Dear Yuvam, if you are new to the code it might be more "robust" to test parts of your system (just a single-cell graphene; separately a single-cell oxide; ... ) before making and relaxing a supercell out of them. The point is, the choice of basis functions might be critical in Siesta because it represents a deal between accuracy you need and the computational effort you would like to spend, and it is easier to get some feeling about this compromise when testing small systems. Exactly for this reason the question "how fast is Siesta", put in general form, is ill-defined. The answer would depend on the accuracy needed (i.e., which properties you are after), moreover – a lot – on the geometry of your system, number of neighbours to each atom, amount of vacuum in the unit cell, etc. This said, an automatic relaxation of all crystallographic degrees of freedom is technically no problem with MD.VariableCell set to T, but, in a complex system, you might need to fix some of parameters (the lattice parameter of the substrate, or else) before engaging in a global uncontrollable relaxation which might lead to weird results. This said, I'd warn you not to expect too much tutorial-like introductions from the mailing list(tutorials being accessible in abundance via the Siesta web page), but answers to specific questions.Your inquiry seems mysterious without indicating who "they" used which scripts for which optimization, and what was confusing in it.Best regardsAndrei Postnikov ----- Yuvam Bhateja <yuvam...@gmail.com> a écrit : >Hello everyone, My name is Yuvam and I am an undergraduate student from Kolkata, India. I am new in siesta and have some experience with softwares like Quantum ESPRESSO. I want to perform geometrical optimization of my unit cell (atomic position as well as cell vectors). I have created my custom made unit cell to accommodate the graphene as well as metal oxide for gas sensing. I was following a tutorial in which they used scripts for optimization but I find it very confusing and very unsuitable as varying all 9 values of lattice vector using loops was very cumbersome. Can someone help provide any other method in which system does it by its own like in QE? And also, how fast is siesta as compared to other codes like QE, as my system consists of 200-300 atoms and using a cluster with 8 cores and 42 GB RAM. Thank you in advance. RegardsYuvam BhatejaB.Tech. 3rd yearE&TcIIEST Shibpur
-- SIESTA is supported by the Spanish Research Agency (AEI) and by the European H2020 MaX Centre of Excellence (http://www.max-centre.eu/)