Amer Marwan El-Samman
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Amer Marwan El-Samman has a total of 8 co-authors. Here are the top co-authors by number of shared publications:
- Stijn De Baerdemacker
University of New Brunswick
6 shared publications - Brooke Morton
University of New Brunswick
2 shared publications - Stefano De Castro
Pitzer Center for Theoretical Chemistry and Department of Chemistry, University of California; Berkeley, United States of America
2 shared publications Ince Amina Husain
University of New Brunswick
1 shared publicationIncé Amina Husain
Department of Chemistry, University of New Brunswick, Fredericton, Canada
1 shared publication- Incé Amina Husain
University of New Brunswick, Department of Chemistry, 30 Dineen Dr, Fredericton, Canada
1 shared publication
Latest Publications
- 8 January 2026
- Published by Springer Nature in Theoretical Chemistry Accounts
- Vol. 145 (1) , 16
Occupancies vs embeddings: internal representations of Ab Initio and graph probabilistic models.
- 8 January 2026
- Published by Springer Nature in Theoretical Chemistry Accounts
- Vol. 145 (1) , 16
Occupancies vs embeddings: internal representations of Ab Initio and graph probabilistic models.
- Open Access
- 15 September 2025
- Published by Springer Nature in Research Square
Occupancies vs Embeddings: Internal representations of Ab Initio and Graph Probabilistic Models*
Chemical modeling has traditionally been separated into two fairly non-overlapping frameworks: the exact framework of \textit{ab initio} models and the data-driven framework of statistical models. However, recent advances in machine learning approaches have intercepted the exact \textit{ab initio} frameworks by using an \textit{end-to-end} probabilistic model of the Schr\"odinger equation in the Born-Oppenheimer approximation---from raw nuclear configurations to quantum observables such as the energy. While probabilistic modelling offers promising advancements to speed up computational predictions, the internal representation and its connection to the underlying quantum data---the wave function or electron density---which they implicitly rely on for training, remains relatively poorly understood. This study seeks to make a first comparison of the internal representation from a probabilistic end-to-end models with those from pure \textit{ab initio} frameworks such as density functional theory. We do this by comparing the internal atomwise representation of graph models, also called embeddings, to their most natural \textit{ab initio} counterpart, electron occupancy values. Our findings show that the embedding representations can be employed to transfer learn atomic occupancy values whereas the reverse mapping appears to be less accurate. We discuss the assumption that the probabilistic model could infer the existence of an underlying electron density underpinning the computational \textit{ab-initio} data.
- 15 September 2025
- Published by Springer Nature in Research Square
Open AccessOccupancies vs Embeddings: Internal representations of Ab Initio and Graph Probabilistic Models*
Chemical modeling has traditionally been separated into two fairly non-overlapping frameworks: the exact framework of \textit{ab initio} models and the data-driven framework of statistical models. However, recent advances in machine learning approaches have intercepted the exact \textit{ab initio} frameworks by using an \textit{end-to-end} probabilistic model of the Schr\"odinger equation in the Born-Oppenheimer approximation---from raw nuclear configurations to quantum observables such as the energy. While probabilistic modelling offers promising advancements to speed up computational predictions, the internal representation and its connection to the underlying quantum data---the wave function or electron density---which they implicitly rely on for training, remains relatively poorly understood. This study seeks to make a first comparison of the internal representation from a probabilistic end-to-end models with those from pure \textit{ab initio} frameworks such as density functional theory. We do this by comparing the internal atomwise representation of graph models, also called embeddings, to their most natural \textit{ab initio} counterpart, electron occupancy values. Our findings show that the embedding representations can be employed to transfer learn atomic occupancy values whereas the reverse mapping appears to be less accurate. We discuss the assumption that the probabilistic model could infer the existence of an underlying electron density underpinning the computational \textit{ab-initio} data.
- Open Access
- 23 April 2025
- Published by Royal Society of Chemistry (RSC) in Chemical Science
- Vol. 16 (24) , 10895-10908
“Amide − amine + alcohol = carboxylic acid.” chemical reactions as linear algebraic analogies in graph neural networks
Graph models in chemistry show NLP-like behavior: atom embeddings capture consistent reaction patterns, like word analogies. These interpretable vectors reveal a structured, syntax-like model of chemical transformations on a linear algebraic level.
- 23 April 2025
- Published by Royal Society of Chemistry (RSC) in Chemical Science
- Vol. 16 (24) , 10895-10908
Open Access“Amide − amine + alcohol = carboxylic acid.” chemical reactions as linear algebraic analogies in graph neural networks
Graph models in chemistry show NLP-like behavior: atom embeddings capture consistent reaction patterns, like word analogies. These interpretable vectors reveal a structured, syntax-like model of chemical transformations on a linear algebraic level.
- Open Access
- 19 July 2024
- Published by American Chemical Society (ACS) in ChemRxiv
"Amide - amine + alcohol = carboxylic acid." Chemical reactions as linear algebraic analogies in graph neural networks.
In deep learning methods, especially in the context of chemistry, there is an increasing urgency to uncover the hidden learning mechanisms often dubbed as ``black box." In this work, we show that graph models built on computational chemical data behave similar to natural language processing (NLP) models built on text data. Crucially, we show that atom-embeddings, a.k.a atom-parsed graph neural activation patterns, exhibit arithmetic properties that represent valid reaction formulas. This is very similar to how word-embeddings can be combined to make word analogies, thus preserving the semantic meaning behind the words, as in the famous example ``King" $-$ ``Man" $+$ ``Woman" = ``Queen." For instance, we show how the reaction from an alcohol to a carbonyl is represented by a constant vector in the embedding space, implicitly representing ``-$\text{H}_{2}$," independent from the particular carbonyl reactant and alcohol product. This reveals a highly-structured vector space, wherein the directions in the embedding space are synonymous with chemical changes (ex. the oxidation direction), and distinct chemical changes are orthogonal. In contrast to natural language processing, we can explain the observed chemical analogies using algebraic manipulations on the local chemical composition that surrounds each atom-embedding. Furthermore, the observations find applications in transfer learning, for instance in the formal structure and prediction of atomistic properties, such as $^{1}$H-NMR and $^{13}$C-NMR. This work is in line with the recent push for interpretable explanations to graph neural network modeling of chemistry and uncovers a latent model of chemistry that is highly structured, consistent, and analogous to chemical syntax.- 19 July 2024
- Published by American Chemical Society (ACS) in ChemRxiv
Open Access"Amide - amine + alcohol = carboxylic acid." Chemical reactions as linear algebraic analogies in graph neural networks.
In deep learning methods, especially in the context of chemistry, there is an increasing urgency to uncover the hidden learning mechanisms often dubbed as ``black box." In this work, we show that graph models built on computational chemical data behave similar to natural language processing (NLP) models built on text data. Crucially, we show that atom-embeddings, a.k.a atom-parsed graph neural activation patterns, exhibit arithmetic properties that represent valid reaction formulas. This is very similar to how word-embeddings can be combined to make word analogies, thus preserving the semantic meaning behind the words, as in the famous example ``King" $-$ ``Man" $+$ ``Woman" = ``Queen." For instance, we show how the reaction from an alcohol to a carbonyl is represented by a constant vector in the embedding space, implicitly representing ``-$\text{H}_{2}$," independent from the particular carbonyl reactant and alcohol product. This reveals a highly-structured vector space, wherein the directions in the embedding space are synonymous with chemical changes (ex. the oxidation direction), and distinct chemical changes are orthogonal. In contrast to natural language processing, we can explain the observed chemical analogies using algebraic manipulations on the local chemical composition that surrounds each atom-embedding. Furthermore, the observations find applications in transfer learning, for instance in the formal structure and prediction of atomistic properties, such as $^{1}$H-NMR and $^{13}$C-NMR. This work is in line with the recent push for interpretable explanations to graph neural network modeling of chemistry and uncovers a latent model of chemistry that is highly structured, consistent, and analogous to chemical syntax. - Open Access
- 1 April 2024
- Published by Canadian Science Publishing in Canadian Journal of Chemistry
- Vol. 102 (4) , 275-288
Transfer learning graph representations of molecules for pKa, 13C-NMR, and solubility
We explore transfer learning models from a pre-trained graph convolutional neural network representation of molecules, obtained from SchNet, to predict 13C-NMR, pKa, and log S solubility. SchNet learns a graph representation of a molecule by associating each atom with an “embedding vector” and interacts the atom-embeddings with each other by leveraging graph convolutional filters on their interatomic distances. We pre-trained SchNet on molecular energy and demonstrate that the pre-trained atomistic embeddings can then be used as a transferable representation for a wide array of properties. On the one hand, for atomic properties such as micro-pK1 and 13C-NMR, we investigate two models, one linear and one neural net, that input pre-trained atom-embeddings of a particular atom (e.g. carbon) and predict a local property (e.g., 13C-NMR). On the other hand, for molecular properties such as solubility, a size-extensive graph model is built using the embeddings of all atoms in the molecule as input. For all cases, qualitatively correct predictions are made with relatively little training data (<1000 training points), showcasing the ease with which pre-trained embeddings pick up on important chemical patterns. The proposed models successfully capture well-understood trends of pK1 and solubility. This study advances our understanding of current neural net graph representations and their capacity for transfer learning applications in chemistry.
- 1 April 2024
- Published by Canadian Science Publishing in Canadian Journal of Chemistry
- Vol. 102 (4) , 275-288
Open AccessTransfer learning graph representations of molecules for pKa, 13C-NMR, and solubility
We explore transfer learning models from a pre-trained graph convolutional neural network representation of molecules, obtained from SchNet, to predict 13C-NMR, pKa, and log S solubility. SchNet learns a graph representation of a molecule by associating each atom with an “embedding vector” and interacts the atom-embeddings with each other by leveraging graph convolutional filters on their interatomic distances. We pre-trained SchNet on molecular energy and demonstrate that the pre-trained atomistic embeddings can then be used as a transferable representation for a wide array of properties. On the one hand, for atomic properties such as micro-pK1 and 13C-NMR, we investigate two models, one linear and one neural net, that input pre-trained atom-embeddings of a particular atom (e.g. carbon) and predict a local property (e.g., 13C-NMR). On the other hand, for molecular properties such as solubility, a size-extensive graph model is built using the embeddings of all atoms in the molecule as input. For all cases, qualitatively correct predictions are made with relatively little training data (<1000 training points), showcasing the ease with which pre-trained embeddings pick up on important chemical patterns. The proposed models successfully capture well-understood trends of pK1 and solubility. This study advances our understanding of current neural net graph representations and their capacity for transfer learning applications in chemistry.
- Open Access
- 15 February 2024
- Published by Royal Society of Chemistry (RSC) in Digital Discovery
- Vol. 3 (3) , 544-557
Global geometry of chemical graph neural network representations in terms of chemical moieties
The embedding vectors from a Graph Neural Network trained on quantum chemical data allow for a global geometric space with a Euclidean distance metric. Moieties that are close in chemical sense, are also close in Euclidean sense.
- 15 February 2024
- Published by Royal Society of Chemistry (RSC) in Digital Discovery
- Vol. 3 (3) , 544-557
Open AccessGlobal geometry of chemical graph neural network representations in terms of chemical moieties
The embedding vectors from a Graph Neural Network trained on quantum chemical data allow for a global geometric space with a Euclidean distance metric. Moieties that are close in chemical sense, are also close in Euclidean sense.
- 5 October 2020
- Published by AIP Publishing in The Journal of Chemical Physics
- Vol. 153 (13) , 134109
Asymptotic behavior of the average local ionization energy in finite basis sets
The average local ionization energy (ALIE) has important applications in several areas of electronic structure theory. Theoretically, the ALIE should asymptotically approach the first vertical ionization energy (IE) of the system, as implied by the rate of exponential decay of the electron density; for one-determinantal wavefunctions, this IE is the negative of the highest-occupied orbital energy. In practice, finite-basis-set representations of the ALIE exhibit seemingly irregular and sometimes dramatic deviations from the expected asymptotic behavior. We analyze the long-range behavior of the ALIE in finite basis sets and explain the puzzling observations. The findings have implications for practical calculations of the ALIE, the construction of Kohn–Sham potentials from wavefunctions and electron densities, and basis-set development.
- 5 October 2020
- Published by AIP Publishing in The Journal of Chemical Physics
- Vol. 153 (13) , 134109
Asymptotic behavior of the average local ionization energy in finite basis sets
The average local ionization energy (ALIE) has important applications in several areas of electronic structure theory. Theoretically, the ALIE should asymptotically approach the first vertical ionization energy (IE) of the system, as implied by the rate of exponential decay of the electron density; for one-determinantal wavefunctions, this IE is the negative of the highest-occupied orbital energy. In practice, finite-basis-set representations of the ALIE exhibit seemingly irregular and sometimes dramatic deviations from the expected asymptotic behavior. We analyze the long-range behavior of the ALIE in finite basis sets and explain the puzzling observations. The findings have implications for practical calculations of the ALIE, the construction of Kohn–Sham potentials from wavefunctions and electron densities, and basis-set development.