Since the 1930s and until recently, the antiferromagnetism prediction in a density functional theory (DFT) had failed. The anion (negative ion) O in Cu oxide anitiferromagnets is stable, but an isolated O is not observed in a vacuum, although an isolated O is observed experimentally in a vacuum. The conventional discrete variational (DV) method in a DFT uses atomic orbital basis functions calculated numerically. The DV method adds a well potential to the potential for electrons to stabilize the basis set, and the radius and depth of the well are not determined uniquely, with the usual well depth having the relatively deep value of 1 Eh. To overcome this difficulty for antiferromagnetism, the present author has developed the SIWB (surrounding or solid Coulomb-potential-induced well for basis set) method to improve the anion basis set. The SIWB method uniquely determines the well depth considering the Coulomb potential from surrounding nuclei and electron clouds for the negative ion. The SIWB method also uniquely sets the well radius to Req, which is the distance between the anion nucleus and a point where the electronic charge density of the anion coincides with that of the nearest cation. The anion well depth determined from SIWB is shallower, improving the basis set for delocalizing electrons on the negative ion site, and reproduces the antiferromagnetism in a DFT.