Backflow transformation

First, how do the particles move in VMC In the continuum it is usually more efficient to move the particles one at a time by adding a random vector to a particle s coordinate, where the vector is either uniform inside of a cube, or is a normally distributed random vector centered around the old position. Unfortunately, this procedure cannot be used with backflow orbitals. This is because the backflow transformation couples all the electronic coordinates in the orbitals so that once a single electron move is attempted the entire Slater determinant needs to be recomputed, an 0 N ) operation. It is much more efficient to move all electrons at once, although global moves could become inefficient for large systems. 

Yet another approach to incorporating correlation effects into an antisymmetric function is to allow the orbitals (or geminals) to depend on the coordinates of the other electrons. This is accomplished by the backflow transformation in which the coordinates of the electrons are modified by a backflow displacement ... 

Another method that is able to modify the nodal surface with a few parameters is the introduction of backflow transformations. In this method the correlated motion of the electrons is built into the coordinates themselves by introducing the quasi-coordinates... 

Therefore, if we switch the field off abruptly, the backflow will bring the system into the twisted n = 2) state. However, if we smoothly reduce the field to zero, the backflow will be negligible and, according to Fig. 12.19, the system will follow curve B (state B) downward and smoothly transform into state n = 0). This selection of the final state has been confirmed experimentally using different forms of the voltage pulse either with the abrupt rear edge or the rear edge consisted of several steps down. 

---