Two equally strong perpendicular linear polarizations are able to add up to anything between a linear +/-45° oriented linear polarized field and a left or right handed circular polarized field only depending on their relative phase or retardation.
Just consider the left and right circular polarizations as two opposite circulating vectors. The linear polarization is a combination of the left and right circular polarizations with zero phase difference. A change of the phase difference causes the polarization rotation:
phase =0deg => pol azimuth =0deg;
phase =90deg => pol azimuth =45deg;
phase =180deg => pol azimuth =90deg;
phase =270deg => pol azimuth =135deg;
phase =360deg => pol azimuth =180deg =0deg.
In other words, the output polarization azimuth equals to half of the phase difference between two equal-amplitude circularly polarized components.
I have a book chapter "Polarization", which could help more
Mathematically one can transform anything into something else by choosing an appropriate basis set.
To physically transform anything that has a physical existence, one has to implement appropriate set of interacting entities. A linearly polarized beam first has to be sent through a polarizing crystal to break it into two linear orthogonally polarized beams and then one has to insert appropriately structured and oriented crystal devices to convert the two beams into left and right circularly polarized beams. Basic books describes this - how to convert a linearly polarized light into left or right circular polarization.
My key point is to underscore that we must adopt "Interaction Process Mapping Thinking" (IPM-T) to understand and do physics properly.
By the way, even when one combines two light beams into one common beam leveraging some of their different parameters, the beam amplitudes keep propagating as a common beam without interacting, or transforming each other's parameters, in the absence of joint interaction with some materials. This is known as "Non-Interaction of Waves" (NIW). This is built into Huygens-Fresnel's LINEAR diffraction Integral, and also in Maxwell's LINEAR wave equation. Amplitude superposition does not create observable phenomenon. The mathematically abstract "+" sign cannot execute the PHYSICAL OPERATION of summation on the amplitudes. When a detector executes the NONLINEAR quadratic operation on the superposed waves to absorb energy out of all the superposed beams, we have OBSERVABLE Superposition Effect. Jones Matrix works precisely because it sums the X-amplitudes and the Y-amplitudes separately through a tandem of polarizing components. At the end of the optical train, the detector displays the total energy, which is the sum of the separate energies due to the sum of X-components and separately the sum of the Y-components. Optical detectors are DIPOLES. They cannot be stimulated in two orthogonal directions at the same time. They will chose to respond to only one linear polarized component. They cannot respond to circularly polarized E-vector. (THINK! THIS IS VERY IMPORTANT). Text books do not underscore the physical processes that a detector has to undergo. This is why IPM-T is important epistemology! See the book, "Causal Physics: Photon by Non-Interaction of Waves", Taylor and Francis, 2014, by C. Roychoudhuri.
is it practically possible to pass light from fiber optic then polarizer and then fiber optic? ( fiber optic will be cutted to put the polarizer and how to collect the light and pass the light through fiber optic again?)