What I understand is the following.
1) There is an abstract ket ( vector) which contains all the information about the system and it lives in an abstract vector space, the Hilbert Space which might or might not be infinite dimensional. The position basis/momentum basis/energy basis/ angular momentum basis is the basis of position/momentum/energy/angular momentum operator in this Hilbert space ( please don't go to rigged Hilbert spaces and right now loosely accept that a position eigen ket can be expressed as a linear combination of energy or momentum eigen kets somehow, not going into the details of how). Project the ket onto the basis kets and out of the scalar product construct the position, momentum, energy, angular momentum, etc spaces. The position space and momentum space share the structure of a function space and so accordingly redefine the inner product in them ( #Shankar). The inner product in the original Hilbert Space in defined without reference to any basis. Here inner product is defined w. r. t position space so it's definition changes ( becomes an integral). The energy/ angular momentum basis can have the same definition of inner product as the original Hilbert Space. The operators in the abstract Hilbert space are themselves abstract and need to be given a structure, which they are in these spaces. Wavefunction is in one to one correspondence with the system ket so quantum mechanics proceeds the same way in this function space as in the abstract Hilbert Space. The operators are written in position basis ( their matrix elements are w. r. t position basis, etc).
My questions are ( please address each of the following in your answer) -
1) Momentum operator can be written in position basis. How is energy operator or angular momentum operator written in position basis. Does writing energy operator in position basis means writing it's matrix elements in position basis or something else.
2) Why we don't do quantum mechanics in energy space rather than position space.
3) Can the complete Hilbert Space be considered as the sum ( not direct sum since space of eigenvectors of energy will not be closed under addition so will not be subspace of the original complete space) of spans of position, momentum, energy, angular momentum spaces, etc or not. I ask this because to incorporate Spin we need to taken tensor product of two spaces but here position and momentum operator don't commute so you can't take take a tensor product of their spaces.
4) How is that position space different from our real 3D space.
I couldn't get the answer till now. The already given answer doesn't doesn't answer my doubts.If anybody could help, it would be great.