We propose a finite-state Markov chain framework for tracking and forecasting the status of project portfolios. This approach enables forecasts of portfolio composition over time and the computation of long-run distributions of project outcomes. It supports strategic planning by identifying project success rates, average durations, and the balance of resource allocation between active and idle projects. From a managerial perspective, the model facilitates early detection of portfolio-level risks and provides a data-driven basis for adjusting resource deployment or re-prioritizing projects. We show that forecasts remain robust under moderate errors in model identification, enhancing the method’s practical applicability in environments with noisy or incomplete data. This work lays the foundation for a scalable, organization-wide mechanism to improve visibility into project dynamics and support evidence-based decision-making.

What is life? The most widely cited definition, adopted by NASA, describes it as “a self-sustaining chemical system capable of Darwinian evolution.” This replicator-centric view has served biology well, but it obscures the deeper architecture underlying adaptation. Genes, replicators, and chemical substrates are not themselves the engines of evolution; they are memory traces of a more general logic. We propose that the essence of life is the capacity to perform signal weighting — the recursive filtering, reinforcement, and recontextualization of signals in ways that sustain and extend adaptive complexity. From immune repertoires and microbial ecologies to neural networks and viral symbioses, systems evolve by refining which signals matter and how they are integrated over time. Crucially, such architectures tend toward mutualism as an attractor, coupling distinct signal-weighting engines into higher-order adaptive units. By reframing life in these terms, we move beyond substrate-bound definitions toward a principle that unites the biological and the artificial: life as the emergence of relevance optimization through recursive signal weighting. Definition: Life is any system in which at least two signal-weighting processes couple through mutualistic attraction, creating an architecture that sustains and extends adaptive complexity.

Melt electrowriting (MEW) is a distinct class of additive manufacturing technologies that generates fibrous and porous macrostructures with microscale resolution from an electrically charged molten polymer. The high-resolution scaffolds produced by MEW have been primarily used for tissue engineering, cancer research, biofabrication and biomaterials. Unfortunately, the commercial MEW device’s prohibitive affordability and lack of standardization of custom devices, represent obstacles to further research. Built on the achievements and affordability of material extrusion 3D printers, we convert an open-source Voron 0.1 printer into a highly capable MEW device, termed as MEWron. To guarantee availability, the use of commercial and affordable components is prioritized, while in the cases where this has not been possible, 3D printed, and easy-to-machine components have been employed. Two main approaches have been followed, the first one focused on the existing material extrusion configuration (i.e., filament-based feeding system and material input) while the second one focuses on a conventional MEW pneumatic feeding system and syringe reservoir. When not including the high voltage supply, both approaches have a final budget below $1000. The manuscript describes all required steps and components to modify a Voron 0.1 printer and provides the computer-aided design (CAD) for required custom components reproduction. Moreover, the MEWron devices' reliability is demonstrated, as well as their potential to extend the MEW field boundaries. We believe that the open-source MEWron device will facilitate unprecedented MEW technology accessibility using a well-established and modifiable platform. (Abstract from https://doi.org/10.1016/j.addma.2023.103604)

To aid researchers in building their own MEWron, we have collected everything in this place. With this release you will find the full source code and build guides to build a MEWron with flat or tubular collector, based on a filament feeding system. Further, here you can find code generators for tubular codes that run on python.

Full Changelog: https://github.com/EPFL-LMIS1/LMIS1_MEWron/commits/MEWron

Strained ring systems such as cyclobutanes offer unique opportunities for synthetic innovation. Its significant ring strain, caused by compressed bond angles, presents synthetic challenges; however, this same strain renders the ring a valuable handle for activation and subsequent functionalization for the construction of complex molecular structures that would otherwise be difficult to obtain through conventional methods. As such, the development of efficient strategies to construct and transform cyclobutane scaffolds remains a compelling area of research, with broad implications for the synthesis of complex, bioactive molecules. Heteroatoms play a major role in medicinal chemistry by modulating the physicochemical properties of drug candidates, including electronic properties, hydrogen bonding potential, and metabolic stability. In this context, heteroatom-substituted cyclobutanes are increasingly used in pharmaceuticals and agrochemicals, where they exhibit strong pharmacological potential. In particular, β-disubstituted cyclobutanes have gained considerable attention due to their conformational rigidity induced by the cyclobutane ring. Nevertheless, general synthetic access to heteroatom-substituted β-cyclobutanes remains largely underexplored and generally lacks efficiency, selectivity, and broad applicability. The first objective of this thesis was to establish a synthesis of novel donor-acceptor amino-cyclobutane monoesters, and to use these new compounds in catalytic, diastereoselective [4+2] annulation reaction with indoles as partners. The silylium-catalyzed methodology developed was applied both inter- and intramolecularly, with the latter yielding divergent alkaloid scaffolds depending on the reaction temperature. DFT studies rationalized this divergent outcome, and the synthetic utility of the method was demonstrated in the divergent synthesis of structurally diverse alkaloids, starting from simple and commercially available starting materials. The second part of this thesis focused on expanding the scope of available aminocyclobutane esters. The previously developed approach, relying on in situ dehydrobromination and subsequent Michael addition, to synthesize donor-acceptor aminocyclobutane monoesters was streamlined and expanded to a wide variety of relevant N-heterocycles, such as imidazoles, azoles, and nucleobase derivatives, allowing the elaboration of a new library of aminocyclobutane esters and amides.
The scope of this approach using other hetero-atom nucleophiles, such as sulfur nucleophiles, was then investigated. As a result, we developed the diastereoselective synthesis of thiocyclobutane esters and amides. Moreover, we achieved the enantioselective version of the sulfa-Michael addition onto cyclobutenes using a chiral cinchona squaramide organocatalyst, affording 1,2-substituted thiocyclobutanes with high enantioselectivity. Finally, attention was turned to the synthesis of multi-substituted aminocyclobutane esters. This led us to the development of the first [2+2] cycloaddition between ynamides and simple acrylates. The resulting stable cyclobutenes obtained were selectively hydrogenated to afford multi-substituted aminocyclobutane monoesters with two distinct stereoconfigurations, offering a practical and modular route to novel β-cyclobutane amino acid derivatives.

Channel coding has become an indispensable component of modern communication systems. By introducing redundancy into the transmitted data, it enables the receiver to detect and correct errors without feedback, thereby significantly improving the efficiency of transceivers. As wireless standards continue to evolve, particularly toward 6G, channel coding faces ever greater demands on performance, flexibility, and implementation efficiency. However, the increasing complexity of advanced decoders, combined with the slowdown of Moore's law, has made it challenging to bridge the gap between coding theory and hardware implementation. Given that polar codes and low-density parity-check (LDPC) codes have been ratified as the 5G New Radio (NR) standard codes, the design of efficient decoders for these two modern codes has become a central bottleneck for 5G-NR and beyond.

In addition to support for varying block sizes, code rates, and code structures, 5G-NR imposes strict demands on polar codes and LDPC codes in terms of reliability, latency, and throughput. These challenges call for a cross-layer design perspective to tightly integrate algorithmic efficiency with architectural scalability. As a newly adopted class of codes in 5G, polar codes face challenges in low-latency and high-reliability of decoders. While node-based successive cancellation list (SCL) decoding has emerged as an effective approach to reduce the latency of SCL decoding, current node-based decoders are often constrained by their limited ability to generalize across diverse node types. This restricts their flexibility and leaves room for improvement in decoding speed. In this thesis, we propose a generalized node-based SCL decoder to minimize latency in single-frame polar decoding. Moreover, we introduce a frame-interleaving architecture to explore the throughput potential of polar decoders. For LDPC codes, the primary challenge in 5G-NR is to support a wide range of code configurations while meeting the required peak throughput. To this end, we develop a fully reconfigurable 5G-NR LDPC decoder architecture to fulfill these requirements. In summary, we present algorithmic and architectural optimizations for 5G-NR codes, along with two decoder implementations that are currently among the most efficient designs meeting the standard requirements.

Meanwhile, as 5G is making inroads to commercial devices, the global research community is actively exploring candidates for 6G channel coding. In contrast to 5G-NR, where code constructions are already fixed, the transition to 6G offers an opportunity to incorporate code design itself into the innovation process, prompting a revisit of coding schemes, decoding algorithms, and hardware architectures. In this thesis, we propose a new spatially-coupled LDPC code family called edge-spreading Raptor-like (ESRL) LDPC codes as a candidate for 6G next-generation mobile broadband. While preserving key features of 5G-NR standard codes, the proposed ESRL codes have advantages in error-rate performance, throughput, and hardware complexity compared to 5G-NR LDPC codes. To effectively realize the theoretical advantages in practice, we further present a fully reconfigurable high-throughput LDPC decoder implementation for ESRL codes. This ASIC can support a wide range of code rates and code lengths (up to five times longer than 5G) and achieve a high peak throughput of more than 100 Gbps, making it a promising solution for 6G wireless systems.

Quantum computing is expected to complement classical von-Neumann architectures in solving problems beyond the reach of conventional digital machines. This new model of computation leverages the rich complexity of many-body quantum systems to perform computations in a space that scales exponentially with the number of quantum bits without incurring an exponential cost in resources such as time, space, or energy. Over the past decade, remarkable experimental progress in quantum processing units has been made possible through the convergence of advances in information theory, quantum physics, materials science, and electrical engineering. Current research is actively benchmarking various platforms to implement scalable quantum architectures with high fidelity operation, aiming to integrate an increasing number of interconnected qubits on a single platform. This thesis investigates technological advancements in semiconductor spin qubits, with a focus on silicon-based quantum device architectures leveraging fully-depleted silicon-on-insulator (FD-SOI) technology. Emphasis is placed on quantum confinement, spin qubit design, and fabrication techniques that enable a compact device integration. Theoretical foundations are provided through an overview of quantum confinement in nanostructures and its application to single-electron transistors (SETs) and spin qubits. Key qubit types based on semiconductor quantum dots including Lossâ DiVincenzo, singletâ triplet, and exchange-only configurations are discussed. Electric-dipole spin resonance (EDSR) is presented as a method for operating single-qubit gates with high frequency and fidelity. Our experimental work explores advanced fabrication strategies for thin and ultra-thin SOI substrates, optimized for reproducibility and scalability. The successful realization of multi-gate FD-SOI SETs, demonstrating Coulomb oscillations at 4K and 10mK, confirms the compatibility of these devices with standard CMOS processes and supports the development of SETâ MOS hybrid circuits for both digital and analog applications. A novel back-gating approach, based on the so-called "Nanomole" fabrication process is introduced to achieve implantation-free dual-gate control at cryogenic temperatures. This method enhances electrostatic tunability and supports symmetric gating in FET devices at millikelvin temperatures. Simulations reveal that back-gate biasing offers in situ control over quantum dot shape and position, enabling tuning of key parameters such as volume inversion, valley splitting, and spinâ orbit coupling. Furthermore, the front-end-of-line (FEOL) integration of cobalt nanomagnets into spin qubit architectures is explored to enable fast, localized EDSR driving. Thin-film characterization and micromagnetic simulations support the design of three spin qubit architectures optimized for high and low magnetic field operation, featuring Co depletion gates, wrap-gates and side-horseshoe nanomagnets. While charge noise and contact resistance in ultra-thin SOI remain challenges for future efforts to benchmark qubit operation, proof-of-concept devices demonstrate the feasibility of FD-SOI-based quantum technologies with integrated ferromagnetic components. This work provides the experimental foundation for the development of scalable, CMOS-compatible quantum devices, highlighting innovative fabrication processes and architectures tailored for the monolithic integration of FD-SOI quantum and classical processing units.

Heparin and heparan sulfate (HS) glycosaminoglycans (GAGs), essential components of the extracellular matrix (ECM), regulate a vast array of biological processes by modulating protein interactions and cellular signaling. Their structural complexity, driven by diverse sulfation patterns, underpins their broad physiological activities, including cell proliferation, differentiation, and neural development. However, the intrinsic heterogeneity of native HS complicates the elucidation of structure-activity relationships (SAR), hindering efforts to fully harness their therapeutic potential. This challenge underscores the need for structurally defined HS glycomimetics that can replicate the functional diversity of native HS with precise control and optimized therapeutic effects. 2 Here, we present a library of HS glycomimetics, rationally designed using molecular modeling and synthesized through a divergent synthesis strategy that allows sulfate groups to be installed at specific positions along the carbohydrate backbone. Biophysical characterizations unveil that these glycomimetics selectively bind and stabilize growth factors, including fibroblast growth factors (FGF-1, FGF-2) and nerve growth factor (NGF), in a sulfation-dependent manner without inducing anticoagulant activity, which is a critical prerequisite for successful clinical translation in nerve regeneration. The lead glycomimetic has neuritogenic ability because in two neuronal cell models, PC12 and SH-SY5Y, it enhances NGF-mediated neural maturation when immobilized on a surface. Moreover, functional studies in primary rat hippocampal neurons reveal that the lead glycomimetic potentiates FGF-2- mediated neurite outgrowth and spontaneous synaptic activity, effectively translating its molecular interactions into measurable cellular responses. By bridging molecular-level insights with functional bioactivity, this work establishes HS glycomimetics as precision tools for neurotrophic signaling. Their ability to fine-tune growth factor activity offers a versatile platform for regenerative medicine, extending beyond neural regeneration to broader tissue repair applications. These findings advance the development of next-generation carbohydrate-based therapeutics, unlocking new opportunities for precise and targeted regenerative strategies.

Recreating real-world objects and scenes in a visually plausible manner and making them editable with intuitive instructions is central to Virtual and Augmented Reality (VR/AR) applications. This research lies at the intersection of neural rendering, 3D content generation, and scene reconstruction. Radiance fields, built on volumetric primitives, effectively model real-world scenes for photorealistic novel view synthesis and are emerging as key tools for 3D representation. However, current methods face limitations in realism, editability, and dynamic modeling.

First, most radiance fields use emissive volumes rendered via rasterization, which restricts accurate light transport modeling and degrades plausibility for reflective and refractive materials. Second, volumetric primitives lack embedded semantic understanding, limiting user editability via text or image prompts. Third, extending radiance fields to 4D for dynamic scenes presents challenges in maintaining spatiotemporal consistency. This thesis addresses these gaps by proposing methods that improve expressiveness and controllability in 3D representations, focusing on transparency modeling, object removal and inpainting, and volumetric stylization.

To better model refractive materials, we introduce an end-to-end pipeline combining implicit Signed Distance Functions (SDFs) with a refraction-aware Ray Bending Network. This allows reconstruction and relighting of transparent objects with complex geometry and unknown indices of refraction. Our method separates geometry from appearance and compensates for the lack of sharp refractive details in volumetric fields, substantially improving novel-view synthesis and relighting quality.

Beyond object reconstruction, the ability to modify and manipulate existing representations is crucial for practical applications in content creation. To this end, we present the first text-guided object-inpainting pipeline for 360-degree Neural Radiance Field based scenes. The proposed pipeline achieves accurate semantic selection through depth-space warping, ensuring multiview-consistent segmentations. It refines inpainted regions using perceptual priors and 3D diffusion-based geometric constraints. This enables seamless object removal while preserving scene coherence, thereby significantly enhancing NeRF's adaptability for real-world scene editing.

We further explore the controllability of dynamic radiance fields by introducing a volumetric variant based on neural cellular automata, termed VNCA. This architecture generates spatiotemporally consistent appearances with naturally emerging motion, driven by input images. Unlike prior approaches, our VNCA model maintains both temporal and multiview consistency by integrating the emergence property of NCAs within an Eulerian framework and supervising motion with optical flow. Beyond smoke simulation, VNCA supports stylization of solid textures on meshes, demonstrating its versatility in dynamic texture synthesis.

By advancing techniques for reconstruction, editing, and stylization, this thesis contributes to the development of more controllable and visually plausible 3D representations. Our work paves the way for enhanced neural-rendering applications, from photorealistic content creation to interactive scene manipulation, hence making 3D content modeling both higher quality and more artist-friendly.