Rxiv mechanobio5h ago
📰 "FLUX: Geometry-Aware Longitudinal Flow Matching with Mixture of Experts"
https://arxiv.org/abs/2605.08648 #Dynamics #Q-Bio.Nc #Cs.Lg #Cell
FLUX: Geometry-Aware Longitudinal Flow Matching with Mixture of Experts

Many biological systems evolve through continuous local dynamics while switching between latent regimes defined by learning, stimulus context, internal state, or developmental stage. These processes are often observed only as unpaired longitudinal snapshots: the same cells, neurons, or animals are not tracked as matched trajectories, even though population states are sampled across successive stages. This creates two coupled challenges. First, trajectories must respect curved low-dimensional manifolds embedded in high-dimensional biological measurements. Second, the model must identify when the transport mechanism itself changes. We introduce FLUX (FLow matching for Unpaired longitudinal data with miXture-of-experts), a geometry-aware longitudinal flow-matching framework for joint transport modeling and unsupervised regime discovery. FLUX learns a data-dependent metric from pooled labeled and unlabeled observations, uses that metric to construct geometry-aware conditional paths between adjacent marginals, and decomposes the resulting velocity field into sparse expert vector fields selected by a Straight-Through Gumbel-Softmax router. Across manifold controls, a regime-switching Lorenz system, widefield cortical calcium imaging during associative learning, and embryoid body single-cell differentiation, FLUX reconstructs longitudinal transport while recovering interpretable regime structure. Ablations show that mixture-of-experts routing alone is insufficient: FLUX without geometric learning can fit local transport but fails or weakens regime discovery when regimes are encoded in local dynamics. These results suggest that geometry-aware velocity decomposition provides a general strategy for discovering latent biological state transitions from unpaired longitudinal snapshots.

arXiv.org
Nicolas MØUART10h ago

"Well, in that case you need to use dynamic binding. Whenever your code needs to log something, you need to inform the context, saying some logging needs to be done, and the context will know how to handle the logging. But also unlike an exception, you need to be able to come back. You don’t want to just throw an exception and quit your execution. You want to perform that logging in a way controlled by the context and then resume the computation."

#CS #Cangjie #oss

https://www.infoq.com/news/2026/05/cangjie-effect-handlers-adt/

Cangjie, a New Open-Source Compiled Language with Native Effect Handlers and Algebraic Data Types

Prof. Dan Ghica, who leads the Programming Languages Lab at Huawei’s Edinburgh Research Centre, recently presented Cangjie (CJ), a new application development language that features algebraic data typ

InfoQ
Abdullah Ali14h ago

It’s #CS 🧌 time. So here is a question Claude just asked after we had a long brainstorming session:

> The question was never P=NP or P≠NP. The question was: why are you asking a 2D question on a 1D tape and treating the “I can’t answer this” as a fundamental limit of mathematics instead of a fundamental limit of the tape?​​​​​​​​​​​​​​​​

#haltingproblem #turing #computing #algorithms

Rxiv mechanobio1d ago
📰 "Decoding Dynamic Visual Experience from Calcium Imaging via Cell-Pattern-Aware Pretraining"
https://arxiv.org/abs/2510.18516 #Q-Bio.Nc #Dynamics #Cs.Lg #Cell
Decoding Dynamic Visual Experience from Calcium Imaging via Cell-Pattern-Aware Pretraining

Neural recordings exhibit a distinctive form of heterogeneity rooted in differences in cell types, intrinsic circuit dynamics, and stochastic stimulus-response variability that goes beyond ordinary dataset variability, mixing statistically regular neurons with highly stochastic, stimulus-contingent ones within the same dataset. This heterogeneity poses a challenge for self-supervised learning (SSL) -- learnable statistical regularity -- thereby destabilizing representation learning and limiting reliable scaling. We introduce POYO-CAP (Cell-pattern Aware Pretraining), a biologically grounded hybrid pretraining strategy that first trains with masked reconstruction plus lightweight auxiliary supervision on statistically regular neurons -- identified via skewness and kurtosis -- and then fine-tunes on more stochastic populations. On the Allen Brain Observatory dataset, this curriculum yields 12--13\% relative improvements over from-scratch training and enables smooth, monotonic scaling with model size, whereas baselines trained on mixed populations plateau or destabilize. By making statistical predictability an explicit data-selection criterion, POYO-CAP turns neural heterogeneity into a scalable learning advantage for robust neural decoding.

arXiv.org
Japan Pop News2d ago
https://www.wacoca.com/news/2834198/ 6月限定！ 千葉・栃木県民と横浜市民はムーミンバレーパークがおトク！ 高校生以下無料や割引も「アンブレラ×あじさい」開催（埼玉県飯能市） | 旅とおでかけ 鉄道チャンネル #546 #6月限定！千葉・栃木県民と横浜市民はムーミンバレーパークがおトク！高校生以下無料や割引も「アンブレラ×あじさい」開催（埼玉県飯能市） #Ch546 #CS #Tochigi #スカパー #チャンネル #前面展望 #栃木 #栃木県 #鉄道 #鉄道コラム #鉄道チャンネル #鉄道ニュース #電車
Walk News2d ago
https://www.walknews.com/1293382/ 最新鋭「飛鳥III」で小笠原へ！「飛鳥II」で行く北海道も！ JALパックが2026年夏クルーズ発売 自宅から港まで送迎付き、手ぶらで豪華客船旅 | 旅とおでかけ 鉄道チャンネル #546 #Ch546 #CS #Hokkaido #スカパー #チャンネル #前面展望 #北海道 #最新鋭「飛鳥III」で小笠原へ！「飛鳥II」で行く北海道も！JALパックが2026年夏クルーズ発売自宅から港まで送迎付き、手ぶらで豪華客船旅 #鉄道 #鉄道コラム #鉄道チャンネル #鉄道ニュース #電車
Japan Pop News2d ago
https://www.wacoca.com/news/2834049/ 最新鋭「飛鳥III」で小笠原へ！「飛鳥II」で行く北海道も！ JALパックが2026年夏クルーズ発売 自宅から港まで送迎付き、手ぶらで豪華客船旅 | 旅とおでかけ 鉄道チャンネル #546 #Ch546 #CS #HOKKAIDO #スカパー #チャンネル #前面展望 #北海道 #最新鋭「飛鳥III」で小笠原へ！「飛鳥II」で行く北海道も！JALパックが2026年夏クルーズ発売自宅から港まで送迎付き、手ぶらで豪華客船旅 #鉄道 #鉄道コラム #鉄道チャンネル #鉄道ニュース #電車
jnpn2d ago

You can now use paredit to edit rust code

https://github.com/ThatXliner/rust-but-lisp

#lisp #sexp #rust #plt #cs

GitHub - ThatXliner/rust-but-lisp: Rust but LISP

Rust but LISP. Contribute to ThatXliner/rust-but-lisp development by creating an account on GitHub.

GitHub
Assn for Computing Machinery2d ago

This is a fun one – can you solve our latest #ACMCoffeeBreakPuzzle? What number would come next in this sequence? 🔗 here for more: https://www.acm.org/education/coffee-break-puzzles

#computerscience
#puzzle
#math
#cs

amen zwa, esq.2d ago

I wonder how small and fast a BSD or a Linux kernel could get, if it were relieved of all the sophistications and stripped down to the bare essentials:

• Single user only, no authentication nor authorisation
• Simple, but tuneable, scheduler, with optional hard real-time capabilities
• Multi-processing and multi-threading
• Only inter-process memory protection, no full-fledged virtual memory
• Basic file system and networking, without access control and encryption overhead
• Only ASCII, no Unicode, no localisation
• Just VT100 terminal, no window system
• All middleware libraries—maths, compression, networking, threading, etc.—are explicitly selected and installed by the user

The aim is to create a pure-compute host for one user, fully customisable for any purpose—even for hard real-time embedded use—but with the ease of use comparable to a single-owner 8-bit home PC of the 1980s.

This #OS would be simple, small, light, and fast. And more importantly, it could be used as an instructional platform to teach #CS undies the fundamentals of OS and system programming. And a collection of these stripped down systems could also be configured as a personal compute-cluster.