Consider the scalar fields $f(\theta, \phi)$ and $g(\theta, \phi)$ defined on the unit surface $S^2$, where $\theta$ is the co-latitude ($0$ at the north pole and $\pi$ at the south) and $\phi$ is the length. Show or a found counterexample such that:
$$\int_{S^2} (\nabla_s g \circ \nabla_s g)f(\theta, \phi)\sin\theta d\theta d\phi=-∫_{S^2}(fg∇^2_sg+g∇_sf \circ \nabla_sg)\sin\theta d\theta d\phi$$
where $\nabla_s$ is the operator gradient on $S^2$ in spherical coordinates $(\theta, \phi)$ $\circ$ denotes the vector or dot product and $\nabla^2_s$ is the Laplacian on $S^2$ (Hint: use Cartesian coordinates).
Usig gradiennt in spherical coordinate and doing dot product I have this: $$\int_{S^2} (\nabla_s g \circ \nabla_s g)f(\theta, \phi)\sin\theta d\theta d\phi=\int_{S^2}(\dfrac{\partial g}{\partial \phi})^2 f(\theta, \phi)sin\theta d\theta +\int_{S^2}(\dfrac{\partial g}{\partial \theta})^2\dfrac{1}{(\sin\phi)^2}f(\theta, \phi)sin\theta d\theta d\phi$$ please help I dont know wheter I'm right or wrong and how to continue, Do I use dot point or vector product?, when and how use the cartesian coordinates.