Inverse Modeling of Molecular Magnetic Quantum Materials

Jonathan Gant

Authors:  Jonathan Gant, William Perry, and Xiaoguang Zhang

Faculty Mentor: Dr. Xiaoguang Zhang

College:  College of Liberal Arts and Sciences


Ligand replacement is an avenue available for engineering single-molecule magnets, but the variable-length Cartesian representation prevents the application of machine learning techniques useful in the discovery of promising molecules. The Atomic Environment Vector (AEV) allows the application of machine learning techniques by mapping a ligand’s Cartesian representation of a ligand onto a fixed-length vector. The AEV loses information about the spatial configuration of the ligand’s atoms and must be mapped back into the Cartesian representation in order to be useful in quantum chemistry codes. The reverse Monte Carlo (RMC) method was implemented to recover the Cartesian representation from an arbitrary AEV. A modified data type, the Conic Atomic Environment Vector (CAEV), was created to correct for the loss of information attributed to the standard AEV by utilizing the conic geometry of ligand’s being attached to magnetic core regions. Though the RMC method had difficulty in reproducing molecules under both AEV representations, the CAEV produced more meaningful outputs. The development of more accurate inverse modeling methods could unlock the potential utility of machine learning for ligand searches and ultimately progress the development of conventional and quantum computing applications of single-molecule magnets.

Poster Pitch

Click the video below to view the student's poster pitch.


Click the image to enlarge.
0 0 vote
Presenter Rating
Newest Most Voted
Inline Feedbacks
View all comments
Mark Law
Mark Law (@guest_130)
1 year ago

Nice work.

Jonathan Gant
Jonathan Gant (@guest_286)
Reply to  Mark Law
1 year ago

Thank you!

Jonathan Gant
Jonathan Gant (@guest_300)
1 year ago

Good afternoon! Thank you for your interest in my research. I will be answering questions via zoom from 2 pm to 3 pm which can be accessed from the following link:

Zachary Player
Zachary Player (@guest_2128)
Reply to  Jonathan Gant
1 year ago

Looks good. Sounds really promising if it gets refined and we can implement machine learning to find the right ligand!

Jonathan Gant
Jonathan Gant (@guest_1226)
1 year ago

Comment test

Jonathan Gant
Jonathan Gant (@guest_1422)
1 year ago

It appears that the other link was not working properly so here is the new link to the zoom conference:

Sophia Paulitz
Sophia Paulitz (@guest_2024)
1 year ago

Great job Jon!!

Jonathan Gant
Jonathan Gant (@guest_2258)
Reply to  Sophia Paulitz
1 year ago

Thank you, hope you enjoyed! Let me know if you have any questions!

Thomas H Mareci
Thomas H Mareci (@guest_6810)
1 year ago

Nice work to address a very complex problem. You study addresses a very significant issue in memory storage and your introduction was very clear: From your presentation, I surmise that temperature (i.e. motion) causes a decay toward a low energy equilibrium state in spin alignment which results is a lose of magnetization. Your approach will take more than 3 minutes to explain but I get a sense of the method you applied to address the problem.