Janina

Your image contains the answer. Your basis set is not linearly independent. Check your basis set and try changing it to something a bit smaller. I am not familiar with Material Studio but some programs allow the user to remove a vector corresponding to the negative eigenvalue of the overlap matrix.

Does Materials Studio automatically check for double atoms? If not, check your structure, if you have two atoms in the identical location, the determinant containing the orbitals cannot be set up properly.

Ajit J. Thakkar Sir, actually im also new in material studio.

Could you please tell me which is basis in that picture? and what is linear dependency in basis set?

thank you

Jürgen Weippert Sir, I didn't understand what you said "'Does Materials Studio automatically check for double atoms''

I'm adsorbing Mg atom in a graphene nanosheet and increasing no of atom, the adsorption site is fixed which is tetragon site. A 2D sheet surface contain 9 tetragon site. So when I increase number of atom,ultimately there sets more than one atom in a site.

so what would be the solution ?

If your input file accidentally contains two atoms in the same location, you have a Pauli-forbidden geometry. You can't see that on a picture, you have to check your list of atoms. If there are more atoms in your input file than you see in your visual output, that would be the cause. There's no guarantee that this is what's going on here, but such a geometry error could cause an error message like that.

I agree with Jürgen Weippert and Ajit J. Thakkar . It is definitely worth checking that there are no atoms at the same point in space (or very close), this is a very common cause of non-positive definite overlap matrices in pretty much all DFT codes.

Your text output from DMol3 is obscured by the graphical windows in the screenshot, but if we take what you quoted it says:

"Consider increasing the atomic cutoff parameter or changing the basis set"

so DMol3 has determined that there are linear dependencies in the basis set (probably because orbitals are overlapping too much) and suggests that you change your cutoff or basis set. Have you tried doing that? If you have, I suggest that you read that DMol3 output carefully, and see if there are other suggestions elsewhere. If you still aren't sure what to do you can always post the full output here.

In short:

1. Check that no atoms are on top of each other, and no atoms are very close to each other

2. Try to change the atomic cutoff parameter

3. Try a different basis set

This well-known problem has been silenced or almost in the literature.

Why? I suppose that the ones who invented the diffuse functions didn't fully discuss the problem and opted for a not-too-elegant solution. Obviate the eigenvalues of the overlap that don't appear really positive.

Please, try to read the attached paper. The problem is a bit discussed.

Also, have a look at the so-called Hilbert matrix eigensystem. It is a simple example of this behavior.

The error message you received indicates that the overlap matrix in your dMol3 calculation is not positive definite. This typically occurs when the basis set used in the calculation contains linear dependencies, which means that some of the basis functions used to describe the electronic wavefunction are not independent and can be expressed as a linear combination of other basis functions. This can lead to an ill-conditioned overlap matrix, which can cause convergence problems in the calculation.

To solve this problem, you can try the following steps:

If none of these steps solve the problem, you may need to seek additional assistance from the support team for the software or consult with an expert in the field.