using x-ray

Hadi Jabbar Alagealy yes sir

but

if we have only one option of CASTEP, in every situation we are to use CASTEP and we have not space group and lattice parameters then another option available ?

Suppose if we use expected space group and lattice parameters CASTEP will traced or not ?

There are many options, but I recommend that first you read the literature on crystal structure prediction -- in combination with ab initio methods like DFT, it is often called ex nihilo crystal structure prediction.

The basic issue is that DFT programs like CASTEP will optimise the crystal structure by minimising the forces and stresses, but this will only lead to the stable structure which is "closest" to the starting structure. For example, if you put carbon atoms in a cubic cell in approximately tetrahedral coordination, CASTEP will find the diamond structure; yet graphite is a lower energy form. The reason CASTEP did not find graphite, is because in order to transform the initial structure into graphite, it is necessary to pass through some intermediate structures which have higher energies, forces and stresses -- in other words, there is an energy barrier between the two stable forms.

In technical terms, CASTEP's built-in methods are known as "local optimisation methods", and they find the structure which is a local energy minimum. When you are doing crystal structure prediction you usually need a "global optimisation method". There are many different methods, including:

Random/Monte Carlo

Genetic Algorithms

Basin hopping (or even simulated annealing)

For the "random" method, you could investigate AIRSS which has a good CASTEP interface (in fact, it was designed for CASTEP): https://www.mtg.msm.cam.ac.uk/Codes/AIRSS

There are many other packages to do genetic algorithms etc, such as XtalOpt or CALYPSO.

Hope that helps,

Phil