Message boards : Projects : QuChemPedIA
Message board moderation
Author | Message |
---|---|
Send message Joined: 1 Jul 16 Posts: 146 |
Some hours showed up on WUProp for a new project called QuChemPedIA. About the most I could find out about it is the GitHub link. Does anyone have more info about it? https://github.com/bharismendy/QuChemPedIA |
Send message Joined: 2 Oct 05 Posts: 404 |
Translation from google... >>> This project is in the fields of data sciences and artificial intelligence. The goal is to revolutionize the field of quantum chemistry and molecular computing. Molecular chemistry is defined as the study of discrete entities (molecules) and corresponds to the widest community of chemists. Hundreds of millions of molecules are known, usually containing less than a hundred atoms and less than a thousand electrons. The chemical properties of these molecules depend on the position of the electrons, which can be calculated in an approximate way using methods from quantum mechanics. The various calculation parameters define the quality of the method and its cost in time (from a few hours to a few thousand hours per molecule). With the democratization of computational power, computer chemistry has become an essential part of chemistry research. Today, it is impossible for a human to take into account the results, even limited to the most important data, for millions of known molecules. For the chemists, the ambition of this project is to radically change the approach using artificial intelligence and data mining methods in order to more effectively explore the immense combinatorial space of the molecular space. The project is therefore at the frontier between artificial learning when it comes to predicting the properties of a molecule and combinatorial optimization when we seek to propose new molecules that maximize one or more objectives. The proposed methodology, which combines deep learning and constraint programming to generate molecules, is new to our knowledge and offers remarkable opportunities. Associating a cost function while predicting the synthesis routes makes it possible, for example, to propose molecules that are less costly for the environment, the basis of the green chemistry of tomorrow. The search for new photovoltaic materials is also the first of the applications we are considering. <<< Wave upon wave of demented avengers march cheerfully out of obscurity into the dream. |
Send message Joined: 8 Nov 10 Posts: 310 |
For the chemists, the ambition of this project is to radically change the approach using artificial intelligence and data mining methods in order to more effectively explore the immense combinatorial space of the molecular space.From what little I know about AI, it could be well suited to distributed computing. We can do the training calculations in small batches on our machines, and they can use the results on their machines (either CPU or GPU) for the final calculations. I don't have any idea if their science is any good, but the only way to find out is to try it. |
Send message Joined: 1 Jul 16 Posts: 146 |
I've read the google translate. Another member messaged me saying all the WUProp hours were from the only admin. So it's probably not open for testing yet. |
Send message Joined: 12 Feb 11 Posts: 419 |
So it's probably not open for testing yet. So, the url of this project is not public? |
Send message Joined: 10 May 07 Posts: 1444 |
I have not found any public link for BOINC yet. |
Send message Joined: 1 Jul 16 Posts: 146 |
The project title at WUProp now includes 'Private'. |
Send message Joined: 10 May 07 Posts: 1444 |
Found what appears to be the public Project web page: https://quchempedia.univ-angers.fr/athome/ Need VBOX64 to run apps. Also need INVITATION CODE TO JOIN/REGISTER APPS: Platform Version Created Average computing Microsoft Windows running on an AMD x86_64 or Intel EM64T CPU 0.81 (vbox64) 30 Jul 2019, 18:06:55 UTC 0 GigaFLOPS Linux running on an AMD x86_64 or Intel EM64T CPU 0.81 (vbox64) 30 Jul 2019, 18:06:18 UTC 15 GigaFLOPS Intel 64-bit Mac OS 10.5 or later 0.81 (vbox64) 30 Jul 2019, 18:06:32 UTC 0 GigaFLOPS |
Send message Joined: 8 Nov 10 Posts: 310 |
The goals and methodology look very nice. I would like to try it out as soon as possible, to check whether it runs better on AMD or Intel (probably Linux better than Windows, but I will check that too). But apparently it used VirtualBox, so there may not be very much difference. |
Send message Joined: 12 Feb 11 Posts: 419 |
Also need INVITATION CODE TO JOIN/REGISTER Uh, waiting for project to be open |
Send message Joined: 12 Feb 11 Posts: 419 |
Found what appears to be the public Project web page: https://quchempedia.univ-angers.fr/athome/ Site is down... |
Send message Joined: 12 Feb 11 Posts: 419 |
Found what appears to be the public Project web page: https://quchempedia.univ-angers.fr/athome/ The site is up (and with work) But still needs an invitation code to register... |
Send message Joined: 12 Feb 11 Posts: 419 |
They has opened forums |
Send message Joined: 8 Nov 10 Posts: 310 |
I just joined, attached, and downloaded work. They run on two cores each. Looks good. |
Copyright © 2024 University of California.
Permission is granted to copy, distribute and/or modify this document
under the terms of the GNU Free Documentation License,
Version 1.2 or any later version published by the Free Software Foundation.