MARICI demands on small memory resources; in fact, the data to search crystal structures is lattice parameters, several kinds of atomic radii, atomic positions, logical decision to assign the types of interatomic distance constraints on each pair of atoms, and logical constraints on the logical decision. That is why MARICI can execute a structural optimization per thread to design one crystal structure if the number of atoms per unit cell is small enough (less than about 30). However, since the simplest threading results in lacking the memory resoures for large chemical systems, the developer should implement parallel structural optimization for saving the consumption.

While MARICI can search complex crystal structures consisting of more than three kinds of chemical elements with small computation, the computational resource is lacked for eahaustive search for crystal structures consisting of more than 30 atoms of several kinds of chemical elements due to too large number of the compilation of chemical compositions. For such large chemical systems, initial structure generation based on space group constraints may be helpful, because it constrains not only spatial order of atoms but also chemical compositions.

As inorganic structural chemistry points out, chemical bonding distance can be estimated by the sum of empirically determined atomic radii, which correspond to the type of interatomic forces. However, since the conventional tables of atomic radii assign one value for each chemical element, the conventional tables are insufficient for mathematical crystal chemistry (MCC). That is why the latest MARICI does not have any table on atomic radii, which make all users fill in the values for each ionic chemical element. Note that MCC need both the minimum and maximum atomic radii for each chemical element. In fact, most of chemical elements have the flexibilities of their atomic radii. The developer must update the conventional tables on atomic radii by analyzing crystal structures registerd in Inorganic Crystal Structure Database for usability of MARICI.

Many face sharing of coordination octahedra are stabilized by the covalent bonds between the central cations, as you can see in the hexagonal Perovskite structure. However, the latest MARICI makes all pairs of cations form the interatomic distance constraints corresponding the ionic repulsion force. Therefore, when you search ionic crystal structures with face sharings of octahedra, you must assign smaller values to ionic-repulsion radii of the central cations, which increasing the number of optimal solutions resulting in increasing the computational cost. That is why the developer must introduce the metallic bonds.

Crystal structures of Zintl phase consist of ionic chemical elements, but covalent bonds are formed between anions or cations, which enables atoms to make different spatial orders of atoms from ionic crystal structures. This is because either anions or cations do not satisfy the octet rule of sp electronic orbitals. Since a covalent bond is formed between explicit pair of atoms as ionic bonds, it is easy to introduce the covalent bond to mathematical crystal chemistry. The developer will implement it to MARICI soon.

We can expect that unknown mixed-anion compounds consist of many kinds of unknown crystal structures owing to a fwo different sizes of anions. Although I have not yet checked completely, the latest MARICI have already been able to search crystal structures of them. However, additional constraints such as Pauling's second rule may be necessary to acceralate discovery of them. Since one of the main target of MARICI is mixed-anion compounds, the developer must formulate and implement the principles of crystal structures of mixed-anion compounds.

MARICI can search complex crystal structures consisting of more than three kinds of chemical elements with small computation. Now the developer try to collaborate with Okamoto and Yamaura laboratory in The Institute for Solid State Physics, The University of Tokyo, to discover unknown crystal structures of quaternary oxides.

Since MARICI just designs several kinds of spatial orders of atomic spheres for each chemical composition, MARICI cannot determine which prototypes of crystal structures are synthesizable. It means you must select the promising crystal prototypes among the optimal solutions, assign chemical elements to each atom site, and apply ab initio calculation for evaluating the stability of crystal structures. However, how to execute the first two parts? You can remove most of (maybe) unstable structures by such as upper limits of the number of atomic environments, but it is very arbitrary decision. Besides, the computational cost of ab initio simulation is not small, the exhaustive assignment of chemical elements to each atom site is infeasible due to finite computational resources.

Crystal structures determine their functionality through electronic structure. For example, if cations constitute the kagome lattice, magnetic flustration is caused. The developer keens to formulate the rule to select promising crystal structures having functionality by mixed-integer nonlinear programming as "Mathematical solid-state chemistry", but there is no idea now.

If you have no ideas on macro economics, I think you must learn it before criticizing the carbon-neutral society. I believe that we can realize Green Transformation (GX) accompanied by economic growth if scientists succeed in discovering very high-performance materials which are necessary for GX such as ferroelectric materials, photovoltaic materials, catalysts for sunlight water splitting, solid-state battery. Note that these innovative materials will be realized by oxides, nitrides, halides, chalcogenides, mixed-anion compounds, or Zintl phase, which MARICI is able to discover. Since it may be necessary for "green economic growth" to collaborate with entrepreneurs and angel investors, the developer keens to receive message from them.