We have used two-fluid dynamics to study the discrepancy between the work of Thouless, Ao, and Niu (TAN) and that of Iordanskii. In TAN no transverse force on a vortex due to normal fluid flow was found, whereas the earlier work found a transverse force proportional to normal fluid velocity $u_n$ and normal fluid density ${\rho }_n.$ We have linearized the time-independent two-fluid equations about the exact solution for a vortex, and find three solutions that are important in the region far from the vortex. Uniform superfluid flow gives rise to the usual superfluid Magnus force. Uniform normal fluid flow gives rise to no forces in the linear region, but does not satisfy reasonable boundary conditions at short distances. A logarithmically increasing normal fluid flow gives a viscous force. As in classical hydrodynamics, and as in the early work of Hall and Vinen, this logarithmic increase must be cut off by nonlinear effects at large distances; this gives a viscous force proportional to $u_n/\ln u_n,$ and a transverse contribution that goes like $u_n/(\ln u_n{)}^2,$ even in the absence of an explicit Iordanskii force. In the limit $u_n\to 0$ the TAN result is obtained, but at nonzero $u_n$ there are important corrections that were not found in TAN. We argue that the Magnus force in a superfluid at nonzero temperature is an example of a topological relation for which finite-size corrections may be large.

• Received 18 January 2001

DOI:https://doi.org/10.1103/PhysRevB.63.224504