Propagation is a very complex subject. To predict using purely mathematical means how a transmission from point A to B is more complex than trying to predict the throw of dice. There are stochastic processes, huge complexity, and many unknowns involved.
Simplifications must be made, and the valid simplifications depend on your objective, the frequency, and the application, among other things. If you want to understand all the physics relevant to propagation, the best I can do is refer you to many concepts to further research.
The most general solution is Maxwell's equations. All of the other concepts discussed here are solutions to these equations with certain assumptions added. If you want an exact model of how propagation occurs in all conditions, this is it. These equations are the foundation of classical electromagnetism, and consequently radio, optics, and electronics.
Of course to understand exactly the path from A to B would require a precise understanding of the geometry of everything in the path and near it, as well as the exact electromagnetic properties of every material within. That's not not realistically possible for complicated paths.
When the paths are simple we have things like the Friis transmission equation. All waves will be subject to this equation, though there may be other things contributing to the path effects. For things like space communications, or microwave links with nothing in the path's Fresnel zone, the Friis equation is an accurate and precise calculation of path losses. There's very little to be understood about this model besides the inverse square law.
Adding a bit of complexity, there's the two ray ground reflection model. This is relevant for many VHF and microwave links where the ground will reflect the signal and create multipath interference due to interference between the waves, but with otherwise clear paths.
Of course multipath interference can occur due to reflections off buildings and other things. You might try to use ray tracing to calculate the interference, though unless the reflecting surfaces are very much larger than the wavelength, diffraction will make this calculation inaccurate. A better calculation uses a field solver to approximate a solution to Maxwell's equations.
Environments often contain lossy materials, resulting in absorption. The ground is one such common lossy material. Steel structures will incur losses due to hysteresis. Resistive losses also exist in metal structures, among other things. Dielectric loss can occur, especially at higher frequencies. Water is a common factor in dielectric loss. At high enough frequencies, gasses in the atmosphere can cause absorption.
In practice, modeling the contour of the Earth, buildings, and trees in the path is infeasible, so some sort of propagation model will be used which approximates typical environments. Often these are a combination of educated guesses about the environment's properties and empirical data.
Skywave propagation such as on HF is very complicated due to conditions which are constantly changing. Here again, Maxwell's equations describe the behavior, but the chaotic and complex nature of the ionosphere makes a simple solution impossible. Many topics in the field of optics are relevant here. In practice empirical models are used, for example VOACAP.
I hope this gives you some things to research.
