arXiv subjects

Sung-Min Park

Publications and source records attributed to Sung-Min Park.

4 recordsLinked to original sources

arXiv2026

Spin-orbit-induced Instability and Finite-Temperature Stabilization of a Triangular-lattice Supersolid

Geometrically frustrated triangular-lattice magnets provide fertile ground for realizing intriguing quantum phases such as spin supersolids. A common expectation is that spin-orbit coupling (SOC), which breaks continuous spin rotational symmetry, destabilizes these phases by gapping their low-energy modes. Revisiting this assumption, we map out the SOC-field phase diagram of a frustrated triangular-lattice magnet using spin-wave theory and infinite density-matrix renormalization group (iDMRG) simulations. We find that while infinitesimally weak SOC indeed drives a zero-temperature instability of the supersolid by opening a gap, certain supersolid states remain thermodynamically stable at non-zero temperatures. This reveals a previously unrecognized mechanism in which thermal fluctuations counteract SOC to stabilize supersolidity. The resulting finite-temperature supersolids retain key responses, including a giant magnetocaloric effect, highlighting their potential relevance to real materials. At larger SOC, the system transitions into distinct magnetic orders, including a skyrmion lattice, completing a unified phase diagram.

cond-mat.str-el↗

arXiv2024

Geometric additivity of modular commutator for multipartite entanglement

A recent surge of research in many-body quantum entanglement has uncovered intriguing properties of quantum many-body systems. A prime example is the modular commutator, which can extract a topological invariant from a single wave function. Here, we unveil novel geometric properties of many-body entanglement via a modular commutator of two-dimensional gapped quantum many-body systems. We obtain the geometric additivity of a modular commutator, indicating that modular commutator for a multipartite system may be an integer multiple of the one for tripartite systems. Using our additivity formula, we also derive a curious identity for the modular commutators involving disconnected intervals in a certain class of conformal field theories. We further illustrate this geometric additivity for both bulk and edge subsystems using numerical calculations of the Haldane and $π$-flux models.

quant-ph↗

arXiv2023

Clinical Decision Transformer: Intended Treatment Recommendation through Goal Prompting

With recent achievements in tasks requiring context awareness, foundation models have been adopted to treat large-scale data from electronic health record (EHR) systems. However, previous clinical recommender systems based on foundation models have a limited purpose of imitating clinicians' behavior and do not directly consider a problem of missing values. In this paper, we propose Clinical Decision Transformer (CDT), a recommender system that generates a sequence of medications to reach a desired range of clinical states given as goal prompts. For this, we conducted goal-conditioned sequencing, which generated a subsequence of treatment history with prepended future goal state, and trained the CDT to model sequential medications required to reach that goal state. For contextual embedding over intra-admission and inter-admissions, we adopted a GPT-based architecture with an admission-wise attention mask and column embedding. In an experiment, we extracted a diabetes dataset from an EHR system, which contained treatment histories of 4788 patients. We observed that the CDT achieved the intended treatment effect according to goal prompt ranges (e.g., NormalA1c, LowerA1c, and HigherA1c), contrary to the case with behavior cloning. To the best of our knowledge, this is the first study to explore clinical recommendations from the perspective of goal prompting. See https://clinical-decision-transformer.github.io for code and additional information.

cs.AI↗

arXiv2017

Prediction of Daytime Hypoglycemic Events Using Continuous Glucose Monitoring Data and Classification Technique

Daytime hypoglycemia should be accurately predicted to achieve normoglycemia and to avoid disastrous situations. Hypoglycemia, an abnormally low blood glucose level, is divided into daytime hypoglycemia and nocturnal hypoglycemia. Many studies of hypoglycemia prevention deal with nocturnal hypoglycemia. In this paper, we propose new predictor variables to predict daytime hypoglycemia using continuous glucose monitoring (CGM) data. We apply classification and regression tree (CART) as a prediction method. The independent variables of our prediction model are the rate of decrease from a peak and absolute level of the BG at the decision point. The evaluation results showed that our model was able to detect almost 80% of hypoglycemic events 15 min in advance, which was higher than the existing methods with similar conditions. The proposed method might achieve a real-time prediction as well as can be embedded into BG monitoring device.

stat.ML↗