不同的体系，不同的自旋多重度，能量有所差别，究竟何种自旋多重度能量最低？AMS2018.105中，将引入该功能，允许用户在不知道自旋多重度的情况下，程序自动找到能量最低的自旋多重度。具体参考英文说明：

r68711: ADF: Implemented spin polarization optimization. The spin polarization can be optimized by minimizing the difference between Fermi level values for spin-alpha and spin-beta MOs. This method works reliably only when a reasonably large finite electron temperature is used (300K is found to be sufficiently large for small transition metal clusters). The polarization value is optimized up to a specified tolerance (fractional values are allowed) by step-wise reduction of the root-braketing interval for the objective function f=FermiLevel(alpha)-FermiLevel(beta). The SCF procedure is converged for each spin polarization. When the interval becomes smaller than the tolerance the optimization is considered converged. After that the SCF may optionally be repeated with the spin polarization set to the nearest even or odd number (depending on the total number of electrons in the system) or, in other words, with the total number of electrons per spin rounded to the nearest integer. The method is invoked by using one of the following Occupations keys:

OCCUPATIONS  ElectronicTemperature=300 OptimizeSpinRound=0.05
OCCUPATIONS  ElectronicTemperature=300 OptimizeSpin=0.05

Here, the ElectronicTemperature sets the electronic temperature, in Kelvin, for calculating the Fermi MO occupations. The OptimizeSpin and OptimizeSpinRound argument specifies the tolerance value. The OptimizeSpinRound key specifies that after convergence the SCF must be completed with the total number of electrons per spin rounded to the nearest integer.

It should be noted that as usual the Unrestricted and Charge keys must be specified as for any other spin-unrestricted calculation. The second argument of the Charge key sets the initial spin polarization value from which the optimization is started. The result of the optimization may depend on the initial value as the objective function may have more than one root.