Tag: artificial intelligence

Sep 22 2024

End of summer, and the start of a new academic year.

  • Filed under blog

  • September 22, 2024

Mid September in Hasselt means the start of a new academic year. It brings to an end a summer of doing some unrestricted research. Especially the last month has been extremely busy.

  • The DFT2024 conference in Paris from August 25th until 30th, where I presented our recent work on the GeV defect, which we will be submitting shortly.
  • The materiomics summer school, where I gave a lecture on performing practical quantum mechanical calculations,
  • The public PhD defense of Emerick Guillaume (QuATOMs group member) on the growth of diamond: congratulations Emerick!
  • A seminar at UNamur on “extreme machine learning”, discussing our work on small datasets and some of the work I did this summer on a spray coating dataset.

Today the first week of academic year ended, and I already had the pleasure teaching quantum mechanics and modelling courses to chemistry and materiomics students. We also welcome 2 MSc materiomics students to the group: Brent Motmans and Eleonora Thomas. Brent will be working on an experimental-theoretical project, where the theoretical side will focus on machine learning of his experimental data. Eleonora on the other hand will be combining DFT and machine learning in her study of diamond. A very warm welcome to the QuATOMs group for both.

 

Permanent link to this article: https://dannyvanpoucke.be/end-of-summer-and-the-start-of-a-new-academic-year/

Dec 22 2023

Materiomics Chronicles: week 13 & 14

Weeks eleven and twelve gave some rest, needed for the last busy week of the semester: week 13. During this week, I have an extra cameo in the first year our materiomics program at UHasselt.

NightCafe's response to the prompt: "Professor teaching quantum chemistry."

NightCafe’s response to the prompt: “Professor teaching quantum chemistry.”

Within the Bachelor of chemistry, the courses introduction to quantum chemistry and quantum and computational chemistry draw to a close, leaving just some last loose to tie up. For the second bachelor students in chemistry, this meant diving into the purely mathematical framework describing the quantum mechanical angular momentum and discovering spin operators are an example, though they do not represent an actual rotating object. Many commutators were calculated and ladder operators were introduced. The third bachelor students in chemistry dove deeper in the quantum chemical modeling of simple molecules, both in theory as well as in computation using a new set of jupyter notebooks during an exercise session.

In the first master materiomics, I had gave the students a short introduction into high-throughput modeling and computational screening approaches during a lecture and exercise class in the course Materials design and synthesis. The students came into contact with materials project via the web-interface and the python API. For the course on Density Functional Theory there was a final guest response lecture, while in the course Machine learning and artificial intelligence in modern materials science a guest lecture on optimal control was provided. During the last response lecture, final questions were addressed.

With week 14 coming to a close, the first semester draws to an end for me. We added another 15h of classes, ~1h of video lecture, and 3h of guest lectures, putting our semester total at 133h of live lectures (excluding guest lectures, obviously). January and February brings the exams for the second quarter and first semester courses.

I wish the students the best of luck with their exams, and I happily look back at surviving this semester.

Permanent link to this article: https://dannyvanpoucke.be/materiomics-chronicles-week-13-14/

Dec 10 2023

Materiomics Chronicles: week 11 & 12

After the exam period in weeks nine and ten, the eleventh and twelfth week of the academic year bring the second quarter of our materiomics program at UHasselt for the first master students. Although I’m not coordinating any courses in this quarter, I do have some teaching duties, including being involved in two of the hands-on projects.

As in the past 10 weeks, the bachelor students in chemistry had lectures for the courses introduction to quantum chemistry and quantum and computational chemistry. For the second bachelor this meant they finally came into contact with the H atom, the first and only system that can be exactly solved using pen and paper quantum chemistry (anything beyond can only be solved given additional approximations.) During the exercise class we investigated the concept of aromatic stabilization in more detail in addition to the usual exercises with simple Schrödinger  equations and wave functions. For the third bachelor, their travel into the world of computational chemistry continued, introducing post-Hartree-Fock methods with also include the missing correlation energy. This is the failure of Hartree-Fock theory, making it a nice framework, but of little practical use for any but the most trivial molecules (e.g. H2 for example already being out of scope). We also started looking into molecular systems, starting with simple diatomic molecules like H2+.

SnV split vacancy defect in diamond.

SnV split vacancy defect in diamond.

In the master materiomics, the course Machine learning and artificial intelligence in modern materials science hosted a guest lecture on Large Language Models, and their use in materials research as well as an exercise session during which the overarching ML study of the QM9 dataset was extended. During the course on  Density Functional Theory there was a second lab, this time on conceptual DFT. For the first master students, the hands-on project kept them busy. One group combining AI and experiments, and a second group combining DFT modeling of SnV0 defects in diamond with their actual lab growth. It was interesting to see the enthusiasm of the students. With only some mild debugging, I was able to get them up and running relatively smoothly on the HPC. I am also truly grateful to our experimental colleagues of the diamond growth group, who bravely set up these experiments and having backup plans for the backup plans.

At the end of week 12, we added another 12h of classes, ~1h of video lecture, ~2h of HPC support for the handson project and 6h of guest lectures, putting our semester total at 118h of live lectures. Upwards and onward to weeks 13 & 14.

Permanent link to this article: https://dannyvanpoucke.be/materiomics-chronicles-week-11-12/

Nov 12 2023

Materiomics Chronicles: week 8

After the complexities of week seven, week eight brings the last lecture week of the first quarter of the academic year. After this week, the students of our materiomics program at UHasselt will start studying for a first batch of exams. It also means with this week, their basic courses come to an end and they have all been brought up to speed and to a similar level, needed for the continuation of their study in the materiomics program.

In the bachelor program, the third bachelor chemistry students ended their detailed study of the He atom in the course quantum and computational chemistry with the investigation of its excited states. They learned about the splitting of in singlet and triplet states as well as Fermi-holes and heaps.

Vulcanoplot

Vulcano-plot of small data model quality of model instances in a large ensemble. Taken from our paper “Small Data Materials Design with Machine Learning: When the Average Model Knows Best“, J. Appl. Phys. 128, 054901 (2020)

The first mater materiomics students got their last lecture in the course Fundamentals of materials modeling, where we looked into some examples of application of machine learning in materials research. We also brought all levels of the course together and imagined how to link these in a multiscale project. Starting from the example of a windmill we discussed the application of computational materials modeling at different scales. For the course Properties of functional materials, the third and final presentation and discussion was held, now focusing on characterization methods. The second master students had response lectures for the courses on Density Functional Theory and Machine learning and artificial intelligence in modern materials science where the various topics of the self study were discussed (e.g., concepts of Neural Networks in case of the latter).

At the end of this week, we have added another 8h of live lectures, putting our semester total at 99h of live lectures. With the workload of the first master materiomics coming to an end, the following chronicles will be biweekly. Upwards and onward to week 9&10.

 

Permanent link to this article: https://dannyvanpoucke.be/materiomics-chronicles-week-8/

Nov 05 2023

Materiomics Chronicles: week 7

After a relatively chill week six, the seventh week of the academic year ended up being complicated. As it was fall-break the week only consisted of two class days at university. However, primary schools are closed entirely so our son was at home having a holiday, while both parents were trying to juggle classes and project proposal deadlines as well as additional administrative reporting. A second evaluation meeting with the students of our materiomics program at UHasselt took place (second master this time), and also these students appreciated the effort put into creating their classes.

Although there were only two days of teaching, this did not mean there was little work there. The students of the second bachelor in chemistry extended their knowledge of a particle in a box to the model of a particle on a ring, during the course introduction to quantum chemistry. Similar as for a particle in a box, this can be considered a simplified model for a circular molecule like benzene, allowing us to estimate the first excitation quite accurately.

For the course Fundamentals of Materials Modeling, there was a lecture introducing the first master students materiomics into the very basics of machine learning, as well as an exercise session. During these, the students learned about linear regression, decision trees and support vector machines. This class was also open to students of the bachelor programs to get a bit of an idea of the content of the materiomics program. Finally, the first master students also presented the results of their lab on finite element modeling as part of the course Fundamentals of Materials Modeling. They presented flow studies around arrows, reef and car models, as well as heat transfer in complex partially insulated systems, as well as sinking boats. They showed they clearly gained insight through this type of hands-on tasks, which is always a joy to note, resulting in grades reflecting their efforts and insights.

Though this week was rather short, we added another 6h of classes, putting our semester total at 91h of live lectures. Upwards and onward to week 8.

Permanent link to this article: https://dannyvanpoucke.be/materiomics-chronicles-week-7/

Oct 22 2023

Materiomics Chronicles: week 5

After week four, this fifth week of the academic year is most arguably the most intense and hectic week of teaching. With 22h of classes and still two classes that needed to be prepared from scratch (even including weekends time was running out), I’m tired but happy it is over. However, all the effort is worth it, and I was happy to hear the students of our materiomics program at UHasselt appreciate the effort put into creating their classes, during an evaluation meeting.

The corral is an artificial structure created from 48 iron atoms (the sharp peaks) on a copper surface. The wave patterns in this scanning tunneling microscope image are formed by copper electrons confined by the iron atoms. Don Eigler and colleagues created this structure in 1993 by using the tip of a low-temperature scanning tunneling microscope (STM) to position iron atoms on a copper surface, creating an electron-trapping barrier. This was the first successful attempt at manipulating individual atoms and led to the development of new techniques for nanoscale construction.source: https://www.nisenet.org/catalog/scientific-image-quantum-corral-top-view

The corral is an artificial structure created from 48 iron atoms (the sharp peaks) on a copper surface. The wave patterns in this scanning tunneling microscope image are formed by copper electrons confined by the iron atoms. Don Eigler and colleagues created this structure in 1993 by using the tip of a low-temperature scanning tunneling microscope (STM) to position iron atoms on a copper surface, creating an electron-trapping barrier. This was the first successful attempt at manipulating individual atoms and led to the development of new techniques for nanoscale construction.
source: https://www.nisenet.org/catalog/scientific-image-quantum-corral-top-view

For the second bachelor students in chemistry the introduction to quantum chemistry finally put them into contact with some “real” quantum mechanics when they were introduced into the world of potential barriers, steps, and wells. Though these are still abstract and toy-model in nature, they provide a first connection with reality, where they can be seen as crude approximations of the potential experienced by valence electrons near the surface, or STM experiments. They were also introduced to my favorite quantum system related to this course: the quantum corral. Without any effort it can be used in half a dozen situations with varying complexity to show and learn basic quantum mechanics. For the third bachelor chemistry students the course quantum and computational chemistry finally provided them the long promised first example of a non-trivial quantum chemical object: The Helium atom. With it’s two electrons, we break free of the H-atom(-like)  world. Using perturbation theory and Slater determinant wave functions, we made our first approximations of its energy. In addition, these students also had a seminar for their course Introductory lectures in preparation to the bachelor project (Kennismakingstraject m.b.t. stage en eindproject, in Dutch). During this lecture I gave a brief introduction and overview of the work I did in the past and the work we do in our research group QuATOMs, which although “quantum” is quite different of what the students experience during their courses on quantum chemistry.

In the materiomics program, the first master students continued their travels into the basics of force-fields during the lecture of the course Fundamentals of materials modelling. The exercise class of this week upped the ante by moving from ASE to LAMMPS for practical modeling of alkane chains, which was also the topic of their second lab session. In the course Properties of functional materials, we investigated the ab initio modelling of vibrations. During the exercise classes we investigated precalculated phonon spectra in the materials-project database, as well as calculated our own vibrational spectrum at the gamma-point of the first Brillouin zone. During the second master course Machine learning and artificial intelligence in modern materials science the central theme was GIGO (Garbage-In-Garbage-Out). How can we make sure our data is suitable and good enough for our models to return useful results. We therefore looked into data-preparation & cleaning, as well as  clustering methods.

At the end of this week, we have added another 22h of live lectures and ~1h of video lectures, putting our semester total at 74h of live lectures. Upwards and onward to week 6.

Permanent link to this article: https://dannyvanpoucke.be/materiomics-chronicles-week-5/

Jul 18 2022

Cover image Polymer International: A machine learning approach for the design of hyperbranched polymeric dispersing agents based on aliphatic polyesters for radiation curable inks

Authors: Danny E.P. Vanpoucke, Marie A.F. Delgove, Jules Stouten, Jurrie Noordijk, Nils De Vos, Kamiel Matthysen, Geert G.P. Deroover, Siamak Mehrkanoon, and Katrien V. Bernaerts
Journal: Polymer International 71(8), i-i (2022)
doi: 10.1002/pi.6434
IF(2015): 3.213
export: bibtex
pdf: <PolymerInt>

Abstract

The cover image is based on the Research Article A machine learning approach for the design of hyperbranched polymeric dispersing agents based on aliphatic polyesters for radiation-curable inks by Danny E.P. Vanpoucke et al., https://doi.org/10.1002/pi.6378.  (The related paper can be found here.)

 

Cover Polymer International: Machine learning on small data sets, application on UV curable inks.

Cover Polymer International: Machine learning on small data sets, application on UV curable inks.

 

Permanent link to this article: https://dannyvanpoucke.be/paper2022_mldannycover-en/

Jul 16 2022

A machine learning approach for the design of hyperbranched polymeric dispersing agents based on aliphatic polyesters for radiation curable inks

Authors: Danny E.P. Vanpoucke, Marie A.F. Delgove, Jules Stouten, Jurrie Noordijk, Nils De Vos, Kamiel Matthysen, Geert G.P. Deroover, Siamak Mehrkanoon, and Katrien V. Bernaerts
Journal: Polymer International 71(8), 966-975 (2022)
doi: 10.1002/pi.6378
IF(2021): 3.213
export: bibtex
pdf: <PI> (Open Access) (Cover Paper)

 

An ensemble based machine learning model for small datasets was used to predict the relationship between the dispersant structure and the pigment dispersion quality (particle size) for radiation curable formulations.
Graphical Abstract:An ensemble based machine learning model for small datasets was used to predict the relationship between the dispersant structure and the pigment dispersion quality (particle size) for radiation curable formulations.

Abstract

Polymeric dispersing agents were prepared from aliphatic polyesters consisting of δ-undecalactone (UDL) and β,δ-trimethyl-ε-caprolactones (TMCL) as biobased monomers, which are polymerized in bulk via organocatalysts. Graft copolymers were obtained by coupling of the polyesters to poly(ethylene imine) (PEI) in the bulk without using solvents. Different parameters that influence the performance of the dispersing agents in pigment based UV-curable matrices were investigated: chemistry of the polyester (UDL or TMCL), weight ratio of polyester/PEI, molecular weight of the polyesters and of PEI. The performance of the dispersing agents was modelled using machine learning in order to increase the efficiency of the dispersant design. The resulting models were presented as analytical models for the individual polyesters and the synthesis conditions for optimal performing dispersing agents were indicated as a preference for high molecular weight polyesters and a polyester dependent maximum weight ratio polyester/PEI.

Animation of TMCL model 6

Animation of TMCL model 6

 

Permanent link to this article: https://dannyvanpoucke.be/paper-ml_inks_um-en/

Feb 16 2022

Wordle-mania: an opportunity for python programming and artificial intelligence.

  • Filed under blog

  • February 16, 2022

Over that last few months the wordle game has become increasingly popular, with people sharing their daily feats on Twitter. Currently the game is hosted by the NY times which bought it the end of January 2022 from its creator.  The game is rather straightforward: you have 6 guesses to find a 5-letter English word. Every guess, the game tells you if a letter is (1) not in the word [grey], (2) in the word at a different position [yellow/orange], or (3) in the word at the exact same position [green].

Wordle 242 4/6
⬛⬛🟨⬛⬛
⬛🟨⬛⬛⬛
⬛⬛🟩🟨🟩
🟩🟩🟩🟩🟩		 An example of the result (as it looks when shared on Twitter). My first guess was the word “PIANO”, which means the A is in the word but at a different position. My second word, “QUERY”, adds the U to the list of letters that are present. With my third guess, “STUCK”, the position of the U and the K are fixed and we now also know the letter C is involved. At this point, I was also stuck, so I got some help of my wordle-assistant program, which taught me there could only be 1 word matching all the information we had: “CAULK“.

This seamlessly brings me to the central topic of this post: writing a program to help win this game as efficiently as possible. Not terribly original, but it’s a means to an end, as this simple project allows us to explore some more advanced topics in programming in python as well as artificial intelligence.

During this exploration I’ll  be including and updating a set of tutorials as well as this post. The python side of the project will focus on efficiency and easy of use and distribution, while the AI side will focus on smart ways predicting the best possible next guess. For the latter, an important caveat is that this means that the program should also work if you’re the last player living on earth, or if you decide to play wordle in a different language or a different number of letters. This means that creating a distribution of the  tweeted results of other players and comparing this with the complete set of brute-forced distributions to guess the wordle of the day in a single guess, would not satisfy my definition of AI. It is an interesting Big-data kaggle competition though.

Python Tutorials

  • Classes in Python. This tutorial provides a simple introduction in the concept of classes in python.
  • Child Classes. Continuing on the previous we deal now with child classes and the intricacies of function overriding and accessibility.
  • Python Library on Github.
  • Jupyter Notebook examples.
  • Slow Python: Profiling python in Jupyter. We look into profiling a Jupyter notebook script, to find the bottlenecks.
  • Slow Python: speeding up copying.

 

Artificial Intelligence

  • Distributions of words and letters.
  • Information theory of wordle?

 

The WordleAssistant Library and notebooks.

All tutorial code and jupyter notebooks can be found in this github repository.

Permanent link to this article: https://dannyvanpoucke.be/wordle-mania-an-opportunity-for-python-programming-and-artificial-intelligence/

May 06 2021

TEDx Talk: The Virtual Lab

  • Filed under blog

  • May 6, 2021

Happy to announce my TEDxUHasselt talk is officially part of the TEDx universe:  https://www.ted.com/talks/danny_vanpoucke_the_virtual_lab .

I enjoyed talking about the VirtualLab. Showed examples from atoms to galaxies, from computer-chips to drug-design and from to opinion-dynamics to epidemiology. I looked at the past and and glanced towards the future, where machine learning and artificial intelligence are the new kids on the block.

 

Permanent link to this article: https://dannyvanpoucke.be/tedx-talk-the-virtual-lab/