Engineering pump-probe schemes with strong sub and few optical fs pulses has made experimentally feasible to imprint and probe electronic coherence in molecules. It is such coherences that allow a manifestation of a purely electronic time scale, before the coupling to the nuclei takes place. Here I will discuss effects on a longer time scale and in particular how the onset of nuclear motion and subsequently the fate of a chemical reaction can be controlled by creating this electronic coherence through a strong ultra short excitation pulse. Our quantum dynamical computations include the electron-nuclei coupling as well as the photoexcitation and photoionization dynamics due to the coupling to the strong field electric field of the pulse. We show how the ultrafast beatings of the electronic coherences in space and in time are modulated by the different periods of the nuclear motion and the non adiabatic couplings between electronic states and by the photoionization process that can occur during the pump and/or the probe pulses. It is nevertheless possible to follow the spatial and temporal localization of the electronic coherence for a large number of vibrational periods using the probe pulse and monitoring angularly resolved photoelectron distribution or transient absorption as a function of the pump-probe delay time. Examples will be discussed for small molecules, LiH and HCN.
Tin-oxo cages belong to a class of organometallic compounds which were recently recognized as promising photoresist materials for the extreme ultraviolet (EUV) lithography − a method proposed for the new generation of integrated circuits. The EUV exposure of tin-oxo cages causes significant structural changes; however, the molecular mechanism of these changes is to a large extent unknown. In the present work, we focus on the photoionization dynamics of the small tin-oxo molecules: the trimethlytin hydroxide (TMTH) and trihydroxymethyl stannane (THMS) molecules. We employed the surface hopping (SH) dynamics on the Floating Occupation Molecular Orbital Complete Active Space Configuration Interaction (FOMO-CASCI) potential energy surfaces accelerated with graphical processor units (GPU). After the ionization of both molecules, we observed an ultrafast relaxation (~100 fs) into the ground and first excited electronic states followed by a rapid dissociation yielding the neutral CH3 and OH radicals and a charged remainder. However, the dissociation dynamic does not occur solely on the hot ground electronic state and reaction outcome partially depends on initially ionized state. Our work also demonstrate that the SH+FOMO-CASSCI based dynamics is suitable for electron-rich open-shell systems.
The atmosphere, ocean, and other components of the climate system behave in a fairly irregular way. A major surprise of the late 20th century was the realization that such behavior could be produced by natural systems with a small number of degrees of freedom, governed by fully deterministic, but nonlinear laws. Still, the climate system has a large number of degrees of freedom, and ample room for random factors to intervene. Can we reconcile a low-order, deterministically nonlinear description of weather or climate with a high-order, possibly linear but random one? This talk will present some steps on the road to such a "grand unification," and implications for climate variability, sensitivity and predictability will be discussed. References: 1. Ghil, M., 2002: Natural climate variability, in Encyclopedia of Global Environmental Change, T. Munn (Ed.), Vol. 1, J. Wiley & Sons, Chichester/New York, pp. 544–549. 2. Ghil, M., 2014: Climate variability: Nonlinear and random aspects, in Encyclopedia of Atmospheric Sciences, 2nd edn., G. R. North, J. Pyle and F. Zhang (Eds.), Elsevier, vol. 2, pp. 38–46. 3. Ghil, M., 2017: The wind-driven ocean circulation: Applying dynamical systems theory to a climate problem, Discr. Cont. Dyn. Syst. – A, 37(1), 189–228, doi:10.3934/dcds.2017008.
Tensor networks are an efficient representation of interesting many-body wavefunctions and underpin powerful algorithms for strongly correlated systems. But tensor networks could be applied much more broadly than just for representing wavefunctions. Large tensors similar to wavefunctions appear naturally in certain families of models studied extensively in machine learning. Decomposing the model parameters as a tensor network leads to interesting algorithms for training models on real-world data which scale better than existing approaches. In addition to training models directly for recognizing labeled data, tensor network real-space renormalization approaches can be used to extract statistically significant "features" for subsequent learning tasks. I will also highlight other benefits of the tensor network approach such as the flexibility to blend different approaches and to interpret trained models.