Comments and additions to "A three-dimensional quantum vacuum-noise/signalbeamsplitter model for nonideal linear optical amplifiers"

In a recent paper (E. Desurvire, 1999, Opt. Fiber Technol., vol. 5, no. 1, 82), we presented a new model for describing nonideal (or realistic) linear optical amplifiers through a three-dimensional quantum beamsplitter (QBS) representation. This quantum model unifies the well-known properties of the passive attenuator and the ideal lossless optical amplifier into a single comprehensive picture. While such a representation stems from the straightf orward cascading of the two fundamental QBS types (i.e., attenuator followe d by ideal amplifier), we have emphasized that the cascading can only be ma de in such order, while the reverse one (attenuation following ideal amplif ication) is not applicable. We have also shown that realistic optical ampli fiers can be represented by discrete or infinite chains of attenuators/ amp lifier sequences, but always in the loss/gain order. The infinite-chain rep resentation raises a question about the relevance of a specific ordering. T he purpose of these comments is to clarify the issue and complement the dis cussion. In particular, we introduce an "elementary noise figure" (ENF) par ameter, corresponding to the noise figure of the QBS element; we show that the ENF can be made arbitrarily small (e.g., 10-50 mdB for k = 1000). Its i ncremental deviation from the nonphysical noiseless limit (F = 1) is shown to decrease in inverse proportion to the number of elements in the QBS chai n, (C) 2000 Academic Press.