Two improvements in Grover's algorithm

When the current Grover algorithm is applied to searching some targets in an unsorted quantum database, the differences of the significance for each target are not taken into consideration. When the targets are more than a quarter of the total items, the probability of finding targets rapidly falls with the increase of the targets, and when the targets are more than half of the total items, the algorithm will be invalidated. Aiming at these problems, firstly, an improvement based on the weighted targets is proposed in which each target is endowed a weight coefficient according to its significance. Using these different weight coefficients, we give a quantum superposition of all target states, which may make the probability for each target equal to its weight coefficient. Secondly, an improved phase matching is proposed in which two directions of phase rotation are same, and two magnitudes are determined by the inner product of the superposition of target states and the initial system state. When this inner product is more than 1/2, using the improved phase matching, The targets can be found with a probability of 100% and by the only one Grover iteration. Finally, the validity of these measures is validated by a simple searching example.