Skip to main content

Questions tagged [machine-learning]

For questions related to the use of machine learning techniques (neural networks, support vector machines, Gaussian process regression, etc) to study material or molecular properties.

Filter by
Sorted by
Tagged with
4 votes
0 answers
33 views

Doubt regarding Ehull (energy above convex hull) and formation energy

I am creating a ML model for predicting thermodynamic-stability of 2-d materials, for predicting stability formation energy and Ehull (Energy above convex hull) is used and this is how it is also ...
ROXY_01092002's user avatar
5 votes
0 answers
53 views

Butina cluster over a 300 million SMILES

I have a dataset with approximately 300 million SMILES strings, for which I want to apply Butina clustering and extract the Tanimoto similarity matrix to visualize the PCA of this matrix. Since I have ...
user avatar
5 votes
0 answers
54 views

Protein structure prediction: levels of coarse-graining in deep learning approaches

I am researching the areas of protein structure prediction, protein + ligand pose prediction, and co-folding algos with deep learning. I am trying to map the current state of the art architectures ...
operator's user avatar
4 votes
1 answer
122 views

Energies from single points vs. AIMD for training Machine Learning Force Fields

I am currently training a force field using machine learning techniques. One way to test the performance of the ML model is to use the force field for a production run (= molecular dynamics) and ...
Lukas's user avatar
  • 303
6 votes
0 answers
123 views

Normalization of the Smooth Overlap of Atomic Positions (SOAP) Kernel

I am using the DScribe implementation (https://singroup.github.io/dscribe/latest/tutorials/descriptors/soap.html) of the Smooth overlap of atomic positions (SOAP) ...
C_Swann22's user avatar
  • 543
6 votes
0 answers
44 views

How useful were the materials found by GNoME?

In November 2023 DeepMind used an AI system Graph Networks for Materials Exploration or GNoME (I have no idea where the ‘o’ comes from) to find 421,000 new materials. Way more than had been found ...
H. de Gracht's user avatar
5 votes
1 answer
119 views

Neural Network to predict atomic forces using SOAP descriptors

I am working on neural network in tensorflow to train a model to predict atomic forces from SOAP descriptors (regression problem). I have 64 020 data points (chemical structures), each structure has ...
jessss's user avatar
  • 73
4 votes
1 answer
72 views

Computing energy and forces directly from machine learned wavefuntion/density

I have been reading up on methods where electron density was estimated directly from some machine learning method, followed by evaluation of energy and forces[1,2]. My understanding is that this is ...
ipcamit's user avatar
  • 615
5 votes
0 answers
59 views

Strongly Correlated Materials: Theoretical Frameworks, Computational Techniques, and Machine Learning Breakthroughs in Matter Modeling

This is more of an opened discussion, recommendations, and experts views on the Strongly correlated materials. Exploring the behavior of strongly correlated materials poses significant challenges in ...
Jaafar Mehrez's user avatar
3 votes
0 answers
78 views

Using GPyTorch for GPR together with the SOAP-Kernel

My goal is to use a Gaussian Process Regression (GPR) model with the GPyTorch library with the SOAP (smooth overlap of atomic positions) descriptor that can be obtained with the DScribe library. I ...
C_Swann22's user avatar
  • 543
9 votes
0 answers
124 views

How to go from zero to hero in Machine Learning for materials modelling

I am interested in exploring the use of machine learning tools to accelerate molecular dynamics simulation probably by training machine learning potentials. While I have found some papers that apply ...
manuelpb's user avatar
  • 453
2 votes
0 answers
213 views

How to fit Smooth Overlap of Atomic Positions (SOAP) into Kernel Ridge Regression (KRR)?

I have a dataset of 64 chemical structures and their calculated energies, which I want to predict using machine learning. I need to use SOAP as a descriptor so that each structure would be an ...
jessss's user avatar
  • 73
3 votes
1 answer
197 views

Saving forces from ASE Molecular Dynamics calculations

I have run some molecular dynamics calculations with a pretrained machine-learning potential (M3GNet) which uses the ASE MolecularDynamics calculator. My goal is to compare the results (energies and ...
Jingyang Wang's user avatar
6 votes
1 answer
135 views

Use of genetic algorithm based methods for geometry optimization

What are the benefits or pitfalls in using a GA based optimizer for geometry optimization instead of a more traditional Hessian based algorithm? It can be argued that performing single point energy ...
Hemanth Haridas's user avatar
5 votes
1 answer
92 views

Training a ML forcefield on a simple bulk material

Lately, I've become quite intrigued by the recent advances in machine learning forcefields within the field of computational chemistry. I've developed a curiosity about training a simple forcefield on ...
Okano's user avatar
  • 1,427
4 votes
1 answer
267 views

Smooth overlap of atomic positions (SOAP) output for Gaussian Process regression in sklearn

I want to use the smooth overlap of atomic positions (SOAP) as a descriptor to represent the atomic environment of a specific atom to predict chemical shifts. I have generated averaged SOAPs for ...
C_Swann22's user avatar
  • 543
4 votes
0 answers
103 views

Predict if a group will be acceptor/donor of H bonds [closed]

Context: I am a student of cheminformatics, and I am trying to solve a problem in predicting whether atom can be an acceptor/donor of hydrogen bonds, using Python programming language and RDKit ...
Gianmarco Luchetti 's user avatar
9 votes
1 answer
1k views

If total energies differ across different software, how do I decide which software to use?

I posted a question previously about why I got different total energy values when I ran a DFT calculation with quantum espresso compared to VASP (link to the question). From what I understood from the ...
Yonatan Kurniawan's user avatar
11 votes
3 answers
602 views

How a beginner should start his studies in ML for chemistry application?

Background I am a chemistry undergrad that wants to learn Machine Learning (ML) for application in my field(Chemistry). However, I have never studied programming or anything related to code in my life....
Everson Gomes's user avatar
8 votes
1 answer
346 views

How can I build the wheels necessary for a quick installation of PySCFad on a "compute node"?

The following is the procedure I have been following: Install PySCFad ...
Nike Dattani - No Free Time's user avatar
4 votes
0 answers
51 views

Specific descriptors for dimers [closed]

I am studying the dependence of the rate of a specific process in a dimer from the relative positioning of molecules in the dimer. Currently, I am using 2000 descriptors from the padelpy for model ...
user avatar
6 votes
2 answers
199 views

Course on machine learning on application in material science

Are there any online courses available that is on the application of ML in material science? Any recommendations?
Paulie Bao's user avatar
  • 3,973
6 votes
1 answer
130 views

Are there packages which calculate normal modes for ML potentials?

While I'm primarily interested in molecular normal modes, the question could also apply to calculating phonon spectra / vibrations in solids. There are many machine learning approaches to calculate ...
Geoff Hutchison's user avatar
8 votes
1 answer
302 views

Developing a machine-learned interatomic potential for molecular dynamics

There are few questions in this SE about the advantages/disadvantages of machine-learned interatomic potentials in molecular dynamics, but there is not much discussion on how to make them. I thought ...
user35952's user avatar
  • 343
6 votes
0 answers
135 views

Books on machine learning and condensed matter? [closed]

What are the best books on the use of machine learning in condensed matter, focusing on density functional theory and many-body physics?
Bekaso's user avatar
  • 245
3 votes
1 answer
2k views

Why is the Alphafold PAE (predicted aligned error) not symmetric?

Cross-posted on AI Stack Exchange. We are running alphafold2 multimer on Google Colab to predict the association between two proteins. It generally works fine, but we get an asymmetric PAE plot for a ...
NKGon's user avatar
  • 31
3 votes
1 answer
196 views

Neural network as a customised XC functional in PySCF?

I've been trying to implement neural network as a customised exchange-correlation functional in PySCF but with no success. Below is the example code for customised XC functional. Anyone has any idea ...
Ken.WS's user avatar
  • 31
5 votes
0 answers
82 views

Why are graph-structured molecules invariant to the ordering of atoms? [closed]

In order to train GANs (Generative Adversarial Networks) on QM9 dataset, the dataset is converted to graph format as graph-structured molecules are invariant to the ordering of atoms. I want to ...
Vinay Sharma's user avatar
3 votes
0 answers
39 views

How to select the best features for a drug discovery model? [closed]

I would like to build a regression model to predict a property of molecules not yet synthesized. I have calculated 5666 alvadesc descriptors for each SMILES in my experimental dataset, but comparing ...
vorsamaqoy's user avatar
12 votes
1 answer
201 views

Best resources for someone going from experimental to computational chemistry

In a few months, I will complete my PhD and I'm considering leaving experimental chemistry and looking for a postdoc on computational chemistry. During my PhD I used some computational tools like ...
manuelpb's user avatar
  • 453
12 votes
3 answers
3k views

Artificial intelligence is a hot topic, but should I pursue it if I'm interested in Matter Modeling?

Since my previous question seems to have a large number of views, I would like to ask my central question. I am a student now and should decide my major to research soon. AI/ML is a very hot research ...
neco's user avatar
  • 1,789
15 votes
2 answers
3k views

What does machine learning learn about DFT?

I'm a student and now studying quantum chemistry but also interested in machine learning (ML) and materials informatics (MI). In order to understand an ML method for MI, I tried to use the smooth ...
neco's user avatar
  • 1,789
5 votes
1 answer
204 views

What is the best way to measure similarity between molecules of the same formula?

I have enumerated a large set of carbocations all of the formula C10H17+, all of course with differing structures. I know there are many different approaches of computing similarity between molecules, ...
BanAckerman's user avatar
4 votes
1 answer
140 views

Understanding use of Machine Learning in Multiscale Enhanced Sampling

Multiscale Enhanced Sampling Using Machine Learning Multiscale enhanced sampling (MSES) allows for enhanced sampling of all-atom protein structures by coupling with the accelerated dynamics of the ...
user366312's user avatar
  • 2,556
5 votes
1 answer
195 views

In which part of the Hartree–Fock algorithm (or DFT one) could a neural network be most effectively used?

I understand that the task of implementing machine learning in DFT and Hartree–Fock (HF) algorithm has already been solved, perhaps to some extent, but it is interesting to think about how to ...
SFriendly's user avatar
  • 1,007
7 votes
2 answers
691 views

Is it possible to build a force field that suits all elements based on VASP's machine learning result?

I have tried to use VASP's machine learning force field calculation during running molecular dynamics simulation with a supercell including some elements of Ti, O, Cu, etc. It does increase the speed ...
Jack's user avatar
  • 2,067
10 votes
2 answers
672 views

Getting interpretable chemical information from hashed molecular fingerprints

I have been using molecular fingerprints like ECFP (extended connectivity fingerprint), APFP (atom-pair fingerprint) etc. in my research to predict spectral properties of organic molecules with ...
S R Maiti's user avatar
  • 7,131
5 votes
1 answer
170 views

PyTorch WL Kernel

I’m learning to use PyTorch Geometric, I tried to replicate the WL kernel written by the PyG developers on GitHub (https://github.com/pyg-team/pytorch_geometric/blob/master/examples/wl_kernel.py), but ...
Gianmarco Luchetti 's user avatar
18 votes
1 answer
374 views

What are the advantages of (semi)-empirical force fields over Machine Learning Potentials?

I am currently working with ReaxFF, an empirical reactive force field that can describe chemical bond forming and breaking. The main advantage over ab initio methods are of course the greatly ...
lcdumort's user avatar
  • 283
6 votes
1 answer
286 views

How to improve my cross-validation R2_score?

I built a prediction model with fingerprints from 300 molecules and got an R2 of 0.9 However when I go to perform a cross-validation I get a very low R2. How can I improve this result? I'm using ...
vorsamaqoy's user avatar
14 votes
1 answer
1k views

How to start a Machine Learning project for chemical properties prediction?

I know that is a very general question but I would like to start a ML project in Python to predict some chemical properties with a large set of experimental data. The compounds I would like to study ...
vorsamaqoy's user avatar
10 votes
1 answer
599 views

Node features matrix with Networkx

I built a function to generate graphs from smiles strings using networkx, inserting various features on the nodes. This is the code: ...
Gianmarco Luchetti 's user avatar
9 votes
0 answers
76 views

Graph Classification via Random Forest [closed]

I’m a medicinal chemistry undergraduate student who is preparing his dissertation. My idea would be to create a classifier that can distinguish anticancer drugs as active or inactive and distinguish ...
Gianmarco Luchetti 's user avatar
11 votes
2 answers
6k views

How to input 3D coordinates from xyz file and connectivity from SMILES in rdkit?

I am working on a QSAR project where the 3D structural descriptors are an input to a machine learning model. I am generating the descriptors using the python Mordred API (which uses rdkit). ...
S R Maiti's user avatar
  • 7,131
13 votes
1 answer
167 views

How are structure descriptors used in regression or machine learning?

I am currently working on prediction of UV-vis spectra from structure of molecules. I have read multiple papers where structure descriptors were used as inputs for machine learning to find various ...
S R Maiti's user avatar
  • 7,131
6 votes
1 answer
213 views

FireWorks for Workflow management or TensorFlow

In computational material science, we need workflows for optimization surrogate models which requires high computation resources. I am actually concerned with why material science community is using ...
gfdsal's user avatar
  • 373
9 votes
0 answers
216 views

Is "Valence Electron Density" and "Electron Density" data of a molecule the same thing? [closed]

I'm wondering specifically in the context of calculating physical properties from valence-electron-density data using DFT, MD, and or ML (machine learning).
Pranoy Ray's user avatar
  • 1,637
9 votes
1 answer
184 views

Is there a database where one can find the Electron Density data of materials?

The database I am looking for may be experimental, computational or user-generated data. If I end up using the data, I will be providing the necessary citations and credits. Thank You.
Pranoy Ray's user avatar
  • 1,637
12 votes
1 answer
424 views

What are the databases of semiconductor properties?

Is there any database listing the experimentally-determined properties of semiconductor materials - things like Band Gap and Electron Mobility. These are easily found for common materials like ...
user0's user avatar
  • 281
7 votes
0 answers
99 views

Should one normalize the "features" in binary fingerprints? [closed]

In regression models to predict chemical compounds' activities, fingerprints are often used as features. Should one normalize a fingerprint feature to be in the 0-1 range?
BND's user avatar
  • 1,341