-
TimeGPT-1
Authors:
Azul Garza,
Cristian Challu,
Max Mergenthaler-Canseco
Abstract:
In this paper, we introduce TimeGPT, the first foundation model for time series, capable of generating accurate predictions for diverse datasets not seen during training. We evaluate our pre-trained model against established statistical, machine learning, and deep learning methods, demonstrating that TimeGPT zero-shot inference excels in performance, efficiency, and simplicity. Our study provides…
▽ More
In this paper, we introduce TimeGPT, the first foundation model for time series, capable of generating accurate predictions for diverse datasets not seen during training. We evaluate our pre-trained model against established statistical, machine learning, and deep learning methods, demonstrating that TimeGPT zero-shot inference excels in performance, efficiency, and simplicity. Our study provides compelling evidence that insights from other domains of artificial intelligence can be effectively applied to time series analysis. We conclude that large-scale time series models offer an exciting opportunity to democratize access to precise predictions and reduce uncertainty by leveraging the capabilities of contemporary advancements in deep learning.
△ Less
Submitted 27 May, 2024; v1 submitted 5 October, 2023;
originally announced October 2023.
-
Adaptive Functional Principal Component Analysis
Authors:
Angel Garcia de la Garza,
Britton Sauerbrei,
Adam Hantman,
Jeff Goldsmith
Abstract:
We introduce Adaptive Functional Principal Component Analysis, a novel method to capture directions of variation in functional data that exhibit sharp changes in smoothness. We first propose a new adaptive scatterplot smoothing technique that is fast and scalable, and then integrate this technique into a probabilistic FPCA framework to adaptively smooth functional principal components. Our simulat…
▽ More
We introduce Adaptive Functional Principal Component Analysis, a novel method to capture directions of variation in functional data that exhibit sharp changes in smoothness. We first propose a new adaptive scatterplot smoothing technique that is fast and scalable, and then integrate this technique into a probabilistic FPCA framework to adaptively smooth functional principal components. Our simulation results show that our approach is better able to model functional data with sharp changes in smoothness compared to standard approaches. We are motivated by the need to identify coordinated patterns of brain activity across multiple neurons during reaching movements prompted by an auditory cue, which enables understanding of the dynamics in the brain during dexterous movement. Our proposed method captures the underlying biological mechanisms that arise in data obtained from a mouse experiment focused on voluntary reaching movements, offering more interpretable activation patterns that reflect sharp changes in neural activity following the cue. We develop accompanying publicly available software for our proposed methodology, along with implementations to reproduce our results.
△ Less
Submitted 2 October, 2023;
originally announced October 2023.
-
Multi-Institutional Audit of FLASH and Conventional Dosimetry with a 3D-Printed Anatomically Realistic Mouse Phantom
Authors:
M Ramish Ashraf,
Stavros Melemenidis,
Kevin Liu,
Veljko Grilj,
Jeannette Jansen,
Brett Velasquez,
Luke Connell,
Joseph B Schulz,
Claude Bailat,
Aaron Libed,
Rakesh Manjappa,
Suparna Dutt,
Luis Soto,
Brianna Lau,
Aaron Garza,
William Larsen,
Lawrie Skinner,
Amy S Yu,
Murat Surucu,
Edward E Graves,
Peter G Maxim,
Stephen F. Kry,
Marie-Catherine Vozenin,
Emil Schüler,
Billy W Loo Jr
Abstract:
We conducted a multi-institutional audit of dosimetric variability between FLASH and conventional dose rate (CONV) electron irradiations by using an anatomically realistic 3D-printed mouse phantom. A CT scan of a live mouse was used to create a 3D model of bony anatomy, lungs, and soft tissue. A dual-nozzle 3D printer was used to print the mouse phantom using acrylonitrile butadiene styrene (…
▽ More
We conducted a multi-institutional audit of dosimetric variability between FLASH and conventional dose rate (CONV) electron irradiations by using an anatomically realistic 3D-printed mouse phantom. A CT scan of a live mouse was used to create a 3D model of bony anatomy, lungs, and soft tissue. A dual-nozzle 3D printer was used to print the mouse phantom using acrylonitrile butadiene styrene ($~1.02 g/cm^3$) and polylactic acid ($~1.24 g/cm^3$) simultaneously to simulate soft tissue and bone densities, respectively. The lungs were printed separately using lightweight polylactic acid ($~0.64 g/cm^3$). Hounsfield units (HU) and densities were compared with the reference CT scan of the live mouse. Print-to-print reproducibility of the phantom was assessed. Three institutions were each provided a phantom, and each institution performed two replicates of irradiations at selected mouse anatomic regions. The average dose difference between FLASH and CONV dose distributions and deviation from the prescribed dose were measured with radiochromic film. Compared to the reference CT scan, CT scans of the phantom demonstrated mass density differences of $0.10 g/cm^3$ for bone, $0.12 g/cm^3$ for lung, and $0.03 g/cm^3$ for soft tissue regions. Between phantoms, the difference in HU for soft tissue and bone was <10 HU from print to print. Lung exhibited the most variation (54 HU) but minimally affected dose distribution (<0.5% dose differences between phantoms). The mean difference between FLASH and CONV from the first replicate to the second decreased from 4.3% to 1.2%, and the mean difference from the prescribed dose decreased from 3.6% to 2.5% for CONV and 6.4% to 2.7% for FLASH. The framework presented here is promising for credentialing of multi-institutional studies of FLASH preclinical research to maximize the reproducibility of biological findings.
△ Less
Submitted 28 September, 2023;
originally announced September 2023.
-
Haptic Feedback Relocation from the Fingertips to the Wrist for Two-Finger Manipulation in Virtual Reality
Authors:
Jasmin E. Palmer,
Mine Sarac,
Aaron A. Garza,
Allison M. Okamura
Abstract:
Relocation of haptic feedback from the fingertips to the wrist has been considered as a way to enable haptic interaction with mixed reality virtual environments while leaving the fingers free for other tasks. We present a pair of wrist-worn tactile haptic devices and a virtual environment to study how various mappings between fingers and tactors affect task performance. The haptic feedback rendere…
▽ More
Relocation of haptic feedback from the fingertips to the wrist has been considered as a way to enable haptic interaction with mixed reality virtual environments while leaving the fingers free for other tasks. We present a pair of wrist-worn tactile haptic devices and a virtual environment to study how various mappings between fingers and tactors affect task performance. The haptic feedback rendered to the wrist reflects the interaction forces occurring between a virtual object and virtual avatars controlled by the index finger and thumb. We performed a user study comparing four different finger-to-tactor haptic feedback mappings and one no-feedback condition as a control. We evaluated users' ability to perform a simple pick-and-place task via the metrics of task completion time, path length of the fingers and virtual cube, and magnitudes of normal and shear forces at the fingertips. We found that multiple mappings were effective, and there was a greater impact when visual cues were limited. We discuss the limitations of our approach and describe next steps toward multi-degree-of-freedom haptic rendering for wrist-worn devices to improve task performance in virtual environments.
△ Less
Submitted 14 November, 2022; v1 submitted 15 September, 2022;
originally announced September 2022.
-
Tangelo: An Open-source Python Package for End-to-end Chemistry Workflows on Quantum Computers
Authors:
Valentin Senicourt,
James Brown,
Alexandre Fleury,
Ryan Day,
Erika Lloyd,
Marc P. Coons,
Krzysztof Bieniasz,
Lee Huntington,
Alejandro J. Garza,
Shunji Matsuura,
Rudi Plesch,
Takeshi Yamazaki,
Arman Zaribafiyan
Abstract:
Tangelo [link: https://github.com/goodchemistryco/Tangelo] is an open-source Python software package for the development of end-to-end chemistry workflows on quantum computers, released under Apache 2.0 license. It aims to support the design of successful experiments on quantum hardware, and to facilitate advances in quantum algorithm development. The software enables quick exploration of differen…
▽ More
Tangelo [link: https://github.com/goodchemistryco/Tangelo] is an open-source Python software package for the development of end-to-end chemistry workflows on quantum computers, released under Apache 2.0 license. It aims to support the design of successful experiments on quantum hardware, and to facilitate advances in quantum algorithm development. The software enables quick exploration of different approaches by assembling reusable building blocks and algorithms, with the flexibility to let users introduce their own. Tangelo is backend-agnostic and enables switching between various backends (Braket, Qiskit, Qulacs, Azure Quantum, QDK, Cirq...) with minimal changes in the code. The package can be used to explore quantum computing applications such as open-shell systems, excited states, or more industrially-relevant systems by leveraging problem decomposition at scale. This paper outlines the design choices, philosophy, and main features of Tangelo.
△ Less
Submitted 24 June, 2022;
originally announced June 2022.
-
Optimizing Electronic Structure Simulations on a Trapped-ion Quantum Computer using Problem Decomposition
Authors:
Yukio Kawashima,
Erika Lloyd,
Marc P. Coons,
Yunseong Nam,
Shunji Matsuura,
Alejandro J. Garza,
Sonika Johri,
Lee Huntington,
Valentin Senicourt,
Andrii O. Maksymov,
Jason H. V. Nguyen,
Jungsang Kim,
Nima Alidoust,
Arman Zaribafiyan,
Takeshi Yamazaki
Abstract:
Quantum computers have the potential to advance material design and drug discovery by performing costly electronic structure calculations. A critical aspect of this application requires optimizing the limited resources of the quantum hardware. Here, we experimentally demonstrate an end-to-end pipeline that focuses on minimizing quantum resources while maintaining accuracy. Using density matrix emb…
▽ More
Quantum computers have the potential to advance material design and drug discovery by performing costly electronic structure calculations. A critical aspect of this application requires optimizing the limited resources of the quantum hardware. Here, we experimentally demonstrate an end-to-end pipeline that focuses on minimizing quantum resources while maintaining accuracy. Using density matrix embedding theory as a problem decomposition technique, and an ion-trap quantum computer, we simulate a ring of 10 hydrogen atoms without freezing any electrons. The originally 20-qubit system is decomposed into 10 two-qubit problems, making it amenable to currently available hardware. Combining this decomposition with a qubit coupled cluster circuit ansatz, circuit optimization, and density matrix purification, we accurately reproduce the potential energy curve in agreement with the full configuration interaction energy in the minimal basis set. Our experimental results are an early demonstration of the potential for problem decomposition to accurately simulate large molecules on quantum hardware.
△ Less
Submitted 28 September, 2021; v1 submitted 13 February, 2021;
originally announced February 2021.
-
Solvation Entropy Made Simple
Authors:
Alejandro J. Garza
Abstract:
The entropies of molecules in solution are routinely calculated using gas phase formulae. It is assumed that, because implicit solvation models are fitted to reproduce free energies, this is sufficient for modeling reactions in solution. However, this procedure exaggerates entropic effects in processes that change molecularity. Here, computationally efficient (i.e., having similar cost as gas phas…
▽ More
The entropies of molecules in solution are routinely calculated using gas phase formulae. It is assumed that, because implicit solvation models are fitted to reproduce free energies, this is sufficient for modeling reactions in solution. However, this procedure exaggerates entropic effects in processes that change molecularity. Here, computationally efficient (i.e., having similar cost as gas phase entropy calculations) approximations for determining solvation entropy are proposed to address this issue.The $S_ω$, $S_ε$, and $S_{εα}$ models are nonempirical and rely only on physical arguments and elementary properties of the medium (e.g., density and relative permittivity). For all three methods, average errors as compared to experiment are within chemical accuracy for 110 solvation entropies, 11 activation entropies in solution, and 32 vaporization enthalpies. The models also make predictions regarding microscopic and bulk properties of liquids which prove to be accurate. These results imply that $ΔH_\text{sol}$ and $ΔS_\text{sol}$ can be described separately and with less reliance on parametrization by a combination of the methods presented here with existing, reparametrized implicit solvation models.
△ Less
Submitted 30 January, 2019;
originally announced January 2019.
-
Reaction Mechanism of the Selective Reduction of CO$_2$ to CO by a Tetraaza [Co$^\text{II}$N$_4$H]$^{2+}$ Complex in the Presence of Protons
Authors:
Alejandro J. Garza,
Srimanta Pakhira,
Alexis T. Bell,
Jose L. Mendoza-Cortes,
Martin Head-Gordon
Abstract:
The tetraaza [Co$^\text{II}$N$_4$H]$^{2+}$ complex (\textbf{1}) is remarkable for its ability to selectively reduce CO$_2$ to CO with 45\% Faradaic efficiency and a CO to H$_2$ ratio of 3:2. We employ density functional theory (DFT) to determine the reasons behind the unusual catalytic properties of \textbf{1} and the most likely mechanism for CO$_2$ reduction. The selectivity for CO$_2$ over prot…
▽ More
The tetraaza [Co$^\text{II}$N$_4$H]$^{2+}$ complex (\textbf{1}) is remarkable for its ability to selectively reduce CO$_2$ to CO with 45\% Faradaic efficiency and a CO to H$_2$ ratio of 3:2. We employ density functional theory (DFT) to determine the reasons behind the unusual catalytic properties of \textbf{1} and the most likely mechanism for CO$_2$ reduction. The selectivity for CO$_2$ over proton reduction is explained by analyzing the catalyst's affinity for the possible ligands present under typical reaction conditions: acetonitrile, water, CO$_2$, and bicarbonate. After reduction of the catalyst by two electrons, formation of [Co$^\text{I}$N$_4$H]$^{+}$-CO$_{2}^{-}$ is strongly favored. Based on thermodynamic and kinetic data, we establish that the only likely route for producing CO from here consists of a protonation step to yield [Co$^\text{I}$N$_4$H]$^{+}$-CO$_{2}$H, followed by reaction with CO$_2$ to form [Co$^\text{II}$N$_4$H]$^{2+}$-CO and bicarbonate. This conclusion corroborates the idea of a direct role of CO$_2$ as a Lewis acid to assist in {C-O} bond dissociation, a conjecture put forward by other authors to explain recent experimental observations. The pathway to formic acid is predicted to be forbidden by high activation barriers, in accordance with the products that are known to be generated by \textbf{1}. Calculated physical observables such as standard reduction potentials and the turnover frequency for our proposed catalytic cycle are in agreement with available experimental data reported in the literature. The mechanism also makes a prediction that may be experimentally verified: that the rate of CO formation should increase linearly with the partial pressure of CO$_2$.
△ Less
Submitted 27 March, 2018;
originally announced March 2018.
-
Predicting Band Gaps with Hybrid Density Functionals
Authors:
Alejandro J. Garza,
Gustavo E. Scuseria
Abstract:
We compare the ability of four popular hybrid density functionals (B3LYP, B3PW91, HSE, and PBE0) for predicting band gaps of semiconductors and insulators over a large benchmark set using a consistent methodology. We observe no significant statistical dference in their overall performance although the screened hybrid HSE is more accurate for typical semiconductors. HSE can improve its accuracy for…
▽ More
We compare the ability of four popular hybrid density functionals (B3LYP, B3PW91, HSE, and PBE0) for predicting band gaps of semiconductors and insulators over a large benchmark set using a consistent methodology. We observe no significant statistical dference in their overall performance although the screened hybrid HSE is more accurate for typical semiconductors. HSE can improve its accuracy for large large band gap materials --without affecting that of semiconductors-- by including a larger portion of Hartree-Fock exchange in its short range. Given that screened hybrids are computationally much less expensive than their global counterparts, we conclude that they are a better option for the black box prediction of band gaps.
△ Less
Submitted 5 October, 2016; v1 submitted 16 August, 2016;
originally announced August 2016.
-
Actinide chemistry using singlet-paired coupled cluster and its combinations with density functionals
Authors:
Alejandro J. Garza,
Ana G. Sousa Alencar,
Gustavo E. Scuseria
Abstract:
Singlet-paired coupled cluster doubles (CCD0) is a simplification of CCD that relinquishes a fraction of dynamic correlation in order to be able to describe static correlation. Combinations of CCD0 with density functionals that recover specifically the dynamic correlation missing in the former have also been developed recently. Here, we assess the accuracy of CCD0 and CCD0+DFT (and variants of the…
▽ More
Singlet-paired coupled cluster doubles (CCD0) is a simplification of CCD that relinquishes a fraction of dynamic correlation in order to be able to describe static correlation. Combinations of CCD0 with density functionals that recover specifically the dynamic correlation missing in the former have also been developed recently. Here, we assess the accuracy of CCD0 and CCD0+DFT (and variants of these using Brueckner orbitals) as compared to well-established quantum chemical methods for describing ground-state properties of singlet actinide molecules. The $f^0$ actinyl series (UO$_2^{2+}$, NpO$_2^{2+}$, PuO$_2^{2+}$), the isoelectronic NUN, and Thorium (ThO, ThO$^{2+}$) and Nobelium (NoO, NoO$_2$) oxides are studied.
△ Less
Submitted 20 October, 2015;
originally announced October 2015.
-
Combinations of coupled cluster, density functionals, and the random phase approximation for describing static and dynamic correlation, and van der Waals interactions
Authors:
Alejandro J. Garza,
Ireneusz W. Bulik,
Ana G. Sousa Alencar,
Jianwei Sun,
John P. Perdew,
Gustavo E. Scuseria
Abstract:
Contrary to standard coupled cluster doubles (CCD) and Brueckner doubles (BD), singlet-paired analogues of CCD and BD (denoted here as CCD0 and BD0) do not break down when static correlation is present, but neglect substantial amounts of dynamic correlation. In fact, CCD0 and BD0 do not account for any contributions from multielectron excitations involving only same-spin electrons at all. We explo…
▽ More
Contrary to standard coupled cluster doubles (CCD) and Brueckner doubles (BD), singlet-paired analogues of CCD and BD (denoted here as CCD0 and BD0) do not break down when static correlation is present, but neglect substantial amounts of dynamic correlation. In fact, CCD0 and BD0 do not account for any contributions from multielectron excitations involving only same-spin electrons at all. We exploit this feature to add---without introducing double counting, self-interaction, or increase in cost---the missing correlation to these methods via meta-GGA density functionals (TPSS and SCAN). Furthermore, we improve upon these CCD0+DFT blends by invoking range separation: the short- and long-range correlations absent in CCD0/BD0 are evaluated with DFT and the direct random phase approximation (dRPA), respectively. This corrects the description of long-range van der Waals forces. Comprehensive benchmarking shows that the combinations presented here are very accurate for weakly correlated systems, while also providing a reasonable description of strongly correlated problems without resorting to symmetry breaking.
△ Less
Submitted 10 September, 2015;
originally announced September 2015.
-
On the equivalence of LIST and DIIS methods for convergence acceleration
Authors:
Alejandro J. Garza,
Gustavo E. Scuseria
Abstract:
Self-consistent field extrapolation methods play a pivotal role in quantum chemistry and electronic structure theory. We here demonstrate the mathematical equivalence between the recently proposed family of LIST methods [J. Chem. Phys. 134, 241103 (2011); J. Chem. Theory Comput. 7, 3045 (2011)] with Pulay's DIIS [Chem. Phys. Lett. 73, 393 (1980)]. Our results also explain the differences in perfor…
▽ More
Self-consistent field extrapolation methods play a pivotal role in quantum chemistry and electronic structure theory. We here demonstrate the mathematical equivalence between the recently proposed family of LIST methods [J. Chem. Phys. 134, 241103 (2011); J. Chem. Theory Comput. 7, 3045 (2011)] with Pulay's DIIS [Chem. Phys. Lett. 73, 393 (1980)]. Our results also explain the differences in performance among the various LIST methods.
△ Less
Submitted 11 November, 2014;
originally announced November 2014.
