We investigate the stability of general-relativistic boson stars by classifying singularities of differential mappings and compare it with the results of perturbation theory. Depending on the particle number, the star has the following regimes of behavior: stable, metastable, pulsation, and collapse.

We reconsider the nonlinear second order Abel equation of Stewart and Lyth, which follows from a nonlinear second order slow-roll approximation. We find a new eigenvalue spectrum in the blue regime. Some of the discrete values of the spectral index n_s have consistent fits to the cumulative COBE data as well as to recent ground-base CMB experiments.

There is accumulating evidence that (fundamental) scalar fields may exist in Nature. The gravitational collapse of such a boson cloud would lead to a boson star (BS) as a new type of a compact object. Similarly as for white dwarfs and neutron stars, there exists a limiting mass, below which a BS is stable against complete gravitational collapse to a black hole. According to the form of the self-interaction of the basic constituents and the spacetime symmetry, we can distinguish mini-, axidilaton, soliton, charged, oscillating and rotating BSs. Their compactness prevents a Newtonian approximation, however, modifications of general relativity, as in the case of Jordan-Brans-Dicke theory as a low energy limit of strings, would provide them with gravitational memory. In general, a BS is a compact, completely regular configuration with structured layers due to the anisotropy of scalar matter, an exponentially decreasing 'halo', a critical mass inversely proportional to constituent mass, an effective radius, and a large particle number. Due to the Heisenberg principle, there exists a completely stable branch, and as a coherent state, it allows for rotating solutions with quantised angular momentum. In this review, we concentrate on the fascinating possibilities of detecting the various subtypes of (excited) BSs: Possible signals include gravitational redshift and (micro-)lensing, emission of gravitational waves, or, in the case of a giant BS, its dark matter contribution to the rotation curves of galactic halos.

A possible equivalence of scalar dark matter, the inflaton, and modified gravity is analyzed. After a conformal mapping, the dependence of the effective Lagrangian on the curvature is not only singular but also bifurcates into several almost Einsteinian spaces, distinguished only by a different effective gravitational strength and cosmological constant. A swallow tail catastrophe in the bifurcation set indicates the possibility for the coexistence of different Einsteinian domains in our Universe. This `triple unification' may shed new light on the nature and large scale distribution not only of dark matter but also on `dark energy', regarded as an effective cosmological constant, and inflation.

Galactic dark matter is modelled by a scalar field in order to effectively modify Kepler's law without changing standard Newtonian gravity. In particular, a solvable toy model with a self-interaction U(Phi) borrowed from non-topological solitons produces already qualitatively correct rotation curves and scaling relations. Although relativistic effects in the halo are very small, we indicate corrections arising from the general relativistic formulation. Thereby, we can also probe the weak gravitational lensing of our soliton type halo. For cold scalar fields, it corresponds to a gravitationally confined Boson-Einstein condensate, but of galactic dimensions.

A non-topological soliton model with a repulsive scalar self-interaction of the Emden type provides a constant density core,similarly as the empirical Burkert profile of dark matter haloes. As a further test, we derive the gravitational lens properties of our model, in particular, the demarcation curves between `weak' and `strong' lensing. Accordingly, strong lensing with typically three images is almost three times more probable for our solitonic model than for the Burkert fit. Moreover, some prospective consequences of a possible flattening of dark matter haloes are indicated.

Galactic dark matter is modelled by a scalar field. In particular, it is shown that an analytically solvable toy model with a non-linear self-interaction potential U(Phi) leads to dark halo models which have the form of quasi-isothermal spheres. We argue that these fit better the observed rotation curves of galaxies than the centrally cusped halos of standard cold dark matter. The scalar field model predicts a proportionality between the central densities of the dark halos and the inverse of their core radii. We test this prediction successfully against a set of rotation curves of low surface brightness galaxies and nearby bright galaxies.

By using the rather stringent nonlinear second order slow-roll approximation, we reconsider the nonlinear second order Abel equation of Stewart and Lyth. We determine a new blue eigenvalue spectrum. Some of the discrete values of the spectral index $n_s$ have consistent fits to the cumulative COBE data as well as to recent ground-base CMB experiments.

A four-parameter class of exact asymptotically flat solutions of the Einstein-Maxwell equations involving only rational functions is presented. It is able to describe the exterior field of a slowly or rapidly rotating neutron star with poloidal magnetic field.

The present surge for the astrophysical relevance of boson stars stems from the speculative possibility that these compact objects could provide a considerable fraction of the non-baryonic part of dark matter within the halo of galaxies. For a very light `universal' axion of effective string models, their total gravitational mass will be in the most likely range of \sim 0.5 M_\odot of MACHOs. According to this framework, gravitational microlensing is indirectly ``weighing" the axion mass, resulting in \sim 10^{-10} eV/c^2. This conclusion is not changing much, if we use a dilaton type self-interaction for the bosons. Moreover, we review their formation, rotation and stability as likely candidates of astrophysical importance.

Using the Hubble parameter as new `inverse time' coordinate ($H$-formalism), a new method of reconstructing the inflaton potential is developed also using older results which, in principle, is applicable to any order of the slow-roll approximation. In first and second order, we need three observational data as inputs: the scalar spectral index $n_s$ and the amplitudes of the scalar and the tensor spectrum. We find constraints between the values of $n_s$ and the corresponding values for the wavelength $\lambda $. By imposing a dependence $\lambda (n_s)$, we were able to reconstruct and visualize inflationary potentials which are compatible with recent COBE and other astrophysical observations. >From the reconstructed potentials, it becomes clear that one cannot find only one special value of the scalar spectral index $n_s$.

The mechanism of the initial inflation of the universe is based on gravitationally coupled scalar fields $\phi$. Various scenarios are distinguished by the choice of an {\it effective self--interaction potential} $U(\phi)$ which simulates a {\it temporarily} non--vanishing {\em cosmological term}. Using the Hubble expansion parameter $H$ as a new ``time" coordinate, we can formally derive the {\it general} Robertson--Walker metric for a {\em spatially flat} cosmos. Our new method provides a classification of allowed inflationary potentials and is broad enough to embody all known {\it exact} solutions involving one scalar field as special cases. Moreover, we present new inflationary and deflationary exact solutions and can easily predict the influence of the form of $U(\phi)$ on density perturbations.