École Polytechnique Fédérale de Lausanne
Infoscience - École polytechnique fédérale de LausanneNot a member yet
191401 research outputs found
Sort by
A Decoupled Approach for Composite Sparse-plus-Smooth Penalized Optimization
No full textWe consider a linear inverse problem whose solution is expressed as a sum of two components: one smooth and the other sparse. This problem is addressed by minimizing an objective function with a least squares data-fidelity term and a different regularization term applied to each of the components. Sparsity is promoted with an ℓ1 norm, while the smooth component is penalized with an ℓ2 norm. We characterize the solution set of this composite optimization problem by stating a Representer Theorem. Consequently, we identify that solving the optimization problem can be decoupled by first identifying the sparse solution as a solution of a modified single-variable problem and then deducing the smooth component. We illustrate that this decoupled solving method can lead to significant computational speedups in applications, considering the problem of Dirac recovery over a smooth background with two-dimensional partial Fourier measurements.LCAVAccepted papers will be published on IEEE Xplore©.
In addition, the proceedings are posted in Open Access on the Eurasip website
Modelling The Effect of Future Uncertainty in Energy Prices on Decarbonization Pathways for Secondary Aluminium Production
No full textAluminium is one of the most widely used metals in the world, but its production is highly energy intensive and largely dependent on the use of fossil fuels. A typical secondary aluminium production facility consumes 700-1,000 kWh of natural gas and 200-400 kWh of electricity per tonne of rolled aluminium sheets. Thus, in order to meet its global environmental targets, the aluminium industry is shifting towards alternative energy resources. Potential decarbonization routes include the use of biomass to replace fossil fuel via thermal gasification, the integration of carbon abatement and utilization units, the use of power-togas and energy storage systems, direct electrification of aluminium furnaces, installation of waste heat recovery units for power generation, among other alternatives. While most of these technologies have lifetimes of around 20-25 years, decisions on their installation must be made today. Biomass, electricity, and natural gas costs can be subject to unpredictable market variations, whereas carbon prices are related to both environmental regulations and future market situation. In this context, future uncertainty in energy prices should be accounted for in today's decisionmaking. This work presents a systemic approach for assessing the effect of uncertain energy prices on the performance of different decarbonization routes for secondary aluminium production. To this end, a mixed integer linear programming (MILP) approach is used to generate a list of optimal system configurations under different economic conditions. Next, Monte-Carlo simulations are applied to predict representative price trends of commodities and to compute the resilience of each decarbonization scenario based on the likelihood of its occurrence under varying monetary circumstances. Results indicate that decarbonization pathways have a 50% probability to be cheaper than fossil-based CO2 emitting configurations. Moreover, the decarbonization option with the highest probability of being the economic best was the "Elec-bio" cluster which relies on a combination of electricity and biofuels to operate the plant's furnaces. A probability of 22-37% was estimated for realizing a cheaper "Elecbio" solution with respect to the natural gas driven base case over a lifetime of 25 years. Additionally, CO2 tax must not be the only economic incentive for emissions reduction but needs to be coupled with increased fossil fuel prices. Finally, this type of analysis allows decision makers to appreciate the potentials and risks associated with future decarbonization routes.SCI-STI-F
Causal imitability under context-specific independence relations
No full textDrawbacks of ignoring the causal mechanisms when performing imitation learning have recently been acknowledged. Several approaches both to assess the feasibility of imitation and to circumvent causal confounding and causal misspecifications have been proposed in the literature. However, the potential benefits of the incorporation of additional information about the underlying causal structure are left unexplored. An example of such overlooked information is context-specific independence (CSI), i.e., independence that holds only in certain contexts. We consider the problem of causal imitation learning when CSI relations are known. We prove that the decision problem pertaining to the feasibility of imitation in this setting is NP-hard. Further, we provide a necessary graphical criterion for imitation learning under CSI and show that under a structural assumption, this criterion is also sufficient. Finally, we propose a sound algorithmic approach for causal imitation learning which takes both CSI relations and data into account.BA
Emergent coherent modes in nonlinear magnonic waveguides detected at ultrahigh frequency resolution
No full textNonlinearity of dynamic systems plays a key role in neuromorphic computing, which is expected to reduce the ever-increasing power consumption of machine learning and artificial intelligence applications. For spin waves (magnons), nonlinearity combined with phase coherence is the basis of phenomena like Bose-Einstein condensation, frequency combs, and pattern recognition in neuromorphic computing. Yet, the broadband electrical detection of these phenomena with high-frequency resolution remains a challenge. Here, we demonstrate the generation and detection of phasecoherent nonlinear magnons in an all-electrical GHz probe station based on coplanar waveguides connected to a vector network analyzer which we operate in a frequency-offset mode. Making use of an unprecedented frequency resolution, we resolve the nonlocal emergence of a fine structure of propagating nonlinear magnons, which sensitively depends on both power and a magnetic field. These magnons are shown to maintain coherency with the microwave source while propagating over macroscopic distances. We propose a multi-band four-magnon scattering scheme that is in agreement with the field-dependent characteristics of coherent nonlocal signals in the nonlinear excitation regime. Our findings are key to enable the seamless integration of nonlinear magnon processes into high-speed microwave electronics and to advance phase-encoded information processing in magnonic neuronal networks.LMG
Software optimization for a RISC-V accelerator: A case study
No full textWriting high-performance software these days is a challenging task. In the past, CPU performance scaled regularly with advances in device technology, leaving much of the work to algorithmic optimizations and reducing overhead. Since then, the scaling of CPU performance has waned significantly, and the hardware world has compensated with increasingly application-specific hardware, often known as accelerators. Despite years of effort in automation, by and large, the burden still falls on the developer to write a program in a manner that can take advantage of this hardware. In compute-bound workloads, the penalty for not doing so is severe: in the example of matrix multiplication, a nearly 1300x speedup was observed optimizing a plain C program to properly use the cache, SIMD, multicore, etc. Worst of all, the work to write such programs is often repeated with little reuse for each custom hardware targeted, creating a massive effort for the developer.
This issue will serve as the focal point for this report. We seek to understand the challenges associated with software development for accelerators and some of the proposed solutions for automation in the literature. However, this report is not intended to provide a comprehensive literature survey. Instead, we will investigate this question through a practical case study on developing software for a simple dense matrix multiplication accelerator. We will focus on solutions addressing the most salient challenges encountered. Given the prominence of this type of hardware to accelerate popular applications like deep learning [1, 2, 5, 6, 8], we hope for the findings of this case study to shed broader insights on this problem.SYSTEM
FINITE TIME BLOW UP FOR THE ENERGY CRITICAL ZAKHAROV SYSTEM II: EXACT SOLUTIONS
No full textBased on the companion paper [23], we show that the 4D energy critical Zakharov system admits finite time type II blow up solutions, similar to the ones constructed in [33]. The main new difficulty this work deals with is the appearance of a term in the linearization around the approximate solution, which is non-local with respect to both space and time. In particular this cannot be handled by straightforward adaptation of the methods developed in [22], [21], [20]. The key new ingredients we use are a type of approximate modulation theory, taking advantage of frequency localisations, and the exploitation of an inhomogeneous wave equation with both a non-local, as well as a local potential term. These terms arise for the main non-perturbative component of the ion density n and can be solved via inversion of a certain Fredholm type operator, as well as by using distorted Fourier methods. Our result relies on a number of numerical non-degeneracy assumptions.PD
Determining structural properties of a prestressed concrete bridge through the combination of static and dynamic load testing
No full textExamining structural safety requires hypotheses on several properties of the bridge structure, such as material properties, boundary conditions, and self-weight. The traditional approach relies on conservative assumptions for each structural property, following the conventional new-design approach. Nonetheless, this approach leads to conservative evaluations of the bridge capacity and may lead to the erroneous conclusion that the structure is deficient. Over-conservatism in structural safety assessments may have large environmental and economic impacts on global infrastructure management. A more advanced approach is to conduct multiple tests and monitoring activities on the structural system to provide more accurate values of these bridge properties. This paper presents a methodology to determine several parameters, including the structural stiffness, the boundary conditions, and the self-weight of concrete bridges based on static and dynamic load testing and robust data-interpretation techniques. The methodology is used on a prestressed concrete bridge in Switzerland. This bridge from 1958 has a single span of 35 meters. Prior to monitoring, conservative evaluations (using the conventional approach) led to the conclusion that the bridge has structural deficiencies. After monitoring, the bridge demonstrates significant reserve capacity, mostly due to the reduction of the self-weight safety factor. This study shows the potential of monitoring techniques for more sustainable and economic infrastructure management.GIS-G
Heat transfer modeling and waste heat management in the furnaces of secondary aluminum production
No full textAluminum recycling significantly contributes to supplying the growing demand for aluminum alloys within the framework of a circular economy. Recycling leads to reducing the negative economic and environmental impact of primary aluminum production. Enhancing energy efficiency to reduce carbon footprint in secondary aluminum production is a necessity for meeting the environmental regulations. This project is conducted during an internship at EPFL, and in collaboration with an industrial company. In the current study, several potential solutions are proposed and analyzed to improve energy efficiency in both remelting and rolling plants of secondary aluminum production. To characterize the thermal performance of the furnaces and evaluate the proposed strategies for waste heat management, a heat transfer model is developed using the finite difference method and machine learning approaches. Four regression models are trained and checked for predicting the heat transfer coefficient in preheater and ACL furnaces. These operational models calculate temperature profile of the aluminum inside the furnaces, as well as the fuel consumptions, and allows achieving an energy analysis for non-predefined operating conditions. A low computational time makes the model suitable for the optimization and real-time controlling applications.
The energy efficiency scenarios for the remelting plant are defined based on the integrated waste heat from the melter and holder stack into the preheater furnace to heat up the charge. Results demonstrate that waste heat recovery can reduce fuel consumption of the preheater furnace up to 80.4% less than energy consumption of the preheater furnace in the business-as-usual case. Potential solutions for improvement of the energy performance in rolling plant are proposed on the furnace of annealing continuous line (ACL). ACL furnace is critical for achieving the desired material properties in the production of high-quality aluminum products. The proposed strategies include the modulation of the furnace temperature profiles and the energy integration via the partial recirculation of the zonal exhaust gases. The results show that the advanced energy integration approach significantly reduces fuel consumption by up to 20.7%. In total, the proposed energy efficiency measures reduce fuel consumption by 2.7 and 5.5 m3NG/tonAl in remelting and rolling plants, respectively.SCI-STI-F