Our Work

Release notes Added New Dataset Sudoku #1067 Native MUS for CPO #987 Native MUS for CPLEX #986 Unified prooflogging #862 PySCIPOpt's native cumulative #998 Typing of abs and mult #1075 Breaking chan…

Release notes Added New Dataset Sudoku #1067 Native MUS for CPO #987 Native MUS for CPLEX #986 Unified prooflogging #862 PySCIPOpt's native cumulative #998 Typing of abs and mult #1075 Breaking changes Model.solve(time_limit=...) now includes transformation time; returns False (ExitStatus.UNKNOWN) if the limit runs out before solving. model.status().runtime now includes transformation time, separate solver time is in the new status().solve_time. Custom solver interfaces should set cpm_status.solve_time instead of runtime #803 NDVarArray uses NumPy's __array_ufunc__: np.equal(x, y) etc. now return constraints instead of Booleans. Ufuncs without a Python operator equivalent (np.square, np.maximum, np.matmul/@, ...), ufunc methods (np.add.reduce, np.mean, ...) and out=/dtype=/keepdims= arguments raise an error; use Python operators, np.dot or cp.* functions instead #1062 InDomain has expression name "indomain" instead of "InDomain" #1087 Proof logging is standardised across solvers: the proof file is passed to the constructor (SolverLookup.get(solver, model, proof="path/to/proof")) and checked with s.verify(). For GCS, the prove=, proof_name=, proof_location=, verify=, ... arguments of solve()/solveAll() are removed, verify() now returns True if valid (previously VeriPB's exit code) and GCSVerificationException is removed #862 Zero-duration tasks never overlap in NoOverlap(Optional) #1105 and consume no resources in Cumulative(Optional) #1102 OR-Tools lower version increased to 9.12 #1074 Models pickled with an earlier version are not guaranteed to work correctly (e.g. InDomain keeps its old name making it always decompose, model.status() lacks solve_time); regenerate your pickle files with v1.1.0 Changed Switch to model.add() in README #1113 Account for transformation time in Model.solve() #803 Following modern Numpy ndarray subclassing with ufunc #1062 Account for transformation time in model.solve() #803 Increase OR-Tools lower bound to 9.12 #1074 Modernise hyperparameter search example #960 Fixed Flatten multidimension args to globals #1112 Analyzing XCSP3 benchmark csvs with empty time metrics #923 Typos in docstrings, comments and error messages #1097 NoOverlap non-strict semantics for zero duration tasks #1105 Cumulative no demand when duration is zero #1102 Logo image link for PyPI #1104 Fix MUS tests #1095 Fix linearize when trivially unsat comparison #1101 Support trivial arities of And and Or #1080 Docs: small typo #1092 Small fixes to Hexaly #983 Rename InDomain to indomain #1087 Typo csemap OR-Tools #1088 Fixes to XCSP3 tools #1068 Marker inheritance and non-parametrised tests with solver dependency #961 Full Changelog: https://github.com/CPMpy/cpmpy/compare/v1.0.0...v1.1.0

Overview This dataset contains the full data related with the aplication of the In-Depht PROFILE framework to Navaluenga, central Spain. This includess both the input and the output data. The input da…

Overview This dataset contains the full data related with the aplication of the In-Depht PROFILE framework to Navaluenga, central Spain. This includess both the input and the output data. The input data include the cleaned survey done building-by-building and face-to-face, the webscraping price outpus, all the inputs used to build the hydraulic HEC-RAS 2D model including the digital msourface model corrected with geoestatistics, and the economic flood loss model outputs both raw and processed. This data is related with the In-Depht PROFILE framework also in Zenodo.   Data size and format The data has been compresed in 81 parts of 1 Gb. All must be uncompressed together. Once the data is uncompressed the total size is 166 GB. Data format are gis related (.shp, .tif, ...), HEC-RAS related (.prj, .u01, ...), python related (.pkl, .ipc, ...) and end-user friendly (.xlsx). Particularlly the complete results fromt the economic flood loss monte carlo model has been partially converted into an .xlsx to allow fast and easy overview of the results. However, compled data must be accesed using Python (mocaloss library is recommended).   Structure of the data Note the structure has been mantained according to the use of the In-Depth PROFILE framework and mocaloss library, resulting in the following structure:   [uncompresed_folder]/ ├── models/                                           <- Monte Carlo model-related data  │    └── _HEC-RAS_6.6/                          <- Hydraulic model-related data │          ├── 0_Base_Project - Manual/   <- Copy of Base model for manual individual runs │          ├── RC_0_Base_Project/             <- Base model used for monte carlo iterations │          ├── RC_1_Sample_Projects/       <- Parallel run workers used for monte carlo │          └── RC_2_Output_WSE/             <- Water depth outputs for all montecarlo iterations │ └── data/                                                <- Economic flood loss model-related data       ├── inputs/                                       <- Generated input data       │     ├── gis/                                     <- Gis related data (HEC-RAS, buildings, geostatistic...)       │     └──Survey.xlsx                          <- Survey data cleaned       ├── intermediate/                            <- Data processed with the framework       │     ├── gis/                                    <- Gis related data (sample area of buildings)       │     ├── Flow_Fitted_Function.pkl   <- Input flow distributions for hydraulic model monte carlo       │     ├── Prices_Content.xlsx            <- Collected price data from monte carlo       │     └──Depht_Samples.pkl             <- Survey data cleaned        └── mocaloss/                                 <- Output data from mocaloss library             ├── .cache/                                <- Automatic folder created by mocaloss             │    ├── mocaloss_results          <- Flood loss model results (4 chunks 154 GB)             │    └── sampling_rules             <- Sampling rules and fitted input distributions             ├── postprocessor/                    <- Economic flood loss model results from mocaloss             └──postprocessor_results.xlsx    <- View of the mocaloss results Website Official project website Funding This work was funded by the Spanish Ministry of Science and Innovation (MCIN/AEI/10.13039/501100011033) and by “ERDF A way of making Europe” through the ENGAGEMENT project (PID2023-151292OB-I00).

We report ongoing work from the final phase of a cooperation project with the German NFDI Consortium Text+. The aim of the project is to design a common data format for different FrameNets across lang…

We report ongoing work from the final phase of a cooperation project with the German NFDI Consortium Text+. The aim of the project is to design a common data format for different FrameNets across languages. Our use case data are the German Frame-semantic Online Lexicon (G-FOL), developed at the University of Texas at Austin (https://frames.coerll.utexas.edu/) and the trilingual (German - French - English) Kicktionary (http://kicktionary.de/), a frame-semantic digital dictionary of football language. Both resources were analysed in an E/R approach for the objects and relations they contain (the analysis was further informed by the larger monolingual resources Berkeley FrameNet and German FrameNet). An XML format was then devised, based on the guidelines of the Text Encoding Initiative and the Lexical markup framework (LMF; ISO 24613) which can represent the two resources on a common basis. We are currently in the process of finalising the full conversion of the resources into that format.  

This data package includes interview notes and transcripts from three semi-structured interviews conducted with members of diaspora organizations operating in Germany. All interviews, conducted in Tur…

This data package includes interview notes and transcripts from three semi-structured interviews conducted with members of diaspora organizations operating in Germany. All interviews, conducted in Turkish, were either transcribed or documented through detailed notes and subsequently translated into English. The interviews explore the evolving dynamics of migration discourse, perceptions of policy, and strategic diaspora activism in relation to irregular migration policies.

This paper presents a model predictive control (MPC)-based energy management strategy for hybrid power systems combining a proton-exchange membrane fuel cell (PEMFC) with a lithium iron phosphate (LFP…

This paper presents a model predictive control (MPC)-based energy management strategy for hybrid power systems combining a proton-exchange membrane fuel cell (PEMFC) with a lithium iron phosphate (LFP) battery storage unit for renewable energy applications. The proposed framework optimizes power allocation between the two sources while respecting operational constraints, including current limits, power balance requirements, and state-of-charge (SOC) bounds with soft constraints to prevent overcharging and deep discharging. Unlike conventional rule-based approaches, the MPC formulation employs a quadratic cost function with tunable weighting factors that enable flexible prioritization of either fuel cell conservation or battery lifetime extension. Accurate yet computationally efficient models are developed for both components: an equivalent circuit model for the LFP battery and a theoretical electrochemical model for the PEMFC. The performance of the proposed strategy is validated through comprehensive simulations under realistic renewable generation and load profiles. Five case studies are examined, each representing different operational scenarios characterized by varying initial SOC conditions and component prioritization weights. The results demonstrate that the MPC-based approach effectively manages power distribution, maintains SOC within safe operating ranges, and adapts to changing system conditions. Quantitative analysis shows that the tunable weighting strategy successfully limits high-current events, reducing high-current operation and potentially mitigating current-related degradations. The proposed framework offers a scalable and flexible solution for improving the reliability of hybrid energy storage in modern renewable grids.

In its pursuit of a leadership role in climate service (CS) development and uptake, the European Union (EU) has invested in numerous innovation initiatives that have significantly advanced the field. …

In its pursuit of a leadership role in climate service (CS) development and uptake, the European Union (EU) has invested in numerous innovation initiatives that have significantly advanced the field. Still, efforts to promote broader uptake are fragmented, which underscores the need for a more systematic approach to stimulate upscaling. Upscaling CSs—the process of transferring, replicating, extending and/or institutionalising CSs in new contexts, thereby increasing the value of services by enhancing impact and adaptation outcomes— presents both theoretical and practical challenges. This perspective article explores the challenge of upscaling CSs and the benefits of a nuanced understanding of contexts in which such services may be expanded. Collaboration between diverse actors—including developers, intermediaries, and users—is recommended within the upscaling process. Against this backdrop, we propose a framework to orient (1) horizontal upscaling (replication to new audiences), (2) vertical upscaling (institutional changes) and (3) functional upscaling (adding functions and/or content). This framework is structured around four guiding questions: ‘why’—to uncover the rationale for upscaling CS, the specific benefits and opportunities; ‘what’—to consider specific components valued and/or demanded to be upscaled; ‘who’ − to consider those involved in the process; and ‘how’—to explore practices and opportunities. We substantiate the suggested framework with a select few real-life CS examples, as well as illustrating hypothetical examples. We propose that adopting an upscaling framework informed by the viewpoints outlined here can help EU‑sponsored initiatives enhance the strategic impact of climate‑service investments and, in turn, improve long‑term climate resilience and adaptation. This article results from a cross-project Thematic Working Group promoted by the EU Mission on Climate Adaptation, fostering the exchange and discussion between 39 different EU projects involving CSs.

Sound Event Classification (SEC) has attracted considerable attention in recent times, with applications in a variety of fields, including environmental acoustics. In particular, there has been a nota…

Sound Event Classification (SEC) has attracted considerable attention in recent times, with applications in a variety of fields, including environmental acoustics. In particular, there has been a notable increase in the level of interest in outdoor measurements, with a view to distinguishing the contributions of a particular source from the background noise. The utilisation of machine learning tools is contingent upon the availability of substantial datasets for the purposes of training or validation. DataSEC is an open-access dataset specifically designed for Sound Event Classification of environmental noise that can be listened in outdoor environments.The collection consists of 18 hours and 26 minutes of authentic, non-synthesized audio recordings, which have been meticulously gathered from two distinct sources: sound level measurements and online repositories. The authors have conducted a comprehensive analysis of the sound samples, encompassing a diverse range of environments, from urban to rural settings.DataSEC comprises a total of 4,292 mono-channel .wav audio samples, with a sampling rate of 44.1 kHz. Each sample represents a single event that has been classified into one of the following 22 defined sound classes and 28 subclasses. The utilisation of the symbol "/" serves to denote the subclasses. The identified classes and subclasses are: Bells; Birds; Cat fights and moans; Chicken coop; Cicadas and crickets \Cicadas \Crickets; Crows seagulls and magpies \Crows \Seagulls \Magpies; Dog barkings and howlings; Glass breaking; Horn; Jet aircrafts; Lawn mower brush cutter and olive shaker \lawn mower \Brush cutter \Olive shaker; Music; Propeller aircrafts \Airplanes, \Helicopters; Sirens and alarms \Sirens \Alarms; Thunder fireworks and gunshot \Thunder \Fireworks \Gunshot; Train; Vacuum cleaner fan and hairdryer \Vacuum cleaner \Fan \Hairdryer; Vehicle idling \Car-truck idling, \Motorbike idling; Vehicle pass-by \Car pass-by, \Motorbike pass-by, \Truck pass-by; Voices; Wind turbine; Workshop \Air compressor, \Drill, Grinder, \Jackhammer, \Saw.The minimum number of entries required for each class has been set at 50, with a minimum of 20 entries stipulated for each sub-class. A thorough examination of all samples has been conducted by the authors, with all files being meticulously preprocessed to eliminate silences and superfluous parts. This preprocessing involved the removal of irrelevant background activity and overlapping sounds.The authors expect that the dataset will contribute to future research in real-world sound event analysis and automated acoustic evaluation by means of machine learning.   V.4 fixes a bug in the uploaded file, where the number of Music files was not correct respect to what described in the text. This is for class balancing.