arXiv subjects

Chirag Garg

Publications and source records attributed to Chirag Garg.

8 recordsLinked to original sources

arXiv2026

multi-Stochastic Core Architecture for Scaling Probabilistic Ising Machines

Ising Machines offer vast potential to solve NP-hard optimization problems efficiently that are intractable to solve using conventional computing architecture. A lot of these optimization problems fall into statistical learnability and involve finding an optimal solution among many possible, near-analogous configurations, by searching in a non-convex energy landscape. In this context, the probabilistic Boltzmann machine architecture especially PASS (Parallel Asynchronous Stochastic Sampler), explores and models the complex probability landscape pertaining to all possible configurations and excels in finding the ground-state energy solution of these intractable problems. Additionally, the noise-based neuron architecture addresses the limitation of conventional annealing methods, which may get stuck around local minima. Here, we demonstrate a stochastic sampling approach based on Block Gibbs Sampling to integrate multiple asynchronous PASS chips (four in this work), enabling improved scalability. Further, we demonstrate the scaling by mapping 784 nodes Max-Cut problem integrating 256 nodes PASS accelerator manufactured in 14 nm CMOS FinFET technology. PASS-enabled system with Block Gibbs Sampling protocol shows approximately 1000 times speedup for Max-Cut optimization compared to state-of-the-art methods implemented on CPUs and GPUs. The general applicability of this approach is further illustrated by solving a quantum spin chain Transverse Ising system and accurately representing complex probability landscapes. Moreover, our results demonstrate the change in the scaling law to constant in the scaled-PASS accelerator as compared to exponential on GPUs enabling at least 4 orders of magnitude improvement in time-to-solution. Hence, the presented methodology enables the pathway for scaling of asynchronous brain-like dynamics systems that do not follow any clock for its operation.

cs.ET↗

arXiv2026

Data Scout: Targeted Web Crawling for Domain-Specific Pretraining Corpora

The dominant approach to building domain-specific pretraining corpora is to filter large web archives such as CommonCrawl. This works well for popular domains but breaks down for specialized ones, where relevant content is sparse and often beyond the reach of popularity-driven crawlers. We present Data Scout, which inverts this: instead of filtering an archive, it directs a targeted crawl. An LLM expands a root topic into a taxonomy and thousands of search queries; the returned URLs (seeds) are grouped by subdomain and screened with a user-supplied classifier (the probe), admitting each subdomain on the basis of a small sample. This works because relevance has a sharp boundary at the subdomain level: in mathematics, a page is 21x more likely to be relevant than one on a sibling subdomain. With the FineMath classifier as the probe, 21.9% of crawled pages are high-quality math content, 70x the 0.31% rate from filtering a comparable web sample, so the crawl wastes far less effort. But the payoff is not just efficiency: 63.2% of these pages are missing from CommonCrawl altogether, yet just as useful for training. Continued pretraining of Llama-3.2-3B on 1.9B Data Scout tokens matches FineMath corpus on GSM8k. Because the probe is the only domain-specific component, Data Scout can in principle apply to any domain with such a classifier.

cs.LG↗

arXiv2025

Efficient Optimization Accelerator Framework for Multistate Ising Problems

Ising Machines are emerging hardware architectures that efficiently solve NP-Hard combinatorial optimization problems. Generally, combinatorial problems are transformed into quadratic unconstrained binary optimization (QUBO) form, but this transformation often complicates the solution landscape, degrading performance, especially for multi-state problems. To address this challenge, we model spin interactions as generalized boolean logic function to significantly reduce the exploration space. We demonstrate the effectiveness of our approach on graph coloring problem using probabilistic Ising solvers, achieving similar accuracy compared to state-of-the-art heuristics and machine learning algorithms. It also shows significant improvement over state-of-the-art QUBO-based Ising solvers, including probabilistic Ising and simulated bifurcation machines. We also design 1024-neuron all-to-all connected probabilistic Ising accelerator on FPGA with the proposed approach that shows ~10000x performance acceleration compared to GPU-based Tabucol heuristics and reducing physical neurons by 1.5-4x over baseline Ising frameworks. Thus, this work establishes superior efficiency, scalability and solution quality for multi-state optimization problems.

cs.AR↗

arXiv2024

PASS: An Asynchronous Probabilistic Processor for Next Generation Intelligence

New computing paradigms are required to solve the most challenging computational problems where no exact polynomial time solution exists.Probabilistic Ising Accelerators has gained promise on these problems with the ability to model complex probability distributions and find ground states of intractable problems. In this context, we have demonstrated the Parallel Asynchronous Stochastic Sampler (PASS), the first fully on-chip integrated, asynchronous, probabilistic accelerator that takes advantage of the intrinsic fine-grained parallelism of the Ising Model and built in state of the art 14nm CMOS FinFET technology. We have demonstrated broad applicability of this accelerator on problems ranging from Combinatorial Optimization, Neural Simulation, to Machine Learning along with up to $23,000$x energy to solution improvement compared to CPUs on probabilistic problems.

cs.DC↗

arXiv2024

Ferrimagnetic Heusler tunnel junctions with fast spin-transfer torque switching enabled by low magnetization

Magnetic random access memory that uses magnetic tunnel junction memory cells is a high performance, non-volatile memory technology that goes beyond traditional charge-based memories. Today its speed is limited by the high magnetization of the memory storage layer. Here we show that fast and highly reliable switching is possible using a very low magnetization ferrimagnetic Heusler alloy, Mn3Ge. Moreover, the tunneling magnetoresistance is the highest yet achieved for a ferrimagnetic material at ambient temperature. Furthermore, the devices were prepared on technologically relevant amorphous substrates using a novel combination of a nitride seed layer and a chemical templating layer. These results show a clear path to the lowering of switching currents using ferrimagnetic Heusler materials and, therefore, to the scaling of high performance magnetic random access memories beyond those nodes possible with ferromagnetic devices.

cond-mat.mtrl-sci↗

arXiv2023

Anisotropic proximity-induced superconductivity and edge supercurrent in Kagome metal, K1-xV3Sb5

Materials with transition metals in triangular lattices are of great interest for their potential combination of strong correlation, exotic magnetism and electronic topology. Kagome nets are of particular importance since the discovery of geometrically frustrated magnetism and topological band structures in crystals like Herbertsmithite and Fe3Sn2, respectively. KV3Sb5 was discovered to be a layered topological metal with a Kagome net of vanadium. Here, we fabricated Josephson Junctions (JJ) of K1-xV3Sb5 and induced superconductivity over long junction lengths. Through magnetoresistance and current vs. phase measurements, we observed magnetic field sweeping direction dependent magnetoresistance, and an anisotropic interference pattern with a Fraunhofer pattern for in-plane magnetic field, but a suppression of critical current for out-of-plane magnetic field. These results indicate an anisotropic internal magnetic field in K1-xV3Sb5 which influences the superconducting coupling in the junction, possibly giving rise to spin-triplet superconductivity. In addition, the observation of long-lived fast oscillations shows evidence of spatially localized conducting channels arising from edge states. These observations pave the way for studying unconventional superconductivity and Josephson device based on Kagome metals with electron correlation and topology.

cond-mat.supr-con↗

arXiv2016

Butterfly Magnetoresistance, Quasi-2D Dirac Fermi Surfaces, and a Topological Phase Transition in ZrSiS

Magnetoresistance (MR), the change of a material's electrical resistance in response to an applied magnetic field, is a technologically important property that has been the topic of intense study for more than a quarter century. Here we report the observation of an unusual "butterfly" shaped titanic angular magnetoresistance (AMR) in the non-magnetic, Dirac material, ZrSiS. The MR is large and positive, reaching nearly 1.8 x 10^5 percent at 9 T and 2 K at an angle of 45o between the applied current (along the a-axis) and the applied field (90o is H parallel to the c-axis). Approaching 90o, a "dip" is seen in the AMR which can be traced to an angle dependent deviation from the H^2 law. By analyzing the SdH oscillations at different angles, we find that ZrSiS has a combination of 2D and 3D Dirac pockets comprising its Fermi surface and that the anomalous transport behavior coincides with a topological phase transition whose robust signature is evident despite transport contributions from other parts of the Fermi surface. We also find that as a function of angle, the temperature dependent resistivity in high field displays a broad peak-like behavior, unlike any known Dirac/Weyl material. The combination of very high mobility carriers and multiple Fermi surfaces in ZrSiS allow for large bulk property changes to occur as a function of angle between applied fields makes it a promising platform to study the physics stemming from the coexistence of 2D and 3D Dirac electrons.

cond-mat.mtrl-sci↗

arXiv2012

Assessing the Value of 3D Reconstruction in Building Construction

3-dimensional (3D) reconstruction is an emerging field in image processing and computer vision that aims to create 3D visualizations/ models of objects/ scenes from image sets. However, its commercial applications and benefits are yet to be fully explored. In this paper, we describe ongoing work towards assessing the value of 3D reconstruction in the building construction domain. We present preliminary results from a user study, where our objective is to understand the use of visual information in building construction in order to determine problems with the use of visual information and identify potential benefits and scenarios for the use of 3D reconstruction.

cs.HC↗