arXiv subjects

Riccardo Murri

Publications and source records attributed to Riccardo Murri.

11 recordsLinked to original sources

arXiv2025

Scalable, Cloud-Based Simulations of Blood Flow and Targeted Drug Delivery in Retinal Capillaries

We investigate the capabilities of cloud computing for large-scale,tightly-coupled simulations of biological fluids in complex geometries, traditionally performed in supercomputing centers. We demonstrate scalable and efficient simulations in the public cloud. We perform meso-scale simulations of blood flow in image-reconstructed capillaries, and examine targeted drug delivery by artificial bacterial flagella (ABFs). The simulations deploy dissipative particle dynamics (DPD) with two software frameworks, Mirheo (developed by our team) and LAMMPS. Mirheo exhibits remarkable weak scalability for up to 512 GPUs. Similarly, LAMMPS demonstrated excellent weak scalability for pure solvent as well as for blood suspensions and ABFs in reconstructed retinal capillaries. In particular, LAMMPS maintained weak scaling above 90% on the cloud for up to 2,000 cores. Our findings demonstrate that cloud computing can support tightly coupled, large-scale scientific simulations with competitive performance.

physics.flu-dyn↗

arXiv2019

Ambitious Data Science Can Be Painless

Modern data science research can involve massive computational experimentation; an ambitious PhD in computational fields may do experiments consuming several million CPU hours. Traditional computing practices, in which researchers use laptops or shared campus-resident resources, are inadequate for experiments at the massive scale and varied scope that we now see in data science. On the other hand, modern cloud computing promises seemingly unlimited computational resources that can be custom configured, and seems to offer a powerful new venue for ambitious data-driven science. Exploiting the cloud fully, the amount of work that could be completed in a fixed amount of time can expand by several orders of magnitude. As potentially powerful as cloud-based experimentation may be in the abstract, it has not yet become a standard option for researchers in many academic disciplines. The prospect of actually conducting massive computational experiments in today's cloud systems confronts the potential user with daunting challenges. Leading considerations include: (i) the seeming complexity of today's cloud computing interface, (ii) the difficulty of executing an overwhelmingly large number of jobs, and (iii) the difficulty of monitoring and combining a massive collection of separate results. Starting a massive experiment `bare-handed' seems therefore highly problematic and prone to rapid `researcher burn out'. New software stacks are emerging that render massive cloud experiments relatively painless. Such stacks simplify experimentation by systematizing experiment definition, automating distribution and management of tasks, and allowing easy harvesting of results and documentation. In this article, we discuss several painless computing stacks that abstract away the difficulties of massive experimentation, thereby allowing a proliferation of ambitious experiments for scientific discovery.

cs.DC↗

arXiv2014

Performance of Python runtimes on a non-numeric scientific code

The Python library FatGHol FatGHoL used in Murri2012 to reckon the rational homology of the moduli space of Riemann surfaces is an example of a non-numeric scientific code: most of the processing it does is generating graphs (represented by complex Python objects) and computing their isomorphisms (a triple of Python lists; again a nested data structure). These operations are repeated many times over: for example, the spaces and are triangulated by 4'583'322 and 747'664 graphs, respectively. This is an opportunity for every Python runtime to prove its strength in optimization. The purpose of this experiment was to assess the maturity of alternative Python runtimes, in terms of: compatibility with the language as implemented in CPython 2.7, and performance speedup. This paper compares the results and experiences from running FatGHol with different Python runtimes: CPython 2.7.5, PyPy 2.1, Cython 0.19, Numba 0.11, Nuitka 0.4.4 and Falcon.

cs.MS↗

arXiv2013

VM-MAD: a cloud/cluster software for service-oriented academic environments

The availability of powerful computing hardware in IaaS clouds makes cloud computing attractive also for computational workloads that were up to now almost exclusively run on HPC clusters. In this paper we present the VM-MAD Orchestrator software: an open source framework for cloudbursting Linux-based HPC clusters into IaaS clouds but also computational grids. The Orchestrator is completely modular, allowing flexible configurations of cloudbursting policies. It can be used with any batch system or cloud infrastructure, dynamically extending the cluster when needed. A distinctive feature of our framework is that the policies can be tested and tuned in a simulation mode based on historical or synthetic cluster accounting data. In the paper we also describe how the VM-MAD Orchestrator was used in a production environment at the FGCZ to speed up the analysis of mass spectrometry-based protein data by cloudbursting to the Amazon EC2. The advantages of this hybrid system are shown with a large evaluation run using about hundred large EC2 nodes.

cs.DC↗

arXiv2012

gcodeml: A Grid-enabled Tool for Detecting Positive Selection in Biological Evolution

One of the important questions in biological evolution is to know if certain changes along protein coding genes have contributed to the adaptation of species. This problem is known to be biologically complex and computationally very expensive. It, therefore, requires efficient Grid or cluster solutions to overcome the computational challenge. We have developed a Grid-enabled tool (gcodeml) that relies on the PAML (codeml) package to help analyse large phylogenetic datasets on both Grids and computational clusters. Although we report on results for gcodeml, our approach is applicable and customisable to related problems in biology or other scientific domains.

cs.DC↗

arXiv2012

Batch-oriented software appliances

This paper presents AppPot, a system for creating Linux software appliances. AppPot can be run as a regular batch or grid job and executed in user space, and requires no special virtualization support in the infrastructure. The main design goal of AppPot is to bring the benefits of a virtualization-based IaaS cloud to existing batch-oriented computing infrastructures. In particular, AppPot addresses the application deployment and configuration on large heterogeneous computing infrastructures: users are enabled to prepare their own customized virtual appliance for providing a safe execution environment for their applications. These appliances can then be executed on virtually any computing infrastructure being in a private or public cloud as well as any batch-controlled computing clusters the user may have access to. We give an overview of AppPot and its features, the technology that makes it possible, and report on experiences running it in production use within the Swiss National Grid infrastructure SMSCG.

cs.DC↗

arXiv2012

Fatgraph Algorithms and the Homology of the Kontsevich Complex

Fatgraphs are multigraphs enriched with a cyclic order of the edges incident to a vertex. This paper presents algorithms to: (1) generate the set of all fatgraphs having a given genus and number of boundary cycles; (2) compute automorphisms of any given fatgraph; (3) compute the homology of the fatgraph complex. The algorithms are suitable for effective computer implementation. In particular, this allows us to compute the rational homology of the moduli space of Riemann surfaces with marked points. We thus compute the Betti numbers of $M_{g,n}$ with $(2g + n) \leq 6$, corroborating known results.

math.AG↗

arXiv2012

A novel parallel algorithm for Gaussian Elimination of sparse unsymmetric matrices

We describe a new algorithm for Gaussian Elimination suitable for general (unsymmetric and possibly singular) sparse matrices, of any entry type, which has a natural parallel and distributed-memory formulation but degrades gracefully to sequential execution. We present a sample MPI implementation of a program computing the rank of a sparse integer matrix using the proposed algorithm. Some preliminary performance measurements are presented and discussed, and the performance of the algorithm is compared to corresponding state-of-the-art algorithms for floating-point and integer matrices.

cs.MS↗

arXiv2011

GridCertLib: a Single Sign-on Solution for Grid Web Applications and Portals

This paper describes the design and implementation of GridCertLib, a Java library leveraging a Shibboleth-based authentication infrastructure and the SLCS online certificate signing service, to provide short-lived X.509 certificates and Grid proxies. The main use case envisioned for GridCertLib, is to provide seamless and secure access to Grid/X.509 certificates and proxies in web applications and portals: when a user logs in to the portal using Shibboleth authentication, GridCertLib can automatically obtain a Grid/X.509 certificate from the SLCS service and generate a VOMS proxy from it. We give an overview of the architecture of GridCertLib and briefly describe its programming model. Its application to some deployment scenarios is outlined, as well as a report on practical experience integrating GridCertLib into portals for Bioinformatics and Computational Chemistry applications, based on the popular P-GRADE and Django softwares.

cs.DC↗

arXiv2003

Matrix Integrals and Feynman Diagrams in the Kontsevich Model

We review some relations occurring between the combinatorial intersection theory on the moduli spaces of stable curves and the asymptotic behavior of the 't Hooft-Kontsevich matrix integrals. In particular, we give an alternative proof of the Witten-Di Francesco-Itzykson-Zuber theorem --which expresses derivatives of the partition function of intersection numbers as matrix integrals-- using techniques based on diagrammatic calculus and combinatorial relations among intersection numbers. These techniques extend to a more general interaction potential.

math.AG↗

arXiv2001

Feynman Diagrams via Graphical Calculus

This paper is an introduction to the language of Feynman Diagrams. We use Reshetikhin-Turaev graphical calculus to define Feynman diagrams and prove that asymptotic expansions of Gaussian integrals can be written as a sum over a suitable family of graphs. We discuss how different kind of interactions give rise to different families of graphs. In particular, we show how symmetric and cyclic interactions lead to ``ordinary'' and ``ribbon'' graphs respectively. As an example, the 't Hooft-Kontsevich model for 2D quantum gravity is treated in some detail.

math.QA↗