Complexity in rock fracture and earthquakes: Spontaneous localisation and catastrophic failure in a granular medium
Sala de graus Eduard Fontserè
Wednesday, December 1st, 10:30-11:30, Sala de graus Eduard Fontserè, Facultat Física UB
Speaker: Ian Main, University of Edinburgh
Abstract: Brittle fracture in the Earth occurs on a variety of scales, from grain scale fractures to plate bounding faults. The structures are remarkably self-similar on a broad bandwidth of scales, with power-law scaling rules emerging spontaneously from the interplay between fluctuations associated with structural disorder and interactions associated with the concentration and redistribution of stress. Here I will present a set of recent results obtained by carrying out live deformation experiments on small rock samples in a synchrotron, using x-rays to ‘see’ the deformation and more recently also acoustic wave monitoring to ‘hear’ the mini-earthquakes it produces.
For crystalline rocks, we find the starting porosity of the material has a significant effect on the nature of the acceleration of precursory damage to failure, with higher porosities leading to an inverse power law in acoustic emission event rate v. time, with a well-defined failure time at the singularity. Power-law scaling of fracture sizes emerges spontaneously, with an exponent that decreases with time. Lower porosity materials also exhibit some precursory behaviour, but failure occurs suddenly, and earlier than you would expect. Using a separate set of experiments on synthetic materials as a guide, the early failure time can be explained by differences in the mean distances between pores and cracks in the starting material, with more disorder (higher porosity and lower inter-flaw distances) leading to more predictable behaviour.
One of the big problems in understanding the processes that lead up to catastrophic failure is that they occur too quickly to be captured in the time it takes to obtain a computed tomography image of the deformation in a synchrotron experiment. Accordingly, we slowed down this process using feedback from the acoustic emission event rate, to capture a series of images like the one above left, and hence observe the different stages of localisation after the point of yield and beyond peak stress in a porous sandstone. Here we see spontaneous localisation of ‘en-echelon’ tensile cracks, initially parallel the vertical maximum principal stress, followed by rotation as a shear band develops. We can then use digital image correlation to observe the local strain and use this to help locate micro-seismic events as on the diagram of the top right. We find shear and tensile deformation are closely correlated. Surprisingly, seismic amplitude is not necessarily correlated with local imaged strain; large local strain often occurs with small acoustic emissions, and vice versa. Local strain is overwhelmingly (>99%) aseismic, explained in part by grain/crack rotation along an emergent shear zone, and the shear fracture energy calculated from local dilation and shear strain on the fault is half of that inferred from the bulk deformation. This improvement in process-based understanding holds out the prospect of reducing systematic errors in forecasting system-sized catastrophic failure in a variety of applications.
You can find the presentation here and the recording here.
Seminari Marta Recasens (Google Inc.): La semàntica de la intel·ligència artificial (obert a tot el públic)
Aula Capella, Edifici Històric UB
Malgrat els avenços recents de la intel·ligència artificial i fins i tot les veus que auguren els seus "perills", la realitat és que la semàntica i pragmàtica del llenguatge segueixen sent un gran repte per als sistemes automàtics de processament del llenguatge natural (PLN). En aquesta xerrada exemplificaré aquest repte a través de dues tasques de PLN: la resolució de la coreferència i els sistemes de diàleg. Oferiré una descripció general d'aquestes tasques juntament amb estudis recents i exemples per il·lustrar quins aspectes semàntics estan resolts i quins són encara el focus de la recerca actual, especialment des de la perspectiva de les aplicacions de PLN en productes comercials.
La Dra. Marta Recasens és investigadora a Google des del 2013. Anteriorment va ser investigadora postdoctoral a la Universitat de Stanford i membre del Stanford NLP Group. Va obtenir el doctorat en Lingüística a la UB l'any 2010 amb una tesi sobre la resolució de la coreferència. La seva investigació se centra en aproximacions empíriques i aplicades a la semàntica i pragmàtica del llenguatge des de la Lingüística Computacional.
James Sharpe (EMBL): Limb development, Turing patterns, and Computer modelling
Sala Eduard Fontseré, Facultat de Física Universitat de Barcelona
Dramatic progress has been made over the last 2 decades in how we access key types of biological data - in particular sequence-based data on genomic information. However, integrating this data to produce dynamic and predictive models of higher level biological phenomena (e.g. development, regeneration, homeostasis and cancer) has been limited. Questions about tissues and organs are still most often tackled at the molecular or cellular level. We tend to ask how individual progenitor cells respond to signals from their “environment”, and thus to focus on signal transduction pathways, gene regulatory events, and epigenetic memory. But an organ is more than just an environment for cells to “act” in – it is an integrated whole, a coherent community, with cells in constant genetic, chemical and mechanical communication with each other. New technical advances such as organoid culture, 3D mesoscopic imaging, multicellular omics and computer modeling are helping us to go beyond the molecular and cellular level, to understand multicellular feedback loops, long-range signalling networks and emergent collective decisions, and thus to see tissues and organs as systems in their own right. Modelling these higher-level processes in vitro and in silico will help us understand these complex processes at a deeper level, and I will discuss our own attempts in this direction, to understand one example of complex organogenesis – namely mammalian limb development.
NanoBioMedicine: Current technology, challenges and future guidelines
Sala Eduard Fontsere, Facultat de Fisica, UB
CODIRECTOR OF THE INSTITUTE OF NANOTECHNOLOGY, UNIVERSITY OF OXFORD
Nanotechnology is a new and exciting field that has the potential to transform the way medical and health solutions are being developed. In the Department of Physics, my group investigates new techniques and materials at a nanometric scale. In the Department of Zoology, my group applies this knowledge directly to know the most relevant biology on a single molecule scale and then use science and technology to solve the most urgent medical problems of the 21st century. During the talk, I will focus on describing the field of nanotechnology, current applications and the potential of future applications. I will also explain the latest basic research projects and medical applications that we are developing.
Bio: The group of Sonia Trigueros focuses on the design of the new system of distribution of genes and drugs to specific cells. It is also developing new nanomedicines to address the problem of resistance to bacterial antibiotics. She is a doctor in molecular biology for the IBMB-CSIC and the University of Barcelona. After his postdoctoral research scholarships at Harvard and Oxford Universities, Trigueros was visiting research at various academic institutions, including NIH-Washington and the University of Havana. He is currently an associate researcher at the Department of Zoology and the Department of Physics and co-director of the Oxford Institute of Martin NanoMedicina at the University of Oxford.
Mapping The Structure Of Meaning Across Human Languages With Large Scale Semantic Networks
Sala de Professors Facultat de Filologia Edifici Josep Carner, 5a planta c/ Aribau, 2 08007 Barcelona
Max Plank Institute for Psycholinguistics
"Mapping the structure of meaning accross human languages with large scale semantic networks"
BE AWARE OF POSSIBLE CHANGE LOCATION IN CASE OF PROBLEMS OF ACCESS TO THE BUILDING
WE WILL KEEP THIS POST UPDATED