-
Models Matter: The Impact of Single-Step Retrosynthesis on Synthesis Planning
Authors:
Paula Torren-Peraire,
Alan Kai Hassen,
Samuel Genheden,
Jonas Verhoeven,
Djork-Arne Clevert,
Mike Preuss,
Igor Tetko
Abstract:
Retrosynthesis consists of breaking down a chemical compound recursively step-by-step into molecular precursors until a set of commercially available molecules is found with the goal to provide a synthesis route. Its two primary research directions, single-step retrosynthesis prediction, which models the chemical reaction logic, and multi-step synthesis planning, which tries to find the correct se…
▽ More
Retrosynthesis consists of breaking down a chemical compound recursively step-by-step into molecular precursors until a set of commercially available molecules is found with the goal to provide a synthesis route. Its two primary research directions, single-step retrosynthesis prediction, which models the chemical reaction logic, and multi-step synthesis planning, which tries to find the correct sequence of reactions, are inherently intertwined. Still, this connection is not reflected in contemporary research. In this work, we combine these two major research directions by applying multiple single-step retrosynthesis models within multi-step synthesis planning and analyzing their impact using public and proprietary reaction data. We find a disconnection between high single-step performance and potential route-finding success, suggesting that single-step models must be evaluated within synthesis planning in the future. Furthermore, we show that the commonly used single-step retrosynthesis benchmark dataset USPTO-50k is insufficient as this evaluation task does not represent model performance and scalability on larger and more diverse datasets. For multi-step synthesis planning, we show that the choice of the single-step model can improve the overall success rate of synthesis planning by up to +28% compared to the commonly used baseline model. Finally, we show that each single-step model finds unique synthesis routes, and differs in aspects such as route-finding success, the number of found synthesis routes, and chemical validity, making the combination of single-step retrosynthesis prediction and multi-step synthesis planning a crucial aspect when developing future methods.
△ Less
Submitted 10 August, 2023;
originally announced August 2023.
-
Mind the Retrosynthesis Gap: Bridging the divide between Single-step and Multi-step Retrosynthesis Prediction
Authors:
Alan Kai Hassen,
Paula Torren-Peraire,
Samuel Genheden,
Jonas Verhoeven,
Mike Preuss,
Igor Tetko
Abstract:
Retrosynthesis is the task of breaking down a chemical compound recursively step-by-step into molecular precursors until a set of commercially available molecules is found. Consequently, the goal is to provide a valid synthesis route for a molecule. As more single-step models develop, we see increasing accuracy in the prediction of molecular disconnections, potentially improving the creation of sy…
▽ More
Retrosynthesis is the task of breaking down a chemical compound recursively step-by-step into molecular precursors until a set of commercially available molecules is found. Consequently, the goal is to provide a valid synthesis route for a molecule. As more single-step models develop, we see increasing accuracy in the prediction of molecular disconnections, potentially improving the creation of synthetic paths. Multi-step approaches repeatedly apply the chemical information stored in single-step retrosynthesis models. However, this connection is not reflected in contemporary research, fixing either the single-step model or the multi-step algorithm in the process. In this work, we establish a bridge between both tasks by benchmarking the performance and transfer of different single-step retrosynthesis models to the multi-step domain by leveraging two common search algorithms, Monte Carlo Tree Search and Retro*. We show that models designed for single-step retrosynthesis, when extended to multi-step, can have a tremendous impact on the route finding capabilities of current multi-step methods, improving performance by up to +30% compared to the most widely used model. Furthermore, we observe no clear link between contemporary single-step and multi-step evaluation metrics, showing that single-step models need to be developed and tested for the multi-step domain and not as an isolated task to find synthesis routes for molecules of interest.
△ Less
Submitted 12 December, 2022;
originally announced December 2022.
-
Validity of Sound-Proof Approaches in Rapidly-Rotating Compressible Convection: Marginal Stability vs. Turbulence
Authors:
Jan Verhoeven,
Gary A. Glatzmaier
Abstract:
The validity of the anelastic approximation has recently been questioned in the regime of rapidly-rotating compressible convection in low Prandtl number fluids (Calkins et al. 2015). Given the broad usage and the high computational efficiency of sound-proof approaches in this astrophysically relevant regime, this paper clarifies the conditions for a safe application. The potential of the alternati…
▽ More
The validity of the anelastic approximation has recently been questioned in the regime of rapidly-rotating compressible convection in low Prandtl number fluids (Calkins et al. 2015). Given the broad usage and the high computational efficiency of sound-proof approaches in this astrophysically relevant regime, this paper clarifies the conditions for a safe application. The potential of the alternative pseudo-incompressible approximation is investigated, which in contrast to the anelastic approximation is shown to never break down for predicting the point of marginal stability. Its accuracy, however, decreases close to the parameters corresponding to the failure of the anelastic approach, which is shown to occur when the sound-crossing time of the domain exceeds a rotation time scale, i.e. for rotational Mach numbers greater than one. Concerning the supercritical case, which is naturally characterised by smaller rotational Mach numbers, we find that the anelastic approximation does not show unphysical behaviour. Growth rates computed with the linearised anelastic equations converge toward the corresponding fully compressible values as the Rayleigh number increases. Likewise, our fully nonlinear turbulent simulations, produced with our fully compressible and anelastic models and carried out in a highly supercritical, rotating, compressible, low Prandtl number regime show good agreement. However, this nonlinear test example is for only a moderately low convective Rossby number of 0.14.
△ Less
Submitted 21 September, 2017; v1 submitted 17 January, 2017;
originally announced January 2017.
-
Turbulent transport by diffusive stratified shear flows: from local to global models. Part I: Numerical simulations of a stratified plane Couette flow
Authors:
P. Garaud,
D. Gagnier,
J. Verhoeven
Abstract:
Shear-induced turbulence could play a significant role in mixing momentum and chemical species in stellar radiation zones, as discussed by Zahn (1974). In this paper we analyze the results of direct numerical simulations of stratified plane Couette flows, in the limit of rapid thermal diffusion, to measure the turbulent diffusivity and turbulent viscosity as a function of the local shear and the l…
▽ More
Shear-induced turbulence could play a significant role in mixing momentum and chemical species in stellar radiation zones, as discussed by Zahn (1974). In this paper we analyze the results of direct numerical simulations of stratified plane Couette flows, in the limit of rapid thermal diffusion, to measure the turbulent diffusivity and turbulent viscosity as a function of the local shear and the local stratification. We find that the stability criterion proposed by Zahn (1974), namely that the product of the gradient Richardson number and the Prandtl number must be smaller than a critical values $(J\Pr)_c$ for instability, adequately accounts for the transition to turbulence in the flow, with $(J\Pr)_c \simeq 0.007$. This result recovers and confirms the prior findings of Prat et al. (2016). Zahn's model for the turbulent diffusivity and viscosity (Zahn 1992), namely that the mixing coefficient should be proportional to the ratio of the thermal diffusivity to the gradient Richardson number, does not satisfactorily match our numerical data when applied as is. It fails (as expected) in the limit of large stratification where the Richardson number exceeds the aforementioned threshold for instability, but it also fails in the limit of low stratification where the turbulent eddy scale becomes limited by the computational domain size. We propose a revised model for turbulent mixing by diffusive stratified shear instabilities, that now properly accounts for both limits, fits our data satisfactorily, and recovers Zahn's 1992 model in the limit of large Reynolds numbers.
△ Less
Submitted 14 October, 2016;
originally announced October 2016.
-
The compressional beta effect: a source of zonal winds in planets?
Authors:
Jan Verhoeven,
Stephan Stellmach
Abstract:
Giant planets like Jupiter and Saturn feature strong zonal wind patterns on their surfaces. Although several different mechanisms that may drive these jets have been proposed over the last decades, the origin of the zonal winds is still unclear. Here, we explore the possibility that the interplay of planetary rotation with the compression and expansion of the convecting fluid can drive multiple de…
▽ More
Giant planets like Jupiter and Saturn feature strong zonal wind patterns on their surfaces. Although several different mechanisms that may drive these jets have been proposed over the last decades, the origin of the zonal winds is still unclear. Here, we explore the possibility that the interplay of planetary rotation with the compression and expansion of the convecting fluid can drive multiple deep zonal jets by a compressional Rhines-type mechanism, as originally proposed by Ingersoll and Pollard (1982). In a certain limit, this deep mechanism is shown to be mathematically analogous to the classical Rhines mechanism possibly operating at cloud level. Jets are predicted to occur on a compressional Rhines length $l_R = (2 Ω\langle H_ρ^{-1} \rangle v_{jet}^{-1} )^{-1/2}$, where $Ω$ is the angular velocity, $\langle H_ρ^{-1} \rangle$ is the mean inverse density scale height and $v_{jet}$ is the typical jet velocity. Two-dimensional numerical simulations using the anelastic approximation reveal that this mechanism robustly generates jets of the predicted width, and that it typically dominates the dynamics in systems deeper than $O(l_R)$. Potential vorticity staircases are observed to form spontaneously and are typically accompanied by unstably stratified buoyancy staircases. The mechanism only operates at large rotation rates, exceeding those typically reached in three-dimensional simulations of deep convection in spherical shells. Applied to Jupiter and Saturn, the compressional Rhines scaling reasonably fits the available observations. Interestingly, even weak vertical density variations such as those in the Earth core can give rise to a large number of jets, leading to fundamentally different flow structures than predicted by the Boussinesq models typically used in this context.
△ Less
Submitted 28 April, 2014;
originally announced April 2014.
-
Comment to the paper "Radiation induced by relativistic electrons propagating through random layered stacks: Numerical simulation results" by A.A.Varfolomeev and et al NIM B 256,705 (2007)
Authors:
Zh. S. Gevorkian,
J. Verhoeven
Abstract:
We show that the numerical code used in the above mentioned paper does not take into account the multiple scattering effects of electromagnetic field properly and is therefore incorrect.
We show that the numerical code used in the above mentioned paper does not take into account the multiple scattering effects of electromagnetic field properly and is therefore incorrect.
△ Less
Submitted 22 July, 2007;
originally announced July 2007.
-
Resonant Diffusive Radiation in Random Multilayered Systems
Authors:
Zh. S. Gevorkian,
J. Verhoeven
Abstract:
We have theoretically shown that the yield of diffuse radiation generated by relativistic electrons passing random multilayered systems can be increased when a resonant condition is met. Resonant condition can be satisfied for the wavelength region representing visible light as well as soft X-rays. The intensity of diffusive soft X-rays for specific multilayered systems consisting of two compone…
▽ More
We have theoretically shown that the yield of diffuse radiation generated by relativistic electrons passing random multilayered systems can be increased when a resonant condition is met. Resonant condition can be satisfied for the wavelength region representing visible light as well as soft X-rays. The intensity of diffusive soft X-rays for specific multilayered systems consisting of two components is compared with the intensity of Cherenkov radiation. For radiation at photon energy of $99.4eV$, the intensity of Resonant Diffusive Radiation (RDR) generated by $5MeV$ electrons passing a $Be/Si$ multilayer exceeds the intensity of Cherenkov radiation by a factor of $\approx 60$ for electrons with the same energy passing a $Si$ foil. For a photon energy of $453eV$ and $13MeV$ electrons passing $Be/Ti$ multilayer generate RDR exceeding Cherenkov radiation generated by electrons passing a $Ti$ foils by a factor $\approx 130$.
△ Less
Submitted 14 February, 2006; v1 submitted 17 October, 2005;
originally announced October 2005.
