Creation of coherent superpositions of Raman qubits by using dissipation

We consider a quantum system of three chainwise connected states 1-2-3, in a "Lambda" linkage pattern, in which the middle state is subjected to strong population decay. We show that this scenario can be used to create coherent superpositions between the two ground states. The idea is to deplete the population of the bright state formed by the two ground states 1 and 3 via the population loss channel. Then the surviving population is trapped in the dark state. The latter can be designed a priori, by selecting suitable couplings, to be equal to any desired coherent superposition of the ground states. In particular, for equal couplings on the 1-2 and 2-3 transitions, equal superposition between states 1 and 3 is created. Moreover, its phase can be controlled by the relative phase between the 1-2 and 2-3 fields. The present concept is an alternative to the slow adiabatic creation of coherent superpositions and may therefore be realized over short times, especially in the case where the middle state has a short life span. However, the price we pay for the fast evolution is associated with an overall 50% population loss. This issue can be removed in an experiment by using postselection. Interestingly, this physical scenario can be used to test the probabilistic interpretation of quantum mechanics in a very simple experiment. Indeed, if the probabilistic interpretation is correct and the decay from the middle state 2 is virtual (until measured), then the coherent superposition of states 1 and 3 is created. However, if the decay from state 2 actually occurs during the process, then there are two possibilities: Either state 2 decays and the population leaves the system completely, or state 2 does not decay and the population remains in the system. By suitably choosing the couplings we can make sure that in the latter case (no decay) the population returns to the initial state 1 with certainty. In both cases state 3 will end up with zero population. Therefore, by measuring the population of state 3 alone, we can find out if the probabilistic interpretation is correct (if its population is 1/4 ) or not (if its population is 0).