MLatom@XACS cloud computing
This page provides a tutorial with required materials (sample input files, data sets etc) for using MLatom@XACS cloud computing service.
Register an account
Go to the XACS website http://xacs.xmu.edu.cn, click “Register” on the top right of the homepage, as shown below (in the red square).
To register an account, the user need to provide a valid email address, then click button “Send” in the next line. An email with verification code will be sent to the provided email address. Fill in the verification code after all other information is provided, and click “Register” at the bottom of the table, the account will be successfully created.
Submit a MLatom job
The input area for MLatom cloud computing is shown below:
The main input file can be uploaded or edited, and some auxiliary files may be needed. User can upload the prepared file from local by clicking the button in red square. By doing that, both the name and content of the auxiliary file will be uploaded and shown below. The user may edit the name and the content as he/she wants.
Alternatively, the user is also allowed to create the auxiliary file online.By clicking the button in blue square, an input area will be shown below.
The user can input the name and the content of the file in the left and middle area respectively, and save the modification by clicking “Save” in the right. By clicking “Cancel” in the right, the modification will be ignored.
For a file which is already uploaded or saved, the user may also edit the name and the content by clicking “Edit” in the right, or delete by clicking “Delete”.
The user may go to the page “My computing” to manage the jobs after submitting a job by clicking “my computing” at the bottom of the cloud computing page.
The same purpose can be achieved by clicking “My computing” in the left side bar after signing in.
There are 4 types of job status shown in column “Status”, namely:
- Waiting: The submitted job is in the queue and waiting for proceeding.
- Running: The submitted job is in computation.
- Completed: The computation is already finished.
- Failed: The job is failed to submit for computation.
Download the results
The user may download the result by clicking the corresponding link “Download” in the last column when the job is in the status of “Completed”, as shown in Figure 16. A zip file including all input and output files will be downloaded to the local.
This example is showcasing the application of out-of-the-box, general-purpose ML methods such as AIQM1 and ANI-1ccx which do not require training of your own models.
The theory behind tutorial is covered in the following publications:
P. Zheng, R. Zubatyuk, W. Wu, O. Isayev, P. O. Dral. Artificial Intelligence-Enhanced Quantum Chemical Method with Broad Applicability. Nat. Commun. 2021, 12, 7022.
Here, we will optimize the geometry of methane (CH4) molecule using AIQM1 and ANI-1ccx. The MLatom input file looks like this:
AIQM1 # or ANI-1ccx if you choose ANI-1ccx xyzfile=init.xyz optxyz=final.xyz geomopt optprog=ase
File init.xyz with the initial geometry of CH4 (you can use your own initial geometry):
On the platform for MLatom cloud computing http://xacs.xmu.edu.cn/records?attribution_program=MLatom, your screen should look like:
In your output file with .log extension, you should see:
In the folder you have downloaded, you should also have file ‘final.xyz’, which is the optimized geometry predicted by the method you used.
After you get the final geometries optimized by AIQM1 and ANI-1ccx, you can view their structures and compare their bond lengths in your favorite visualization program. CH4 geometry at AIQM1 is shown below: