Monday, February 25, 2019

190302 How Does Omnivory Influence the Effects of Consumer Diversity on Resource Prey Biomass?

Title:
How Does Omnivory Influence the Effects of Consumer Diversity on Resource Prey Biomass?

Speaker:
張峰勳 (Oscar Feng-Hsun Chang), PhD candidate, University of Michigan

Time:
03/02 (Sat.) 7 pm PST, 8 pm MST, 9 pm CST, 10 pm EST
03/03 (Sun.) 11 am Taiwan

Keywords:
Ecology, Community ecology, Ecosystem ecology


Abstract:
It is crucial to understand what drives and regulates the biomass production of plants and animals since plant and consumer biomass provide the resources necessary to sustain numerous aspects of human life. Over the past two decades, scientists have generally agreed that plant diversity, i.e. the variety of plant species, genes and functional groups coexisting in an ecosystem, has a positive effect on plant biomass production presumably due to resource partitioning. In contrast, we still have a rudimentary understanding of how resource prey captured by consumers and, in turn, secondary production, is affected by consumer diversity. Similar to plants, consumer diversity can positively affect prey capture when consumers partition their resource prey. However, consumer diversity can also have a greater variety of impacts on prey consumption due to more complex interspecific interactions like omnivory. Omnivory occurs when animals eat each other, which is rare among plants. More complex inter-specific interactions in consumers could either weaken or strengthen the effects of consumer diversity on prey consumption depending on the strength of omnivorous consumption. To mechanistically understand how consumer diversity affects secondary production, we need to understand how omnivory interacts with resource partitioning to regulate prey consumption.

Sunday, February 17, 2019

190223 Next Generation Low-Mass Dark Matter Search

Title:
Next Generation Low-Mass Dark Matter Search

Speaker:
張硯詠 (Yen-Yung Chang), PhD candidate, Caltech

Time:
02/23 (Sat.) 7 pm PST, 8 pm MST, 9 pm CST, 10 pm EST
02/24 (Sun.) 11 am Taiwan

Keywords:
Physics, High energy physics, Dark matter, Cryogenic detector, Kinetic Inductance Detector



Abstract:
In recent years, development in dark matter (DM) phenomenology below weak scale has been booming. I will begin by discussing the rich phenomenology and rising interest in low-mass (< 10 GeV) DM candidates. I will then introduce SuperCDMS SNOLAB, one of the leading next generation experiment for DM direct detection. It will search for DM mass in 0.5-10 GeV range with a projected sensitivity down to 10-43 cm2 nucleon scattering cross section. It utilizes cryogenic Ge/Si crystal with superconducting Transition Edge Sensor (TES) for phonon-mediated recoil detection, and it will begin data taking in 2020. The second half of the talk will be focused on the ongoing R&D and proposed concepts beyond SuperCDMS SNOLAB. I will introduce Kinetic Inductance Detector (KID) as a promising alternative to TES for future larger scale, high resolution, sub-GeV DM searches. Finally, I will briefly discuss recent proposals and challenges, and conclude with future prospects toward the complete search for low-mass thermal relic DM.

Sunday, February 10, 2019

190216 Scheduling Benefit-generating Jobs to Capacitated Machines with a Consideration of Fairness

Title:
Scheduling Benefit-generating Jobs to Capacitated Machines with a Consideration of Fairness

Speaker:
孔令傑 (Ling-Chieh Kung), PhD, NTU

Time:
*請注意:本週演講時間調整至週日 (US) / 週一 (TW)

*演講時間調整回週六 (US) / 週日 (TW)

02/16 (Sat.) 5 pm PST, 6 pm MST, 7 pm CST, 8 pm EST
02/17 (Sun.) 9 am Taiwan

Keywords:
Information Management, Operations Research, Scheduling, Approximation Algorithm, Benefit-workload Relationship, Capacitated Machine, Fairness


Abstract:
We consider a problem of allocating jobs to identical machines. Each job has an amount of workload and a benefit collected upon completion. All machines have the same capacity. Under the constraints that jobs cannot be split and that machines have limited capacity, the objective is to maximize the benefit earned by the machine with the least benefit. Our problem is thus a capacitated job allocation problem with consideration of fairness. After showing that this problem is NP-hard, we propose an approximation algorithm modified from the longest processing time (LPT) rule. The algorithm is proved to have a worst-case performance guarantee 1/2 when job benefits are linear to workloads. Different approximation factors are also derived for convex and concave relationships. Finally, numerical studies illustrate the average performances of the algorithm and demonstrates that this algorithm works well when the jobs exhibit economy of scale but not so well when they exhibit diminishing marginal benefits.