You are on page 1of 9

Hanoi University of Science

Physics faculty

The Use of ODE in Simulating


Particle Motion
Group 1:

Phm B Duy

Phm Xun Huy

Bi Vn Tnh

L Xun Vng
1

Short conclusion about second-order ODE

The goal of finite difference method is to determine the value of


X(n+1) and V(n+1) at time t(n+1) = tn + t .
Euler method and modified Euler methods approximate
X(n+1) and V(n+1) from Xn and Vn by Taylor series
expanding.

Euler method:

Euler-Comer method:

Errors in these two above methods


~

Local truncation error (LTE)

Global truncation error (GTE)

Rounding Error (RE) (made by computer):

(f''(x)/2)*(t)^2
2*M*t

with M=max(f''(t)/2)

* Set of number of computer are discrete (not continuous) and finite


(not infinite) .
* A lot decimal numbers has indefinite representation in binary form,
so computer must truncate a quantity = (epsilon).
EXP: 0.1(10) = 0.00011001100110011001100....(2)
Hence, Rounding Error

*f(t)*(1/t)

We say that a method is nth order if its GTE is order (t)^n.


Hence, Euler and Euler-Comer method is first order method.

Errors in these two above methods


~

Local truncation error (LTE)

Global truncation error (GTE)

Rounding Error (RE) (made by computer):

(f''(x)/2)*(t)^2
2*M*t

with M=max(f''(x)/2)

when represent real number in binary and floating point number in


computer each real number may has trund off error = (epsilon).
Hence, Rounding Error

*f(t)*(1/t)

We say that a method is nth order if its GTE is order (t)^n.


Hence, Euler and Euler-cromer method is first order method.
4

Optimal choice for t

From slide above, we refer that if we reduce t then GTE will


also reduce. But if t is too small (several order than
) then RE increases.
Total Error = M*t + *f(t)*(1/t)

Minimized

t =sqrt(2f(t)/M)

Some reason for why Euler Comer is better than


Euler method.
5

Euler-Richardson Method
h

Euler-Richardson Method (cont): Error analysis

Expression:

Error analysis:

LTE

f'''(t)*t^3 + (y2-y1)

GTE

f'''(t)*t^2

RE (similar to Euler method)


RE

*f(t)*(1/t)

From this we can calculate minimized t


7

Euler-Richardson Method (cont): Error analysis

Expression:

Error analysis:

LTE

f'''(t)*t^3 + (y2-y1)

GTE

f'''(t)*t^2

RE (similar to Euler method)


RE

*f(t)*(1/t)

From this we can calculate minimized t


8

Conclusion

To compute error (LTE) at each step, we can follow some basic


step:
step 1: construct function f'''(t), and expression of LTE
step 2: calculate RE at each step
step 3: total error = LTE + RE

We can also find minimized value for t, by find value of t


at which total error minimizes.

You might also like