integrating gives:

 x = 0.5t + c

which includes a constant of integration (c). This is the general solution to the differential equation. It’s called the general solution because we still can’t use it since we don’t know what c is. We need to find the specific solution for this particular problem.

In order to find c we need to know the actual position of the ball is at one point in time. Fortunately, the problem states that the ball starts at the origin where x=0 so we know that:

  • at t = 0, x = 0

So we plug these values into the general solution to get:

 0 = 0.5(0) + c
solving for c gives:

 c = 0

Therefore our specific solution is simply:

 x = 0.5t

And we can write a simple python program to print out the position of the ball every second for 20 seconds:

motion-01-analytic.py

for t in range(21):
     x = 0.5 * t
     print t, x

which gives the result:

>>> 
0 0.0
1 0.5
2 1.0
3 1.5
4 2.0
5 2.5
6 3.0
7 3.5
8 4.0
9 4.5
10 5.0
11 5.5
12 6.0
13 6.5
14 7.0
15 7.5
16 8.0
17 8.5
18 9.0
19 9.5
20 10.0

Numerical Solution:
Finding the numerical solution to the differential equation involves not integrating, which is particularly good if the differential equation can’t be integrated.

We start with the same differential equation for velocity:
 \frac{dx}{dt} = 0.5

but instead of trying to solve it we’ll just approximate a solution by recognizing that we use dx/dy to represent when the change in x and t are really, really small. If we were to assume they weren’t infinitesimally small we would rewrite the equations using deltas instead of d’s:
 \frac{\Delta x}{\Delta t} = 0.5

now we can manipulate this equation using algebra to show that:
 \Delta x = 0.5 \Delta t

so the change in the position at any given moment is just the velocity (0.5 m/s) times the timestep. Therefore, to keep track of the position of the ball we need to just add the change in position to the old position of the ball:

 x_{new} = x_{old} + \Delta x

Now we can write a program to calculate the position of the ball using this numerical approximation.

motion-01-numeric.py

from visual import *

# Initialize
x = 0.0
dt = 1.0

# Time loop
for t in arange(dt, 21, dt):
     v = 0.5
     dx = v * dt
     x = x + dx
     print t, x

I’m sure you’ve noticed a couple inefficiencies in this program. Primarily, that the velocity v, which is a constant, is set inside the loop, which just means it’s reset to the same value every time the loop loops. However, I’m putting it in there because when we get working on acceleration the velocity will change with time.

I also import the visual library (vpython.org) because it imports the numpy library and we’ll be creating and moving 3d balls in a little bit as well.

Finally, the two statements for calculating dx and x could easily be combined into one. I’m only keeping them separate to be consistent with the math described above.

A Program with both Analytical and Numerical Solutions
For constant velocity problems the numerical approach gives the same results as the analytical solution, but that’s most definitely not going to be the case in the future, so to compare the two results more easily we can combine the two programs into one:

motion-01.py

from visual import *
# Initialize
x = 0.0
dt = 1.0

# Time loop
for t in arange(dt, 21, dt):
     v = 0.5

     # Analytical solution
     x_a = v * t

     # Numerical solution
     dx = v * dt
     x = x + dx

     # Output
     print t, x_a, x

which outputs:

>>> 
1.0 0.5 0.5
2.0 1.0 1.0
3.0 1.5 1.5
4.0 2.0 2.0
5.0 2.5 2.5
6.0 3.0 3.0
7.0 3.5 3.5
8.0 4.0 4.0
9.0 4.5 4.5
10.0 5.0 5.0
11.0 5.5 5.5
12.0 6.0 6.0
13.0 6.5 6.5
14.0 7.0 7.0
15.0 7.5 7.5
16.0 8.0 8.0
17.0 8.5 8.5
18.0 9.0 9.0
19.0 9.5 9.5
20.0 10.0 10.0

Solving a problem involving acceleration comes next.

Leave a Reply