Datasets:

CCRss

/

arxiv_papers_cs

like 1

research on the laws of multimodal perception and cognition from a cross-cultural perspective -- taking overseas chinese gardens as an example

2312.17642

this study aims to explore the complex relationship between perceptual andcognitive interactions in multimodal data analysis,with a specific emphasis onspatial experience design in overseas chinese gardens. it is found thatevaluation content and images on social media can reflect individuals' concernsand sentiment responses, providing a rich data base for cognitive research thatcontains both sentimental and image-based cognitive information. leveragingdeep learning techniques, we analyze textual and visual data from social media,thereby unveiling the relationship between people's perceptions and sentimentcognition within the context of overseas chinese gardens. in addition, ourstudy introduces a multi-agent system (mas)alongside ai agents. each agentexplores the laws of aesthetic cognition through chat scene simulation combinedwith web search. this study goes beyond the traditional approach of translatingperceptions into sentiment scores, allowing for an extension of the researchmethodology in terms of directly analyzing texts and digging deeper intoopinion data. this study provides new perspectives for understanding aestheticexperience and its impact on architecture and landscape design across diversecultural contexts, which is an essential contribution to the field of culturalcommunication and aesthetic understanding.

cs.ai cs.cl cs.cv cs.si

nan

['chen', 'yao', 'zhao', 'xu', 'zhang', 'mao']

https://arxiv.org/abs/2312.17642

b-it-bots robocup@work team description paper 2023

2312.17643

this paper presents the b-it-bots robocup@work team and its current hardwareand functional architecture for the kuka youbot robot. we describe theunderlying software framework and the developed capabilities required foroperating in industrial environments including features such as reliable andprecise navigation, flexible manipulation, robust object recognition and taskplanning. new developments include an approach to grasp vertical objects,placement of objects by considering the empty space on a workstation, and theprocess of porting our code to ros2.

cs.ro

nan

['patel', 'kalagaturu', 'mannava', 'selvaraju', 'shinde', 'bakaraniya', 'nair', 'wasil', 'thoduka', 'awaad', 'schneider', 'hochgeschwender', 'plöger']

https://arxiv.org/abs/2312.17643

bridging modality gap for visual grounding with effecitve cross-modal distillation

2312.17648

visual grounding aims to align visual information of specific regions ofimages with corresponding natural language expressions. current visualgrounding methods leverage pre-trained visual and language backbones separatelyto obtain visual features and linguistic features. although these two types offeatures are then fused via delicately designed networks, the heterogeneity ofthe features makes them inapplicable for multi-modal reasoning. this problemarises from the domain gap between the single-modal pre-training backbone usedin current visual grounding methods, which can hardly be overcome by thetraditional end-to-end training method. to alleviate this, our work proposes anempowering pre-trained model for visual grounding (epmvg) framework, whichdistills a multimodal pre-trained model to guide the visual grounding task.epmvg is based on a novel cross-modal distillation mechanism, which caneffectively introduce the consistency information of images and texts in thepre-trained model, to reduce the domain gap existing in the backbone networks,thereby improving the performance of the model in the visual grounding task.extensive experiments are carried out on five conventionally used datasets, andresults demonstrate that our method achieves better performance thanstate-of-the-art methods.

cs.cv cs.ai

nan

['wang', 'hu', 'liu', 'wu', 'qiu', 'cai']

https://arxiv.org/abs/2312.17648

investigating the effects of sparse attention on cross-encoders

2312.17649

cross-encoders are effective passage and document re-rankers but lessefficient than other neural or classic retrieval models. a few previous studieshave applied windowed self-attention to make cross-encoders more efficient.however, these studies did not investigate the potential and limits ofdifferent attention patterns or window sizes. we close this gap andsystematically analyze how token interactions can be reduced without harmingthe re-ranking effectiveness. experimenting with asymmetric attention anddifferent window sizes, we find that the query tokens do not need to attend tothe passage or document tokens for effective re-ranking and that very smallwindow sizes suffice. in our experiments, even windows of 4 tokens still yieldeffectiveness on par with previous cross-encoders while reducing the memoryrequirements to at most 78% / 41% and being 1% / 43% faster at inference timefor passages / documents.

cs.ir

nan

['schlatt', 'fröbe', 'hagen']

https://arxiv.org/abs/2312.17649

grasping, part identification, and pose refinement in one shot with a tactile gripper

2312.17650

the rise in additive manufacturing comes with unique opportunities andchallenges. rapid changes to part design and massive part customizationdistinctive to 3d-print (3dp) can be easily achieved. customized parts that areunique, yet exhibit similar features such as dental moulds, shoe insoles, orengine vanes could be industrially manufactured with 3dp. however, theopportunity for massive part customization comes with unique challenges for theexisting production paradigm of robotics applications, as the current roboticsparadigm for part identification and pose refinement is repetitive, wheredata-driven and object-dependent approaches are often used. thus, a bottleneckexists in robotics applications for 3dp parts where massive customization isinvolved, as it is difficult for feature-based deep learning approaches todistinguish between similar parts such as shoe insoles belonging to differentpeople. as such, we propose a method that augments patterns on 3dp parts sothat grasping, part identification, and pose refinement can be executed in oneshot with a tactile gripper. we also experimentally evaluate our approach fromthree perspectives, including real insertion tasks that mimic robotic sortingand packing, and achieved excellent classification results, a high insertionsuccess rate of 95%, and a sub-millimeter pose refinement accuracy.

cs.ro

nan

['lim', 'pham']

https://arxiv.org/abs/2312.17650

larp: language-agent role play for open-world games

2312.17653

language agents have shown impressive problem-solving skills within definedsettings and brief timelines. yet, with the ever-evolving complexities ofopen-world simulations, there's a pressing need for agents that can flexiblyadapt to complex environments and consistently maintain a long-term memory toensure coherent actions. to bridge the gap between language agents andopen-world games, we introduce language agent for role-playing (larp), whichincludes a cognitive architecture that encompasses memory processing and adecision-making assistant, an environment interaction module with afeedback-driven learnable action space, and a postprocessing method thatpromotes the alignment of various personalities. the larp framework refinesinteractions between users and agents, predefined with unique backgrounds andpersonalities, ultimately enhancing the gaming experience in open-worldcontexts. furthermore, it highlights the diverse uses of language models in arange of areas such as entertainment, education, and various simulationscenarios. the project page is released at https://miao-ai-lab.github.io/larp/.

cs.ai

nan

['yan', 'li', 'zhang', 'wang', 'yang', 'yan']

https://arxiv.org/abs/2312.17653

visual point cloud forecasting enables scalable autonomous driving

2312.17655

in contrast to extensive studies on general vision, pre-training for scalablevisual autonomous driving remains seldom explored. visual autonomous drivingapplications require features encompassing semantics, 3d geometry, and temporalinformation simultaneously for joint perception, prediction, and planning,posing dramatic challenges for pre-training. to resolve this, we bring up a newpre-training task termed as visual point cloud forecasting - predicting futurepoint clouds from historical visual input. the key merit of this task capturesthe synergic learning of semantics, 3d structures, and temporal dynamics. henceit shows superiority in various downstream tasks. to cope with this newproblem, we present vidar, a general model to pre-train downstream visualencoders. it first extracts historical embeddings by the encoder. theserepresentations are then transformed to 3d geometric space via a novel latentrendering operator for future point cloud prediction. experiments showsignificant gain in downstream tasks, e.g., 3.1% nds on 3d detection, ~10%error reduction on motion forecasting, and ~15% less collision rate onplanning.

cs.cv

nan

['yang', 'chen', 'sun', 'li']

https://arxiv.org/abs/2312.17655

solar radiation prediction in the uteq based on machine learning models

2312.17659

this research explores the effectiveness of various machine learning (ml)models used to predicting solar radiation at the central campus of the statetechnical university of quevedo (uteq). the data was obtained from apyranometer, strategically located in a high area of the campus. thisinstrument continuously recorded solar irradiance data since 2020, offering acomprehensive dataset encompassing various weather conditions and temporalvariations. after a correlation analysis, temperature and the time of day wereidentified as the relevant meteorological variables that influenced the solarirradiance. different machine learning algorithms such as linear regression,k-nearest neighbors, decision tree, and gradient boosting were compared usingthe evaluation metrics mean squared error (mse), root mean squared error(rmse), mean absolute error (mae), and the coefficient of determination($r^2$). the study revealed that gradient boosting regressor exhibited superiorperformance, closely followed by the random forest regressor. these modelseffectively captured the non-linear patterns in solar radiation, as evidencedby their low mse and high $r^2$ values. with the aim of assess the performanceof our ml models, we developed a web-based tool for the solar radiationforecasting in the uteq available athttp://https://solarradiationforecastinguteq.streamlit.app/. the resultsobtained demonstrate the effectiveness of our ml models in solar radiationprediction and contribute a practical utility in real-time solar radiationforecasting, aiding in efficient solar energy management.

cs.lg

nan

['troncoso', 'quijije', 'oviedo', 'zambrano-vega']

https://arxiv.org/abs/2312.17659

gemini in reasoning: unveiling commonsense in multimodal large language models

2312.17661

the burgeoning interest in multimodal large language models (mllms), such asopenai's gpt-4v(ision), has significantly impacted both academic and industrialrealms. these models enhance large language models (llms) with advanced visualunderstanding capabilities, facilitating their application in a variety ofmultimodal tasks. recently, google introduced gemini, a cutting-edge mllmdesigned specifically for multimodal integration. despite its advancements,preliminary benchmarks indicate that gemini lags behind gpt models incommonsense reasoning tasks. however, this assessment, based on a limiteddataset (i.e., hellaswag), does not fully capture gemini's authenticcommonsense reasoning potential. to address this gap, our study undertakes athorough evaluation of gemini's performance in complex reasoning tasks thatnecessitate the integration of commonsense knowledge across modalities. wecarry out a comprehensive analysis of 12 commonsense reasoning datasets,ranging from general to domain-specific tasks. this includes 11 datasetsfocused solely on language, as well as one that incorporates multimodalelements. our experiments across four llms and two mllms demonstrate gemini'scompetitive commonsense reasoning capabilities. additionally, we identifycommon challenges faced by current llms and mllms in addressing commonsenseproblems, underscoring the need for further advancements in enhancing thecommonsense reasoning abilities of these models.

cs.cl cs.ai cs.cv

nan

['wang', 'zhao']

https://arxiv.org/abs/2312.17661

fast interpolation of sparse multivariate polynomials

2312.17664

consider a sparse multivariate polynomial f with integer coefficients. assumethat f is represented as a "modular black box polynomial", e.g. via analgorithm to evaluate f at arbitrary integer points, modulo arbitrary positiveintegers. the problem of sparse interpolation is to recover f in its usualsparse representation, as a sum of coefficients times monomials. for the firsttime we present a quasi-optimal algorithm for this task.

cs.sc math.ac

nan

['van der hoeven', 'lecerf']

https://arxiv.org/abs/2312.17664

user strategization and trustworthy algorithms

2312.17666

many human-facing algorithms -- including those that power recommendersystems or hiring decision tools -- are trained on data provided by theirusers. the developers of these algorithms commonly adopt the assumption thatthe data generating process is exogenous: that is, how a user reacts to a givenprompt (e.g., a recommendation or hiring suggestion) depends on the prompt andnot on the algorithm that generated it. for example, the assumption that aperson's behavior follows a ground-truth distribution is an exogeneityassumption. in practice, when algorithms interact with humans, this assumptionrarely holds because users can be strategic. recent studies document, forexample, tiktok users changing their scrolling behavior after learning thattiktok uses it to curate their feed, and uber drivers changing how they acceptand cancel rides in response to changes in uber's algorithm. our work studies the implications of this strategic behavior by modeling theinteractions between a user and their data-driven platform as a repeated,two-player game. we first find that user strategization can actually helpplatforms in the short term. we then show that it corrupts platforms' data andultimately hurts their ability to make counterfactual decisions. we connectthis phenomenon to user trust, and show that designing trustworthy algorithmscan go hand in hand with accurate estimation. finally, we provide aformalization of trustworthiness that inspires potential interventions.

cs.cy cs.gt cs.lg stat.ml

nan

['cen', 'ilyas', 'madry']

https://arxiv.org/abs/2312.17666

aijack: security and privacy risk simulator for machine learning

2312.17667

this paper introduces aijack, an open-source library designed to assesssecurity and privacy risks associated with the training and deployment ofmachine learning models. amid the growing interest in big data and ai,advancements in machine learning research and business are accelerating.however, recent studies reveal potential threats, such as the theft of trainingdata and the manipulation of models by malicious attackers. therefore, acomprehensive understanding of machine learning's security and privacyvulnerabilities is crucial for the safe integration of machine learning intoreal-world products. aijack aims to address this need by providing a librarywith various attack and defense methods through a unified api. the library ispublicly available on github (https://github.com/koukyosyumei/aijack).

cs.lg cs.cr

nan

['takahashi']

https://arxiv.org/abs/2312.17667

vocalics in human-drone interaction

2312.17668

as the presence of flying robots continues to grow in both commercial andprivate sectors, it necessitates an understanding of appropriate methods fornonverbal interaction with humans. while visual cues, such as gesturesincorporated into trajectories, are more apparent and thoroughly researched,acoustic cues have remained unexplored, despite their potential to enhancehuman-drone interaction. given that additional audiovisual and sensoryequipment is not always desired or practicable, and flight noise often maskspotential acoustic communication in rotary-wing drones, such as through aloudspeaker, the rotors themselves offer potential for nonverbal communication.in this paper, quadrotor trajectories are augmented by acoustic informationthat does not visually affect the flight, but adds audible information thatsignificantly facilitates distinctiveness. a user study (n=192) demonstratesthat sonically augmenting the trajectories of two aerial gestures makes themmore easily distinguishable. this enhancement contributes to human-droneinteraction through onboard means, particularly in situations where the humancannot see or look at the drone.

cs.ro

nan

['lieser', 'schwanecke']

https://arxiv.org/abs/2312.17668

qoe-oriented dependent task scheduling under multi-dimensional qos constraints over distributed networks

2312.17674

task scheduling as an effective strategy can improve application performanceon computing resource-limited devices over distributed networks. however,existing evaluation mechanisms fail to depict the complexity of diverseapplications, which involve dependencies among tasks, computing resourcerequirements, and multi-dimensional quality of service (qos) constraints.furthermore, traditional qos-oriented task scheduling strategies struggle tomeet the performance requirements without considering differences insatisfaction and acceptance of application, leading application failures andresource wastage. to tackle these issues, a quality of experience (qoe) costmodel is designed to evaluate application completion, depicting therelationship among application satisfaction, communications, and computingresources in the distributed networks. specifically, considering thesensitivity and preference of qos, we model the different dimensional qosdegradation cost functions for dependent tasks, which are then integrated intothe qoe cost model. based on the qoe model, the dependent task schedulingproblem is formulated as the minimization of overall qoe cost, aiming toimprove the application performance in the distributed networks, which isproven np-hard. moreover, a heuristic hierarchical multi-queue task schedulingalgorithm (hmtsa) is proposed to address the qoe-oriented task schedulingproblem among multiple dependent tasks, which utilizes hierarchical multiplequeues to determine the optimal task execution order and location according todifferent dimensional qos priorities. finally, extensive experimentsdemonstrate that the proposed algorithm can significantly improve thesatisfaction of applications.

cs.dc

nan

['fan', 'cheng', 'chen', 'huang', 'wang']

https://arxiv.org/abs/2312.17674

extending simple monotone drawings

2312.17675

we prove the following variant of levi's enlargement lemma: for an arbitraryarrangement $\mathcal{a}$ of $x$-monotone pseudosegments in the plane and apair of points $a,b$ with distinct $x$-coordinates and not on the samepseudosegment, there exists a simple $x$-monotone curve with endpoints $a,b$that intersects every curve of $\mathcal{a}$ at most once. as a consequence,every simple monotone drawing of a graph can be extended to a simple monotonedrawing of a complete graph. we also show that extending an arrangement ofcylindrically monotone pseudosegments is not always possible; in fact, thecorresponding decision problem is np-hard.

math.co cs.cg

nan

['kynčl', 'soukup']

https://arxiv.org/abs/2312.17675

prompt fuzzing for fuzz driver generation

2312.17677

writing high-quality fuzz drivers is time-consuming and requires a deepunderstanding of the library. however, the performance of the state-of-the-artautomatic fuzz driver generation techniques leaves a lot to be desired. fuzzdrivers, which are learned from consumer code, can reach deep states but arerestricted to their external inputs. on the other hand, interpretative fuzzingcan explore most apis but requires numerous attempts in a vast search space. wepropose promptfuzz, a coverage-guided fuzzer for prompt fuzzing thatiteratively generates fuzz drivers to explore undiscovered library code. toexplore api usage in fuzz drivers during prompt fuzzing, we proposed severalkey techniques: instructive program generation, erroneous program sanitization,coverage-guided prompt mutation, and constrained fuzzer fusion. we implementedpromptfuzz and evaluated its effectiveness on 14 real-world libraries,comparing it against oss-fuzz and the state-of-the-art fuzz driver generationsolution (i.e., hopper). the experiment results demonstrate that the fuzzdrivers generated by promptfuzz achieve higher branch coverage that is 1.61times greater than that of oss-fuzz and 1.67 times greater than that of hopper.in addition, the fuzz drivers generated by promptfuzz successfully detect 33true bugs out of a total of 44 crashes, which were previously unknown, and 27of these bugs have been confirmed by the respective communities.

cs.cr cs.se

nan

['lyu', 'xie', 'chen', 'chen']

https://arxiv.org/abs/2312.17677

data augmentation for supervised graph outlier detection with latent diffusion models

2312.17679

graph outlier detection is a prominent task of research and application inthe realm of graph neural networks. it identifies the outlier nodes thatexhibit deviation from the majority in the graph. one of the fundamentalchallenges confronting supervised graph outlier detection algorithms is theprevalent issue of class imbalance, where the scarcity of outlier instancescompared to normal instances often results in suboptimal performance.conventional methods mitigate the imbalance by reweighting instances in theestimation of the loss function, assigning higher weights to outliers and lowerweights to inliers. nonetheless, these strategies are prone to overfitting andunderfitting, respectively. recently, generative models, especially diffusionmodels, have demonstrated their efficacy in synthesizing high-fidelity images.despite their extraordinary generation quality, their potential in dataaugmentation for supervised graph outlier detection remains largelyunderexplored. to bridge this gap, we introduce godm, a novel data augmentation formitigating class imbalance in supervised graph outlier detection with latentdiffusion models. specifically, our proposed method consists of three keycomponents: (1) variantioanl encoder maps the heterogeneous informationinherent within the graph data into a unified latent space. (2) graph generatorsynthesizes graph data that are statistically similar to real outliers fromlatent space, and (3) latent diffusion model learns the latent spacedistribution of real organic data by iterative denoising. extensive experimentsconducted on multiple datasets substantiate the effectiveness and efficiency ofgodm. the case study further demonstrated the generation quality of oursynthetic data. to foster accessibility and reproducibility, we encapsulategodm into a plug-and-play package and release it at the python package index(pypi).

cs.lg cs.si

nan

['liu', 'zhang', 'hu', 'wang', 'yu']

https://arxiv.org/abs/2312.17679

flowvid: taming imperfect optical flows for consistent video-to-video synthesis

2312.17681

diffusion models have transformed the image-to-image (i2i) synthesis and arenow permeating into videos. however, the advancement of video-to-video (v2v)synthesis has been hampered by the challenge of maintaining temporalconsistency across video frames. this paper proposes a consistent v2v synthesisframework by jointly leveraging spatial conditions and temporal optical flowclues within the source video. contrary to prior methods that strictly adhereto optical flow, our approach harnesses its benefits while handling theimperfection in flow estimation. we encode the optical flow via warping fromthe first frame and serve it as a supplementary reference in the diffusionmodel. this enables our model for video synthesis by editing the first framewith any prevalent i2i models and then propagating edits to successive frames.our v2v model, flowvid, demonstrates remarkable properties: (1) flexibility:flowvid works seamlessly with existing i2i models, facilitating variousmodifications, including stylization, object swaps, and local edits. (2)efficiency: generation of a 4-second video with 30 fps and 512x512 resolutiontakes only 1.5 minutes, which is 3.1x, 7.2x, and 10.5x faster than codef,rerender, and tokenflow, respectively. (3) high-quality: in user studies, ourflowvid is preferred 45.7% of the time, outperforming codef (3.5%), rerender(10.2%), and tokenflow (40.4%).

cs.cv cs.mm

nan

['liang', 'wu', 'wang', 'yu', 'li', 'zhao', 'misra', 'huang', 'zhang', 'vajda', 'marculescu']

https://arxiv.org/abs/2312.17681

latent idiom recognition for a minimalist functional array language using equality saturation

2312.17682

accelerating programs is typically done by recognizing code idioms matchinghigh-performance libraries or hardware interfaces. however, recognizing suchidioms automatically is challenging. the idiom recognition machinery isdifficult to write and requires expert knowledge. in addition, slightvariations in the input program might hide the idiom and defeat the recognizer. this paper advocates for the use of a minimalist functional array languagesupporting a small, but expressive, set of operators. the minimalist designleads to a tiny sets of rewrite rules, which encode the language semantics.crucially, the same minimalist language is also used to encode idioms. thisremoves the need for hand-crafted analysis passes, or for having to learn acomplex domain-specific language to define the idioms. coupled with equality saturation, this approach is able to match the corefunctions from the blas and pytorch libraries on a set of computationalkernels. compared to reference c kernel implementations, the approach producesa geometric mean speedup of 1.46x for c programs using blas, when generatingsuch programs from the high-level minimalist language.

cs.pl

nan

['van der cruysse', 'dubach']

https://arxiv.org/abs/2312.17682

malware detection in iot systems using machine learning techniques

2312.17683

malware detection in iot environments necessitates robust methodologies. thisstudy introduces a cnn-lstm hybrid model for iot malware identification andevaluates its performance against established methods. leveraging k-foldcross-validation, the proposed approach achieved 95.5% accuracy, surpassingexisting methods. the cnn algorithm enabled superior learning modelconstruction, and the lstm classifier exhibited heightened accuracy inclassification. comparative analysis against prevalent techniques demonstratedthe efficacy of the proposed model, highlighting its potential for enhancingiot security. the study advocates for future exploration of svms asalternatives, emphasizes the need for distributed detection strategies, andunderscores the importance of predictive analyses for a more powerful iotsecurity. this research serves as a platform for developing more resilientsecurity measures in iot ecosystems.

cs.cr cs.lg cs.ni

nan

['mehrban', 'ahadian']

https://arxiv.org/abs/2312.17683

multiscale vision transformers meet bipartite matching for efficient single-stage action localization

2312.17686

action localization is a challenging problem that combines detection andrecognition tasks, which are often addressed separately. state-of-the-artmethods rely on off-the-shelf bounding box detections pre-computed at highresolution and propose transformer models that focus on the classification taskalone. such two-stage solutions are prohibitive for real-time deployment. onthe other hand, single-stage methods target both tasks by devoting part of thenetwork (generally the backbone) to sharing the majority of the workload,compromising performance for speed. these methods build on adding a detr headwith learnable queries that, after cross- and self-attention can be sent tocorresponding mlps for detecting a person's bounding box and action. however,detr-like architectures are challenging to train and can incur in bigcomplexity. in this paper, we observe that a straight bipartite matching loss can beapplied to the output tokens of a vision transformer. this results in abackbone + mlp architecture that can do both tasks without the need of an extraencoder-decoder head and learnable queries. we show that a single mvit-sarchitecture trained with bipartite matching to perform both tasks surpassesthe same mvit-s when trained with roi align on pre-computed bounding boxes.with a careful design of token pooling and the proposed training pipeline, ourmvitv2-s model achieves +3 map on ava2.2. w.r.t. the two-stage counterpart.code and models will be released after paper revision.

cs.cv

nan

['ntinou', 'sanchez', 'tzimiropoulos']

https://arxiv.org/abs/2312.17686

simple client-side encryption of personal information with web assembly

2312.17689

the https protocol has enforced a higher level of robustness to severalattacks; however, it is not easy to set up the required certificates onintranets, nor is it effective in the case the server confidentiality is notreliable, as in the case of cloud services, or it could be compromised. asimple method is proposed to encrypt the data on the client side, using webassembly. it never transfers data to the server as clear text. searching fieldsin the server is made possible by an encoding scheme that ensures a stableprefix correspondence between ciphertext and plaintext. the method has beendeveloped for a semantic medical database, and allows accessing personal datausing an additional password while maintaining non-sensitive information inclear form. web assembly has been chosen to guarantee the fast and efficientexecution of encrypting/decrypting operations and because of its characteristicof producing modules that are very robust against reverse engineering. the codeis available at https://github.com/mfalda/client-encdec.

cs.cr

nan

['falda', 'grassi']

https://arxiv.org/abs/2312.17689

a fixed-stress type splitting method for nonlinear poroelasticity

2312.17698

in this paper we consider a nonlinear poroelasticity model that describes thequasi-static mechanical behaviour of a fluid-saturated porous medium whosepermeability depends on the divergence of the displacement. such nonlinearmodels are typically used to study biological structures like tissues, organs,cartilage and bones, which are known for a nonlinear dependence of theirpermeability/hydraulic conductivity on solid dilation. we formulate (extend to the present situation) one of the most popularsplitting schemes, namely the fixed-stress split method for the iterativesolution of the coupled problem. the method is proven to converge linearly forsufficiently small time steps under standard assumptions. the error contractionfactor then is strictly less than one, independent of the lam\'{e} parameters,biot and storage coefficients if the hydraulic conductivity is a strictlypositive, bounded and lipschitz-continuous function.

math.na cs.na

nan

['kraus', 'kumar', 'lymbery', 'radu']

https://arxiv.org/abs/2312.17698

tupy-e: detecting hate speech in brazilian portuguese social media with a novel dataset and comprehensive analysis of models

2312.17704

social media has become integral to human interaction, providing a platformfor communication and expression. however, the rise of hate speech on theseplatforms poses significant risks to individuals and communities. detecting andaddressing hate speech is particularly challenging in languages like portuguesedue to its rich vocabulary, complex grammar, and regional variations. toaddress this, we introduce tupy-e, the largest annotated portuguese corpus forhate speech detection. tupy-e leverages an open-source approach, fosteringcollaboration within the research community. we conduct a detailed analysisusing advanced techniques like bert models, contributing to both academicunderstanding and practical applications

cs.cl

nan

['oliveira', 'reis', 'ebecken']

https://arxiv.org/abs/2312.17704

the six ways to build trust and reduce privacy concern in a central bank digital currency (cbdc)

2312.17708

central bank digital currencies (cbdcs) have been implemented by only ahandful of countries, but they are being explored by many more. cbdcs aredigital currencies issued and backed by a central bank. consumer trust canencourage or discourage the adoption of this currency, which is also a paymentsystem and a technology. this research attempts to understand consumer trust incbdcs so that the development and adoption stages are more effective andsatisfying for all the stakeholders.

cs.hc cs.cy

10.1007/978-3-031-33665-2_6

['zarifis', 'cheng']

https://arxiv.org/abs/2312.17708

principled gradient-based markov chain monte carlo for text generation

2312.17710

recent papers have demonstrated the possibility of energy-based textgeneration by adapting gradient-based sampling algorithms, a paradigm of mcmcalgorithms that promises fast convergence. however, as we show in this paper,previous attempts on this approach to text generation all fail to samplecorrectly from the target language model distributions. to address thislimitation, we consider the problem of designing text samplers that arefaithful, meaning that they have the target text distribution as its limitingdistribution. we propose several faithful gradient-based sampling algorithms tosample from the target energy-based text distribution correctly, and studytheir theoretical properties. through experiments on various forms of textgeneration, we demonstrate that faithful samplers are able to generate morefluent text while adhering to the control objectives better.

cs.cl cs.lg

nan

['du', 'amini', 'hennigen', 'yu', 'eisner', 'lee', 'cotterell']

https://arxiv.org/abs/2312.17710

an experimental study of satisfaction response: evaluation of online collaborative learning

2312.17722

on the one hand, a growing amount of research discusses support for improvingonline collaborative learning quality, and many indicators are focused toassess its success. on the other hand, thinklets for designing reputable andvaluable collaborative processes have been developed for more than ten years.however, few studies try to apply thinklets to online collaborative learning.this paper introduces thinklets to online collaborative learning andexperimentally tests its effectiveness with participants' responses on theirsatisfaction. yield shift theory (yst), a causal theory explaining innersatisfaction, is adopted. in the experiment, 113 students from universities inbeijing, china are chosen as a sample. they were divided into two groups,collaborating online in a simulated class. then, yst in student groups underonline collaborative learning is validated, a comparison study of onlinecollaborative learning with and without thinklets is implemented, and thesatisfaction response of participants are analyzed. as a result of thiscomparison, yst is proved applicable in this context, and satisfaction ishigher in online collaborative learning with thinklets.

cs.hc

10.19173/irrodl.v17i1.2110

['cheng', 'wang', 'huang', 'zarifis']

https://arxiv.org/abs/2312.17722

comparing effectiveness and efficiency of interactive application security testing (iast) and runtime application self-protection (rasp) tools in a large java-based system

2312.17726

security resources are scarce, and practitioners need guidance in theeffective and efficient usage of techniques and tools available in thecybersecurity industry. two emerging tool types, interactive applicationsecurity testing (iast) and runtime application self-protection (rasp), havenot been thoroughly evaluated against well-established counterparts such asdynamic application security testing (dast) and static application securitytesting (sast). the goal of this research is to aid practitioners in makinginformed choices about the use of interactive application security testing(iast) and runtime application self-protection (rasp) tools through an analysisof their effectiveness and efficiency in comparison with differentvulnerability detection and prevention techniques and tools. we apply iast andrasp on openmrs, an open-source java-based online application. we compare theefficiency and effectiveness of iast and rasp with techniques applied onopenmrs in prior work. we measure efficiency and effectiveness in terms of thenumber and type of vulnerabilities detected and prevented per hour. our studyshows iast performed relatively well compared to other techniques, performingsecond-best in both efficiency and effectiveness. iast detected eight top-10owasp security risks compared to nine by smpt and seven for empt, dast, andsast. iast found more vulnerabilities than smpt. the efficiency of iast (2.14vph) is second to only empt (2.22 vph). these findings imply that our studybenefited from using iast when conducting black-box security testing. in thecontext of a large, enterprise-scale web application such as openmrs, rasp doesnot replace vulnerability detection, while iast is a powerful tool thatcomplements other techniques.

cs.cr cs.se

nan

['seth', 'bhattacharya', 'elder', 'zahan', 'williams']

https://arxiv.org/abs/2312.17726

murp: multi-agent ultra-wideband relative pose estimation with constrained communications in 3d environments

2312.17731

inter-agent relative localization is critical for many multi-robot systemsoperating in the absence of external positioning infrastructure or priorenvironmental knowledge. we propose a novel inter-agent relative 3d poseestimation system where each participating agent is equipped with severalultra-wideband (uwb) ranging tags. prior work typically supplements noisy uwbrange measurements with additional continuously transmitted data, such asodometry, leading to potential scaling issues with increased team size and/ordecreased communication network capability. by equipping each agent withmultiple uwb antennas, our approach addresses these concerns by using onlylocally collected uwb range measurements, a priori state constraints, anddetections of when said constraints are violated. leveraging our learned meanranging bias correction, we gain a 19% positional error improvement giving usexperimental mean absolute position and heading errors of 0.24m and 9.5 degreesrespectively. when compared to other state-of-the-art approaches, our workdemonstrates improved performance over similar systems, while remainingcompetitive with methods that have significantly higher communication costs.additionally, we make our datasets available.

cs.ro

nan

['fishberg', 'quiter', 'how']

https://arxiv.org/abs/2312.17731

physics-informed graphical neural network for power system state estimation

2312.17738

state estimation is highly critical for accurately observing the dynamicbehavior of the power grids and minimizing risks from cyber threats. however,existing state estimation methods encounter challenges in accurately capturingpower system dynamics, primarily because of limitations in encoding the gridtopology and sparse measurements. this paper proposes a physics-informedgraphical learning state estimation method to address these limitations byleveraging both domain physical knowledge and a graph neural network (gnn). weemploy a gnn architecture that can handle the graph-structured data of powersystems more effectively than traditional data-driven methods. thephysics-based knowledge is constructed from the branch current formulation,making the approach adaptable to both transmission and distribution systems.the validation results of three ieee test systems show that the proposed methodcan achieve lower mean square error more than 20% than the conventionalmethods.

eess.sy cs.sy

nan

['ngo', 'nguyen', 'vu', 'zhang', 'ngo']

https://arxiv.org/abs/2312.17738

learning vision from models rivals learning vision from data

2312.17742

we introduce synclr, a novel approach for learning visual representationsexclusively from synthetic images and synthetic captions, without any realdata. we synthesize a large dataset of image captions using llms, then use anoff-the-shelf text-to-image model to generate multiple images corresponding toeach synthetic caption. we perform visual representation learning on thesesynthetic images via contrastive learning, treating images sharing the samecaption as positive pairs. the resulting representations transfer well to manydownstream tasks, competing favorably with other general-purpose visualrepresentation learners such as clip and dino v2 in image classification tasks.furthermore, in dense prediction tasks such as semantic segmentation, synclroutperforms previous self-supervised methods by a significant margin, e.g.,improving over mae and ibot by 6.2 and 4.3 miou on ade20k for vit-b/16.

cs.cv

nan

['tian', 'fan', 'chen', 'katabi', 'krishnan', 'isola']

https://arxiv.org/abs/2312.17742

k-perm: personalized response generation using dynamic knowledge retrieval and persona-adaptive queries

2312.17748

personalizing conversational agents can enhance the quality of conversationsand increase user engagement. however, they often lack external knowledge toappropriately tend to a user's persona. this is particularly crucial forpractical applications like mental health support, nutrition planning,culturally sensitive conversations, or reducing toxic behavior inconversational agents. to enhance the relevance and comprehensiveness ofpersonalized responses, we propose using a two-step approach that involves (1)selectively integrating user personas and (2) contextualizing the response withsupplementing information from a background knowledge source. we develop k-perm(knowledge-guided personalization with reward modulation), a dynamicconversational agent that combines these elements. k-perm achievesstate-of-the-art performance on the popular focus dataset, containingreal-world personalized conversations concerning global landmarks. we show thatusing responses from k-perm can improve performance in state-of-the-art llms(gpt 3.5) by 10.5%, highlighting the impact of k-perm for personalizingchatbots.

cs.ir

nan

['raj', 'roy', 'gaur']

https://arxiv.org/abs/2312.17748

performance of data networks with random links

adap-org/9909006

we investigate simplified models of computer data networks and examine howthe introduction of additional random links influences the performance of thesenet works. in general, the impact of additional random links on the performanceof the network strongly depends on the routing algorithm used in the network.significant performance gains can be achieved if the routing is based on"geometrical distance" or shortest path reduced table routing. with shortestpath full table routing degradation of performance is observed.

adap-org cs.ni nlin.ao

10.1016/s0378-4754(99)00125-1

['fuks', 'lawniczak']

https://arxiv.org/abs/adap-org/9909006

packet delay in models of data networks

cs/0207092

we investigate individual packet delay in a model of data networks withtable-free, partial table and full table routing. we present analyticalestimation for the average packet delay in a network with small partial routingtable. dependence of the delay on the size of the network and on the size ofthe partial routing table is examined numerically. consequences for networkscalability are discussed.

cs.ni nlin.cg

10.1145/502109.502110

['fuks', 'lawniczak', 'volkov']

https://arxiv.org/abs/cs/0207092

an operational foundation for delimited continuations in the cps hierarchy

cs/0508048

we present an abstract machine and a reduction semantics for thelambda-calculus extended with control operators that give access to delimitedcontinuations in the cps hierarchy. the abstract machine is derived from anevaluator in continuation-passing style (cps); the reduction semantics (i.e., asmall-step operational semantics with an explicit representation of evaluationcontexts) is constructed from the abstract machine; and the control operatorsare the shift and reset family. we also present new applications of delimitedcontinuations in the cps hierarchy: finding list prefixes and normalization byevaluation for a hierarchical language of units and products.

cs.lo cs.pl

10.2168/lmcs-1(2:5)2005

['biernacka', 'biernacki', 'danvy']

https://arxiv.org/abs/cs/0508048

distributed control of microscopic robots in biomedical applications

cs/0611111

current developments in molecular electronics, motors and chemical sensorscould enable constructing large numbers of devices able to sense, compute andact in micron-scale environments. such microscopic machines, of sizescomparable to bacteria, could simultaneously monitor entire populations ofcells individually in vivo. this paper reviews plausible capabilities formicroscopic robots and the physical constraints due to operation in fluids atlow reynolds number, diffusion-limited sensing and thermal noise from brownianmotion. simple distributed controls are then presented in the context ofprototypical biomedical tasks, which require control decisions on millisecondtime scales. the resulting behaviors illustrate trade-offs among speed,accuracy and resource use. a specific example is monitoring for patterns ofchemicals in a flowing fluid released at chemically distinctive sites.information collected from a large number of such devices allows estimatingproperties of cell-sized chemical sources in a macroscopic volume. themicroscopic devices moving with the fluid flow in small blood vessels candetect chemicals released by tissues in response to localized injury orinfection. we find the devices can readily discriminate a single cell-sizedchemical source from the background chemical concentration, providinghigh-resolution sensing in both time and space. by contrast, such a sourcewould be difficult to distinguish from background when diluted throughout theblood volume as obtained with a blood sample.

cs.ro cs.ma

10.1007/978-1-4471-5113-5_8

['hogg']

https://arxiv.org/abs/cs/0611111

an algorithmic and a geometric characterization of coarsening at random

math/0510276

we show that the class of conditional distributions satisfying the coarseningat random (car) property for discrete data has a simple and robust algorithmicdescription based on randomized uniform multicovers: combinatorial objectsgeneralizing the notion of partition of a set. however, the complexity of agiven car mechanism can be large: the maximal "height" of the neededmulticovers can be exponential in the number of points in the sample space. theresults stem from a geometric interpretation of the set of car distributions asa convex polytope and a characterization of its extreme points. the hierarchyof car models defined in this way could be useful in parsimonious statisticalmodelling of car mechanisms, though the results also raise doubts in appliedwork as to the meaningfulness of the car assumption in its full generality.

math.st cs.ai stat.me stat.th

10.1214/07-aos532

['gill', 'grunwald']

https://arxiv.org/abs/math/0510276

matrix-forest theorems

math/0602575

the laplacian matrix of a graph $g$ is $l(g)=d(g)-a(g)$, where $a(g)$ is theadjacency matrix and $d(g)$ is the diagonal matrix of vertex degrees. accordingto the matrix-tree theorem, the number of spanning trees in $g$ is equal to anycofactor of an entry of $l(g)$. a rooted forest is a union of disjoint rootedtrees. we consider the matrix $w(g)=i+l(g)$ and prove that the $(i,j)$-cofactorof $w(g)$ is equal to the number of spanning rooted forests of $g$, in whichthe vertices $i$ and $j$ belong to the same tree rooted at $i$. the determinantof $w(g)$ equals the total number of spanning rooted forests, therefore the$(i,j)$-entry of the matrix $w^{-1}(g)$ can be considered as a measure ofrelative ''forest-accessibility'' of vertex $i$ from $j$ (or $j$ from $i$).these results follow from somewhat more general theorems we prove, whichconcern weighted multigraphs. the analogous theorems for (multi)digraphs arealso established. these results provide a graph-theoretic interpretation forthe adjugate to the laplacian characteristic matrix.

math.co cs.dm math.ra

nan

['chebotarev', 'shamis']

https://arxiv.org/abs/math/0602575

beyond stabilizer codes ii: clifford codes

quant-ph/0010076

knill introduced a generalization of stabilizer codes, in this note calledclifford codes. it remained unclear whether or not clifford codes can besuperior to stabilizer codes. we show that clifford codes are stabilizer codesprovided that the abstract error group has an abelian index group. inparticular, if the errors are modelled by tensor products of pauli matrices,then the associated clifford codes are necessarily stabilizer codes.

quant-ph cs.et

nan

['klappenecker', 'roetteler']

https://arxiv.org/abs/quant-ph/0010076

beyond stabilizer codes i: nice error bases

quant-ph/0010082

nice error bases have been introduced by knill as a generalization of thepauli basis. these bases are shown to be projective representations of finitegroups. we classify all nice error bases of small degree, and all nice errorbases with abelian index groups. we show that in general an index group of anice error basis is necessarily solvable.

quant-ph cs.et

nan

['klappenecker', 'roetteler']

https://arxiv.org/abs/quant-ph/0010082

universal simulation of hamiltonians using a finite set of control operations

quant-ph/0109063

any quantum system with a non-trivial hamiltonian is able to simulate anyother hamiltonian evolution provided that a sufficiently large group of unitarycontrol operations is available. we show that there exist finite groups withthis property and present a sufficient condition in terms of group characters.we give examples of such groups in dimension 2 and 3. furthermore, we show thatit is possible to simulate an arbitrary bipartite interaction by a given oneusing such groups acting locally on the subsystems.

quant-ph cs.et

nan

['wocjan', 'roetteler', 'janzing', 'beth']

https://arxiv.org/abs/quant-ph/0109063

simulating hamiltonians in quantum networks: efficient schemes and complexity bounds

quant-ph/0109088

we address the problem of simulating pair-interaction hamiltonians in n nodequantum networks where the subsystems have arbitrary, possibly different,dimensions. we show that any pair-interaction can be used to simulate any otherby applying sequences of appropriate local control sequences. efficient schemesfor decoupling and time reversal can be constructed from orthogonal arrays.conditions on time optimal simulation are formulated in terms of spectralmajorization of matrices characterizing the coupling parameters. moreover, weconsider a specific system of n harmonic oscillators with bilinear interaction.in this case, decoupling can efficiently be achieved using the combinatorialconcept of difference schemes. for this type of interactions we present optimalschemes for inversion.

quant-ph cs.et

10.1103/physreva.65.042309

['wocjan', 'roetteler', 'janzing', 'beth']

https://arxiv.org/abs/quant-ph/0109088

discrete cosine transforms on quantum computers

quant-ph/0111038

a classical computer does not allow to calculate a discrete cosine transformon n points in less than linear time. this trivial lower bound is no longervalid for a computer that takes advantage of quantum mechanical superposition,entanglement, and interference principles. in fact, we show that it is possibleto realize the discrete cosine transforms and the discrete sine transforms ofsize nxn and types i,ii,iii, and iv with as little as o(log^2 n) operations ona quantum computer, whereas the known fast algorithms on a classical computerneed o(n log n) operations.

quant-ph cs.et

nan

['klappenecker', 'roetteler']

https://arxiv.org/abs/quant-ph/0111038

on the irresistible efficiency of signal processing methods in quantum computing

quant-ph/0111039

we show that many well-known signal transforms allow highly efficientrealizations on a quantum computer. we explain some elementary quantum circuitsand review the construction of the quantum fourier transform. we derive quantumcircuits for the discrete cosine and sine transforms, and for the discretehartley transform. we show that at most o(log^2 n) elementary quantum gates arenecessary to implement any of those transforms for input sequences of length n.

quant-ph cs.et

nan

['klappenecker', 'roetteler']

https://arxiv.org/abs/quant-ph/0111039

engineering functional quantum algorithms

quant-ph/0208130

suppose that a quantum circuit with k elementary gates is known for a unitarymatrix u, and assume that u^m is a scalar matrix for some positive integer m.we show that a function of u can be realized on a quantum computer with at mosto(mk+m^2log m) elementary gates. the functions of u are realized by a genericquantum circuit, which has a particularly simple structure. among otherresults, we obtain efficient circuits for the fractional fourier transform.

quant-ph cs.et

10.1103/physreva.67.010302

['klappenecker', 'roetteler']

https://arxiv.org/abs/quant-ph/0208130

efficient quantum circuits for non-qubit quantum error-correcting codes

quant-ph/0211014

we present two methods for the construction of quantum circuits for quantumerror-correcting codes (qecc). the underlying quantum systems are tensorproducts of subsystems (qudits) of equal dimension which is a prime power. fora qecc encoding k qudits into n qudits, the resulting quantum circuit haso(n(n-k)) gates. the running time of the classical algorithm to compute thequantum circuit is o(n(n-k)^2).

quant-ph cs.et

10.1142/s0129054103002011

['grassl', 'roetteler', 'beth']

https://arxiv.org/abs/quant-ph/0211014

on the monomiality of nice error bases

quant-ph/0301078

unitary error bases generalize the pauli matrices to higher dimensionalsystems. two basic constructions of unitary error bases are known: an algebraicconstruction by knill, which yields nice error bases, and a combinatorialconstruction by werner, which yields shift-and-multiply bases. an open problemposed by schlingemann and werner (seehttp://www.imaph.tu-bs.de/qi/problems/6.html) relates these two constructionsand asks whether each nice error basis is equivalent to a shift-and-multiplybasis. we solve this problem and show that the answer is negative. however, wealso show that it is always possible to find a fairly sparse representation ofa nice error basis.

quant-ph cs.et

nan

['klappenecker', 'roetteler']

https://arxiv.org/abs/quant-ph/0301078

comment on "probabilistic quantum memories"

quant-ph/0303091

this is a comment on two wrong phys. rev. letters papers by c.a.trugenberger. trugenberger claimed that quantum registers could be used asexponentially large "associative" memories. we show that his scheme is nobetter than one where the quantum register is replaced with a classical one ofequal size. we also point out that the holevo bound and more recent bounds on "quantumrandom access codes" pretty much rule out powerful memories (for classicalinformation) based on quantum states.

quant-ph cs.et

10.1103/physrevlett.91.209801

['brun', 'klauck', 'nayak', 'roetteler', 'zalka']

https://arxiv.org/abs/quant-ph/0303091

constructions of mutually unbiased bases

quant-ph/0309120

two orthonormal bases b and b' of a d-dimensional complex inner-product spaceare called mutually unbiased if and only if |<b|b'>|^2=1/d holds for all b in band b' in b'. the size of any set containing (pairwise) mutually unbiased basesof c^d cannot exceed d+1. if d is a power of a prime, then extremal setscontaining d+1 mutually unbiased bases are known to exist. we give a simplifiedproof of this fact based on the estimation of exponential sums. we discussconjectures and open problems concerning the maximal number of mutuallyunbiased bases for arbitrary dimensions.

quant-ph cs.et

nan

['klappenecker', 'roetteler']

https://arxiv.org/abs/quant-ph/0309120

quantum software reusability

quant-ph/0309121

the design of efficient quantum circuits is an important issue in quantumcomputing. it is in general a formidable task to find a highly optimizedquantum circuit for a given unitary matrix. we propose a quantum circuit designmethod that has the following unique feature: it allows to construct efficientquantum circuits in a systematic way by reusing and combining a set of highlyoptimized quantum circuits. specifically, the method realizes a quantum circuitfor a given unitary matrix by implementing a linear combination of representingmatrices of a group, which have known fast quantum circuits. we motivate andillustrate this method by deriving extremely efficient quantum circuits for thediscrete hartley transform and for the fractional fourier transforms. the soundmathematical basis of this design method allows to give meaningful and naturalinterpretations of the resulting circuits. we demonstrate this aspect by givinga natural interpretation of known teleportation circuits.

quant-ph cs.et

nan

['klappenecker', 'roetteler']

https://arxiv.org/abs/quant-ph/0309121

on optimal quantum codes

quant-ph/0312164

we present families of quantum error-correcting codes which are optimal inthe sense that the minimum distance is maximal. these maximum distanceseparable (mds) codes are defined over q-dimensional quantum systems, where qis an arbitrary prime power. it is shown that codes with parameters[[n,n-2d+2,d]]_q exist for all 3 <= n <= q and 1 <= d <= n/2+1. we also presentquantum mds codes with parameters [[q^2,q^2-2d+2,d]]_q for 1 <= d <= q whichadditionally give rise to shortened codes [[q^2-s,q^2-2d+2-s,d]]_q for some s.

quant-ph cs.et

10.1142/s0219749904000079

['grassl', 'beth', 'roetteler']

https://arxiv.org/abs/quant-ph/0312164

remarks on clifford codes

quant-ph/0312228

clifford codes are a class of quantum error control codes that form a naturalgeneralization of stabilizer codes. these codes were introduced in 1996 byknill, but only a single clifford code was known, which is not already astabilizer code. we derive a necessary and sufficient condition that allows todecide when a clifford code is a stabilizer code, and compile a table of alltrue clifford codes for error groups of small order.

quant-ph cs.et

nan

['klappenecker', 'roetteler']

https://arxiv.org/abs/quant-ph/0312228

implementation of group-covariant povms by orthogonal measurements

quant-ph/0407054

we consider group-covariant positive operator valued measures (povms) on afinite dimensional quantum system. following neumark's theorem a povm can beimplemented by an orthogonal measurement on a larger system. accordingly, ourgoal is to find an implementation of a given group-covariant povm by a quantumcircuit using its symmetry. based on representation theory of the symmetrygroup we develop a general approach for the implementation of group-covariantpovms which consist of rank-one operators. the construction relies on a methodto decompose matrices that intertwine two representations of a finite group. wegive several examples for which the resulting quantum circuits are efficient.in particular, we obtain efficient quantum circuits for a class of povmsgenerated by weyl-heisenberg groups. these circuits allow to implement anapproximative simultaneous measurement of the position and crystal momentum ofa particle moving on a cyclic chain.

quant-ph cs.et

10.1063/1.1827924

['decker', 'janzing', 'roetteler']

https://arxiv.org/abs/quant-ph/0407054

efficient decoupling schemes based on hamilton cycles

quant-ph/0408078

decoupling the interactions in a spin network governed by a pair-interactionhamiltonian is a well-studied problem. combinatorial schemes for decoupling andfor manipulating the couplings of hamiltonians have been developed which useselective pulses. in this paper we consider an additional requirement on thesepulse sequences: as few {\em different} control operations as possible shouldbe used. this requirement is motivated by the fact that optimizing eachindividual selective pulse will be expensive, i. e., it is desirable to use asfew different selective pulses as possible. for an arbitrary $d$-dimensionalsystem we show that the ability to implement only two control operations issufficient to turn off the time evolution. in case of a bipartite system withlocal control we show that four different control operations are sufficient.turning to networks consisting of several $d$-dimensional nodes which aregoverned by a pair-interaction hamiltonian, we show that decoupling can beachieved if one is able to control a number of different control operationswhich is logarithmic in the number of nodes.

quant-ph cs.et

nan

['roetteler']

https://arxiv.org/abs/quant-ph/0408078

equivalence of decoupling schemes and orthogonal arrays

quant-ph/0409135

we consider the problem of switching off unwanted interactions in a givenmulti-partite hamiltonian. this is known to be an important primitive inquantum information processing and several schemes have been presented in theliterature to achieve this task. a method to construct decoupling schemes forquantum systems of pairwise interacting qubits was introduced by m.stollsteimer and g. mahler and is based on orthogonal arrays. another approachbased on triples of hadamard matrices that are closed under pointwisemultiplication was proposed by d. leung. in this paper, we show that bothmethods lead to the same class of decoupling schemes. moreover, we establish acharacterization of orthogonal arrays by showing that they are equivalent todecoupling schemes which allow a refinement into equidistant time-slots.furthermore, we show that decoupling schemes for networks of higher-dimensionalquantum systems with t-local hamiltonians can be constructed from classicalerror-correcting codes.

quant-ph cs.et

10.1109/tit.2006.880059

['roetteler', 'wocjan']

https://arxiv.org/abs/quant-ph/0409135

mutually unbiased bases are complex projective 2-designs

quant-ph/0502031

mutually unbiased bases (mubs) are a primitive used in quantum informationprocessing to capture the principle of complementarity. while constructions ofmaximal sets of d+1 such bases are known for systems of prime power dimensiond, it is unknown whether this bound can be achieved for any non-prime powerdimension. in this paper we demonstrate that maximal sets of mubs come with arich combinatorial structure by showing that they actually are the same objectsas the complex projective 2-designs with angle set {0,1/d}. we also give a newand simple proof that symmetric informationally complete povms are complexprojective 2-designs with angle set {1/(d+1)}.

quant-ph cs.et

nan

['klappenecker', 'roetteler']

https://arxiv.org/abs/quant-ph/0502031

thresholds for linear optics quantum computing with photon loss at the detectors

quant-ph/0502101

we calculate the error threshold for the linear optics quantum computingproposal by knill, laflamme and milburn [nature 409, pp. 46--52 (2001)] underan error model where photon detectors have efficiency <100% but all othercomponents -- such as single photon sources, beam splitters and phase shifters-- are perfect and introduce no errors. we make use of the fact that the errormodel induced by the lossy hardware is that of an erasure channel, i.e., theerror locations are always known. using a method based on a markov chaindescription of the error correction procedure, our calculations show that, withthe 7 qubit css quantum code, the gate error threshold for fault tolerantquantum computation is bounded below by a value between 1.78% and 11.5%depending on the construction of the entangling gates.

quant-ph cs.et

10.1103/physreva.72.032307

['silva', 'roetteler', 'zalka']

https://arxiv.org/abs/quant-ph/0502101

new tales of the mean king

quant-ph/0502138

the mean king's problem asks to determine the outcome of a measurement thatis randomly selected from a set of complementary observables. we review thisproblem and offer a combinatorial solution. more generally, we show thatwhenever an affine resolvable design exists, then a state reconstructionproblem similar to the mean king's problem can be defined and solved. as anapplication of this general framework we consider a problem involving threequbits in which the outcome of nine different measurements can be determinedwithout using ancillary qubits. the solution is based on a measurement derivedfrom hadamard designs.

quant-ph cs.et

nan

['klappenecker', 'roetteler']

https://arxiv.org/abs/quant-ph/0502138

on the power of random bases in fourier sampling: hidden subgroup problem in the heisenberg group

quant-ph/0503114

the hidden subgroup problem (hsp) provides a unified framework to studyproblems of group-theoretical nature in quantum computing such as order findingand the discrete logarithm problem. while it is known that fourier samplingprovides an efficient solution in the abelian case, not much is known forgeneral non-abelian groups. recently, some authors raised the question as towhether post-processing the fourier spectrum by measuring in a randomorthonormal basis helps for solving the hsp. several negative results on theshortcomings of this random strong method are known. in this paper however, weshow that the random strong method can be quite powerful under certainconditions on the group g. we define a parameter r(g) for a group g and showthat o((\log |g| / r(g))^2) iterations of the random strong method give enoughclassical information to identify a hidden subgroup in g. we illustrate thepower of the random strong method via a concrete example of the hsp over finiteheisenberg groups. we show that r(g) = \omega(1) for these groups; hence thehsp can be solved using polynomially many random strong fourier samplingsfollowed by a possibly exponential classical post-processing without furtherqueries. the quantum part of our algorithm consists of a polynomial computationfollowed by measuring in a random orthonormal basis. this gives the firstexample of a group where random representation bases do help in solving the hspand for which no explicit representation bases are known that solve the problemwith (\log g)^o(1) fourier samplings. as an interesting by-product of our work,we get an algorithm for solving the state identification problem for a set ofnearly orthogonal pure quantum states.

quant-ph cs.et

nan

['radhakrishnan', 'roetteler', 'sen']

https://arxiv.org/abs/quant-ph/0503114

on approximately symmetric informationally complete positive operator-valued measures and related systems of quantum states

quant-ph/0503239

we address the problem of constructing positive operator-valued measures(povms) in finite dimension $n$ consisting of $n^2$ operators of rank one whichhave an inner product close to uniform. this is motivated by the relatedquestion of constructing symmetric informationally complete povms (sic-povms)for which the inner products are perfectly uniform. however, sic-povms arenotoriously hard to construct and despite some success of constructing themnumerically, there is no analytic construction known. we present twoconstructions of approximate versions of sic-povms, where a small deviationfrom uniformity of the inner products is allowed. the first construction isbased on selecting vectors from a maximal collection of mutually unbiased basesand works whenever the dimension of the system is a prime power. the secondconstruction is based on perturbing the matrix elements of a subset of mutuallyunbiased bases. moreover, we construct vector systems in $\c^n$ which are almost orthogonaland which might turn out to be useful for quantum computation. ourconstructions are based on results of analytic number theory.

quant-ph cs.et

10.1063/1.1998831

['klappenecker', 'roetteler', 'shparlinski', 'winterhof']

https://arxiv.org/abs/quant-ph/0503239

limitations of quantum coset states for graph isomorphism

quant-ph/0511148

it has been known for some time that graph isomorphism reduces to the hiddensubgroup problem (hsp). what is more, most exponential speedups in quantumcomputation are obtained by solving instances of the hsp. a common feature ofthe resulting algorithms is the use of quantum coset states, which encode thehidden subgroup. an open question has been how hard it is to use these statesto solve graph isomorphism. it was recently shown by moore, russell, andschulman that only an exponentially small amount of information is availablefrom one, or a pair of coset states. a potential source of power to exploit areentangled quantum measurements that act jointly on many states at once. we showthat entangled quantum measurements on at least \omega(n log n) coset statesare necessary to get useful information for the case of graph isomorphism,matching an information theoretic upper bound. this may be viewed as a negativeresult because highly entangled measurements seem hard to implement in general.our main theorem is very general and also rules out using joint measurements onfew coset states for some other groups, such as gl(n, f_{p^m}) and g^n where gis finite and satisfies a suitable property.

quant-ph cs.et

nan

['hallgren', 'roetteler', 'sen']

https://arxiv.org/abs/quant-ph/0511148

computing local invariants of qubit systems

quant-ph/9712040

we investigate means to describe the non-local properties of quantum systemsand to test if two quantum systems are locally equivalent. for this we considerquantum systems that consist of several subsystems, especially multiple qubits.we compute invariant polynomials, i. e., polynomial functions of the entries ofthe density operator which are invariant under local unitary operations. as an example, we consider a system of two qubits. we compute the molienseries for the corresponding representation which gives information about thenumber of linearly independent invariants. furthermore, we present a set ofpolynomials which generate all invariants (at least) up to degree 23. finally,the use of invariants to check whether two density operators are locallyequivalent is demonstrated.

quant-ph cs.et

10.1103/physreva.58.1833

['grassl', 'roetteler', 'beth']

https://arxiv.org/abs/quant-ph/9712040

fast quantum fourier transforms for a class of non-abelian groups

quant-ph/9807064

an algorithm is presented allowing the construction of fast fouriertransforms for any solvable group on a classical computer. the specialstructure of the recursion formula being the core of this algorithm makes it agood starting point to obtain systematically fast fourier transforms forsolvable groups on a quantum computer. the inherent structure of the hilbertspace imposed by the qubit architecture suggests to consider groups of order2^n first (where n is the number of qubits). as an example, fast quantumfourier transforms for all 4 classes of non-abelian 2-groups with cyclic normalsubgroup of index 2 are explicitly constructed in terms of quantum circuits.the (quantum) complexity of the fourier transform for these groups of size 2^nis o(n^2) in all cases.

quant-ph cs.et

nan

['pueschel', 'roetteler', 'beth']

https://arxiv.org/abs/quant-ph/9807064