- NumPI - Mailing list — Dipartimento di Matematica, UniPI

Fwd: Talk by Justo Puerto - May 13th
by Fabio Durastante 04 May '26

04 May '26

Dear Everybody, Please find attached the announcement for an upcoming FLOPtalks from our friends in Florence. The seminar may be of interest to many of you. Best, Fabio -------- Messaggio Inoltrato -------- Oggetto: Talk by Justo Puerto - May 13th Data: Mon, 4 May 2026 06:34:09 -0700 (PDT) Mittente: FLOPtalks <floptalks-group(a)unifi.it> A: FLOPtalks <floptalks-group(a)unifi.it> CC: JUSTO PUERTO ALBANDOZ <puerto(a)us.es> Dear colleagues and friends, we are pleased to announce next seminar of the FLorence OPtimization talkS (FLOPS) series. Justo Puerto <https://www.us.es/trabaja-en-la-us/directorio/justo-puerto-albandoz>, Professor at Universidad de Sevilla,will give a talk entitled "Ordered optimization: theory and applications". The talk will be on *May** 13th at 2:00 p.m. *in**Aula Caminetto, Scuola di ingegneria Santa Marta <https://maps.app.goo.gl/iWi2TtY2rM49epux5> The abstract of the talk is available at the following link <https://sites.google.com/unifi.it/floptalks/home?authuser=3> (as well as in the attached poster). Best regards, Marianna & Margherita -- You received this message because you are subscribed to the Google Groups "FLOPtalks" group. To unsubscribe from this group and stop receiving emails from it, send an email to floptalks-group+unsubscribe(a)unifi.it. To view this discussion, visit https://groups.google.com/a/unifi.it/d/msgid/floptalks-group/970f5d90-f649-… <https://groups.google.com/a/unifi.it/d/msgid/floptalks-group/970f5d90-f649-…>.

1 0

[Indico] [Event reminder] A Riemannian Framework for Optimization Problems in Reversible Markov Chains (19/05/2026, 11:00 Europe/Rome)
by Fabio Durastante 04 May '26

04 May '26

Title: A Riemannian Framework for Optimization Problems in Reversible Markov Chains, Speaker(s): Miryam Gnazzo, , Date and time: 19 May 2026, 11:00 (Europe/Rome), Lecture series: Seminar on Numerical Analysis, Venue: Dipartimento di Matematica (Aula Magna). Address: Largo Bruno Pontecorvo, 5 – 56127 Pisa You can access the full event here: https://events.dm.unipi.it/e/342 Abstract -------- An ergodic Markov chain with transition matrix $P$ and stationary distribution ${\pi}$ is said to be reversible if $D_{{\pi}} P = P^\top D_{{\pi}}$, where $D_{{\pi}}$ denotes the diagonal matrix with the components of ${\pi}$ on its diagonal. Reversibility is a key property in Markov chain theory, with applications ranging from computational biology to network models, such as those arising in power grid analysis. In this talk, we present an approach to optimization problems over reversible Markov chains, with prescribed stationary distribution ${\pi}$. Our framework is based on Riemannian optimization over suitable manifolds of stochastic matrices associated with reversible chains, and it allows us to employ efficient and reliable Riemannian solvers. Within this framework, we address the approximation of the reversible Markov chain closest to a given one. In addition, we describe an application for minimizing Kemeny’s constant, which measures the efficiency of a Markov chain in traversing its states. Note ---- Announcing the next seminar in the NUMPI series. -- Indico :: Email Notifier https://events.dm.unipi.it/e/342

1 0

[Indico] [Event reminder] Efficient edge-centrality measures based on matrix functions of the line graph (28/04/2026, 11:00 Europe/Rome)
by Fabio Durastante 27 Apr '26

27 Apr '26

Title: Efficient edge-centrality measures based on matrix functions of the line graph, Speaker(s): Francesco Zigliotto, , Date and time: 28 Apr 2026, 11:00 (Europe/Rome), Lecture series: Seminar on Numerical Analysis, Venue: Dipartimento di Matematica (Aula Magna). Address: Largo Bruno Pontecorvo, 5 – 56127 Pisa You can access the full event here: https://events.dm.unipi.it/e/341 Abstract -------- While network centrality is predominantly studied from a node perspective, here we investigate edge-centrality measures based on matrix functions. Specifically, we examine centrality within the context of the line graph, a structure where the roles of nodes and edges are inverted. Although any standard node-centrality metric can theoretically be applied to a line graph to assess edge importance, line graphs generally possess significantly more nodes than their original counterparts, which makes computing matrix functions of their adjacency matrices computationally infeasible. To address this bottleneck, we introduce novel matrix function identities that reduce the dimensionality of the required calculations. By operating on matrices of a much smaller size, these identities yield edge-centrality algorithms that are, in practice, nearly as fast as their node-level equivalents. Both directed and undirected graphs are treated. Note ---- Reminder of tomorrow's seminar -- Indico :: Email Notifier https://events.dm.unipi.it/e/341

1 0

[Indico] [Event reminder] Efficient edge-centrality measures based on matrix functions of the line graph (28/04/2026, 11:00 Europe/Rome)
by Fabio Durastante 13 Apr '26

13 Apr '26

Title: Efficient edge-centrality measures based on matrix functions of the line graph, Speaker(s): Francesco Zigliotto, , Date and time: 28 Apr 2026, 11:00 (Europe/Rome), Lecture series: Seminar on Numerical Analysis, Venue: Dipartimento di Matematica (Aula Magna). Address: Largo Bruno Pontecorvo, 5 – 56127 Pisa You can access the full event here: https://events.dm.unipi.it/e/341 Abstract -------- While network centrality is predominantly studied from a node perspective, here we investigate edge-centrality measures based on matrix functions. Specifically, we examine centrality within the context of the line graph, a structure where the roles of nodes and edges are inverted. Although any standard node-centrality metric can theoretically be applied to a line graph to assess edge importance, line graphs generally possess significantly more nodes than their original counterparts, which makes computing matrix functions of their adjacency matrices computationally infeasible. To address this bottleneck, we introduce novel matrix function identities that reduce the dimensionality of the required calculations. By operating on matrices of a much smaller size, these identities yield edge-centrality algorithms that are, in practice, nearly as fast as their node-level equivalents. Both directed and undirected graphs are treated. Note ---- Next NUMPI Seminar by Francesco Zigliotto, SNS (Pisa) -- Indico :: Email Notifier https://events.dm.unipi.it/e/341

1 0

Short Course by Alice Cortinovis
by Nicola Guglielmi 13 Apr '26

13 Apr '26

Dear all, This is a gentle reminder of the short course Introduction to randomized numerical linear algebra by Professor Alice Cortinovis (University of Pisa). Notice that the first lecture is scheduled at 4.30 today (April 13) in Aurora (ex-ISEF) building MLH room. It will be also possible to follow the course online. Best wishes, Nicola https://indico.gssi.it/event/938/ Ps details can be found here ----------------------------------------------------------------------------------------------------- Nicola Guglielmi Professor of Numerical Analysis Deputy Rector, Gran Sasso Science Institute https://www.gssi.it/people/professors/lectures-maths/item/545-guglielmi-nic… ------------------------------------------------------------------------------------------------------- -- You received this message because you are subscribed to the Google Groups "nomads-list" group. To unsubscribe from this group and stop receiving emails from it, send an email to nomads-list+unsubscribe(a)gssi.it. To view this discussion visit https://groups.google.com/a/gssi.it/d/msgid/nomads-list/CAMe-9vEPyTEV14OU2W…. For more options, visit https://groups.google.com/a/gssi.it/d/optout.

1 1

[Indico] [Event reminder] Stochastic Optimisation algorithms in CIL (14/04/2026, 11:00 Europe/Rome)
by Fabio Durastante 13 Apr '26

13 Apr '26

Title: Stochastic Optimisation algorithms in CIL, Speaker(s): Prof. Edoardo Pasca, UK Science and Technology Facilities Council (STFC), Date and time: 14 Apr 2026, 11:00 (Europe/Rome), Lecture series: Seminar on Numerical Analysis, Venue: Dipartimento di Matematica (Aula Magna (DM)). Address: Largo Bruno Pontecorvo, 5 – 56127 Pisa You can access the full event here: https://events.dm.unipi.it/e/340 Abstract -------- Stochastic optimization algorithms are commonly used in the training of machine learning models because they ensure progress towards the solution while reducing the per-iteration computational costs. This seminar is about the Stochastic Optimization algorithms available in the Core Imaging Library (CIL), including usage on real data at scale. CIL is an open source mainly Python framework for tomographic imaging originally developed for X-ray Computed Tomography. CIL has been used in various fields such as Magnetic Resonance Imaging (MRI), Positron Emission Tomography (PET), and Synthetic Aperture Radar (SAR). Note ---- Reminder for tomorrow's seminar -- Indico :: Email Notifier https://events.dm.unipi.it/e/340

1 0

Training session and seminar on Computational Imaging
by Silvia Gazzola 08 Apr '26

08 Apr '26

Dear all, Next Tuesday (14/04/2026) we will host Dr. Edoardo Pasca<https://ccpi.ac.uk/people/edoardo-pasca/>, Head of the Computed Tomography and Imaging group at the Scientific Computing division of the national UK Science and Technology Facilities Council (STFC<https://www.ukri.org/councils/stfc/>). He is an expert in solution methods and software for inverse problems in image reconstruction. See below for a more complete biography. Dr. Edoardo Pasca leads a team of scientists that develop the Core Imaging Library software (CIL<https://ccpi.ac.uk/cil/>). CIL is an open source, mainly Python, framework for tomographic imaging originally developed for X-ray Computed Tomography. CIL has been used in various fields such as Magnetic Resonance Imaging (MRI), Positron Emission Tomography (PET), and Synthetic Aperture Radar (SAR). Here are the events scheduled for the day: Workshop (hands on session) on CIL for CT reconstruction. TIME: 9:15-11:00 (morning session), 14:00-16:30 (afternoon session) LOCATION: TBD, Dipartimento di Matematica Address: Largo Bruno Pontecorvo, 5 – 56127 Pisa The event is free but registration is mandatory: if you’d like to attend, please fill in this form https://docs.google.com/forms/d/e/1FAIpQLSezb6BzvAwyIvGW-lLiQSnDVP0U655XW3D… by 12/04/2026. ABSTRACT: We will run a group-based practical session using CIL in Python for X-ray CT data manipulation and reconstruction. An introduction on concepts of X-ray CT data acquisition, optimization, iterative reconstruction algorithms and regularization will be given, followed by exercises in CIL. No software installation will be required: the sessions will be based on our Jupyter notebook training material<https://github.com/TomographicImaging/CIL-Demos>, and participants will have access to the material in the cloud. Knowledge of Python or object oriented programming is not mandatory to attend the event. By the end of the session you will know more about computational imaging and potentially be ready to use CIL for your research. Seminar (part of the Numerical Analysis seminar series) TITLE: Stochastic Optimisation algorithms in CIL TIME: 11:00-12:00 LOCATION: Dipartimento di Matematica (Aula Magna (DM)) Address: Largo Bruno Pontecorvo, 5 – 56127 Pisa ABSTRACT: Stochastic optimization algorithms are commonly used in the training of machine learning models because they ensure progress towards the solution while reducing the per-iteration computational costs. In this seminar, I will talk about the Stochastic Optimization algorithms available in the Core Imaging Library (CIL) and show some usage on real data at scale. Biography: Edoardo Pasca<https://ccpi.ac.uk/people/edoardo-pasca/> graduated in Physics and obtained a PhD in material science and technology at the University of Florence. After working as a post-doc in experimental condensed matter physics, he transitioned to software development and imaging. He now leads the Computed Tomography and Imaging group in the Scientific Computing of the Science Technology Facilities Council (STFC). The group supports material-science and medical-imaging scientists as well as UK national facilities such as the ISIS Neutron and Muon source, the UK synchrotron Diamond Light Source, the Central Laser Facility (CLF) and their users. Please do not hesitate to contact me if you need any additional information. Best wishes, Silvia Gazzola ============= Dr Silvia Gazzola Dipartimento di Matematica Università di Pisa https://sites.google.com/view/silviaspersonalwebpage/silvias-homepage ---------------------------------------------------------------------------------------------------------------------------------------------------------- You may receive this email outside my normal working hours — no response is expected outside your working hours ----------------------------------------------------------------------------------------------------------------------------------------------------------

1 0

Fwd: Talk by Britt van Veggel - April 29th
by Fabio Durastante 08 Apr '26

08 Apr '26

Dear all, I am forwarding the announcement below from the FLorence OPtimization talkS (FLOPS) mailing list, as the topic may be of interest to many of you. Please find the details of the seminar hereunder. Best, Fabio -------- Messaggio Inoltrato -------- Oggetto: Talk by Britt van Veggel - April 29th Data: Wed, 8 Apr 2026 04:18:43 -0700 (PDT) Mittente: FLOPtalks <floptalks-group(a)unifi.it> A: FLOPtalks <floptalks-group(a)unifi.it> CC: b.w.j.r.vanveggel(a)uva.nl Dear colleagues and friends, we are pleased to announce next seminar of the FLorence OPtimization talkS (FLOPS) series. Britt van Veggel <https://www.uva.nl/en/profile/v/e/b.w.j.r.vanveggel/b.w.j.r.van-veggel.html>,PhD student at University of Amsterdam,will give a talk entitled "Heatmaps for maximum coverage facility location optimization". The talk will be on *April** 29th at 2:00 p.m. *in**Room 221 Centro didattico Morgagni <https://g.co/kgs/uuxpeDA>. The abstract of the talk is available at the following link <https://sites.google.com/unifi.it/floptalks/home?authuser=3> (as well as in the attached poster). Best regards, Marianna & Margherita -- You received this message because you are subscribed to the Google Groups "FLOPtalks" group. To unsubscribe from this group and stop receiving emails from it, send an email to floptalks-group+unsubscribe(a)unifi.it. To view this discussion, visit https://groups.google.com/a/unifi.it/d/msgid/floptalks-group/b996480f-7c2e-… <https://groups.google.com/a/unifi.it/d/msgid/floptalks-group/b996480f-7c2e-…>.

1 0

[Indico] [Event reminder] Stochastic Optimisation algorithms in CIL (14/04/2026, 11:00 Europe/Rome)
by Fabio Durastante 08 Apr '26

08 Apr '26

Title: Stochastic Optimisation algorithms in CIL, Speaker(s): Prof. Edoardo Pasca, UK Science and Technology Facilities Council (STFC), Date and time: 14 Apr 2026, 11:00 (Europe/Rome), Lecture series: Seminar on Numerical Analysis, Venue: Dipartimento di Matematica (Aula Magna (DM)). Address: Largo Bruno Pontecorvo, 5 – 56127 Pisa You can access the full event here: https://events.dm.unipi.it/e/340 Abstract -------- Stochastic optimization algorithms are commonly used in the training of machine learning models because they ensure progress towards the solution while reducing the per-iteration computational costs. This seminar is about the Stochastic Optimization algorithms available in the Core Imaging Library (CIL), including usage on real data at scale. CIL is an open source mainly Python framework for tomographic imaging originally developed for X-ray Computed Tomography. CIL has been used in various fields such as Magnetic Resonance Imaging (MRI), Positron Emission Tomography (PET), and Synthetic Aperture Radar (SAR). Note ---- Next seminar in the NUMPI series -- Indico :: Email Notifier https://events.dm.unipi.it/e/340

1 0

[Indico] [Event reminder] A TAME method for the Inverse Laplace Transform of tame functions (31/03/2026, 11:00 Europe/Rome)
by Fabio Durastante 30 Mar '26

30 Mar '26

Title: A TAME method for the Inverse Laplace Transform of tame functions, Speaker(s): Nikita Deniskin, Scuola Normale Superiore, Date and time: 31 Mar 2026, 11:00 (Europe/Rome), Lecture series: Seminar on Numerical Analysis, Venue: Dipartimento di Matematica (Aula Magna). You can access the full event here: https://events.dm.unipi.it/e/339 Abstract -------- The Laplace transform and its inverse are widely used tools in both theoretical and applied contexts. However, while the direct Laplace transform is stable, the inverse Laplace transform is an inherently ill-posed problem, which makes its accurate computation challenging; this has led to extensive research in numerical methods for the inverse Laplace transform. In this talk, we focus on a family of algorithms, called Abate-Whitt methods, which recover the original function via linear combinations of evaluations of the transform. We relate the accuracy, with theoretical bounds, of the Abate-Whitt method to an approximation problem: constructing a rational approximation of the exponential $e^z$ on certain domains of the complex plane. We propose a new method, dubbed TAME, based on the AAA algorithm for rational approximation. TAME is especially effective for problems in queueing theory, in particular in analyzing phase-type distributions (Markov chains), and in computing the first return times in fluid queues. Here evaluations of the Laplace transform are (relatively) expensive, so it is highly important to minimize the number of needed evaluations. Joint work with Federico Poloni, arXiv:2510.14799. Note ---- Reminder of tomorrow’s seminar -- Indico :: Email Notifier https://events.dm.unipi.it/e/339

1 0