Our Work

Universities play a crucial role in the production and dissemination of knowledge, acting as pillars of intellectual, scientific, and cultural progress. As dynamic hubs of research and innovation, the…

Universities play a crucial role in the production and dissemination of knowledge, acting as pillars of intellectual, scientific, and cultural progress. As dynamic hubs of research and innovation, they generate groundbreaking ideas, technologies, and discoveries that shape societies, stimulate economic growth, and tackle pressing global challenges. The knowledge produced within universities not only advances science, medicine, and technology but also enriches the humanities, social sciences, and the arts, fostering a deeper understanding of the world and human experience. Beyond creating knowledge, universities bear the important responsibility of transmitting this wealth of information to students, equipping them with critical thinking skills, specialized expertise, and the ability to address complex societal issues. By nurturing creativity, curiosity, and intellectual independence, universities prepare individuals not only for careers but also for active and informed citizenship in an ever-changing world. However, access to education should not be hindered by financial barriers. The democratization of knowledge is essential to ensuring equal opportunities for all, regardless of socioeconomic background. Providing higher education without costs upholds the principle that knowledge is a public good, meant to benefit society as a whole, rather than being restricted to those who can afford it. Free education promotes social mobility, empowers individuals, and strengthens economies by cultivating a skilled and knowledgeable workforce. More importantly, it affirms the fundamental belief that education is a right, not a privilege—one that drives innovation, supports social justice, and creates a more equitable and prosperous future. These guidelines are dedicated to university presses, particularly open access diamond ones, and focus solely on the publication of textbooks, a specific type of publication. They do not address organizational and management issues. The guidelines may also be useful for university presses that are not diamond but are interested in publishing open textbooks, potentially as a step toward transforming the press into a diamond one. This second edition is a revision of the previous text with a chapter on Accessibility and multilinguism

For over seven decades, computer architecture has operated under a single, unstated dogma: that physical space, continuous motion, field mechanics, and multi-agent coordination must be forcibly flatte…

For over seven decades, computer architecture has operated under a single, unstated dogma: that physical space, continuous motion, field mechanics, and multi-agent coordination must be forcibly flattened into the unoriented, one-dimensional byte array of the von Neumann machine. While this linear abstraction democratized software engineering, it severed computation from the native geometry and topology of the physical universe, imposing artificial concurrency locks, memory wall thrashing, and empirical termination heuristics upon spatial problems. Topological and Geometric Native Coding fundamentally reframes the nature of computation. Rather than executing sequences of instructions against passive, flat memory, execution proceeds directly over graded, cell-oriented topological complexes governed by the fundamental discrete boundary law: $$\partial \circ \partial = 0$$ In this paradigm, state is distributed across multi-dimensional cells (points, lines, surfaces, volumes, hypercells) carrying coordinate-free geometric fibers belonging to a Clifford algebra $\mathrm{Cl}(p,q,r)$. Control flow is not driven by arbitrary counter loops or boolean flags; it is driven by topological invariants. Execution iterates until spatial obstructions and non-contractible voids collapse, quantified by the kernel nullity of a discrete combinatorial Hodge Laplacian: $$L_k = B_k^\top B_k + B_{k+1} B_{k+1}^\top, \qquad \beta_k = \dim(\ker(L_k))$$ This repository contains the complete theoretical foundation, intermediate representation (MLIR dialect), formal verification suite (Lean 4), C-ABI specifications, and bare-metal execution runtime for ToposLang-Cubical—the compiler architecture that resolves the fundamental execution bottleneck of higher-dimensional programming languages. The Core Duality: Solving the Homotopy Scalability Dilemma ToposLang-Cubical operationalizes a profound structural duality long recognized in pure category theory but hitherto unexploited in systems engineering: discrete homological algebra and continuous constructive homotopy are not competing paradigms, but multi-scale expressions of the same underlying geometric reality. Prior to this work, implementations of topological computing faced an unyielding trade-off: The Global Rational Integer Explosion Problem ($\mathbb{Q}$-RREF): Standard discrete cell complexes evaluate exact invariants via Gaussian elimination on boundary matrices. Over rational fields ($\mathbb{Q}$), row pivot cross-multiplication induces Hadamard determinant growth, causing intermediate numerators and denominators to explode past thousands of bits ($O(n^3)$ operations). Execution throughput degrades by four orders of magnitude, saturating memory bandwidth. The Unbounded Kan Memory Problem (Cubical Type Theory): Constructive Cubical Type Theory (CTT) synthesizes continuous homotopies dynamically on-demand via Kan filling operations (hcomp). However, in higher dimensions, resolving boundary coherences generates infinite towers of higher homotopies ($\pi_1 \Rightarrow \pi_2 \Rightarrow \dots \Rightarrow \pi_\infty$). Allocated on dynamic heaps, these towers exhaust physical RAM and lack the global spatial routing required for local updates. The ToposLang-Cubical Solution: Global macro-spatial structure is governed by sparse integer boundary operators on physical memory. Whenever a localized topological obstruction arises ($L_k c \neq 0$), a Three-Tier Topological Trigger Engine extracts its minimal downward support and lowers it via a formal Realization Functor ($\mathcal{R}$) into a local, stack-allocated, $n$-truncated Kan composition engine. By evaluating continuous Lie group motor sandwiches ($PGA(3,0,1)$) inside zero-heap, cache-aligned 192-byte C-ABI stack frames (KanStackFrame), the runtime synthesizes terminal filling cells in $O(1)$ stack memory, completely bypassing global rational matrix inversion. Architectural & Theoretical Breakthroughs The Formal Realization Functor ($\mathcal{R}$): Mathematically bridges discrete chain complexes with constructive cubical sets: $$\mathcal{R} : \mathrm{Chain}_{\mathrm{Cl}(p,q,r)}(U) \longrightarrow \mathrm{CubSet}_{/X}$$ Compiles sparse boundary columns $(B_k)_{*j}$ into bitwise boolean face constraint masks $\varphi \in \mathrm{Face}(I^{k+1})$ without loss of algebraic orientation. Three-Tier Topological Trigger Engine: Eliminates the $O(n^3)$ cycle-detection loop bottleneck during invariant evaluation (until betti(k) == 0): Tier 1 ($O(1)$ Fast Path): Local support patching tracks boundary changes in $<50\,\mu\mathrm{s}$. Tier 2 ($O(\alpha(N))$ Filter): Lock-free Union-Find connectivity checks suppress $>77\%$ of false-alarm trigger checks. Tier 3 ($\mathbb{Z}_p$ Finite-Field Reduction): Sparse elimination over prime field $\mathbb{Z}_p$ ($p = 10007$) calculates exact homology ranks in fixed 64-bit machine integers, resolving macro Betti jumps in $\sim 1.6\text{ ms}$. Dual Cache-Aligned C-ABI (KanStackFrame): A strict 64-byte aligned, 192-byte C-compatible thread structure (3 cache lines) featuring a 64-bit face constraint mask (uint64_t) supporting complex polyhedral cells with up to 32 boundary faces ($k \le 31$) with zero heap allocation. Lie Manifold Projection Operator ($\pi_{\PGA}$): In-line projection ($\hat{r} = r/\Vert{}r\Vert{}$) during continuous Projective Geometric Algebra ($\mathrm{PGA}(3,0,1)$) versor sandwich evaluations ($u(t) = M(t) u_0 \widetilde{M}(t)$), holding floating-point norm drift to exact machine precision ($0.00 \times 10^0$) over $50{,}000+$ consecutive transformation steps. The topos_cubical MLIR Dialect: Lowering infrastructure providing native operations for Kan box extraction (bridge_lower), stack composition (hcomp), and dynamic degree pruning (truncate). Lock-Free Spatial Concurrency: Spatially disjoint wavefronts ($\mathrm{supp}(r_1) \cap \mathrm{supp}(r_2) = \varnothing$) enable parallel cellular rewriting with zero lock contention and zero cache invalidation across 128 physical cores. Empirical Benchmark Performance Evaluated on a dual-socket workstation equipped with two AMD EPYC 7763 processors (128 physical cores, 256 logical threads, 256 MB L3 cache per socket), 512 GB DDR4 ECC RAM, running Linux kernel 6.8: Metric / Scenario Classical $\mathbb{Q}$-RREF ToposLang-Cubical Bridge Performance Gain Cycle Resolution Latency ($N = 4,096$ edges) 4,112.5 ms 0.152 ms 27,055.9$\times$ Speedup Max Intermediate Bit-Length 8,192 bits Fixed 64-bit / Double Zero Bit-Length Growth Execution Memory Footprint Dynamic Heap ($O(n^3)$) 192 Bytes Stack ($O(1)$) Bounded Memory Invariant Parallel Rewrite Throughput (128 Cores) Unscalable (Lock Bottlenecks) 72,100 Rules / sec 87.9$\times$ Scaling Efficiency  

A Book of Abstracts of 1st International Conference on Magnetic Soft Matter for Robotics

The Stakeholder Engagement Toolkit takes a step forward from the experience and practical knowledge developed throughout the GRANULAR project, particularly through the Multi-Actor Labs and their initi…

The Stakeholder Engagement Toolkit takes a step forward from the experience and practical knowledge developed throughout the GRANULAR project, particularly through the Multi-Actor Labs and their initiatives. Rather than presenting a specific project methodology, the toolkit provides, especially for the beginners, a set of principles and practical guidance that can be used flexibly in different territorial contexts and alongside other materials and resources. This is a toolkit that can support territorial actors during their stakeholder engagement process. It has theambition to support the “leading actor” that can be a civil servant, a researcher, a local actor or a consultantduring these processes and guiding you through the different stages and operational strategies, from the set-up of the group to the monitoring activities, within advice and practical examples for finding a common solution based on the continuous engagement and process of knowledge exchange. The toolkit is not a linear guideline, instead is a guided process that can be revisited as new knowledge, actors and circumstances emerge. It is organised around principles emerging from the territorial experiences, including context-based, pluralistic and interactive approach. Its contents address key aspects of stakeholder engagement, such as understanding the territory, managing the process, the skills and knowledge required to work effectively with actors avoiding technicalities.

A lo largo del presente artículo nos disponemos, en primer lugar, a difundir la aparición de un corpus actual de literatura de migración escrita por mujeres de origen latinoameric…

A lo largo del presente artículo nos disponemos, en primer lugar, a difundir la aparición de un corpus actual de literatura de migración escrita por mujeres de origen latinoamericano en Europa, si bien fundamentalmente en España, para detallar sus particularidades y vincular así su existencia con el aumento de la presencia migrante y la feminización de los flujos migratorios desde Latinoamérica hacia el país ibérico. Así pues, podremos apreciar cómo en los textos aparece una hegemonía de materiales narrativosde índole cultural y, de manera más precisa, una minuciosa descripción de las relaciones de verticalidad surgidas entre el grupo cultural de poder al que intenta integrarse la mujer migrada en su nuevo medio y aquel del que esta procede. Por último, las teorías del investigador canadiense Erving Goffman actuarán como marco teórico frente al testimonio procedente de la experiencia de la escritora migrante de origen hondureño Cinthya Maldonado, recogido en su obra Mujeres Migradas(2018), para examinar, en última instancia, la resignificación del estigma social que emerge del contacto entre sendos grupos culturales imposibilitando así la integración de la desplazada en su nuevarealidad, así como la aparición colateral de violencias físicas y simbólicas.   This work focuses on the emergence of a corpus of contemporary migration literature written by Latin American women in Europe, especially in Spain. It describes its main characteristics and connects its origins with the increased presence of migrants and the feminization of the migratory flows coming from Latin America. From this perspective, it identifies a clear hegemony of cultural materials in these narratives and, more precisely, adetailed description of the relationships between the dominant cultural power group and the subaltern one. It also analyzes the literary presence of the social stigma described by Professor Erving Goffman (1963), which is now resignificated by the experience of these migrants. Thus, Goffman’s approach is used to examine Cinthya Maldonado’s testimony in Mujeres Migradas(2018) and prove how social stigma is reframed through the contact between both cultural power groups. Finally, it discusses how such stigma makes it impossible for female migrants to integrate into their new environment and leads to collateral physical and symbolic violences.

Nutqni diarizatsiya qilish ko‘p so‘zlovchili audioyozuvlarda “kim qachon gapirdi?” savoliga javob berib, nutq segmentlarini alohida so‘zlovchilarga ajratish vazifasidir. …

Nutqni diarizatsiya qilish ko‘p so‘zlovchili audioyozuvlarda “kim qachon gapirdi?” savoliga javob berib, nutq segmentlarini alohida so‘zlovchilarga ajratish vazifasidir. Ushbu maqolada x-vector/VBx, EEND, EEND-EDA, TS-VAD, EEND-vector klasterlash va WavLM asosidagi zamonaviy diarizatsiya usullari tahlil qilinadi. DER va JER baholash mezonlari hamda CALLHOME, AMI, DIHARD va VoxConverse mezoniy sinov to‘plamlari ko‘rib chiqiladi. Asosiy e’tibor kam resursli qoraqalpoq tiliga qaratiladi. 50 soatlik va 25 nafar ona tilida so‘zlashuvchini qamrab oluvchi Karakalpak Speech Corpus asosan avtomatik nutqni tanib olish uchun mo‘ljallangan bo‘lib, individual nutq ifodalaridan tashkil topgan. Shu sababli qoraqalpoq nutqini diarizatsiya qilish uchun so‘zlovchilar va ustma-ust nutq segmentlari belgilangan maxsus ko‘p-so‘zlovchili korpusni yaratish hamda oldindan o‘qitilgan modellarni qoraqalpoq nutqiga moslashtirish zarur.

This exploitation roadmap is a background document supporting the development of the Exploitation Strategy of the GRANULAR project, focusing on the European Open Accessibility Indicators and Methodolo…

This exploitation roadmap is a background document supporting the development of the Exploitation Strategy of the GRANULAR project, focusing on the European Open Accessibility Indicators and Methodological Framework, developed to support transparent, comparable and reproducible analyses of accessibility to services across European territories. The result combines an open European accessibility dataset, a methodological framework for calculating accessibility indicators, reproducible analytical workflows and tools for visualising and communicating the results. It addresses the difficulties researchers and public authorities face in obtaining consistent accessibility data across European regions, particularly when existing data sources are fragmented, methodologies are not transparent or analytical solutions depend on proprietary platforms. The data and visualisation solution were already available during the exploitation planning process. The exploitation strategy therefore focuses primarily on supporting their reuse, maintaining the underlying software and documentation, and extending the methodology to new applications. This exploitation roadmap follows the exploitation methodology and templates provided by the Horizon Results Booster for building exploitation. 

Vibe coding responsable es una guía para el profesorado que crea materiales educativos conversando con una inteligencia artificial que escribe el código, lo que se conoce como vibe coding. Reúne diez …

Vibe coding responsable es una guía para el profesorado que crea materiales educativos conversando con una inteligencia artificial que escribe el código, lo que se conoce como vibe coding. Reúne diez recomendaciones para publicarlos de forma responsable: revisar el contenido, no enviar datos personales a servicios ajenos al centro, entender qué hace el material, no depender de servicios que pueden desaparecer, hacerlo accesible, citar la autoría de lo ajeno, guardar el rastro de las decisiones, declarar el uso de IA, publicar con una licencia libre y ofrecer el código para que otras personas lo adapten. Cada recomendación distingue lo mínimo, al alcance de cualquier herramienta, y lo recomendado.El depósito contiene la guía completa en PDF, en castellano, catalán, gallego, euskera e inglés, y en cada idioma dos archivos de instrucciones para la IA: uno para crear un material siguiendo la guía y otro para evaluar un recurso ya hecho con la rúbrica VCER (vibe coding educativo responsable). La edición web está en https://vibe-coding-educativo.github.io/vibe-responsable/.Elaborada con IA generativa en el nivel 4 del MIAE (cocreación): el autor ha dirigido y corregido el contenido, ha contrastado las afirmaciones con sus fuentes y ha revisado el resultado. Las traducciones se han hecho de forma automática con la misma IA, a partir del original en castellano, y no las ha revisado ningún traductor profesional.