arXiv subjects

Ke Fu

Publications and source records attributed to Ke Fu.

3 recordsLinked to original sources

arXiv2026

Thermal-acoustic fermionic dynamics and primordial perturbations in warm Yukawa inflation

We investigate fermion production and its cosmological consequences in a warm inflationary scenario with a Yukawa interaction between the inflaton field and Dirac fermions. Treating the produced fermions as a thermalized relativistic fluid, we derive the coupled perturbation equations incorporating dissipative effects and thermal fluctuations. By solving the resulting system through Green's function methods, we identify distinct contributions from thermal noise, thermal-acoustic propagation, and higher-order scattering processes to the scalar perturbations. We show that acoustic propagation of fermionic fluctuations becomes significant in the relativistic regime, where the competition between gravitational attraction and Pauli degeneracy pressure allows collective fermionic modes to influence the primordial spectrum. Furthermore, we calculate the fermion number density and acoustic velocity, revealing that strong Yukawa interactions enhance non-adiabatic fermion production and modify the thermodynamic properties of the fermion bath. The sourced scalar and tensor perturbations are calculated, indicating that fermion production can substantially amplify primordial fluctuations, particularly in the strong-coupling regime. Since scalar perturbations receive more efficient enhancement than tensor modes, the tensor-to-scalar ratio is suppressed relative to the standard single-field warm inflation prediction. Our results show that thermalized fermions provide an additional mechanism for modifying primordial perturbations and offer new insights into the role of particle production in warm inflationary cosmology.

astro-ph.CO↗

arXiv2024

Learning Decisions Offline from Censored Observations with ε-insensitive Operational Costs

Many important managerial decisions are made based on censored observations. Making decisions without adequately handling the censoring leads to inferior outcomes. We investigate the data-driven decision-making problem with an offline dataset containing the feature data and the censored historical data of the variable of interest without the censoring indicators. Without assuming the underlying distribution, we design and leverage ε-insensitive operational costs to deal with the unobserved censoring in an offline data-driven fashion. We demonstrate the customization of the ε-insensitive operational costs for a newsvendor problem and use such costs to train two representative ML models, including linear regression (LR) models and neural networks (NNs). We derive tight generalization bounds for the custom LR model without regularization (LR-εNVC) and with regularization (LR-εNVC-R), and a high-probability generalization bound for the custom NN (NN-εNVC) trained by stochastic gradient descent. The theoretical results reveal the stability and learnability of LR-εNVC, LR-εNVC-R and NN-εNVC. We conduct extensive numerical experiments to compare LR-εNVC-R and NN-εNVC with two existing approaches, estimate-as-solution (EAS) and integrated estimation and optimization (IEO). The results show that LR-εNVC-R and NN-εNVC outperform both EAS and IEO, with maximum cost savings up to 14.40% and 12.21% compared to the lowest cost generated by the two existing approaches. In addition, LR-εNVC-R's and NN-εNVC's order quantities are statistically significantly closer to the optimal solutions should the underlying distribution be known.

cs.LG↗

arXiv2023

Optimal investment problem for a hybrid pension with intergenerational risk-sharing and longevity trend under model uncertainty

This paper studies the optimal investment problem for a hybrid pension plan under model uncertainty, where both the contribution and the benefit are adjusted depending on the performance of the plan. Furthermore, an age and time-dependent force of mortality and a linear maximum age are considered to capture the longevity trend. Suppose that the plan manager is ambiguity averse and is allowed to invest in a risk-free asset and a stock. The plan manager aims to find optimal investment strategies and optimal intergenerational risk-sharing arrangements by minimizing the cost of unstable contribution risk, the cost of unstable benefit risk and discontinuity risk under the worst-case scenario. By applying the stochastic optimal control approach, closed-form solutions are derived under a penalized quadratic cost function. Through numerical analysis and three special cases, we find that the intergeneration risk-sharing is achieved in our collective hybrid pension plan effectively. And it also shows that when people live longer, postponing the retirement seems a feasible way to alleviate the stress of the aging problem.

math.OC↗