NumPI

numpi@lists.dm.unipi.it

2 participants
265 discussions

Seminario G. Barbarino - poll
by Federico Poloni 21 Apr '18

21 Apr '18

Buongiorno! Vorremmo far tenere un seminario a Giovanni Barbarino nelle prossime settimane, sul tema "Higher Order Spectral Symbols and Eigenvalues Approximation for Toeplitz Matrices", e volevo chiedervi scegliere tra le varie possibilità per l'orario del seminario. Se siete interessati a seguire il seminario, segnate le vostre disponibilità di massima su https://doodle.com/poll/vv96x5zbth5cbss4 e poi sceglierò l'opzione più gettonata. Grazie e a risentirci per la data definitiva. Cordialmente, -- --federico poloni Dipartimento di Informatica, Università di Pisa http://www.di.unipi.it/~fpoloni/ tel:+39-050-2213143

1 2

Avviso di seminario
by Dario A. Bini 30 Mar '18

30 Mar '18

Buongiorno, vi invio gli estremi del seminario di Raf Vandebril. Dario --- Giorno: 6 Aprile, ore 10:00 Luogo: Aula Magna, DIpartimento di Matematica Autore: Raf Vandebril, KU Leuven Titolo: On classical, extended, and rational Krylov and the associated QR algorithms Abstract: In this lecture we will discuss three main classes of Krylov and QR type methods: the classical, the extended, and the rational versions. We will initially focus on the classical QR method and reintepret all ingredients such as: associated Krylov spaces, associated Hessenberg matrices, implicit theorems, bulge chasing, eigenvalue swapping, and subspace iteration. Another look at these essentials will allow us to straightforwardly generalize all of this to the extended and rational case. More precisely, we will start by deducing the structure of the recurrence pencil associated to all Krylov methods. It will be shown that the rational Krylov method produces the most general pencil, namely two Hessenberg matrices, which implicitly store the poles of the rational Krylov method. We will revisit the implicit Q-Theorem and link it to its dual theorem, the implicit HK-Theorem, allowing us to provide a pole chasing algorithm for two Hessenberg matrices. This generalizes the classical and extended QR algorithms. Correctness will follow directly from the previous analysis of the classical QR algorithm. The rational QR algorithm allows for additional freedom, which is the introduction of poles. Even without a good pole-strategy we will see that the iteration count reduces with almost 10 procent.

1 0

Fwd: opportunità nella geofisica computazionale - INGV Pisa
by Dario A. Bini 25 Feb '18

25 Feb '18

Mi viene chiesto di diffondere questa informazione. Provvedete anche voi. Dario --- -------- Forwarded Message -------- Subject: opportunità nella geofisica computazionale - INGV Pisa Date: Sat, 24 Feb 2018 10:35:55 -0700 From: Mattia De' Michieli Vitturi <mattia.demichielivitturi(a)ingv.it> To: stefano.galatolo(a)unipi.it, giovanni.federico.gronchi(a)unipi.it, dario.bini(a)unipi.it CC: Tomaso Esposti Ongaro <tomaso.espostiongaro(a)ingv.it> Buongiorno, Vi disturbo perché avremonei prossimi mesi diverse opportunità per assegni di ricerca pre/post-dottorali annuali su tematiche di geofisica computazionale all'INGV di Pisa. Prima di procedere con i bandi, vorremmo esplorare possibili candidature. In particolare, per: 1) data assimilation in atmospheric dispersal models 2) High Performance Computing of multiphase volcanic flows 3) Modeling/simulation of flow and mechanics of porous media 4) Development of automated procedures for the integration of monitoring data into numerical models of volcanic plumes Il profilo 1) sarebbe più matematico applicato, il 2) più computazionale, il 3) più fisico/numerico, il 4) informatico. Se avete in mente persone potenzialmente interessate potete indirizzarle a me o al collega Tomaso Esposti Ongaro per maggiori informazioni: mattia.demichielivitturi(a)ingv.it <mailto:mattia.demichielivitturi@ingv.it> tomaso.espostiongaro(a)ingv.it <mailto:tomaso.espostiongaro@ingv.it> Tutte le attività sarebbero presso l'INGV a Pisa, ma nel caso 2) la borsa è finanziata da OGS (Istituto Naz di Oceanografia e Geofisica Sperimentale) di Trieste. Grazie, buona giornata e buon lavoro Mattia e Tomaso -- Mattia de' Michieli Vitturi, PhD Istituto Nazionale di Geofisica e Vulcanologia Sezione di Pisa Via della Faggiola 32, I-56126 Pisa, Italia ph.: +39-050-8311923 fax.: +39-050-8311942 email: mattia.demichielivitturi(a)ingv.it <mailto:mattia.demichielivitturi@ingv.it> skype: demichie

1 0

Numerical analysis seminars - Durastante
by Federico Poloni 05 Feb '18

05 Feb '18

Buongiorno, inoltro questo avviso di seminario. > il giorno 14 Febbraio alle ore 11 in sala seminari est presso il dipartimento di informatica > il Dott. Fabio Durastante assegnista presso il Dipartimento di Informatica dell'Universita` di Pisa > terra` un seminario dal titolo "Preconditioning Techniques for Some Large Localized or Structured Linear Systems with Application to Discrete Fractional Calculus". Cordialmente, -- --federico poloni Dipartimento di Informatica, Università di Pisa http://www.di.unipi.it/~fpoloni/ tel:+39-050-2213143

1 0

Posizione PhD a Strathclyde
by Federico Poloni 20 Dec '17

20 Dec '17

Ricevo e inoltro in caso ci siano persone interessate. > PhD Studentship on novel preconditioned iterative solvers for radial basis function linear systems > > Applications are invited for a PhD studentship on novel preconditioners for radial basis function linear systems, under the supervision of Dr Jennifer Pestana and Dr Alison Ramage, at the Department of Mathematics and Statistics at the University of Strathclyde, Glasgow, Scotland. The project will suit students with interests in numerical analysis and linear algebra. > > The studentship covers UK/EU tuition fees and comes with an annual tax-free stipend at the standard UK rate. International students who can fund the difference between UK/EU and International fee rates are also encouraged to apply. > > The project will begin 1 March, 2018, or earlier by arrangement. > > Further details are available at > > http://personal.strath.ac.uk/jennifer.pestana/RBF_Project.pdf > > Informal enquiries to Jennifer Pestana (jennifer.pestana(a)strath.ac.uk) are welcome. Buone feste a tutti, -federico -- --federico poloni Dipartimento di Informatica, Università di Pisa http://www.di.unipi.it/~fpoloni/ tel:+39-050-2213143

1 0

due giorni a Padova
by Dario Bini 27 Nov '17

27 Nov '17

Carissimi, vi ricordo che l'8 e il 9 febbraio 2018 si terra' la "Due Giorni di Algebra Lineare Numerica e Applicazioni" www.math.unipd.it/~pd2ggaln Se interessati vi invito a comunicare all'indirizzo pd2ggaln(a)math.unipd.it la vostra manifestazione di interesse possibilmente specificando se si desidera tenere una comunicazione. Ciao, Dario ---

1 0

assegno di ricerca UNIPI
by Federico Poloni 21 Nov '17

21 Nov '17

Luca Gemignani mi comunica che è stato pubblicato il seguente bando per un assegno di ricerca sul progetto "Modelli e algoritmi innovativi per problemi strutturati e sparsi di grandi dimensioni" (scadenza domande 20 dicembre; necessario avere il dottorato). Info su https://www.unipi.it/ateneo/bandi/assegni/pra2017/inf/20dic2017/index.htm Un saluto a tutti, -- --federico poloni Dipartimento di Informatica, Università di Pisa http://www.di.unipi.it/~fpoloni/ tel:+39-050-2213143

1 0

Seminar
by Dario Bini 02 Nov '17

02 Nov '17

Avviso di seminario: Speaker: Samandar Babaev (Bukhara State University, Uzbekistan) Date: Wednesday, November 8 Time: 11:00 Place: Sala Seminari, Dipartimento di Matematica Title: Optimal Interpolation Formulas in W_2^{(m,m-1)} space Abstract: In this talk, using S.L.Sobolev’s method, optimal interpolation formulas are constructed in /W_{2} ^/(/m,m//−/1)(0/,/1) space. Explicit formulas for coefficients of optimal interpolation formulas are obtained. Some numerical results are presented.

1 0

Avviso di Seminario
by Dario Bini 26 Oct '17

26 Oct '17

Avviso di seminario: Speaker: Federico Poloni (Università di Pisa) Date: Tuesday 2017-10-31 Time: 11:00 Place: Sala Seminari Ovest, Dipartimento di Informatica Title: Counting Fiedler pencils using diagrams Fiedler pencils are a family of matrix pencils which generalizes the well-known companion matrix: they are*linearizations*, i.e., they provide a method to construct, given a matrix polynomial, a linear eigenvalue problem with the same eigenvalues and multiplicities. Fiedler pencils are constructed as products of special block matrices that act nontrivially only on two contiguous blocks. They have a rich structure that give rise to many combinatorial properties. In this talk, we introduce a notation that associates to each pencil a diagram that depicts its action on the blocks. Using this notation, we can obtain visual proofs of several statements in the theory, and we can solve several counting properties (such as "how many distinct Fiedler pencils with repetitions of a given dimension exist"). Among them, in particular, we are interested in counting Fiedler pencils associated to symmetric and palindromic matrix polynomials which preserve the same structure. This talk is based on a joint work with Gianna Del Corso.

1 0

Avviso di seminario
by Dario A. Bini 13 Oct '17

13 Oct '17

Invio il seguente avviso di seminario Dario --- Gorno: Martedi 17 ottobre Ora: 11:00 Aula: Seminari, Dipartimento di Matematica Speaker: Leonardo Robol, ISTI --- CNR Titolo: Backward error analysis for polynomial rootfinders. Abstract: The most common methods to compute roots of polynomials is to rephrase the problem in linear algebra terms: one constructs a companion matrix (or pencil) that has the roots of the polynomial as eigenvalues. However, even when the eigenvalue method used is backward stable, it's non-trivial to map the backward error back on the polynomial. This problem has been studied extensively in the past and it has been shown that relying on the QR method does not provide a normwise backward-stable rootfinder. For this reason, one needs to rely on QZ + appropriate scaling. However, companion matrices are unitary plus rank 1, and we claim that relying on fast QR method that exploits this structure can lead to a backward-stable method. To this end, one needs to ensure that the errors on the unitary and rank 1 parts satisfy different bounds. We show how to perform this analysis of "mixed perturbation backward-error", as well as a method that satisfies these constrains. Time permitting, we discuss some more details that one should consider. We show, for example, that normwise backward stable operations (i.e., unitary rotations in our case) do not necessarily imply backward stable operations on the unitary and low-rank factors, unless some particular identities involving sines and cosines are preserved. This is joint work with Jared Aurentz, Thomas Mach, Raf Vandebril, and David Watkins.

1 0

← Newer
1
...
18
19
20
21
22
23
24
...
27
Older →

Jump to page:

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

NumPI