1-compute the airghbors of each data point.
2-compute the weihts Wi,j that best reconstruct each data point X from its neighbor.
3-compute the vector Y best reconstructed by the weights Wi,j . [4] 
              And we have other  dimention reduction such us :
 
          Now I will speak for a use local tangent space for nonlinear dimention reduction .  So  what's local tangent  ??