SearcharxivSearch

arXiv subjects

Milena Filipovic

Publications and source records attributed to Milena Filipovic.

5 recordsLinked to original sources

Effect of uniaxial magnetic anisotropy on charge transport in a junction with a precessing anisotropic molecular spin

Anisotropic magnetic molecules can be employed to manipulate charge transport in molecular nanojunctions. Charge transport through a molecular orbital connected to two leads and exchange-coupled with a precessing anisotropic molecular spin in a constant magnetic field is studied here. Both the magnetic field and the uniaxial magnetic anisotropy parameter of the molecular spin modulate the total precession frequency. The precessing molecular magnetization drives inelastic tunneling processes between electronic quasienergy levels. The dc-bias voltages allow to unveil the quasienergy levels, Larmor frequency, and the anisotropy parameter, through characteristics of charge-transport measurements involving features such as steps, peaks and dips. Quantum interference effects between states connected with spin-flip events are reflected in the shot noise as peak-dip (dip-peak) features, resembling Fano-like resonance profiles, and are controlled by the anisotropy parameter and Larmor frequency. Under zero bias, the increase of the anisotropy parameter enables the decrease of the precession frequency or alters the precession direction, and shot noise is reduced. Furthermore, it is possible to adjust the anisotropy parameter to suppress the precession frequency, leading to the suppression of shot noise. The results show that in the given setup, the charge current and shot noise can be controlled by the magnetic anisotropy parameter of the molecular spin.

cond-mat.mes-hall

Modeling Online Behavior in Recommender Systems: The Importance of Temporal Context

Recommender systems research tends to evaluate model performance offline and on randomly sampled targets, yet the same systems are later used to predict user behavior sequentially from a fixed point in time. Simulating online recommender system performance is notoriously difficult and the discrepancy between online and offline behaviors is typically not accounted for in offline evaluations. This disparity permits weaknesses to go unnoticed until the model is deployed in a production setting. In this paper, we first demonstrate how omitting temporal context when evaluating recommender system performance leads to false confidence. To overcome this, we postulate that offline evaluation protocols can only model real-life use-cases if they account for temporal context. Next, we propose a training procedure to further embed the temporal context in existing models. We use a multi-objective approach to introduce temporal context into traditionally time-unaware recommender systems and confirm its advantage via the proposed evaluation protocol. Finally, we validate that the Pareto Fronts obtained with the added objective dominate those produced by state-of-the-art models that are only optimized for accuracy on three real-world publicly available datasets. The results show that including our temporal objective can improve recall@20 by up to 20%.

cs.IR

Momentum-based Gradient Methods in Multi-Objective Recommendation

Multi-objective gradient methods are becoming the standard for solving multi-objective problems. Among others, they show promising results in developing multi-objective recommender systems with both correlated and conflicting objectives. Classic multi-gradient~descent usually relies on the combination of the gradients, not including the computation of first and second moments of the gradients. This leads to a brittle behavior and misses important areas in the solution space. In this work, we create a multi-objective model-agnostic Adamize method that leverages the benefits of the Adam optimizer in single-objective problems. This corrects and stabilizes~the~gradients of every objective before calculating a common gradient descent vector that optimizes all the objectives simultaneously. We evaluate the benefits of Multi-objective Adamize on two multi-objective recommender systems and for three different objective combinations, both correlated or conflicting. We report significant improvements, measured with three different Pareto front metrics: hypervolume, coverage, and spacing. Finally, we show that the \textit{Adamized} Pareto front strictly dominates the previous one on multiple objective pairs.

cs.LG

Shot noise of charge and spin transport in a junction with a precessing molecular spin

Magnetic molecules and nanomagnets can be used to influence the electronic transport in mesoscopic junction. In a magnetic field the precessional motion leads to resonances in the dc- and ac-transport properties of a nanocontact, in which the electrons are coupled to the precession. Quantities like the dc-conductance or the ac-response provide valuable information like the level structure and the coupling parameters. Here, we address the current noise properties of such contacts. This encompasses the charge current and spin-torque shot noise, which both show a step-like behavior as functions of bias voltage and magnetic field. The charge current noise shows pronounced dips around the steps, which we trace back to interference effects of electron in quasienergy levels coupled by the molecular spin precession. We show that some components of the noise of the spin-torque currents are directly related to the Gilbert damping and, hence, are experimentally accessible. Our results show that the noise characteristics allow to investigate in more detail the coherence of spin transport in contacts containing magnetic molecules.

cond-mat.mes-hall

Spin transport and tunable Gilbert damping in a single-molecule magnet junction

We study time-dependent electronic and spin transport through an electronic level connected to two leads and coupled with a single-molecule magnet via exchange interaction. The molecular spin is treated as a classical variable and precesses around an external magnetic field. We derive expressions for charge and spin currents by means of the Keldysh non-equilibrium Green's functions technique in linear order with respect to the time-dependent magnetic field created by this precession. The coupling between the electronic spins and the magnetization dynamics of the molecule creates inelastic tunneling processes which contribute to the spin currents. The inelastic spin currents, in turn, generate a spin-transfer torque acting on the molecular spin. This back-action includes a contribution to the Gilbert damping and a modification of the precession frequency. The Gilbert damping coefficient can be controlled by the bias and gate voltages or via the external magnetic field and has a non-monotonic dependence on the tunneling rates.

cond-mat.mes-hall