arXiv subjects

Anastasios Kyrillidis

Publications and source records attributed to Anastasios Kyrillidis.

At least 19 recordsLinked to original sources

arXiv2026

Identity-Paired Progressive Depth Training: When Trainability Persists Beyond Expressibility

Variational Quantum Algorithms (VQAs) are a leading paradigm for near-term quantum computing, yet their training suffers from sensitivity to circuit depth, initialization, and landscape pathologies such as barren plateaus. We study \emph{progressive depth training} (PDT) -- a layerwise curriculum that trains a shallow circuit before appending new layers -- and identify a fundamental obstacle: fixed entangling gates (CNOTs) in hardware-efficient ans\"atze cause \emph{initialization shock}, an energy spike when new layers are added. We propose \emph{identity-paired progressive depth training} (IP-PDT), which appends forward/inverse block pairs -- each consisting of a standard rotation$+$CNOT block followed by its reverse -- that compose to the identity at initialization. Because the adjacent CNOT rings cancel, the effective circuit retains only \textit{a single entangling layer} surrounded by \textit{overparameterized local rotations}. We prove a simple \textit{Reachable Set Saturation Theorem}: under this construction the variational manifold expands exactly once (when post-entangler rotations are first introduced) and then \emph{saturates}; all subsequent depth increases provide pure overparameterization of single-qubit unitaries. Despite this saturation, progressive addition of rotation parameters can continue to improve optimization outcomes -- a phenomenon we term \emph{trainability beyond expressibility}. We formalize IP-PDT as a continuation method on nested manifolds, prove monotone energy guarantees under an acceptance rule, and connect energy error to ground-state fidelity through spectral-gap inequalities. A detailed resource analysis shows that IP-PDT achieves lower total gate cost than both baselines by eliminating most CNOT gates.

quant-ph↗

arXiv2026

SuperThoughts: Reasoning Tokens in Superposition

Long Chain-of-Thought (CoT) reasoning improves LLM problem-solving but is computationally expensive due to sequential token generation. While recent works explore reasoning in continuous latent spaces to bypass discrete token generation, they often struggle with training stability and fail to scale to complex, long-horizon tasks due to lack of supervision signal. We propose SuperThoughts, which compresses pairs of consecutive CoT tokens into single latent representations and decodes two tokens per step via a lightweight Multi-Token Prediction (MTP) module. This preserves discrete token supervision at training time while doubling throughput at inference time. We finetune Qwen2.5-Math-1.5B-Instruct, Qwen2.5-Math-7B-Instruct, Qwen2.5-Math-14B-Instruct, and evaluate on MATH500, AMC, OlympiadBench, and GPQA-Diamond. With a confidence-based adaptive mechanism that falls back to standard decoding when uncertain, SuperThoughts achieves $\sim$20--30\% CoT length reduction while maintaining accuracy with minimal degradation (1-2 points accuracy drop on most tasks).

cs.LG↗

arXiv2026

One Model, Two Roles: Emergent Specialization in a Shared Recurrent Transformer

Can a shared-weight recurrent Transformer develop distinct internal roles without being partitioned into separate modules? We study this in Asymmetric Input Recurrence (AIR), a minimal two-state reasoning architecture in which the same Transformer model is reused for both updates (per literature, L and H) and the only built-in difference in the update rule is that the encoded input is injected during L-updates but not H-updates. Across Sudoku-Extreme and Maze, decoded rollouts reveal a consistent split: $\zH$ behaves like a fully committed proposal state, whereas $\zL$ retains local uncertainty and shifting intermediate structure. Freeze experiments show that this split is, in practice, related to the model's state dynamics: in Sudoku, freezing $\zH$ reduces $\zL$'s content changes whereas freezing $\zL$ increases $\zH$'s, while in Maze, freezing either state increases content changes in the other state. Ablations show that to induce specialization, the shared model needs to be able to tell the two update types apart, either from input injection asymmetry or from a separate level token. Mechanistically, attention analysis shows that L-updates are consistently more local than H-updates in both Sudoku and Maze. Together, these results show that, in a two-state recurrent setting, a clear state-identity signal can induce stable, related functional roles inside a shared-parameter recurrent Transformer. Code is available at \href{https://github.com/juchengshen/air}{\textcolor{blue}{https://github.com/juchengshen/air}}.

cs.LG↗

arXiv2026

AdaPaD: Adaptive Parallel Deflation for PEFT with Self-Correcting Rank Discovery

Fine-tuning large language models with LoRA requires choosing a rank r before training starts. Existing approaches either extract rank-1 components sequentially, freezing each component's error permanently into every subsequent residual, or optimize the full low-rank factorization jointly with guarantees that describe only the joint update, not individual rank-1 directions. We present AdaPaD (Adaptive Parallel Deflation), which trains all rank-1 components simultaneously: each worker refines its component against a deflation target built from the latest estimates of all predecessors, and as those estimates improve, the targets improve too. We call this property self-correction: deflation errors converge to zero over rounds rather than persisting as fixed residuals. On top of this backbone, AdaPaD adds advance learning (private pre-training before activation) and per-module dynamic rank discovery (importance-based growth until a shared budget is exhausted), making the rank distribution an output rather than an input. We prove that every component's error decays exponentially after a warm-up period, with a generalization bound that splits into a vanishing algorithmic term and an irreducible statistical floor. Empirically, AdaPaD is competitive with adaptive-rank LoRA baselines on GLUE with DeBERTaV3-base at matched parameter budgets, and competitive with fixed-rank LoRA on Qwen3-0.6B SQuAD/SQuAD v2 while deploying an adapter that is on average 30.7% smaller.

cs.LG↗

arXiv2026

SGD at the Edge of Stability: The Stochastic Sharpness Gap

When training neural networks with full-batch gradient descent (GD) and step size $\eta$, the largest eigenvalue of the Hessian -- the sharpness $S(\boldsymbol{\theta})$ -- rises to $2/\eta$ and hovers there, a phenomenon termed the Edge of Stability (EoS). \citet{damian2023selfstab} showed that this behavior is explained by a self-stabilization mechanism driven by third-order structure of the loss, and that GD implicitly follows projected gradient descent (PGD) on the constraint $ S(\boldsymbol{\theta})\leq 2/\eta$. For mini-batch stochastic gradient descent (SGD), the sharpness stabilizes below $2/\eta$, with the gap widening as the batch size decreases; yet no theoretical explanation exists for this suppression. We introduce stochastic self-stabilization, extending the self-stabilization framework to SGD. Our key insight is that gradient noise injects variance into the oscillatory dynamics along the top Hessian eigenvector, strengthening the cubic sharpness-reducing force and shifting the equilibrium below $2/\eta$. Following the approach of \citet{damian2023selfstab}, we define stochastic predicted dynamics relative to a moving projected gradient descent trajectory and prove a stochastic coupling theorem that bounds the deviation of SGD from these predictions. We derive a closed-form equilibrium sharpness gap: $\Delta S = \eta \beta \sigma_{\boldsymbol{u}}^{2}/(4\alpha)$, where $\alpha$ is the progressive sharpening rate, $\beta$ is the self-stabilization strength, and $\sigma_{ \boldsymbol{u}}^{2}$ is the gradient noise variance projected onto the top eigenvector. This formula predicts that smaller batch sizes yield flatter solutions and recovers GD when the batch equals the full dataset.

cs.LG↗

arXiv2026

Exploiting Low-Rank Objective Structure in Discrete Quadratic Optimization

We study the problem of maximizing a complex-valued quadratic form over the $K^{\text{th}}$ roots of unity. We show that when the objective matrix $\mathbf{Q}^\star \in \mathbb{C}^{n \times n}$ of the quadratic has rank $r$, the global maximizer belongs to a candidate set of size $O(rn^{2r-1})$. This set can be constructed deterministically in $O(rn^{2r+1})$ time by enumerating the vertices of a hyperplane arrangement in $\mathbb{R}^{2r}.$ The algorithm is embarrassingly parallel; with~$P$ processors, the time complexity drops to $O(r n^{2r+1}/P)$. For approximately low-rank settings, where the objective matrix is a noise-perturbed variant of a rank-$r$ matrix, we prove that applying our framework to a spectral truncation yields a multiplicative $(1 - O(\left\|\mathbf{H}\right\|_2 / \delta^{\star}))$-approximation guarantee, where $\delta^{\star}$ denotes the eigengap of the underlying rank-$r$ matrix and $\mathbf{H}$ represents the perturbation. To scale to high-dimensional problems, we establish a randomized sampling variant. We prove that uniformly sampling $S \geq O(1/\varepsilon^{r-1})$ candidates achieves a $(1-\varepsilon)\cos^2(\pi / K)$-approximation of the optimal rank-$r$ solution with high probability. Crucially, this sample size is entirely independent of $n$, reducing the overall runtime to $O(S \cdot n^2)$. Computational experiments on synthetic benchmarks and large-scale graphs for \textsc{Max-3-Cut} confirm that our algorithms match or exceed semi-definite programming solution quality on structured instances while enabling massive parallelization across heterogeneous hardware and scaling seamlessly to problems where $n \geq 10^6$.

cs.DS↗

arXiv2026

Convergence Analysis of Two-Layer Neural Networks under Gaussian Input Masking

We investigate the convergence guarantee of two-layer neural network training with Gaussian randomly masked inputs. This scenario corresponds to Gaussian dropout at the input level, or noisy input training common in sensor networks, privacy-preserving training, and federated learning, where each user may have access to partial or corrupted features. Using a Neural Tangent Kernel (NTK) analysis, we demonstrate that training a two-layer ReLU network with Gaussian randomly masked inputs achieves linear convergence up to an error region proportional to the mask's variance. A key technical contribution is resolving the randomness within the non-linear activation, a problem of independent interest.

cs.LG↗

arXiv2026

GHOST: Unmasking Phantom States in Mamba2 via Grouped Hidden-state Output-aware Selection & Truncation

While Mamba2's expanded state dimension enhances temporal modeling, it incurs substantial inference overhead that saturates bandwidth during autoregressive generation. Standard pruning methods fail to address this bottleneck: unstructured sparsity leaves activations dense, magnitude-based selection ignores runtime dynamics, and gradient-based methods impose prohibitive costs. We introduce GHOST (Grouped Hidden-state Output-aware Selection and Truncation), a structured pruning framework that approximates control-theoretic balanced truncation using only forward-pass statistics. By jointly measuring controllability and observability, GHOST rivals the fidelity of gradient-based methods without requiring backpropagation. As a highlight, on models ranging from 130M to 2.7B parameters, our approach achieves a 50\% state-dimension reduction with approximately 1 perplexity point increase on WikiText-2. Code is available at https://anonymous.4open.science/r/mamba2_ghost-7BCB/.

cs.AI↗

arXiv2025

On The Finetuning of MLIPs Through the Lens of Iterated Maps With BPTT

Accurate structural relaxation is critical for advanced materials design. Traditional approaches built on physics-derived first-principles calculations are computationally expensive, motivating the creation of machine-learning interatomic potentials (MLIPs), which strive to faithfully reproduce first-principles computed forces. We propose a fine-tuning method to be used on a pretrained MLIP in which we create a fully-differentiable end-to-end simulation loop that optimizes the predicted final structures directly. Trajectories are unrolled and gradients are tracked through the entire relaxation. We show that this method consistently improves performance across all evaluated pretrained models; resulting in an average of roughly 32% reduction in prediction error. Interestingly, we show the process is robust to substantial variation in the relaxation setup, achieving negligibly different results across varied hyperparameter and procedural modifications.

cond-mat.mtrl-sci↗

arXiv2025

Completion of partial structures using Patterson maps with the CrysFormer machine learning model

Protein structure determination has long been one of the primary challenges of structural biology, to which deep machine learning (ML)-based approaches have increasingly been applied. However, these ML models generally do not incorporate the experimental measurements directly, such as X-ray crystallographic diffraction data. To this end, we explore an approach that more tightly couples these traditional crystallographic and recent ML-based methods, by training a hybrid 3-d vision transformer and convolutional network on inputs from both domains. We make use of two distinct input constructs / Patterson maps, which are directly obtainable from crystallographic data, and ``partial structure'' template maps derived from predicted structures deposited in the AlphaFold Protein Structure Database with subsequently omitted residues. With these, we predict electron density maps that are then post-processed into atomic models through standard crystallographic refinement processes. Introducing an initial dataset of small protein fragments taken from Protein Data Bank entries and placing them in hypothetical crystal settings, we demonstrate that our method is effective at both improving the phases of the crystallographic structure factors and completing the regions missing from partial structure templates, as well as improving the agreement of the electron density maps with the ground truth atomic structures.

physics.bio-ph↗

arXiv2025

TwIST: Rigging the Lottery in Transformers with Independent Subnetwork Training

We introduce TwIST, a distributed training framework for efficient large language model (LLM) sparsification. TwIST trains multiple subnetworks in parallel, periodically aggregates their parameters, and resamples new subnetworks during training. This process identifies high-quality subnetworks ("golden tickets") without requiring post-training procedures such as calibration or Hessian-based recovery. As a result, TwIST enables zero-cost pruning at deployment time while achieving perplexity competitive with state-of-the-art post-training sparsification methods. The benefits are most pronounced under aggressive sparsity (e.g., 50%+), where TwIST significantly outperforms baseline methods; for example, reaching 23.14 PPL compared to 31.64 for the closest prior approach. Unlike unstructured pruning, TwIST produces structured, dense matrices that offer practical inference speedups and memory reductions on commodity hardware (e.g., CPUs) that do not support efficient sparse computation. TwIST provides an efficient training-time path to deployable sparse LLMs without additional fine-tuning or recovery overhead.

cs.LG↗

arXiv2025

Three Birds with One Stone: Improving Performance, Convergence, and System Throughput with Nest

Variational quantum algorithms (VQAs) have the potential to demonstrate quantum utility on near-term quantum computers. However, these algorithms often get executed on the highest-fidelity qubits and computers to achieve the best performance, causing low system throughput. Recent efforts have shown that VQAs can be run on low-fidelity qubits initially and high-fidelity qubits later on to still achieve good performance. We take this effort forward and show that carefully varying the qubit fidelity map of the VQA over its execution using our technique, Nest, does not just (1) improve performance (i.e., help achieve close to optimal results), but also (2) lead to faster convergence. We also use Nest to co-locate multiple VQAs concurrently on the same computer, thus (3) increasing the system throughput, and therefore, balancing and optimizing three conflicting metrics simultaneously.

quant-ph↗

arXiv2025

Guided by the Experts: Provable Feature Learning Dynamic of Soft-Routed Mixture-of-Experts

Mixture-of-Experts (MoE) architectures have emerged as a cornerstone of modern AI systems. In particular, MoEs route inputs dynamically to specialized experts whose outputs are aggregated through weighted summation. Despite their widespread application, theoretical understanding of MoE training dynamics remains limited to either separate expert-router optimization or only top-1 routing scenarios with carefully constructed datasets. This paper advances MoE theory by providing convergence guarantees for joint training of soft-routed MoE models with non-linear routers and experts in a student-teacher framework. We prove that, with moderate over-parameterization, the student network undergoes a feature learning phase, where the router's learning process is ``guided'' by the experts, that recovers the teacher's parameters. Moreover, we show that a post-training pruning can effectively eliminate redundant neurons, followed by a provably convergent fine-tuning process that reaches global optimality. To our knowledge, our analysis is the first to bring novel insights in understanding the optimization landscape of the MoE architecture.

cs.LG↗

arXiv2025

ADAPT: Lightweight, Long-Range Machine Learning Force Fields Without Graphs

Point defects play a central role in driving the properties of materials. First-principles methods are widely used to compute defect energetics and structures, including at scale for high-throughput defect databases. However, these methods are computationally expensive, making machine-learning force fields (MLFFs) an attractive alternative for accelerating structural relaxations. Most existing MLFFs are based on graph neural networks (GNNs), which can suffer from oversmoothing and poor representation of long-range interactions. Both of these issues are especially of concern when modeling point defects. To address these challenges, we introduce the Accelerated Deep Atomic Potential Transformer (ADAPT), an MLFF that replaces graph representations with a direct coordinates-in-space formulation and explicitly considers all pairwise atomic interactions. Atoms are treated as tokens, with a Transformer encoder modeling their interactions. Applied to a dataset of silicon point defects, ADAPT achieves a roughly 33 percent reduction in both force and energy prediction errors relative to a state-of-the-art GNN-based model, while requiring only a fraction of the computational cost.

cs.LG↗

arXiv2025

Thinking Out of the Box: Hybrid SAT Solving by Unconstrained Continuous Optimization

The Boolean satisfiability (SAT) problem lies at the core of many applications in combinatorial optimization, software verification, cryptography, and machine learning. While state-of-the-art solvers have demonstrated high efficiency in handling conjunctive normal form (CNF) formulas, numerous applications require non-CNF (hybrid) constraints, such as XOR, cardinality, and Not-All-Equal constraints. Recent work leverages polynomial representations to represent such hybrid constraints, but it relies on box constraints that can limit the use of powerful unconstrained optimizers. In this paper, we propose unconstrained continuous optimization formulations for hybrid SAT solving by penalty terms. We provide theoretical insights into when these penalty terms are necessary and demonstrate empirically that unconstrained optimizers (e.g., Adam) can enhance SAT solving on hybrid benchmarks. Our results highlight the potential of combining continuous optimization and machine-learning-based methods for effective hybrid SAT solving.

cs.LO↗

arXiv2025

One Rank at a Time: Cascading Error Dynamics in Sequential Learning

Sequential learning -- where complex tasks are broken down into simpler, hierarchical components -- has emerged as a paradigm in AI. This paper views sequential learning through the lens of low-rank linear regression, focusing specifically on how errors propagate when learning rank-1 subspaces sequentially. We present an analysis framework that decomposes the learning process into a series of rank-1 estimation problems, where each subsequent estimation depends on the accuracy of previous steps. Our contribution is a characterization of the error propagation in this sequential process, establishing bounds on how errors -- e.g., due to limited computational budgets and finite precision -- affect the overall model accuracy. We prove that these errors compound in predictable ways, with implications for both algorithmic design and stability guarantees.

cs.LG↗

arXiv2025

Exploring How LLMs Capture and Represent Domain-Specific Knowledge

We study whether Large Language Models (LLMs) inherently capture domain-specific nuances in natural language. Our experiments probe the domain sensitivity of LLMs by examining their ability to distinguish queries from different domains using hidden states generated during the prefill phase. We reveal latent domain-related trajectories that indicate the model's internal recognition of query domains. We also study the robustness of these domain representations to variations in prompt styles and sources. Our approach leverages these representations for model selection, mapping the LLM that best matches the domain trace of the input query (i.e., the model with the highest performance on similar traces). Our findings show that LLMs can differentiate queries for related domains, and that the fine-tuned model is not always the most accurate. Unlike previous work, our interpretations apply to both closed and open-ended generative tasks

cs.LG↗

arXiv2025

Quantum EigenGame for excited state calculation

Computing the excited states of a given Hamiltonian is computationally hard for large systems, but methods that do so using quantum computers scale tractably. This problem is equivalent to the PCA problem where we are interested in decomposing a matrix into a collection of principal components. Classically, PCA is a well-studied problem setting, for which both centralized and distributed approaches have been developed. On the distributed side, one recent approach is that of EigenGame, a game-theoretic approach to finding eigenvectors where each eigenvector reaches a Nash equilibrium either sequentially or in parallel. With this work, we extend the EigenGame algorithm for both a $0^\text{th}$-order approach and for quantum computers, and harness the framework that quantum computing provides in computing excited states. Results show that using the Quantum EigenGame allows us to converge to excited states of a given Hamiltonian without the need of a deflation step. We also develop theory on error accumulation for finite-differences and parameterized approaches.

quant-ph↗