.

Analysis of Lateral Offset Invariance in Parallel Parking Maneuver: A Geometric Simulation Study

Kevin Luo and Hsinghan Meng,Sunny Hills High School, Fullerton, CA 92833, SAT Professionals, Diamond Bar,CA 91765

ABSTRACT

This paper investigates whether the initial lateral offset of a vehicle from theparking space, denoted ∆y, affects the vehicle’s ability to successfully complete a parallelparking maneuver. Using the geometric mathematical model developed by Wahab et al.[1] and implemented via a simulation developed in a Java-based environment to allow forhigh-fidelity kinematic modelling and real-time geometric validation, a series of trials wereconducted in which ∆y was systematically varied while all other vehicle and parking spaceparameters were held constant. The results demonstrate that, under the constraints of thismodel, the parallel parking maneuver can be completed successfully across a wide range of ∆yvalues. Specifically, varying ∆y causes the geometric solution to self-adjust: the intermediatequantities a, c, y1, and θ all recalculate such that a valid two-arc trajectory always exists.It is concluded that ∆y does not prevent a vehicle from parking; rather, it only shifts thestarting position and arc geometry whi e preserving the feasibility of the maneuver

KEYWORDS

parallel parking, path planning, lateral offset, ∆y, simulation, Ackermann steer ing, bicycle model.

The Illusion of Cybersecurity A Systematic Review of Dynamic Threat Profiles, Attacker Economics, and Adaptive Defense Equilibria

David Mpunwa, ESAMI, Namibia

ABSTRACT

Security teams love the word “hardened.” It is comforting — it suggests permanence, the sense that once a firewall goes up and multi-factor authentication is switched on, the job is finished. That comfort is the problem. This review argues that cybersecurity is not a fixed state but a moving equilibrium, one where attacker economics and defender investment continuously recalibrate against each other. Drawing on the Gordon–Loeb framework and recent evidence on corporate disclosure and reconnaissance costs, the paper traces how commodified cybercrime — ransomware-as-a-service, initial access brokers, dark-web data markets — has collapsed the technical barrier to entry for serious attacks. Core controls (cryptography, multi-factor authentication, VPNs) are mapped against the CIA Triad to show why compliance checklists routinely mistake a snapshot for a state. A layered, Zero-Trust-anchored mitigation roadmap closes the paper.

KEYWORDS

Cybersecurity, Zero Trust Architecture, Threat Actor Economics, Defense-in-Depth, Digital Forensics

The Pyramidal Harmonic Series: Empirical Verification of the Universal Moiré Matrix and Wave Pattern Mathematics

Mark Lance Moody , United States of America

ABSTRACT

Classical wave mechanics rely on continuous algebraic approximations that compress and obscure underlying geometric structures. This paper presents a deterministic framework that natively generates topological wave functions by observing the explicit, uncompressed arithmetic expansions of cyclical partial fractions. By analyzing the reciprocal expansions of specific generating functions—namely the symmetric spatial dilator (10^9 - 1)^2 = 999,999,998,000,000,001 and the asymmetric phase-shift engine (10^8 - 1)(10^9 - 1) = 99,999,998,900,000,001—we demonstrate that explicit arithmetic cascades act as exact mathematical analogs for amplitude stacking and phase precession. When these uncompressed data streams are evaluated within a modulo-72 boundary condition, the output autonomously renders a two-dimensional Moiré interference lattice. Furthermore, by factoring the cyclic limit, we derive a rigid 81-node phase-space combinatorial grid governed by the prime factors 3, 37, and 333,667. This framework proves that positional arithmetic, when fully expanded rather than algebraically compressed, functions as a zero-entropy holographic tensor network.

Use Of Ai-Based Conversational Agents And Postsecondary Student Adjustment And Persistence: Evidence From A Comparative Study Of Ali And Chatgpt

Bruno Kesangana, ¹Department of Teaching and Learning Studies, Faculty of Education Sciences, Université Laval, Quebec City, Quebec, Canada

ABSTRACT

Conversational agents (CAs) are increasingly deployed in postsecondary institutions as accessible, stigma-free supplements to mental health and academic support services. However, the literature rarely distinguishes between institutionally designed, specialized educational CAs and general-purpose large language model (LLM)-based tools such as ChatGPT, treating them as a homogeneous category. This study addresses this gap by comparing Ali — a specialized CA anchored in the Quebec collegiate network — with ChatGPT, a general-purpose LLM, on their relationships with postsecondary student academic adjustment, socio-emotional adjustment, and persistence intentions. Using a quantitative cross-sectional design with 151 students (M_age = 21.86, SD = 1.62; 58.3% female), we conducted ANCOVAs and hierarchical multiple regression analyses. ChatGPT users reported significantly higher competence expectations, perceived value, academic adjustment, socio-emotional adjustment, and persistence intentions. However, regression analyses revealed a more nuanced picture: perceived cost uniquely predicted academic adjustment (β = .196, p = .018); usage duration negatively predicted socio-emotional adjustment (β = −.199, p = .018); and competence expectations predicted persistence intentions (β = .255, p = .012). Moderation analyses showed structurally divergent prediction patterns by CA type: for Ali users, duration, value, and cost positively predicted outcomes, whereas for ChatGPT users, duration negatively predicted socio-emotional adjustment and value showed a negative trend for persistence. These findings challenge the assumption that higher perceived utility translates linearly into better student outcomes and call for tool-specific policies for AI integration in higher education.

KEYWORDS

conversational agents, ChatGPT, Ali, academic adjustment, student persistence, expectancy-value theory, postsecondary education, AI in education

Adaptive Learner Modelling Through Multimodal Behavioural Signals: A Framework For Real-Time Personalization In Higher Education

George Amanortsu ^1,2, Richard Kobla Nyamalor¹,¹ Department of Information Technology, Ghana Communication Technology University (GCTU), Accra,² Ghana Accra Technical University, Accra, Ghana

ABSTRACT

Personalized constrained characterizes learning by static learner profiles that miss moment-to-moment shifts in engagement, comprehension, and motivation. This paper proposes a multimodal learner-modelling framework that fuses clickstream information, response-time patterns, and self-reported affect into continuously updated learner profiles within an AI-driven tutoring environment. Unlike knowledge-tracing approaches that rely on correctness alone, the framework infers latent states such as confusion, disengagement, and cognitive overload to adapt content sequencing, feedback tone, and task difficulty in near real time. The paper detail the architecture, feature set, and a planned semester-long deployment across three undergraduate courses, and specify anticipated outcome ranges grounded in prior meta-analytic evidence. The paper close with design principles for balancing personalization with interpretability and student information privacy.

KEYWORDS

Learner Modelling, AI-Driven Personalization, Adaptive Learning, Multimodal Analytics, Intelligent Tutoring Systems

An Intelligent Web Application to Accelerate Art Reference Discovery Using Pose Estimation and Generative Image Synthesis

1Claremont High School, 1601 N Indian Hill Blvd, Claremont, CA 91711 2University of California, Irvine, Irvine, CA 92697

ABSTRACT

Artists depend on reference images to study proportion, lighting, and movement, yet locating a reference that matches a specific visual intention remains slow because mainstream image search matches keywords rather than visual structure. This paper presents Reference Gallery, a web application that lets artists retrieve and general art references through three channels: a written description, a full-body pose built on a draggable mannequin, and a color chosen from a color wheel. Library poses are indexed with a pose-estimation model and ranked by joint distance, while colors are ranked by distance from each image’s dominant palette. When no stored reference is suitable, a moderated generation service synthesises a new one and adds it to the shared library. An internal ranking study over ten queries produced a mean relevance score of 7.2 out of 10, with pose queries scoring a perfect 10.0 and description and color queries averaging 6.0, indicating that structural channels outperform semantic ones here.

KEYWORDS

Art Reference Retrieval, Pose Estimation, Generative Image Synthesis, Content-Based Image Retrieval, Creativity Support Tools

A Community-Centered Mobile System to Improve Access to Mental Health Support for Students using RealTime Peer Matching, Location-Based Counselor Discovery, and AI-Assisted Conversation

1Lawrence academy, 26 powderhouse road, Groton, MA 01450 2University of California, Santa Cruz, 1156 High St, Santa Cruz, CA 95064

ABSTRACT

Students facing anxiety, depression, or everyday emotional distress often lack a single, low-friction starting point for support, caught between under-resourced school counselors, intimidating crisis hotlines, and costly professional therapy. This paper presents Haven Circle, a mobile application built with Flutter, Firebase Authentication, CloudFirestore, and Firebase Cloud Functions that unifies mood tracking, journaling, real-time peer volunteer matching, a self-help resource library, a multi-source counselor directory, and a safety-scoped AI conversational assistant in one system [6]. Key engineering challenges included preventing duplicate volunteer matches through Firestore transactions, grounding free-text location input in real coordinates before querying Google Places, and constraining an AI assistant to reliably redirect crisis language toward emergency resources. Two experiments evaluated the AI Helper’s crisis-response reliability (92.5% accuracy across 40 test messages) and the effect of explicit location bias on search proximity (average result distance fell from about 42 to 6 miles) [7]. Compared to existing single-purpose peer-support, AI-chatbot, and crisis-monitoring apps, Haven Circle’s integration of all three into one real-time backend offers students a more complete and realistic path toward help.

KEYWORDS

Mobile mental health, Peer support, Flutter, Firebase Firestore, Real-time matching, Conversational AI, Geolocation search, Adolescent well-being

An Auditable Mobile Decision-Support System for Equitable Playing Time and Team Cohesion in School Soccer

Ying Yee Yu ^1,2, Kyler Harris¹,¹ 1Basis Independent Brooklyn, 556 Columbus St, NY 11231, ² 2University of California, Santa Cruz, 1156 High St, Santa Cruz, CA 95064

ABSTRACT

Middle- and high-school soccer coaches decide who plays, for how long, and at what cost to a student’s rest and schoolwork, yet they make those decisions from memory and a paper roster while validated monitoring tools remain confined to professional academies. This paper presents FairPlay Lab, a cross-platform mobile application that turns the records a volunteer coach can realistically keep — training attendance, a seven-day perceived-exertion average, match minutes, and tallied pitch events — into three transparent indicators: a squad fairness index, a heroball versus team-ball cohesion split, and a per-athlete academic crash-risk band. Every recommendation is deliberately rule-based rather than learned, and each substitution suggestion opens into an itemized explanation card naming the factors that produced it. An analysis of the shipped decision surfaces shows the rotation engine responds most strongly to perceived exertion, confines all recommendations to a 24–42 minute window, and contains a discontinuity at the high-load threshold that inverts the intended penalty.

KEYWORDS

youth soccer, playing-time equity, coaching decision support, explainable analytics, rating of perceived exertion, training load monitoring, passing networks, team cohesion, athlete wellbeing, mobile application, offline-first architecture, rule-based scoring

Cyber Override: A Layered Simulated-Operating-System Game for Teaching Evidence-Based Cybersecurity Judgment

Huanyu Cao ^1,2, Quincy Stokes¹,¹ 1Beijing National Day School, 66 Yuquan Rd, Haidian District, Beijing, China, 100141, ² 2University of California, Irvine, Irvine, CA 92697

ABSTRACT

Phishing and information-stealing attacks succeed because people commit judgment errors under time pressure, not because they lack definitions of the threat. Most anti-phishing instruction reinforces those definitions through isolated recognition drills, leaving the harder skill — assembling scattered evidence into a defensible conclusion — largely untrained. This paper presents Cyber Override, a two-dimensional serious game built in Unity that places the player inside a sandboxed simulated operating system and requires a conclusion to be proven from objective evidence, including mail headers, file hashes, process trees, and certificates, before any action is taken. Three subsystems carry the design: a window manager that treats sibling order as z-order, a de-duplicated evidence board that records only committed judgments, and an event-driven failure adjudicator that pairs every fatal action with a corrective hint. A ten-item post-play survey of five players returned to a mean of 4.36 out of 5, and a construct-level decomposition located the sole weak point in difficulty pacing rather than in the evidence mechanics themselves.

KEYWORDS

cybersecurity education, serious games, phishing awareness, simulated operating system, evidence-based reasoning, game-based learning, window management, event-driven architecture, instructional feedback, security awareness training

ChemSafe AI: Design and Development of an Intelligent Mobile Application for Laboratory Safety Education and Hazard Support

Wentao Zhou ^1,2, Cesar Magana¹,¹ 1Portola High School, 1001 Cadence, Irvine, CA 92618 ² 2California State University Long Beach, 1250 Bellflower Blvd, Long Beach, CA 90840

ABSTRACT

With a large portion of educators and students having experienced laboratory accidents, there’s been a clear shortage of laboratory safety education, leaving young chemists defenseless against hazardous situations [1]. To address this, I developed ChemSafe AI, a mobile application combining digital safety education with real-time hazard support [2]. The app features 3 key systems: A Lab Planning Assistant for reviewing experiments, an Emergency Assistant for immediate emergency guidance, and a Camera System for visual hazard recognition. Development challenges arose when differentiating visually similar compounds with images proved to be impractical, forcing a redesign of the camera system to generate accurate safety protocols by instead analyzing contextual clues in photos. This change proved to be successful, as experimental testing revealed that the system had high detection and classification accuracy across various images. Ultimately, through guidance and education, ChemSafe AI hopes to become a reliable digital mentor, used to reduce lab accidents while restoring confidence within young aspiring chemists to safely pursue individual experimentation [3].

KEYWORDS

Laboratory Safety, Artificial Intelligence, Hazard Recognition, Mobile Application

An Intelligent Mobile Application to Connect Tennis Players by Skill, Location, and Availability Using Flutter and Firebase

Mingcheng Tian ^1,2, Rodrigo Onate¹,¹ 1Santa Margarita Catholic High, 22062 Antonio Parkway, Rancho Santa Margarita, CA 92688 ² 2California State University, Fullerton, 800 N State College Blvd, Fullerton, CA 92831

ABSTRACT

Tennis players often struggle to find compatible people to play with because they need someone nearby, at a similar skill level, and available at the same time. This project proposes a mobile application that helps players find rally partners through structured profiles and search filters. The app was built using Flutter for the interface and Firebase for authentication and profile storage [11]. The main components are the profile system, search and filter system, and settings system. Users can store details such as UTR rating, location, gender, preferred courts, playing hand, available days, and available times. Challenges included keeping profile data consistent across settings, search, and profile displays. Experiments tested search accuracy and availability updates. Results showed strong performance overall, although city and availability formatting affected some results. This idea is useful because it gives tennis players a faster and more organized way to connect.

KEYWORDS

Tennis, Community, Social, Exercise, Mobile Application, Flutter

A TWO-STAGE VEDIC FORECASTING FRAMEWORK AND DAILY TARABALAM: OBSERVATIONAL EVALUATION OF POSITIVE EMOTIONAL OUTCOMES IN VEDICVISIONS.AI

Venkata Duvvuri1, Nikhil Reddy Thokala1, Jaikalyan Tatineni2, Abhishek Srinivas3 and Priya Ganesh Phapale1 1Siriusmindshare Labs 2Research Department, Siriusmindshare Lab LLC, Santa Clara, USA 3MS Analytics, MSIS & Researcher, Northeastern University & Trine University, Austin, USA

ABSTRACT

Within the broader context of a Vedic horoscope, VedicVisions.ai is an AI-based two-stage Vedic forecasting system combining deterministic Vedic calculations with generative AI. Natal-chart factors, planetary transits, dashas, houses, Nakshatra, and forecast scores provide the chart and timing foundation; generative AI converts these calculated inputs into a personalized six-month narrative and context-aware guidance for career, finances, health, relationships, and family. The AI narrative is fetched in the background, displayed in the application, and incorporated into the downloadable PDF report alongside the deterministic astrology calculations. Stage 2 provides granular daily Tarabalam classifications – Excellent, Good, Caution, Unfavorable, or Avoid – based on the relationship between the user’s Janma Nakshatra and the daily Nakshatra. This study evaluates the system using 161 daily observations. Mood assessment was blinded to the day’s Tarabalam classification: mood was recorded before the participant viewed that day’s label. Positive mood occurred on 77.6% of observed days. Good days were positive in 43/55 observations (78.2%, p < .001 versus a 50% reference), Excellent days in 28/40 (70.0%, p = .008), and Good/Excellent days together in 71/95 (74.7%). These favorable-day results show substantial absolute positive-emotional experience and are directionally consistent with the traditional preference for auspicious timing, although they did not significantly exceed the subject’s already-high 77.6% period base rate. Caution, Unfavorable, and Avoid labels matched negative/meh mood in 22.2%, 18.2%, and 11.8% of observations, respectively; Avoid days were positive in 15/17 observations (88.2%), and the overall ordinal association was small and nonsignificant (Spearman rho = -0.12, p = 0.132). The broader VedicVisions.ai forecasts characterized much of 2025-2026 as Mostly favorable, with displayed scores around 6.4-6.7/10, providing context for the high overall positive-mood rate. Together, the findings provide encouraging preliminary observational evidence for the layered framework: Stage 1 supplies AI-enriched period-level forecast context, while Stage 2 offers a granular timing layer whose incremental predictive value requires prospective validation. The findings are consistent with the traditional practice of referring to a favorable day in advance; however, the observational design cannot establish causation, and prospective multi-user studies are needed to establish generalizability, domain-specific outcomes, and incremental predictive value.

KEYWORDS

Tarabalam, Vedic astrology, Nakshatra, VedicVisions.ai, generative AI, mood, ecological momentary assessment, hierarchical forecasting, positive outcomes

A Unified Mobile Application to Connect, Organize, and Safeguard the High School Community using Cloud Services and On-Device Content Safety

Jacob Lee1 , Tyler Boulom2 1Troy High School, 2200 Dorothy Ln, Fullerton, CA 92831 2Woodbury University, 7500 N Glenoaks Blvd, Burbank, CA 91504

ABSTRACT

High schools run on information, yet students must track announcements, clubs, schedules, and events across disconnected channels, missing the participation and belonging that improve academic outcomes, while the general social platforms they fall back on expose them to unsafe content. This paper presents Quilt, a single mobile application that unifies a school's communication, organization, and community into one school-scoped and moderated space. Built with Flutter and Google's Firebase platform, Quilt ties membership to a verified school identity, lets clubs post and poll, shares a common bell schedule and event calendar, connects parents to their students through a consentbased QR link, and keeps the feed safe by pairing an on-device content-safety filter with a human reporting-andmoderation workflow. Key challenges around moderation accuracy, real-time synchronization at school scale, and the privacy of a parent-student link were addressed through a layered filter-and-report design, a real-time database with local caching, and an in-person QR exchange. By consolidating scattered channels and treating safety as a firstclass concern, Quilt targets the informational, social, and safety gaps that make school life harder than it needs to be.

KEYWORDS

School Community, Mobile Application, Content Moderation, Firebase, QR Authentication, Student Engagement

An Adaptive Immersive Program to Assist Students with Attention Difficulties in Standardized–Test Vocabulary Acquisition using a First-Person Heist Game

Yibo Hu, ¹1Saratoga High School, 20300 Herriman Avenue, Saratoga, CA 95070

Moddwyn Andaya, ²2California State University, Sacramento, 6000 Jed Smith Dr, Sacramento, CA 95819

ABSTRACT

Vocabulary knowledge predicts academic achievement and gates access to higher education through the SAT and ACT, yet conventional preparation depends on sustained attention and verbal working memory — the capacities most impaired in the roughly one in nine United States children who have received an ADHD diagnosis. This paper presents a three-dimensional first-person heist game, built in Unity, in which level progression is gated by recall of standardized-test vocabulary, so that studying constitutes play rather than accompanying it. The system comprises a difficulty-tiered question provider over a pre-generated bank drawn from a twelve-thousand-word corpus, a vaultdoor state machine binding recall to a consequential in-game action, and a guard behavior system that surfaces additional prompts. A one-week retention study with ten participants found the game group recalled 4.8 words of ten against 3.8 for a matched flashcard control. Although the difference did not reach statistical significance given the limited sample size, the observed effect size suggests promising potential for game-based vocabulary learning and warrants further investigation with a larger participant pool.

KEYWORDS

Game-based learning; serious games; Unity; vocabulary acquisition; attention-deficit/hyperactivity disorder; standardized test preparation; retrieval practice; self-determination theory