Source author record

Stephen D. H. Hsu

Stephen D. H. Hsu appears in the imported research catalog. Authorship, coauthor and topic links are available while profile ownership is still unclaimed.

ResearcherUnclaimed source record

Catalog footprint

What is connected

35works
14topics
4close collaborators

Actions

Connect this record

Log in to claim

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

35 published item(s)

preprint2026arXiv

Embedded AI Companion System on Edge Devices

Computational resource constraints on edge devices make it difficult to develop a fully embedded AI companion system with a satisfactory user experience. AI companion and memory systems detailed in existing literature cannot be directly used in such an environment due to lack of compute resources and latency concerns. In this paper, we propose a memory paradigm that alternates between active and inactive phases: during phases of user activity, the system performs low-latency, real-time dialog using lightweight retrieval over existing memories and context; whereas during phases of user inactivity, it conducts more computationally intensive extraction, consolidation, and maintenance of memories across full conversation sessions. This design minimizes latency while maintaining long-term personalization under the tight constraints of embedded hardware. We also introduce an AI Companion benchmark designed to holistically evaluate the AI Companion across both its conversational quality and memory capabilities. In our experiments, we found that our system (using a very weak model: Qwen2.5-7B-Instruct quantized int4) outperforms the equivalent raw LLM without memory across most metrics, and performs comparably to GPT-3.5 with 16k context window.

preprint2022arXiv

A Brief History of Hawking's Information Paradox

In this invited review, we describe Hawking's information paradox and a recently proposed resolution of it. Explicit calculations demonstrate the existence of quantum hair on black holes, meaning that the quantum state of the external graviton field depends on the internal state of the black hole. Simple quantum mechanics then implies that Hawking radiation amplitudes depend on the internal state, resulting in a pure final radiation state that preserves unitarity and, importantly, violates a factorization assumption which is central to the original paradox. Black hole information is encoded in entangled macroscopic superposition states of the radiation.

preprint2022arXiv

Quantum Hair and Black Hole Information

It has been shown that the quantum state of the graviton field outside a black hole horizon carries information about the internal state of the hole. We explain how this allows unitary evaporation: the final radiation state is a complex superposition which depends linearly on the initial black hole state. Under time reversal, the radiation state evolves back to the original black hole quantum state. Formulations of the information paradox on a fixed semiclassical geometry describe only a small subset of the evaporation Hilbert space, and do not exclude overall unitarity.

preprint2022arXiv

Quantum Hair from Gravity

We explore the relationship between the quantum state of a compact matter source and of its asymptotic graviton field. For a matter source in an energy eigenstate, the graviton state is determined at leading order by the energy eigenvalue. Insofar as there are no accidental energy degeneracies there is a one to one map between graviton states on the boundary of spacetime and the matter source states. A typical semiclassical matter source results in an entangled asymptotic graviton state. We exhibit a purely quantum gravitational effect which causes the subleading asymptotic behavior of the graviton state to depend on the internal structure of the source. These observations establish the existence of ubiquitous quantum hair due to gravitational effects.

preprint2021arXiv

From Genotype to Phenotype: polygenic prediction of complex human traits

Decoding the genome confers the capability to predict characteristics of the organism(phenotype) from DNA (genotype). We describe the present status and future prospects of genomic prediction of complex traits in humans. Some highly heritable complex phenotypes such as height and other quantitative traits can already be predicted with reasonable accuracy from DNA alone. For many diseases, including important common conditions such as coronary artery disease, breast cancer, type I and II diabetes, individuals with outlier polygenic scores (e.g., top few percent) have been shown to have 5 or even 10 times higher risk than average. Several psychiatric conditions such as schizophrenia and autism also fall into this category. We discuss related topics such as the genetic architecture of complex traits, sibling validation of polygenic scores, and applications to adult health, in vitro fertilization (embryo selection), and genetic engineering.

preprint2020arXiv

Discrete Hilbert Space, the Born Rule, and Quantum Gravity

Quantum gravitational effects suggest a minimal length, or spacetime interval, of order the Planck length. This in turn suggests that Hilbert space itself may be discrete rather than continuous. One implication is that quantum states with norm below some very small threshold do not exist. The exclusion of what Everett referred to as maverick branches is necessary for the emergence of the Born Rule in no collapse quantum mechanics. We discuss this in the context of quantum gravity, showing that discrete models (such as simplicial or lattice quantum gravity) indeed suggest a discrete Hilbert space with minimum norm. These considerations are related to the ultimate level of fine-graining found in decoherent histories (of spacetime geometry plus matter fields) produced by quantum gravity.

preprint2020arXiv

Theta terms and asymptotic behavior of gauge potentials in (3+1) dimensions

We describe paths in the configuration space of (3+1) dimensional QED whose relative quantum phase (or relative phase in the functional integral) depends on the value of the theta angle. The final configurations on the two paths are related by a gauge transformation but differ in magnetic helicity or Chern-Simons number. Such configurations must exhibit gauge potentials that fall off no faster than 1/r in some region of finite solid angle, although they need not have net magnetic charge (i.e., are not magnetic monopoles). The relative phase is proportional to theta times the difference in Chern-Simons number. We briefly discuss some possible implications for QCD and the strong CP problem.

preprint2016arXiv

The measure problem in no-collapse (many worlds) quantum mechanics

We explain the measure problem (cf. origin of the Born probability rule) in no-collapse quantum mechanics. Everett defined maverick branches of the state vector as those on which the usual Born probability rule fails to hold -- these branches exhibit highly improbable behaviors, including possibly the breakdown of decoherence or even the absence of an emergent semi-classical reality. Derivations of the Born rule which originate in decision theory or subjective probability (i.e., the reasoning of individual observers) do not resolve this problem, because they are circular: they assume, a priori, that the observer occupies a non-maverick branch. An ab initio probability measure is sometimes assumed to explain why we do not occupy a maverick branch. This measure is constrained by, e.g., Gleason's Theorem or envariance to be the usual Hilbert measure. However, this ab initio measure ultimately governs the allocation of a self or a consciousness to a particular branch of the wavefunction, and hence invokes primitives which lie beyond the Everett wavefunction and beyond what we usually think of as physics. The significance of this leap has been largely overlooked, but requires serious scrutiny.

preprint2015arXiv

Astrophysical Constraints on Dark Energy

Dark energy (i.e., a cosmological constant) leads, in the Newtonian approximation, to a repulsive force which grows linearly with distance and which can have astrophysical consequences. For example, the dark energy force overcomes the gravitational attraction from an isolated object (e.g., dwarf galaxy) of mass $10^7 M_\odot$ at a distance of $~ 23$ kpc. Observable velocities of bound satellites (rotation curves) could be significantly affected, and therefore used to measure or constrain the dark energy density. Here, {\it isolated} means that the gravitational effect of large nearby galaxies (specifically, of their dark matter halos) is negligible; examples of isolated dwarf galaxies include Antlia or DDO 1903

preprint2015arXiv

Determination of Nonlinear Genetic Architecture using Compressed Sensing

We introduce a statistical method that can reconstruct nonlinear genetic models (i.e., including epistasis, or gene-gene interactions) from phenotype-genotype (GWAS) data. The computational and data resource requirements are similar to those necessary for reconstruction of linear genetic models (or identification of gene-trait associations), assuming a condition of generalized sparsity, which limits the total number of gene-gene interactions. An example of a sparse nonlinear model is one in which a typical locus interacts with several or even many others, but only a small subset of all possible interactions exist. It seems plausible that most genetic architectures fall in this category. Our method uses a generalization of compressed sensing (L1-penalized regression) applied to nonlinear functions of the sensing matrix. We give theoretical arguments suggesting that the method is nearly optimal in performance, and demonstrate its effectiveness on broad classes of nonlinear genetic models using both real and simulated human genomes.

preprint2015arXiv

Instability of Quantum de Sitter Spacetime

Quantized fields (e.g., the graviton itself) in de Sitter (dS) spacetime lead to particle production: specifically, we consider a thermal spectrum resulting from the dS (horizon) temperature. The energy required to excite these particles reduces slightly the rate of expansion and eventually modifies the semiclassical spacetime geometry. The resulting manifold no longer has constant curvature nor time reversal invariance, and back-reaction renders the classical dS background unstable to perturbations. In the case of AdS, there exists a global static vacuum state; in this state there is no particle production and the analogous instability does not arise.

preprint2015arXiv

Locality and Nonlinear Quantum Mechanics

Nonlinear modifications of quantum mechanics generically lead to nonlocal effects which violate relativistic causality. We study these effects using the functional Schrodinger equation for quantum fields and identify a type of nonlocality which causes nearly instantaneous entanglement of spacelike separated systems. We describe a simple example involving widely separated wave-packet (coherent) states, showing that nonlinearity in the Schrodinger evolution causes spacelike entanglement, even in free field theory.

preprint2014arXiv

Application of compressed sensing to genome wide association studies and genomic selection

We show that the signal-processing paradigm known as compressed sensing (CS) is applicable to genome-wide association studies (GWAS) and genomic selection (GS). The aim of GWAS is to isolate trait-associated loci, whereas GS attempts to predict the phenotypic values of new individuals on the basis of training data. CS addresses a problem common to both endeavors, namely that the number of genotyped markers often greatly exceeds the sample size. We show using CS methods and theory that all loci of nonzero effect can be identified (selected) using an efficient algorithm, provided that they are sufficiently few in number (sparse) relative to sample size. For heritability h2 = 1, there is a sharp phase transition to complete selection as the sample size is increased. For heritability values less than one, complete selection can still occur although the transition is smoothed. The transition boundary is only weakly dependent on the total number of genotyped markers. The crossing of a transition boundary provides an objective means to determine when true effects are being recovered; we discuss practical methods for detecting the boundary. For h2 = 0.5, we find that a sample size that is thirty times the number of nonzero loci is sufficient for good recovery.

preprint2014arXiv

Does the BICEP2 Observation of Cosmological Tensor Modes Imply an Era of Nearly Planckian Energy Densities?

BICEP2 observations, interpreted most simply, suggest an era of inflation with energy densities of order ($10^{16}\, {\rm GeV})^4$, not far below the Planck density. However, models of TeV gravity with large extra dimensions might allow a very different interpretation involving much more modest energy scales. We discuss the viability of inflation in such models, and conclude that existing scenarios do not provide attractive alternatives to single field inflation in four dimensions. Because the detection of tensor modes strengthens our confidence that inflation occurred, it disfavors models of large extra dimensions, at least for the moment.

preprint2014arXiv

On the genetic architecture of intelligence and other quantitative traits

How do genes affect cognitive ability or other human quantitative traits such as height or disease risk? Progress on this challenging question is likely to be significant in the near future. I begin with a brief review of psychometric measurements of intelligence, introducing the idea of a "general factor" or g score. The main results concern the stability, validity (predictive power), and heritability of adult g. The largest component of genetic variance for both height and intelligence is additive (linear), leading to important simplifications in predictive modeling and statistical estimation. Due mainly to the rapidly decreasing cost of genotyping, it is possible that within the coming decade researchers will identify loci which account for a significant fraction of total g variation. In the case of height analogous efforts are well under way. I describe some unpublished results concerning the genetic architecture of height and cognitive ability, which suggest that roughly 10k moderately rare causal variants of mostly negative effect are responsible for normal population variation. Using results from Compressed Sensing (L1-penalized regression), I estimate the statistical power required to characterize both linear and nonlinear models for quantitative traits. The main unknown parameter s (sparsity) is the number of loci which account for the bulk of the genetic variation. The required sample size is of order 100s, or roughly a million in the case of cognitive ability.

preprint2013arXiv

Factorization of unitarity and black hole firewalls

Unitary black hole evaporation necessarily involves a late-time superposition of decoherent states, including states describing distinct spacetimes (e.g., different center of mass trajectories of the black hole). Typical analyses of the black hole information problem, including the argument for the existence of firewalls, assume approximate unitarity ("factorization of unitarity") on each of the decoherent spacetimes. This factorization assumption is non-trivial, and indeed may be incorrect. We describe an ansatz for the radiation state that violates factorization and which allows unitarity and the equivalence principle to coexist (no firewall). Unitarity without factorization provides a natural realization of the idea of black hole complementarity.

preprint2013arXiv

Macroscopic superpositions and black hole unitarity

We discuss the black hole information problem, including the recent claim that unitarity requires a horizon firewall, emphasizing the role of decoherence and macroscopic superpositions. We consider the formation and evaporation of a large black hole as a quantum amplitude, and note that during intermediate stages (e.g., after the Page time), the amplitude is a superposition of macroscopically distinct (and decohered) spacetimes, with the black hole itself in different positions on different branches. Small but semiclassical observers (who are themselves part of the quantum amplitude) that fall into the hole on one branch will miss it entirely on other branches and instead reach future infinity. This observation can reconcile the subjective experience of an infalling observer with unitarity. We also discuss implications for the nice slice formulation of the information problem, and to complementarity.

preprint2012arXiv

Quantum fluctuations in planar domain-wall space-times: A possible origin of primordial preferred direction

We study the gravitational effects of a planar domain wall on quantum fluctuations of a massless scalar field during inflation. By obtaining an exact solution of the scalar field equation in de Sitter space, we show that the gravitational effects of the domain wall break the rotational invariance of the primordial power spectrum without affecting the translational invariance. The strength of rotational violation is determined by one dimensionless parameter $β$, which is a function of two physical parameters, the domain wall surface tension $σ$ and cosmological constant $Λ$. In the limit of small $β$, the leading effect of rotational violation of the primordial power spectrum is scale-invariant.

preprint2010arXiv

Asymptotic Safety, Singularities, and Gravitational Collapse

Asymptotic safety (an ultraviolet fixed point with finite-dimensional critical surface) offers the possibility that a predictive theory of quantum gravity can be obtained from the quantization of classical general relativity. However, it is unclear what becomes of the singularities of classical general relativity, which, it is hoped, might be resolved by quantum effects. We study dust collapse with a running gravitational coupling and find that a future singularity can be avoided if the coupling becomes exactly zero at some finite energy scale. The singularity can also be avoided (pushed off to infinite proper time) if the coupling approaches zero sufficiently rapidly at high energies. However, the evolution deduced from perturbation theory still implies a singularity at finite proper time.

preprint2010arXiv

Dark Energy, with Signatures

We propose a class of simple dark energy models which predict a late-time dark radiation component and a distinctive time-dependent equation of state $w(z)$ for redshift $z < 3$. The dark energy field can be coupled strongly enough to Standard Model particles to be detected in colliders, and the model requires only modest additional particle content and little or no fine-tuning other than a new energy scale of order milli-electron volts.

preprint2010arXiv

Data Mining the University: College GPA Predictions from SAT Scores

We analyze a data set comprised of academic records of undergraduates at the University of Oregon from 2000-2004. We find correlations of roughly 0.35 to 0.5 between SAT scores and upper division, in-major GPA (henceforth, GPA). Interestingly, low SAT scores do not preclude high performance in most majors. That is, the distribution of SAT scores after conditioning on high GPA (e.g., 3.5 or even 4.0) typically extends below 1000 (the average among test takers). We hypothesize that "overachievers" overcome cognitive deficits through hard work, and discuss to what extent they can be identified from high school records. Only a few majors seem to exhibit a "cognitive threshold" -- such that high GPA (mastery of the subject matter) is very unlikely below a certain SAT threshold (i.e., no matter how dedicated or hard working the student). Our results suggest that almost any student admitted to university can achieve academic success, if they work hard enough. In addition to our primary result, we find that the best predictor of GPA is a roughly equally weighted sum of SAT and high school GPA, measured in standard deviation units. Using a sub-population of honors college students, we can estimate how students at elite universities would fare at a typical state university, allowing us to comment on issues such as grade inflation. Finally, we observe that 1) SAT scores fluctuate little on retest (very high reliability), 2) SAT and GRE scores (where available) correlate at roughly 0.75 (consistent with the notion that both tests measure a stable general cognitive ability) and 3) the SAT distribution of students that obtained a degree does not differ substantially from that of the entering class.

preprint2010arXiv

Grand unification through gravitational effects

We systematically study the unification of gauge couplings in the presence of (one or more) effective dimension-5 operators cHGG/4MPl, induced into the grand unified theory by gravitational interactions at the Planck scale MPl. These operators alter the usual condition for gauge coupling unification, which can, depending on the Higgs content H and vacuum expectation value, result in unification at scales MX significantly different than naively expected. We find non-supersymmetric models of SU(5) and SO(10) unification, with natural Wilson coefficients c, that easily satisfy the constraints from proton decay. Furthermore, gauge coupling unification at scales as high as the Planck scale seems feasible, possibly hinting at simultaneous unification of gauge and gravitational interactions. In the Appendix we work out the group theoretical aspects of this scenario for SU(5) and SO(10) unified groups in detail; this material is also relevant in the analysis of non-universal gaugino masses obtained from supergravity.

preprint2010arXiv

Nonlinear Psychometric Thresholds for Physics and Mathematics

We analyze 5 years of student records at the University of Oregon to estimate the probability of success (as defined by superior undergraduate record; sufficient for admission to graduate school) in Physics and Mathematics as a function of SAT-M score. We find evidence of a nonlinear threshold: below SAT-M score of roughly 600, the probability of success is very low. Interestingly, no similar threshold exists in other majors, such as Sociology, History, English or Biology, whether on SAT combined, SAT-R or SAT-M. Our findings have significant implications for the demographic makeup of graduate populations in mathematically intensive subjects, given the current distribution of SAT-M scores.

preprint2010arXiv

Physical consequences of the QED theta angle

We describe a simple gedanken experiment which illustrates the physical effects of the QED theta angle, a fundamental parameter of Nature that has yet to be measured. The effects are manifest in quantum phases analogous to those in the Aharonov-Bohm effect, although they are not intrinsically topological. We also derive the quantum phases using a functional Schrodinger approach, and generalize the results to non-Abelian gauge theories.

preprint2010arXiv

The black hole information paradox and macroscopic superpositions

We investigate the experimental capabilities required to test whether black holes destroy information. We show that an experiment capable of illuminating the information puzzle must necessarily be able to detect or manipulate macroscopic superpositions (i.e., Everett branches). Hence, it could also address the fundamental question of decoherence versus wavefunction collapse.

preprint2009arXiv

Dark radiation as a signature of dark energy

We propose a simple dark energy model with the following properties: the model predicts a late-time dark radiation component that is not ruled out by current observational data, but which produces a distinctive time-dependent equation of state w(z) for z < 3. The dark energy field can be coupled strongly enough to Standard Model particles to be detected in colliders, and the model requires only modest additional particle content and little or no fine-tuning other than a new energy scale of order milli-electron volts.

preprint2009arXiv

Monsters, black holes and the statistical mechanics of gravity

We review the construction of monsters in classical general relativity. Monsters have finite ADM mass and surface area, but potentially unbounded entropy. From the curved space perspective they are objects with large proper volume that can be glued on to an asymptotically flat space. At no point is the curvature or energy density required to be large in Planck units, and quantum gravitational effects are, in the conventional effective field theory framework, small everywhere. Since they can have more entropy than a black hole of equal mass, monsters are problematic for certain interpretations of black hole entropy and the AdS/CFT duality. In the second part of the paper we review recent developments in the foundations of statistical mechanics which make use of properties of high-dimensional (Hilbert) spaces. These results primarily depend on kinematics -- essentially, the geometry of Hilbert space -- and are relatively insensitive to dynamics. We discuss how this approach might be adopted as a basis for the statistical mechanics of gravity. Interestingly, monsters and other highly entropic configurations play an important role.

preprint2009arXiv

On the sign problem in dense QCD

We investigate the Euclidean path integral formulation of QCD at finite baryon density and temperature. We show that the partition function Z can be written as a difference between two sums Z+ and Z-, each of which defines a partition function with positive weights. We call the sign problem severe if the ratio Z-/Z+ is nonzero in the infinite volume limit. This occurs only if, and generically always if, the associated free energy densities F+ and F- are equal in this limit. We present strong evidence here that the sign problem is severe at almost all points in the phase diagram, with the exception of special cases like exactly zero chemical potential (ordinary QCD), which requires a particular order of limits. Part of our reasoning is based on the analyticity of free energy densities within their open phase regions. Finally, we describe a Monte Carlo technique to simulate finite-density QCD in regions where Z-/Z+ is small.

preprint2005arXiv

Minimum Length from First Principles

We show that no device or gedanken experiment is capable of measuring a distance less than the Planck length. By "measuring a distance less than the Planck length" we mean, technically, resolve the eigenvalues of the position operator to within that accuracy. The only assumptions in our argument are causality, the uncertainty principle from quantum mechanics and a dynamical criteria for gravitational collapse from classical general relativity called the hoop conjecture. The inability of any gedanken experiment to measure a sub-Planckian distance suggests the existence of a minimal length.

preprint2004arXiv

Opening the Window for Technicolor

Recently a new class of technicolor models are proposed, using technifermions of symmetric second-rank tensor. In the models, one can make reasonable estimates of physical quantities like the Higgs mass and the size of oblique corrections, using a correspondence to super Yang-Mills theory in the Corrigan-Ramond limit. The models predict a surprisingly light Higgs of mass, $m_H=150\sim 500 {\rm GeV}$ and have naturally small $S$ parameter.

preprint2003arXiv

High Temperature Superfluid and Feshbach Resonance

We study an effective field theory describing cold fermionic atoms near a Feshbach resonance. The theory gives a unique description of the dynamics in the limit that the energy of the Feshbach resonance is tuned to be twice that of the Fermi surface. We show that in this limit the zero temperature superfluid condensate is of order the Fermi energy, and obtain a critical temperature $T_C \simeq 0.43 T_F$

preprint1999arXiv

Magnetic Interactions, the Renormalization Group and Color Superconductivity in High Density QCD

We investigate the effect of long range magnetic interactions on the renormalization group (RG) evolution of local Cooper pairing interactions near the Fermi surface in high density QCD. We use an explicit cut-off on momentum modes, with special emphasis on screening effects such as Landau damping, to derive the RG equations in a gauge invariant, weak coupling expansion. We obtain the Landau pole $Δ\sim μg^{-5} \exp(- \frac{3 π^2}{\sqrt{2} g})$, although the structure of our equations differs from previous results. We also investigate the gap equation, including condensates of higher angular momentum. We show that rotational invariance is unbroken at asymptotically high density, and verify that $Δ$ is the correct value of the gap when higher modes are included in the analysis.

preprint1993arXiv

Spacetime Symmetries and Semiclassical Amplitudes

We examine the spacetime symmetries of forward $2 \rightarrow 2$ scattering. These symmetries have non-trivial consequences for any class of configurations which might dominate the amplitude in the semiclassical approximation. We derive some dynamical results regarding the stability of configurations which arise solely from reflection symmetry and positivity of the (Euclidean) path integral action. We show that in the case of on-shell scattering, any semiclassical configuration must have $O(3)$ symmetric asymptotic behavior. We argue that non--$O(3)$ symmetric configurations are irrelevant to the semiclassical exponent.