Identical particles

Consider two identical particles.  IF EÄE denotes the state space of the two particles and |ui,uj> a state vector in EÄE, then P21|ui,uj>=|uj,ui> must denote the same physical state, since we cannot distinguish the particles.  If |y> is a normalized state vector in EÄE then P21|y>=e|y>.  Since P212=I, e=±1.  Any physical state of two identical particles must be either symmetric or anti-symmetric with respect to the exchange of the particles.  Any physical state can be as a linear combination of basis vectors of the state space.  The state space of a two particle system therefore is not EÄE, but either the subspace or symmetric or the subspace of anti-symmetric states.  It is restricted.  This extends naturally to an N-particle system.

Postulate:

Problem:

Fermions, bosons, and classical particles obey different probability laws.  Assume that two particles are placed at random into one of two boxes.  What are the probabilities of finding the following distributions?

How do we construct a physical ket that correctly describes the physical state of a system of N identical particle?

Examples:

How do we find a basis for ES or EA?

{|yS>} or {|yA>} do not form a basis, because not all the vectors are linearly independent.

Example:

Applying S to a basis of EÄEÄE we can obtain the same vector in ES more than once.  We introduce the occupation number nk of the individual state |uk>, equal to the number of times that state appears in the tensor product |ui>Ä|uj>Ä… .

.

Two different tensor product vectors, which have the same occupation numbers, are equal to each other within a factor of ±1 after applying S or A.  We denote the basis states by |n1,n2,n3,…,nk,..>, ånk=N.  For bosons the nk are arbitrary, subject only to the restriction ånk=N, for fermions nk=0, or nk=1.


Consequences of the indistinguishability on the calculations of physical predictions:

Consider two particles.  We know one is in the state |f> and one is in the state |c>.  The state of the two particle system is

|f,c>=|1:f,2:c> for distinguishable particles,
|f,c>=(1/2)1/2(1+P21)|1:f,2:c> for indistinguishable bosons,
|f,c>=(1/2)1/2(1-P21)|1:f,2:c> for indistinguishable fermions.

Let {|ui>} be an orthonormal basis of eigenvectors of some operator B.  B|ui>=bi|ui>.  What is the probability that a measurement of B yields bn for one particle and bn' for the other particle?

The eigenstate of B associated with this measurement is

|1:un,2:un'> or |1:un',2:un> for distinguishable particles,
|un,un'> =(1/2)1/2(1+eP21)|1:un,2:un'> for indistinguishable particles,
with e=1 for bosons
and e=-1 for fermions.

We therefore have for distinguishable particles,

P(bn,bn')=| <1:un,2:un'|1:f,2:c>|2+| <1:un',2:un|1:f,2:c>|2.

P(bn,bn') is the sum of the probabilities in of finding the particles in either of the two distinguishable states.

For indistinguishable particles there exists only one state |un,un'>,  and we have

P(bn,bn')=|(1/2) <1:un,2:un'|(1+eP21T)(1+eP21)|1:f,2:c>|2
=|<1:un,2:un'|(1+eP21)|1:f,2:c>|2
=|<un|f><un'|c>+e<un|c><un'|f>|2

(direct term           exchange term)

=|<un|f><un'|c>|2+|<un|c><un'|f>|2
+2e(
Re(<un|f><un'|c>)Re(<un|c><un'|f>)+Im(<un|f><un'|c>)Im(<un|c><un'|f>)).
(interference terms)

Indistinguishability causes "interference terms" to appear when we calculate physical predictions.

If |un>=|un'>, bn=bn', then

P(bn,bn)=|<un|f><un|c>|2 for distinguishable particles,
P(bn,bn)=0 for indistinguishable fermions,
P(bn,bn=|<un|f><u'|c>+<un|c><un|f>|2
=2|<un|f><un|c>|2
for indistinguishable bosons.