spin deviation operator

[jackychenp]

Hi,

In Holstein-Primakoff process, nl = al*al and Nl Psinl = nl Psinl
Why (1-Nl/2S)1/2 Psinl = (1-nl/2S)1/2 Psinl ? Since the operator (1-Nl/2S)1/2 is not linear, I think it should not work as a linear operator. Please let me know if I misunderstood something.

[kanato]

You can expand the operator (1-n_l/2S)^{1/2} in a Taylor series. Each term will have a power of n_l. These evaluate directly to N_l when operating on a wavefunction if it is an eigenstate of the n_l operator. In that case, you wind up with a Taylor series which is the same as your original Taylor series, but with the operator replaced by its eigenvalue. The linearity of the operator will not matter, although you will have significant complications if you have multiple non-commuting operators within your series.

[jackychenp]

Kanato, thanks a lot. That can explain.