13979 research outputs found
Sort by
Nondegradative Synthetic Molecular Glues Enter the Clinic
Full text linkMolecular glues are small molecules that can induce or stabilize protein–protein interactions between proteins inside cells. Unlike classical small molecule drugs, molecular glues can target challenging disease‐causing proteins lacking well‐defined binding pockets. Nature has repeatedly used this mode of action, but identifying molecular glues for new target proteins has been a major challenge. Recently, manmade molecular glues, inspired by natural products, for KRas, entered clinical trials although KRas is a major cancer target long thought to be undruggable. Here, how these molecules are initially discovered and optimized to provide several advanced drug candidates for various KRas‐dependent cancer types are outlined. The major insights obtained for this new class of drug modalities are further summarized. These results showcase how molecular glues that do not rely on protein degradation can provide clinical benefits for challenging drug targets
The Clash of the Policies: The Joint Effect of EU Cohesion Policy and Common Agricultural Policy on the Public Support for European Integration
Full text linkA vital element of the European Union (EU) political system is the idea of further European integration. Whilst most existing works investigate the effects of the Cohesion Policy (CP), only a handful of studies consider the effects of the Common Agricultural Policy (CAP) on citizens' support for European integration. In this contribution, I argue that these two largest EU redistribution policies neutralise each other in their effects on the citizens' support for European integration: the CP improves public support due to its positively perceived redistributive nature and positive externalities, whilst the CAP decreases public support due to due to a limited target audience, somewhat elitarian distribution of resources and bureaucratic complexities. Using regional‐level CP and CAP data for 2017 and Eurobarometer surveys for 2018 and 2019, the empirical findings of multilevel regressions show that the CAP counteracts the positive effects of the CP on citizens' support for further European integration. The findings are robust to including a wide range of control variables, sample size change and alternative specifications of key independent variables. These findings also hint that policy evaluations should not disregard the differences between various policy instruments
Conformational Plasticity and Binding Affinity Enhancement Controlled by Linker Derivatization in Macrocycles
Full text linkMacrocycles are abundantly used by nature to enable cell‐permeable bioactive molecules. Synthetic non‐peptidic macrocycles are also increasingly considered as modalities for difficult‐to‐bind proteins but guidelines for macrocyclization are only beginning to emerge. Macrocycles are thought to constrain the available conformations but also to allow for residual flexibility, the latter being poorly understood. Here we show that even medium‐sized macrocycles display an unexpected high conformational plasticity, even when bound to their protein target. Minor modification of the linker region of macrocycles can shift the conformational ensemble to distinct conformational subclasses, each constituting distinct three‐dimensional scaffolds for further optimization. This led to several new ligands with improved affinity and beneficial physicochemical parameters for the FK506‐binding protein 51, a promising target for depression, obesity and chronic pain. Importantly, none of the beneficial modifications could have been identified by classical medicinal chemistry as they only work in the macrocyclic context.
Our results show that macrocyclization can do more than keeping loose ends together but rather provide a platform for multiple series of macrocycles with distinct binding modes
Integrating regional survey data into life cycle assessment: prospective environmental consequences of directing apple pomace to insect farming
Full text linkPurpose: Insect cultivation on bio-residues exemplifies the circular bioeconomy (CBE) concept by integrating three core CBE strategies: the use of bio-residues, multi-output production chains, and cascading. The sustainability of CBE technologies using bio-residues needs to be evaluated on a case-by-case basis, taking into account regional aspects and the technologies’ future potential. This study provides methodological guidance for assessing the environmental consequences of diverting a by-product from its current to a future utilization pathway through a prospective and consequential life cycle assessment (LCA) at a regional level.
Methods: We illustrate the developed methods with a case study of black soldier fly larvae (BSFL) rearing on an apple pomace (AP)–based feed in Hesse, Germany. We analyze the regional AP situation regarding availability, seasonality, spatial distribution, and market situation, through an industry survey among Hessian press houses and integrate this information into our LCA model to scale it to the regional level, consider regional transport scenarios, and reflect regional market effects of diverting AP. Furthermore, we systematically upscale the process chain to an early industrial scale.
Results and discussion: The annual generation of 11,300 tFM AP in Hesse could supply a small industrial BSFL plant. AP occurs with high spatial density in the south of Hesse. Press houses are concerned that the cost of AP disposal will increase and that it will be difficult to find a recipient for AP in the future. These aspects support a new and centralized use of AP. However, its high seasonality and unstable nature is a hurdle. Diverting the AP annually generated in Hesse from its current utilization as biogas substrate or ruminant feed to insect farming results in a reduction compared to the status quo in land use by 6.4E + 05 to 6.8E + 05 m²a crop eq and in freshwater eutrophication by 3.7E + 03 to 3.8 E + 03 kg P eq when BSFL replace soybean meal. However, most environmental impacts show an increase, for example, global warming potential increases by 3.6E + 07 to 3.8E + 07 kg CO₂ eq. The LCA results identify improvement options such as reducing the feed conversion rate and optimizing the use of feed additives.
Conclusion: This study shows how the regional situation of an industrial by-product can be investigated through a regional industry survey and how this information can be integrated into a cLCA model using a methodological framework. We show that the market situation of by-products can differ regionally, emphasizing the need of region-specific integration into cLCA models
Phase structure and thermodynamics of cold and dense quark matter
No full textThe phase structure of quantum chromodynamics at low temperatures and finite density challenges our understanding ever since the theory of the strong interaction was formulated for the first time. While the vacuum phase structure is already non-trivial, i.e., governed by chiral symmetry breaking and the dynamical generation of hadrons, the ground state at high densities and low temperatures is by now commonly accepted to be a color superconductor. In this high-density regime, the thermodynamic properties are strongly affected by the color-superconducting gap in the excitation spectrum of the quarks.
To gain a deeper understanding of the phase structure of quantum chromodynamics at finite density, we develop advanced calculation techniques to treat the limit of vanishing temperature and finite chemical potential in a well-defined manner. In particular, we focus on a careful ordering of mathematical operations. In addition to that, we develop the Mathematica tool TensorBases, which simplifies the systematic derivation of the correlation functions required to study non-perturbative phenomena.
Using these computational techniques, we study quantum chromodynamics with two massless quark flavors from first principles using the functional renormalization group approach. This includes the characterization of the phase structure at zero temperature over a wide range of chemical potentials, solely via the identification of resonances in four-quark interaction channels generated by the fundamental quark-gluon interactions. Allowing for the dynamical formation of composite diquark and meson states in the renormalization group flow, which mimics the continuous transition in effective degrees of freedom from high to low energies, we calculate both the chiral condensate in the vacuum and the chirally symmetric two-flavor color-superconducting gap at large chemical potential in a unified framework.
We demonstrate that gap-induced corrections to the thermodynamic properties of quantum chromodynamics can be sizable at high densities. For this purpose, we expand the pressure about the color-superconducting ground state and the strong coupling constant up to the next-to-leading order. At very high densities, we find that the speed of sound and the pressure approach the conformal limit associated with the free quark gas from below. This is in agreement with standard perturbative calculations. Towards lower densities, we then find that the presence of a gap pushes the speed of sound above the conformal value. Taking into account results from chiral effective field theory, our findings suggest a maximum in the speed of sound at supranuclear densities that exceeds the conformal limit.
The astrophysically more relevant case of three-flavor quantum chromodynamics at high densities is expected to be a color superconductor in the color-flavor-locked phase. We extend our framework to this case and implement neutron-star conditions in terms of β-equilibrium, charge and color neutrality, within an expansion of the pressure in terms of the gap, the strong coupling, and the strange quark mass up to next-to-leading order. Utilizing a Bayesian analysis and taking into account constraints from chiral effective field theory and astrophysical observations, we constrain the size of the color-flavor-locking gap at perturbative densities. Even in this scenario, the color-superconducting ground state has a substantial effect on the thermodynamic properties of three-flavor quantum chromodynamics at high densities and enhances both the pressure and the speed of sound in accordance with constraints from astrophysical observations
Ein Beitrag zum dynamischen Verhalten beim Gewindebohren durch die Entwicklung eines analytischen Modells sowie den Einsatz eines sensorischen Werkzeughalters
No full textDas Gewindebohren zählt nach wie vor zu den am häufigsten eingesetzten Fertigungsverfahren zur Herstellung von Innengewinden.
Da in der Regel das Gewindebohren in den letzten Bearbeitungsschritten in der Fertigung eines Bauteils ausgeführt
wird, stellt es einen kritischen Fertigungsprozess dar, da infolge von Prozessfehlern, wie z. B. dem Werkzeugbruch, hohe
wirtschaftliche Schäden entstehen können. Aus dem gegenwärtigen Stand der Forschung geht zwar hervor, dass einige Forschungsarbeiten
zum Gewindebohrprozess im Hinblick auf die Verbesserung des Prozessverständnisses und der Zuverlässigkeit
des Gewindebohrens existieren, jedoch hier eine Wissenslücke zum dynamischen Verhalten beim Gewindebohren
vorliegt. Die wenigen existierenden Forschungsarbeiten zum dynamischen Verhalten beim Gewindebohren lassen unter anderem
den Einfluss des instationären Werkzeugeingriffs bzw. des Werkzeug-Werkstück-Kontaktes auf das dynamische Verhalten
unbeleuchtet. Um diese Forschungslücke zu beleuchten, wird im Rahmen der vorliegenden Arbeit ein Beitrag zum
Verständnis über das dynamische Verhalten von Gewindebohrern durch analytische und experimentelle Untersuchungen der
Werkzeugschwingungen des Gewindebohrwerkzeuges während des Gewindebohrprozesses geleistet. Um dieses Forschungsziel
zu erreichen, wurden drei Forschungsfragen abgeleitet und untersucht.
Die erste Forschungsfrage beschäftigt sich mit der analytischen Berechnung der instationären Eigenfrequenzen der Werkzeugschwingungen
des Gewindebohrwerkzeuges unter Berücksichtigung des instationären Werkzeugeingriffs. Zur Entwicklung
des analytischen Modells wurde die komplexe Werkzeuggeometrie des Gewindebohrwerkzeuges zunächst durch ein
segmentiertes Balkensystem mit kontinuierlicher Masse modelliert. Basierend auf den Erkenntnissen der Voruntersuchungen
konnte das System des analytischen Modells auf das Gewindebohrwerkzeug eingegrenzt und als ein einseitig eingespanntes
Balkensystem betrachtet werden, sofern kein Werkzeugeingriff vorliegt. Der instationäre Werkzeugeingriff wurde im analytischen
Modell durch eine Schiebehülse modelliert, um die zeitabhängige Länge des modellierten Gewindebohrwerkzeuges
infolge der axialen Vorschubbewegung der Prozesskinematik zu berücksichtigen. Die aufgrund der Prozesskinematik resultierenden
komplexen Bewegungsgleichungen konnten durch Berücksichtigung der zugrunde liegenden niedrigen Vorschubgeschwindigkeit
sowie durch eine quasi-statische Betrachtung der zeitabhängigen Länge des Balkensystems vereinfacht werden.
Hierdurch konnte die erste und zweite instationäre Biegeeigenfrequenz sowie die erste instationäre
Torsionseigenfrequenz des modellierten Gewindebohrwerkzeuges analytisch berechnet werden.
Zur Untersuchung der zweiten und dritten Forschungsfrage wurde zunächst ein sensorischer Werkzeughalter entwickelt,
welcher die hochfrequente und hochauflösende Erfassung der Werkzeugschwingungen am Gewindebohrer während des Gewindebohrprozesses
ermöglicht. Hierbei kam ein eigens entwickelter Werkzeugschwingungssensor auf Basis von MEMSBeschleunigungssensoren
zum Einsatz, welcher unmittelbar an das Gewindebohrwerkzeug appliziert wird. Nach Validierung
des sensorischen Werkzeughalters im Feld wurde die zweite Forschungsfrage untersucht, welche sich mit der experimentellen
Bestimmung der Eigenfrequenzen des Gewindebohrwerkzeuges während des Gewindebohrprozesses unter Verwendung des
sensorischen Werkzeughalters beschäftigt. Auf Basis der Auswertung der experimentellen Ergebnisse im Zeit-Frequenz-Bereich
sowie der hinreichend guten Beschreibung der analytisch bestimmten instationären ersten Biegeeigenfrequenz konnte
gezeigt werden, dass während des Gewindebohrprozesses die erste Biegeeigenfrequenz angeregt wird und diese ein instationäres
Verhalten aufweist, wobei diese beim Hingang bis zum vollständigen Werkzeugeingriff steigt und beim Rückgang
entsprechend abnimmt. Die mithilfe des analytischen Modells berechnete zweite instationäre Biegeeigenfrequenz sowie die
instationäre erste Torsionseigenfrequenz konnten im Rahmen der experimentellen Untersuchungen unter Verwendung des
sensorischen Werkzeughalters nicht ermittelt werden.
Die dritte Forschungsfrage beschäftigt sich mit der Untersuchung des Einflusses des Werkzeugverschleißes auf die Werkzeugschwingungen
des Gewindebohrwerkzeuges. Hierbei wurde ein Verschleißversuch durchgeführt, wobei bis zum Auftreten
des Werkzeugbruchs die Werkzeugschwingungen unter Verwendung des sensorischen Werkzeughalters aufgezeichnet
wurden. Durch eine Auswertung der Ergebnisse im Zeit-Frequenz-Bereich sowie einer qualitativen Auswertung des Verschleißzustandes
des Gewindebohrwerkzeuges anhand mikroskopischer Aufnahmen konnte gezeigt werden, dass mit fortschreitendem
Werkzeugverschleiß eine hochfrequente Schwingungsanregung des Gewindebohrwerkzeuges beim Rückgang
auftritt. Mithilfe einer numerischen Modalanalyse konnte gezeigt werden, dass es sich bei dieser hochfrequenten Schwingungsanregung
um die erste Torsionseigenfrequenz des Gewindebohrwerkzeuges handelt. Im Ausblick der Arbeit werden
zum gegenwärtigen Stand der Forschung unbeleuchtete Forschungsfragen hinsichtlich des Einflusses anderer Prozessfehler
beim Gewindebohrprozess auf die Werkzeugschwingungen des Gewindebohrwerkzeuges formuliert
Advancing Machine Ethics: A Multi-Stage Approach to Revising AI Models
No full textAdvances in large-scale pretrained AI ("foundation") models such as GPT, CLIP, and Stable Diffusion have significantly transformed our technological landscape. Significant breakthroughs, for instance, in creative expression and commonsense reasoning, result primarily from these models' self-supervised training on vast, uncurated datasets. However, this powerful approach to learning inevitably captures not only human knowledge but also captures and intensifies inherent human biases, leading models to propagate harmful stereotypes and associations. Consequently, the promise of foundation models is accompanied by profound ethical and societal challenges, raising questions about whose values these systems reflect, what risks they entail, and how they can be responsibly governed.
Addressing these challenges, this thesis develops and systematically explores strategies for integrating machine ethics throughout the AI model pipeline, with interventions at three critical stages—data curation, model training, and inference-time adaptation.
First, at the data level, we present methods such as LlavaGuard, a vision-language framework designed to automatically audit visual datasets for unsafe, harmful, or stereotypical content, facilitating dataset filtering before training. Additionally, we explore synthetic data augmentation guided by vision-language models to diversify and debias dataset representations, highlighting its potential and limitations in mitigating biases and gender stereotypes.
Next, at the training stage, we introduce a typology for Explanatory Interactive Learning (XIL). This approach leverages human-provided explanations or automated explanatory feedback to prevent harmful shortcut learning and spurious correlations in AI model training. We demonstrate that even limited targeted interventions during training can significantly reduce shortcut learning and improve robustness and interpretability, thereby aligning model behavior more closely with human feedback.
Finally, recognizing that biases cannot always be fully mitigated earlier, we propose dynamic inference-time strategies including Revision Transformers (RiT), which enable post-training alignment of language models to ethical norms via retrieval-based, targeted human feedback, and FairDiffusion, allowing real-time steering of text-to-image models to ensure more equitable outcomes and mitigate inappropriate portrayals.
Together, these findings emphasize the necessity of a coordinated, holistic, multi-stage approach to machine ethics. We further highlight key practical, methodological, and ethical challenges, including resource constraints, limitations of synthetic augmentation techniques, oversimplified fairness assessments, and questions concerning value alignment. We outline promising avenues for future research, such as enhancing human oversight, context-sensitive interventions, and frameworks supporting more inclusive and representative machine ethics
Essays on the Interrelation between Corporate Finance and Corporate Sustainability
No full textSustainability is a paradigm permeating all aspects of business, including financial performance. Startups as innovation drivers are of particular interest in this context. However, our knowledge of how sustainability and performance are connected is limited for startups compared to large corporations. This doctoral dissertation addresses related research gaps with three empirical studies at the intersection of corporate finance and corporate sustainability.
The first study examines the relationship between startup sustainability signaling and financing success. The results provide the first large-scale empirical evidence for a U-shaped connection between startup sustainability and funding success: those that signal the most and the least raise the most venture capital. Sustainability signaling is, therefore, not exclusively positive or negative for startups but rather part of their strategic differentiation.
The second study explores how investors and startups select each other based on sustainability preferences, how investors influence the sustainability efforts of their startup investments, and the consequences for startup valuation. It provides empirical evidence that startups and investors prefer partners with a similar view on sustainability. As a result, investors are willing to pay a premium for an investment in a similar-minded firm. Moreover, it shows that startup sustainability communication increases after a green investor joins as an investor, while there is no change after a brown investor does so. Overall, the study contributes to a more nuanced understanding of treatment and selection mechanisms between investors and startups.
The last study delves into how the maturity of a startup’s business model influences the founding team’s exit options. It identifies that startups pioneering a new business model are the least likely to perform an initial public offering. However, they are also the most likely to exit via an acquisition. This study contributes to a more nuanced understanding of how firstmoving and business models affect founder’s exit prospects
Karlshof: seit 40 Jahren studentischer Lebensort
Full text linkIm August feiert Darmstadts wohl bekanntestes Studierendenwohnheim sein 40-jähriges Bestehen. Doch die Wurzeln des Karlshofs reichen viel weiter zurück in die Vergangenheit
Dauerhafte Freundschaft – Die TU-Freunde feiern 2018 ihren 100. Geburtstag
Full text link»Gäbe es die Freunde nicht ...« betitelte das Darmstädter Echo am 27. Mai 1967 einen Bericht über die vielfältigen Aktivitäten an der TH Darmstadt, die von den TU-Freunden unterstützt oder durch sie überhaupt erst möglich werden. Es ist eine immer wieder erwähnte Feststellung, die auf wirtschaftliche und politische Krisenzeiten besonders zutrifft