Comments on: Point-and-Compute: Quantum Mechanics in “Auto” Mode https://hackingmaterials.com/2013/12/06/point-and-compute-quantum-mechanics-in-auto-mode/ applying computing to materials design Fri, 03 Jan 2014 20:12:20 +0000 hourly 1 http://wordpress.com/ By: Anubhav https://hackingmaterials.com/2013/12/06/point-and-compute-quantum-mechanics-in-auto-mode/comment-page-1/#comment-32 Fri, 03 Jan 2014 20:12:20 +0000 http://hackingmaterials.com/?p=345#comment-32 Thanks! In terms of methods, many of them are similar. The Electronic Structure Project uses slightly different methods than the rest (FP-LMTO/LDA), and the Computational Materials Repository uses a new GLLB-SC functional (which they claim give better band gaps) for much of its data. There are also a few differences in the way these projects employ a +U correction, and whether they ‘post-process’ results (Materials Project does this for many of its apps). However, the purpose of these databases doesn’t seem so much to validate and compare different methods as much as to provide ‘accurate’ data on many compounds.

Writing about the differences is a bit difficult, because many of these projects are in their infancy and will likely evolve substantially in the coming years. However, there are differences with regard to:

* Data coverage – OQMD has many metals/alloys and predicted compounds, Materials Project has many large oxides and ceramics and is currently mostly ICSD compounds, CMR has specialized data sets like complete coverage of perovskites
* Properties – AFLOWlib allows you to search on thermoelectric properties, Materials Project lets you filter on thermodynamic stability (e above hull), most (but not all) let you filter on band gap
* Data export – OQMD lets you download all the data, AFLOWlib is difficult to get a large amount of data from, and Materials Project provides a programmatic interface to the data.
* Additional features – Materials Project has many additional “apps” for using and exploring the data sets, e.g. Phase Diagram creation, Pourbaix app creation, or Li Battery Explorer
* Ease of use

Probably the best strategy is to explore and get results from a few of these databases and to monitor how they progress over time, since no database currently supersedes all others and many of these will overcome some of their major limitations in the near future.

]]>
By: Bi-Cheng Zhou https://hackingmaterials.com/2013/12/06/point-and-compute-quantum-mechanics-in-auto-mode/comment-page-1/#comment-31 Fri, 03 Jan 2014 19:23:32 +0000 http://hackingmaterials.com/?p=345#comment-31 Hi Anubhav,

Thanks for the interesting blog. I think your blog is the first of its kind in the field of high-throughput computing of materials.

Just a follow-up question of the above discussion. What are the main differences of the DFT databases listed above? It seems to me that most of them all use VASP as the ab-initio computing engine.

]]>
By: Anubhav https://hackingmaterials.com/2013/12/06/point-and-compute-quantum-mechanics-in-auto-mode/comment-page-1/#comment-28 Wed, 01 Jan 2014 16:23:45 +0000 http://hackingmaterials.com/?p=345#comment-28 In reply to Julen Larrucea.

Great! If you’re starting out, a few of these DFT databases include:

* Materials Project (https://www.materialsproject.org) – the one I work on
* AFLOWLib (http://aflowlib.org)
* Computational Materials Repository (https://cmr.fysik.dtu.dk/cmr/index.php)
* OQMD (http://oqmd.org)
* Electronic Structure Project (http://gurka.fysik.uu.se/ESP/)

Happy materials hunting!

]]>
By: Julen Larrucea https://hackingmaterials.com/2013/12/06/point-and-compute-quantum-mechanics-in-auto-mode/comment-page-1/#comment-27 Wed, 01 Jan 2014 13:02:50 +0000 http://hackingmaterials.com/?p=345#comment-27 Hi Anubhav,
Thanks for the post. It’s the first time I hear about the DFT result databases, and after browsing a little bit, I find it extremely convenient and efficient.
I am thinking on the hours I spent trying to reverse engineer “computational details” sections into inputs that reproduce the actual results accurately…

]]>