is it possible to have magnetic flux density B not in the same direction of magnetic field intensity H? it is said that direction of magnetic flux density B in the same direction of magnetic field intensity H for isotropic media so what is isotropic media and is it possible to have B not in the same direction of H 
 A: In general the relation between the magnetic induction vector $\vec{B}$ and the magnetic field vector $\vec{H}$ is 
$$\vec{B}=\mu\vec{H}$$
where $\mu$ is a second rank tensor (i.e a matrix).
In case the medium is isotropic then this matrix is proportional to the identity, let us say $$\mu=a\mathbb{1}_3$$
In this case, computing the matrix product you will see that $\vec{B}$ and $\vec{H}$ will have the same direction.
However, there exist media in which $\mu$ is a matrix non proportional to the identity and therefore the application of this matrix to vector $\vec{H}$ gives a vector $\vec{B}$ which is also rotated.
These media are the one you call non isotropic.
Actually having a medium isotropic around a given point $p$ means that an observer sitting in $p$ will find that properties of the media are the same anywhere he looks.
A non isotropic medium around a given point $p$ is a medium which properties depend on the direction.
We always assume the empty space (at least in a non relativistic context) as isotropic around ANY point. However, if you are in a medium made up by matter, nonisotropies can arise.
Note that also $\epsilon$, the electrical permittivity is a tensor, ans therefore you could have $\vec{D}$ non proportional to $\vec{E}$ but rotated as well.
Usually in introductory courses in electromagnetism one assumes isotropic media and treats $\mu$ and $\epsilon$ as constant real numbers. The way for doing this is purely pedagogical: it is easier to start with the easiest case, and also Maxwell equation in the general case of $\mu$ and $\epsilon$ matrices are much harder to be solved.
However, in the real world there exist tons on non isotropic media.
Note also that is dissipation is present, you can have complex numbers as the entries of those matrices. In this way you can take account of dissipation, for example in the case of waves propagating in dissipative media.
Note also that before I wrote "tensors" to not create confusion. They are actually tensor fields, (i.e sections of the (1,1) tensor bundle over spacetime).
In simpler words, the entries of $\mu$ and $\epsilon$ may depend on the position in space, and also in time. So it is actually better to write
$$\vec{B}(\vec{x})=\mu(\vec{x})\vec{H}(\vec{x})$$
so the angle at which $\vec{B}$ and $\vec{H}$ are rotated may not be costant in each place in space or instant of time.
You may also want to look at permeability and permittivity on wikipedia. The way this argument is treated there is quite simple.
Hope you found this answer clear enough.
I tried to write it as clear as possible.
