Elastic collisions at the microscopic level The Kinetic theory of gases states many postulates one of them being

Molecules collide elastically and there is no loss of energy.   

But as it was later seen that a lot of it's postulates were assumptions and were not completely accurate. My question is whether this postulate is also one such assumption. 
I am unable to understand how can any two bodies collide with no loss of energy in the real world even at the microscopic level...Is this a hypothetical assumption or does this really happen at the microscopic level?
 A: Strictly speaking an ideal gas is defined by (this is from Wikipedia):

An ideal gas is a theoretical gas composed of many randomly moving point particles that do not interact except when they collide elastically.

In practise most gases at STP are very close to ideal even though their collisions are almost always inelastic. In practice the requirement for elastic collisions doesn't matter that much.
Most collisions in real gases are inelastic because the sort of gases we encounter every day are polyatomic and therefore have rotational and vibrational excitations. In the vast majority of collisions the total kinetic energy of the colliding molecules after the collision won't be the same as the total kinetic energy before the collision because some energy will be transferred to or from rotational and vibrational excitations. Note however that the collisions will be elastic on average because at equilibrium the number of collisions that transfer energy to internal modes will be the same as the number that get energy from internal modes.
But all this does is change the specific heat, because each rotational mode gets a $\tfrac{1}{2}kT$ of energy and each vibrational mode gets a $kT$ of energy. The average kinetic energy of the gas molecules remains $\tfrac{3}{2}kT$ just as in an ideal gas, and the gas ends up obeying the ideal gas equation of state to a good approximation:
$$ PV = nRT $$
The deviations from this equation of state come mainly from the fact that real molecules are not point particles. They have a non-zero volume and they have long range interactions that also change their effective volume. Hence you get phenomenological equations like the Van der Waals equation of state that attempt to take this into account.
The point of all this is that the postulate of elastic collisions is routinely violated, but it doesn't matter.
