What are the best packages that can generate candidate structures for quantum chemical optimization (DFT or CCSD) to find the global minimum? I am working on the gas-phase small inorganic clusters (number of atoms less than 30).
source documentation: https://open-babel.readthedocs.io/en/latest/3DStructureGen/multipleconformers.html
obabel <inputfile> -O <outputfile> --confab [confab options]
Some of the options include ecutoff and rcutoff (which help eliminate duplicates)
OpenBabel can be easily downloaded through Anaconda, the python package manager.
However, the energy cutoff criterion may be unreliable for inorganic molecules (you'll have to see) because the molecular mechanics parameters may be unreliable for those elements.
The best bet is to try to generate as many unique conformers as possible and optimize all of them to find which is the lowest energy.
The Cambridge Crystallographic Data Centre organizes a regular organic crystal structure prediction blind test as detailed in their website. These started in 1999 and the latest (the sixth one) was held in 2015. Participants, which include leading research groups in structure prediction, are asked to predict the structure of a range of molecules, the latest test including compounds like a small rigid molecule, drug candidates with multiple polymorphs, and salt hydrates. Although the title of the blind test specifies "organic" crystals, the methods used should be applicable to what you are trying to do.
Each blind test has an associated paper, and you can find the latest here.