Observations by the Wilkinson Microwave Anisotropy Probe and the Planck mission suggest a hemispherical power amplitude asymmetry in the cosmic microwave background, with a correlation length on the order of the size of the observable Universe. We find that this anomaly can be naturally explained by an axion-like particle (ALP) cosmic string formed near our visible Universe. The field variation associated to this cosmic string creates particle density fluctuations after inflation, which consequently decay into radiation before the Big Bang Nucleosynthesis (BBN) era and resulted in the observed power asymmetry. We find in this scenario that the hemispherical power amplitude asymmetry is strongly scale dependent: $A(k)\propto {\rm exp}(-kl)/k$. Admittedly, typical inflation models predict a relic number density of topological defects of order one per observable Universe and so in our model the cosmic string must be tuned to have an impact factor of order $1/H_0$. Interestingly, the constraints based on purely cosmological considerations also give rise to a Peccei-Quinn scale $F_a$ of order $10^3$ larger then the Hubble scale of inflation $H_I$. Assuming $H_I\sim 10^{13}$GeV, we then have an ALP with $F_a\sim 10^{16}$GeV, which coincides with the presumed scale of grand unification. As we require ALP decays occur before the BBN era, which implies a relatively heavy mass or strong self-coupling, and considering that the associated potential should break the shift symmetry softly in order to protect the system from radiative corrections, we also conclude that the required ALP potential should be monodromic in nature.

The next generation of space-borne gravitational wave detectors may detect gravitational waves from extreme mass-ratio inspirals with primordial black holes. To produce primordial black holes which contribute a non-negligible abundance of dark matter and are consistent with the observations, a large enhancement in the primordial curvature power spectrum is needed. For a single field slow-roll inflation, the enhancement requires a very flat potential for the inflaton, and this will increase the number of $e$-folds. To avoid the problem, an ultra-slow-roll inflation at the near inflection point is required. We elaborate the conditions to successfully produce primordial black hole dark matter from single field inflation and propose a toy model with polynomial potential to realize the big enhancement of the curvature power spectrum at small scales while maintaining the consistency with the observations at large scales. The power spectrum for the second order gravitational waves generated by the large density perturbations at small scales is consistent with the current pulsar timing array observations.

The hidden sector U(1) vector bosons created from inflationary fluctuations can be a substantial fraction of dark matter if their mass is around $10^{-5}$eV. The creation mechanism makes the vector bosons' energy spectral density $\rho_{cdm}/\Delta E$ very high. Therefore, the dark electric dipole transition rate in atoms is boosted if the energy gap between atomic states equals the mass of the vector bosons. By using the Zeeman effect, the energy gap between the 2S state and the 2P state in hydrogen atoms or hydrogen like ions can be tuned. The $2S$ state can be populated with electrons due to its relatively long life, which is about $1/7$s. When the energy gap between the semi-ground $2S$ state and the 2P state matches the mass of the cosmic vector bosons, induced transitions occur and the 2P state subsequently decays into the 1S state. The $2P\to1S$ decay emitted Lyman-$\alpha$ photons can then be registered. The choices of target atoms depend on the experimental facilities and the mass ranges of the vector bosons. Because the mass of the vector boson is connected to the inflation scale, the proposed experiment may provide a probe to inflation.

Observations indicate that large-scale anomalies exist in the fluctuations of the cosmic microwave background. In these anomalies, the hemispherical power amplitude asymmetry has a correlation length comparable to that of the observable universe. We propose that a topological defect created by spontaneous breaking of the U(1) symmetry prior to inflation generated an initial phase variation, $\delta \theta$, across the observable region of the universe. The amplitude of this phase fluctuation is protected by topology if the defect is inside the horizon, and is frozen by causality if the defect exits the horizon. After inflation, the phase-corresponding boson field started to oscillate, when the Hubble rate decreased to a level comparable to the mass of the boson field. The energy density of the newly created boson particles varied across the observable universe. The bosons subsequently decayed into radiation prior to the BBN epoch, and the resulting fluctuations in the energy density produced the observed power asymmetry. This scenario predicts a scale-dependent modulation amplitude power asymmetry and in addition, as topological defects created by phase transitions are a very general phenomenon, the observed hemispherical asymmetry may be seen as an evidence for the cosmological inflation.

We explore a scenario that the dark matter is a boson condensate created by the misalignment mechanism, in which a spin 0 boson (an axion-like particle) and a spin 1 boson (the dark photon) are considered, respectively. We find that although the sub-MeV dark matter boson is extremely stable, the huge number of dark matter particles in a galaxy halo makes the decaying signal detectable. A galaxy halo is a large structure bounded by gravity with a typical $\sim10^{12}$ solar mass, and the majority of its components are made of dark matter. For the axion-like particle case, it decays via $\phi\to \gamma\gamma$, therefore the photon spectrum is monochromatic. For the dark photon case, it is a three body decay $A'\to\gamma\gamma\gamma$. However, we find that the photon spectrum is heavily peaked at $M/2$ and thus can facilitate observation. We also suggest a physical explanation for the three body decay spectrum by comparing the physics in the decay of orthopositronium. In addition, for both cases, the decaying photon flux can be measured for some regions of parameter space using current technologies.

If axion-like particles (ALPs) constitute a major part of dark matter (DM), due to the bosonic nature and a relative small mass, they could behave differently from the point like dark matter particles on the formation of the cosmic structures. When studying the structure formation, it is often useful to consider DM as a special fluid with a given density and a given velocity. ALP fluid obeys a same continuity equation comparing to the point-like collisionless DM but has a different first-order velocity equation. In the linear regime of structure formation, the resulted observational differences are negligible for the QCD axions but can be interesting for very light ALPs.