CHAPTER 4 : STATISTICS FOR SYSTEMS OF MANY ELEMENTS

A. Fluctuations For a large number of identically prepared systems (an ensemble), the average number of elements per system that satisfy the criterion is

n = pN
p=

p = probability that a single element satisfies it N = # of elements per system

Example : If a large number of coins were flipped many times, how many would land heads up, on the average? N

1 = probability of 1 coin to land heads 2

n = pN =

2 N n= 6

Example : If a large number of dice are rolled many times, how many would land with 1 dot up, on the average?

1 p= 6

Standard Deviation square root of the average fluctuations squared

= nn

12

= Npq

Half-width at half maximum

Relative Fluctuation

Npq = Np

q Np

Jitter about the mean

Example : 100 flipped coins and 10,000 flipped coins 100 flipped coins n100 = pN = 1 (100 ) = 50

n , , and

=?

for heads

100

n 100

5 = 10 % 50

) = 25 = 5 100 = Npq= (100

1 1 2 2

10,000= Npq= (10,000) = 2500= 50

1 n10,000 = pN = (10,000) = 5,000 2

1 1 2 2

n 10 , 000

50 = = 1% 5 , 000

For 100 flipped coins, the number of heads will usually be (i.e., 68% of the time) between 40 and 60% of the total, while for the 10,000 coins system, the number of heads will usually be between 49 and 51% of the total. The larger the system, its behavior will be more predictable. For large N, the distribution of n about the mean is given by a GAUSSIAN distribution. The probability that n is within 1 standard deviation of n is 0.68.

B. The Gaussian Distribution For systems with large N, the probability that n elements satisfy the criterion and the remaining (N-n) do NOT is

PN (n ) =
Where :

1 [( n n ) 2 e 2
= Npq

(if n n << )
2

n = pN

Example : 3000 air molecules in an empty room. P=? that exactly 1000 of them will be in the front third of the room at any time. (N=3,000 ; n=1,000 ; p=1/3 ; q=2/3 )

1 n = pN = (3000) = 1000 3

(n n )2
2
2

1 2 = Npq = (3000 ) = 25.8 3 3

=0

1 P3000(1000 ) = e0 = 1.6 x10 2 2 (25 . 8)

Example : All the same except n = 1100

(n n )2
2
2

2 ( 100 ) = 2 2 (25.8 )

= 7. 5

1 P3000(1100) = e7. 5 = 8.8x106 2 (25. 8)

Example : How many different combinations of the air molecules are there, such that 1,100 are in the front third of the room?

N! 3000 ! = = 10 858 n !( N n ) ! 1100 ! 1900 !

(by Stirlings formula)

Binomial coefficient C. Random Walk

Binomial distribution is correct for all systems but is cumbersome for large N

study of the net motion of objects where the motion occurs in individual discrete steps with random direction and is independent of the direction of the preceding or succeeding steps Examples of random walk problems Diffusion of an ammonia molecule through air Travel of electrons through a metal Diffusion of a hole through a semiconductor

For the random walk problem in any 1-dimension, if P(s)ds is the probability that a single step falls in the range between s and s+ds, then after N steps The average position of the objects, The square of the standard deviation of their positions, Where :

S = Ns
2

= N (s )
2

s = sP (s )ds

s = ( s s ) =
2 2

( s s ) P (s )ds = s
2

s = s P (s )ds

+ 2

(s )

Example : migration of a conduction electron in a metal 1012 collisions per second with atoms and other electrons (N=1012). 1 root-meansquare distance between collisions (s )2 = (1 angstrom )2

with E 0, s = 10 4 angstrom

Average distance covered after 1012 steps Standard deviation for N steps

S = Ns = 1012 10 4 angstrom = 10 2 meter

) ( ) = N ( s ) = (10 ) (1 angstrom )
2 12

= 10 4 meter

ASSIGNMENT (yellow pad, due Jan 19) 4-4 (Stowe, 1984) For air at room temperature, the probability of one molecule being in an excited electronic state is about 10-10 (i.e., p=10 10, q=1). In a typical room there are about 1028 air molecules. Using these numbers, calculate the following. (a) The mean number of electronically excited air molecules in a room at any time. (b) The standard deviation about this value. (c) The relative fluctuation about this value. 4-5 (Stowe, 1984) Consider identically prepared systems, each having 600 rolled dice. Suppose you are interested in the number of dice per system, n, with six dots showing upward. (a) What is the average value of n? (b) What is the standard deviation, ? (c) In the distribution of probabilities over the various possible states,

P600 (n ) = Ae

B ( n n )2

What are the values of the coefficients A and B? (d) If you rolled 600 dice, what is the probability of exactly 100 landing with 6 dots up? (e) What is the probability of exactly 96 landing with 6 dots up?

4-9 (Stowe, 1984) An ammonia bottle is opened very briefly in the center of a large room releasing many ammonia molecules into the air. These ammonia molecules go an average of so = 10 5 m in any direction between collisions with air (and other ammonia) molecules, and collide on the average of 107 times per second. After each collision they are equally likely to go in any direction. (a) What is the average displacement in one dimension (say the z-dimension) for any one step? (Hint. sz=socos in spherical coordinates. Taking the average over all solid angles gives

sz =

1 4

(s o

cos sin d d

(b) What is the square of the standard deviation for any one step? (Hint. sz is obtained by squaring the answer to part (a), and s 2 is obtained using
2 sz

( )2 = sz2 (sz )2

1 = 4

(s o cos )

sin d d

(c) What is the average displacement in the z-direction of the escaped NH3 molecules after 2 s? (d) What is the standard deviation about the value obtained in part (c)? (e) If you were standing 6 m from the bottle, how long would it take before more than 32% of the ammonia molecules were farther from the bottle than you?