arXiv subjects

Sunil Tiwari

Publications and source records attributed to Sunil Tiwari.

8 recordsLinked to original sources

arXiv2026

Novel Memory Forgetting Techniques for Autonomous AI Agents: Balancing Relevance and Efficiency

Long-horizon conversational agents require persistent memory for coherent reasoning, yet uncontrolled accumulation causes temporal decay and false memory propagation. Benchmarks such as LOCOMO and LOCCO report performance degradation from 0.455 to 0.05 across stages, while MultiWOZ shows 78.2% accuracy with 6.8% false memory rate under persistent retention. This work introduces an adaptive budgeted forgetting framework that regulates memory through relevanceguided scoring and bounded optimization. The approach integrates recency, frequency, and semantic alignment to maintain stability under constrained context. Comparative analysis demonstrates improved long-horizon F1 beyond 0.583 baseline levels, higher retention consistency, and reduced false memory behavior without increasing context usage. These findings confirm that structured forgetting preserves reasoning performance while preventing unbounded memory growth in extended conversational settings.

cs.AI↗

arXiv2026

3D Architect: An Automated Approach to Three-Dimensional Modeling

The aim of our paper is to render an object in 3-dimension using a set of its orthographic views. Corner detector (Harris Detector) is applied on the input views to obtain control points. These control points are projected perpendicular to respective views, in order to construct an envelope. A set of points describing the object in 3-dimension, are obtained from the intersection of these mutually perpendicular envelopes. These set of points are used to regenerate the surfaces of the object using computational geometry. At the end, the object in 3-dimension is rendered using OpenGL

cs.CV↗

arXiv2026

Developing Adaptive Context Compression Techniques for Large Language Models (LLMs) in Long-Running Interactions

Large Language Models (LLMs) often experience performance degradation during long-running interactions due to increasing context length, memory saturation, and computational overhead. This paper presents an adaptive context compression framework that integrates importance-aware memory selection, coherence-sensitive filtering, and dynamic budget allocation to retain essential conversational information while controlling context growth. The approach is evaluated on LOCOMO, LOCCO, and LongBench benchmarks to assess answer quality, retrieval accuracy, coherence preservation, and efficiency. Experimental results demonstrate that the proposed method achieves consistent improvements in conversational stability and retrieval performance while reducing token usage and inference latency compared with existing memory and compression-based approaches. These findings indicate that adaptive context compression provides an effective balance between long-term memory preservation and computational efficiency in persistent LLM interactions

cs.CV↗

arXiv2026

Multi-Layered Memory Architectures for LLM Agents: An Experimental Evaluation of Long-Term Context Retention

Long-horizon dialogue systems suffer from semanticdrift and unstable memory retention across extended sessions. This paper presents a Multi-Layer Memory Framework that decomposes dialogue history into working, episodic, and semantic layers with adaptive retrieval gating and retention regularization. The architecture controls cross-session drift while maintaining bounded context growth and computational efficiency. Experiments on LOCOMO, LOCCO, and LoCoMo show improved performance, achieving 46.85 Success Rate, 0.618 overall F1 with 0.594 multi-hop F1, and 56.90% six-period retention while reducing false memory rate to 5.1% and context usage to 58.40%. Results confirm enhanced long-term retention and reasoning stability under constrained context budgets.

cs.CV↗

arXiv2025

The pickup and delivery problem with time windows and scheduling on the edges

We introduce the Pickup and Delivery Problem with Time Windows and Scheduling on the Edges (PDPTW-SE), a generalization of the PDPTW that integrates vehicle routing and machine scheduling. The problem involves defining routes for transportation requests with specific pickup and delivery locations using a heterogeneous vehicle fleet, while machines must be scheduled to traverse certain edges. The objective is to minimize the total completion time subject to capacity, time window, and precedence constraints. We propose a mixed-integer linear programming (MIP) formulation, including preprocessing and valid inequalities, and a multi-start heuristic with a linear programming (LP) improvement procedure. A benchmark set with two instance families is also introduced: (i) coordination of pickups and deliveries across islands requiring cargo ships, and (ii) transport across multiple floors, as in hospitals, requiring elevator scheduling. Computational experiments show that the solver on the MIP formulation solves instances with up to 12 requests and finds feasible solutions for 95.0% of the 320 instances with up to 12 requests. For these, the heuristic consistently provides feasible solutions with low deviations, often matching or outperforming the MIP results. For the remaining 160 instances with 40 and 60 requests, only the heuristic finds feasible solutions. We thus recommend the MIP for short-horizon instances (up to 12 requests) and the heuristic for larger or long-horizon instances. Results also highlight the LP improvement procedure's relevance, reducing solution values by at least 5% on average in general. For larger, tightly constrained instances, an additional machine helps with feasibility and solution quality.

math.OC↗

arXiv2025

Moisture-driven CO2 direct air capture and delivery for cultivating cyanobacteria

A moisture-driven air capture system was developed and demonstrated for cultivating cyanobacteria and microalgae at the flask (50 mL), bench (12 L) and small pilot (840 L) scale. Purolite A501 anion exchange resin beads were found to be biocompatible and rapidly deliver air-captured CO2 when immersed directly in an alkaline cultivation medium containing cyanobacteria or microalgae. Flask-scale cultivation trials showed A501 could sustain rapid growth (190 mg/L/d) of the cyanobacterium Synechocystis sp. PCC 6803 strain engineered to produce laurate. A bench-scale system installed in a laminar flow hood was able to deliver 2 g CO2/d into abiotic alkaline cultivation medium and 0.5 g/d in the presence of Synechocystis to support vigorous growth (39 mg/L/d) limited by the CO2 delivered by the sorbent. A small pilot-scale system installed in a 4.2 m2 outdoor raceway pond in Mesa, Arizona was able to deliver 100 g CO2/d into abiotic alkaline cultivation medium. Exopolysaccharides and other products excreted by Synechocystis 6803 covered the sorbent beads, reducing their capacity to 25%, which could be partially restored to 70% capacity using a wash protocol, but the CO2 delivery kinetics remained 3-4 fold slower. Analysis of the sorbent beads used as part of four separate outdoor cultivation trials with over 300 days of outdoor wet and dry cycling over four seasons showed significant mechanical fracturing. Infrared spectroscopy and thermogravimetric analysis showed a significant loss of NR4+ functional groups necessary for CO2 capture correlated with extended use. Under the assumption that abiotic performance eventually can be retained by delivering CO2 into the media recycle stream in a way that avoids biofouling, technoeconomic and life cycle analyses show the viability of a small first-of-a-kind biorefinery producing 500 barrels per day of biofuel.

cond-mat.soft↗

arXiv2025

Design and demonstration of a direct air capture system with moisture-driven CO2 delivery into aqueous medium

A moisture-driven air capture (DAC) system was designed and demonstrated. A laboratory-scale system delivering ~1 g CO2 per day was demonstrated in a laminar flow hood and a small pilot-scale system that could deliver ~100 g CO2 daily was operated outdoors in a 4.2 m2 (areal surface area) raceway pond. Elongated mesh tube packets were designed to contain AER beads with high surface area for contacting the air and were found to reduce drying and CO2 loading time ~4-fold over larger mesh bags. Whereas this system was designed for CO2 delivery for cultivating photosynthetic microbes, its potential uses are much broader and include CO2 use in the food and beverage industry, conversion to fuels and chemicals, and sequestration. Techno-economic assessments for a practical scenario based on current results are \$670/tonne to capture CO2 into an alkaline solution and an additional \$280/tonne to extract CO2 from solution, purify and compress to 15 MPa for sequestration. An aspirational scenario modelling reasonable improvements to develop AER sorbents with a capacity of 4 mmol CO2 per gram of sorbent and water uptake of 50 wt.%, which leads to sorbent drying and loading within 1 h, shows a potential to reach \$51/tonne to capture CO2 into an alkaline solution and an additional \$109/tonne to get to 15 MPa for sequestration. Life cycle analysis shows the aspirational moisture-driven process uses up to 87% less energy than thermal and/or vacuum swing DAC by using energy from water evaporation; however, ~330 wt.% water uptake by the sorbent contained in a hydrophilic mesh packets leads to ~33-fold higher water use than the thermodynamic limits, which emphasizes future research is needed to increase sorbent hydrophobicity while maintaining and further increasing ion exchange capacity needed to bind CO2.

cond-mat.soft↗

arXiv2019

Comments on the "Optimal strategy of deteriorating items with capacity constraints under two-levels of trade credit policy"

This technical note rectified the mathematical and conceptual errors present in Liao et al. (2014). Liao et al. (2014) proposed an EOQ model under two-levels trade credit policy considering limited storage capacity whereby the supplier provides a permissible delay period (M) to the retailer, and the retailer also offers a permissible delay period (N) (whereM > N) to its customers. In the current technical note, we point out some defects of their model from the logical viewpoints of mathematics regarding both interest charged and interest earned. Furthermore, as an example, one of the affected numerical results is re-evaluated.

math.OC↗