The effects of radiatively decaying, long-lived particles on big-bang nucleosynthesis (BBN) are discussed. If high-energy photons are emitted after BBN, they may change the abundances of the light elements through photodissociation processes, which may result in a significant discrepancy between the BBN theory and observation. We calculate the abundances of the light elements, including the effects of photodissociation induced by a radiatively decaying particle, but neglecting the hadronic branching ratio. Using these calculated abundances, we derive a constraint on such particles by comparing our theoretical results with observations. Taking into account the recent controversies regarding the observations of the light-element abundances, we derive constraints for various combinations of the measurements. We also discuss several models which predict such radiatively decaying particles, and we derive constraints on such models.

We discuss a new mechanism which can solve the crisis in standard big-bang nucleosynthesis. A long-lived particle $X$ ($10^4 sec \lesssim \tau_X \lesssim 10^6 sec$) which decays into photon(s) will induce cascade photons, and destroy significant amounts of D and $^3$He without destroying $^4$He or too much $^7$Li. We numerically investigate this process and derive a constraint on the properties of $X$ such that the theoretical values of the primordial light-element abundances agree with observation. We also present some candidates for the unstable particle $X$.

We describe and compare two types of microwave sky simulations which are good for small angular scales. The first type uses expansions in spherical harmonics, and the second one is based on plane waves and the Fast Fourier Transform. The angular power spectrum is extracted from maps corresponding to both types of simulations, and the resulting spectra are appropriately compared. In this way, the features and usefulness of Fourier simulations are pointed out. For $\ell \geq 100$, all the simulations lead to similar accuracies; however, the CPU cost of Fourier simulations is $\sim 10$ times smaller than that for spherical harmonic simulations. For $\ell \leq 100$, the simulations based on spherical harmonics seem to be preferable.