The field of LBVs is one in which observations far outstrip plausible theories. For many years, the core was believed to be driving these pulsations through a mechanism similar to the mechanism that drives Ledoux-Schwarzschild pulsations (Ledoux 1941; Schwarzschild and Harm 1959). The mechanism has since been rejected as a probable cause since it inhibited by internal composition gradients and rotation, both of which are suspected to be characteristics of LBVs [](Stothers and Chin 1983)
Current theories on LBV instabilities favor radiation pressure and photospheric opacity as probable mechanisms. The classical Eddington limit, where opacity is due to electron scattering only, has already been explained to depict an LBV's tenuous hold on its outer layers. However, it does little to explain the sudden eruptions witnessed. This is why Humphreys and Davidson [](1984) and later Lamers [](1986) proposed the ``modified'' Eddington limit. In this limit, opacity is taken to be density and temperature dependent. A reduction in temperature favors the existence of hydrogen atoms and ions in the upper atmosphere of the star; both enhance surface opacity. For a star that is already near the Eddington limit, this increase in opacity could cause radiation pressure to overwhelm gravity.
For models incorporating the modified Eddington luminosity, , a star is only in a stable state if: -- the luminosity of the star is below the modified Eddington luminosity; and -- the star is in a local minimum of opacity with respect to effective temperature, i.e. small perturbations in temperature will not cause a runaway. Appenzeller [](1989) noted that opacity reaches a peak at , and that there are two local minima on either side. The atmosphere of a LBV may have two quasi-stable states, one for and one for . These could correspond to the quiescent and eruptive states of the star, respectively. The star oscillates between these two stable points until it loses nearly all of its envelope and evolves into a Wolf-Rayet star [](Maeder 1983).
Unfortunately, stars are not often this simple. Another relevant mode of energy transport, which has not been studied in detail (for LBVs), is convection. Although most high mass stars are usually assumed to have radiative envelopes, if the outer layers of LBVs are convective, then the total pressure must include both radiation and turbulence. In theory, this works in ways similar to the Eddington limit -- the inward force of gravity must be able to offset the outward turbulent pressure. While this is a mechanism that has not yet been calculated for LBVs, it is possible that convection is working together with radiation to produce the resultant variability. Other possible mechanisms involve sub-photospheric density inversions [](Maeder 1989). Observed bipolar structure have inspired variations on the previous discussion for close-binary systems, and rapid rotation [](Lamers and Pauldrach 1991).
