BZPEERPapers, people, evidence and research paths
Menu

Discover

SearchFind papers, people, topics, institutions and opportunities from one place.GraphSee how papers, authors, topics and evidence connect.

Research tools

HomeContinue from the papers, people and topics that matter now.CollectionsTurn saved papers into reading lists, evidence maps and shareable context.ResearchReconstruct the path that led you from search to paper to question.CollaborationMove from a paper, topic or expert request into a focused thread.PublishCreate full-text research articles with metadata, formulas and versions.

Network

ExpertsFind specialists whose work and reviews explain why they can help.CommunitiesJoin research groups organized around papers, topics and shared evidence.

Opportunities

ListingsBrowse roles, calls and collaborations connected to real research context.

Account

Log inReturn to your saved papers, graph views and active threads.Create accountStart saving papers, building a research trail and joining teams.
Log inCreate account

Source author record

Zhenhua Han

Zhenhua Han appears in the imported research catalog. Authorship, coauthor and topic links are available while profile ownership is still unclaimed.

ResearcherUnclaimed source record
astro-ph.GAastro-ph.HEComputation and LanguageArtificial Intelligenceeess.ASNetworking and Internet ArchitectureSoftware Engineering

Catalog footprint

What is connected

6works
7topics
4close collaborators

Actions

Connect this record

Log in to claim
Open graphBrowse works

Research graph

See the researcher in context

Open full explorer

Inspect adjacent papers, topics, institutions and collaborators without losing the researcher page.

Building this map preview

BZPEER is loading the nearby papers, people, topics and institutions for this page.

Published work

6 published item(s)

preprint2026arXiv

Agentic Harness Engineering: Observability-Driven Automatic Evolution of Coding-Agent Harnesses

Harnesses are now central to coding-agent performance, mediating how models interact with tools and execution environments. Yet harness engineering remains a manual craft, because automating it faces a heterogeneous action space across editable components, voluminous trajectories that bury actionable signal, and edits whose effect is hard to attribute. We introduce Agentic Harness Engineering (AHE), a closed loop that addresses these challenges through three matched observability pillars: (1) component observability gives every editable harness component a file-level representation so the action space is explicit and revertible; (2) experience observability distills millions of raw trajectory tokens into a layered, drill-down evidence corpus that an evolving agent can actually consume; and (3) decision observability pairs every edit with a self-declared prediction, later verified against the next round's task-level outcomes. Together, these pillars turn every edit into a falsifiable contract, so harness evolution proceeds autonomously without collapsing into trial-and-error. Empirically, ten AHE iterations lift pass@1 on Terminal-Bench 2 from 69.7% to 77.0%, surpassing the human-designed harness Codex-CLI (71.9%) and the self-evolving baselines ACE and TF-GRPO. The frozen harness transfers without re-evolution: on SWE-bench-verified it tops aggregate success at 12% fewer tokens than the seed, and on Terminal-Bench 2 it yields +5.1 to +10.1pp cross-family gains across three alternate model families, indicating the evolved components encode general engineering experience rather than benchmark-specific tuning. Ablations localize the gain to tools, middleware, and long-term memory rather than the system prompt, suggesting factual harness structure transfers while prose-level strategy does not.

0 citations0 reviewsNo code linkedOpen access
preprint2026arXiv

What Makes a Good Speech Tokenizer for LLM-Centric Speech Generation? A Systematic Study

Speech-language models (SLMs) offer a promising path toward unifying speech and text understanding and generation. However, challenges remain in achieving effective cross-modal alignment and high-quality speech generation. In this work, we systematically investigate the role of speech tokenizer designs in LLM-centric SLMs, augmented by speech heads and speaker modeling. We compare coupled, semi-decoupled, and fully decoupled speech tokenizers under a fair SLM framework and find that decoupled tokenization significantly improves alignment and synthesis quality. To address the information density mismatch between speech and text, we introduce multi-token prediction (MTP) into SLMs, enabling each hidden state to decode multiple speech tokens. This leads to up to 12$\times$ faster decoding and a substantial drop in word error rate (from 6.07 to 3.01). Furthermore, we propose a speaker-aware generation paradigm and introduce RoleTriviaQA, a large-scale role-playing knowledge QA benchmark with diverse speaker identities. Experiments demonstrate that our methods enhance both knowledge understanding and speaker consistency.

0 citations0 reviewsNo code linkedOpen access
preprint2016arXiv

Dynamic Virtual Machine Management via Approximate Markov Decision Process

Efficient virtual machine (VM) management can dramatically reduce energy consumption in data centers. Existing VM management algorithms fall into two categories based on whether the VMs' resource demands are assumed to be static or dynamic. The former category fails to maximize the resource utilization as they cannot adapt to the dynamic nature of VMs' resource demands. Most approaches in the latter category are heuristical and lack theoretical performance guarantees. In this work, we formulate dynamic VM management as a large-scale Markov Decision Process (MDP) problem and derive an optimal solution. Our analysis of real-world data traces supports our choice of the modeling approach. However, solving the large-scale MDP problem suffers from the curse of dimensionality. Therefore, we further exploit the special structure of the problem and propose an approximate MDP-based dynamic VM management method, called MadVM. We prove the convergence of MadVM and analyze the bound of its approximation error. Moreover, MadVM can be implemented in a distributed system, which should suit the needs of real data centers. Extensive simulations based on two real-world workload traces show that MadVM achieves significant performance gains over two existing baseline approaches in power consumption, resource shortage and the number of VM migrations. Specifically, the more intensely the resource demands fluctuate, the more MadVM outperforms.

0 citations0 reviewsNo code linkedOpen access
preprint2015arXiv

The physical fundamental plane of black hole activity: revisited

The correlation between the jet power and accretion disk luminosity is investigated for active galactic nuclei (AGNs) and black hole X-ray binaries (BHXBs) from the literature. The power-law correlation index is steep ($μ\sim$ 1.0--1.4) for radio loud quasars and the `outliers' track of BHXBs, and it is flatter ($μ\sim$ 0.3--0.6) for radio loud galaxies and the standard track of BHXBs. The steep-index groups are mostly at higher accretion rates (peaked at Eddington ratio $>$ 0.01) and the flatter-index groups are at relatively low accretion rates (peaked at Eddington ratio $<$ 0.01), implying that the former groups could be dominated by the inner disk accretion of black hole, while the jet in latter groups would be a hybrid production of the accretion and black hole spin. We could still have a fundamental plane of black hole activity for the BHXBs and AGNs with diverse (maybe two kinds of) correlation indices. It is noted that the fundamental plane of black hole activity should be referred to the correlation between the jet power and disk luminosity or equivalently to the correlation between jet power, Eddington ratio and black hole mass, rather than the jet power, disk luminosity and black hole mass.

0 citations0 reviewsNo code linkedOpen access
preprint2014arXiv

Is radio jet power linearly proportional to the product of central black hole mass and Eddington ratio in AGN?

A model for the relation between radio jet power and the product of central black hole (BH) mass and Eddington ratio of AGN is proposed, and the model is examined with data from the literature. We find that radio jet power positively correlates but not linearly with the product of BH mass ($m$ in solar mass) and Eddington ratio ($λ$), and the power law indices ($μ$) are significantly less than unity for relatively low accretion ($λ<0.1$) AGN, $P_{j}\propto (λm)^μ$, in the radio galaxies and the Seyfert galaxies. This leads to a negative correlation between radio loudness and $λm$ for the low luminosity AGN, i.e. $R\propto (λm)^ρ$ with $ρ=(7/6)μ-1<0$, which may be attributed to a contribution of BH spin to total jet power assuming that the spin induced jet is gradually suppressed as the accretion rate increases. Whereas, for the high-z quasars which often show the slope $μ\geq1$, a positive correlation between the radio loudness and disc luminosity is predicted. We discuss that the jet powers of the high-z FRII quasars are likely dominated by the accretion disc rather than by the BH spin.

0 citations0 reviewsNo code linkedOpen access
preprint2014arXiv

On the relation of accretion rate and spin induced jet power in low luminosity AGN

From Liu and Han (2014), the accretion-dominated jet power has a linear proportionality with the accretion rate, whereas the power law index is <=0.5 at lower accretion rate. Attributing the jet power in low accretion rate AGN to the black hole spin, it implies that the jet power has a flatter spectrum than the accretion-dominated jet versus the accretion rate. The black hole must be spinning rapidly for producing such jet power efficiently, and this may allow us to find high-spin black holes in the radio-loud low-luminosity AGN.

0 citations0 reviewsNo code linkedOpen access