arXiv subjects

Philippe Bonnet

Publications and source records attributed to Philippe Bonnet.

18 recordsLinked to original sources

arXiv2025

DuckDB on xNVMe

DuckDB is designed for portability. It is also designed to run anywhere, and possibly in contexts where it can be specialized for performance, e.g., as a cloud service or on a smart device. In this paper, we consider the way DuckDB interacts with local storage. Our long term research question is whether and how SSDs could be co-designed with DuckDB. As a first step towards vertical integration of DuckDB and programmable SSDs, we consider whether and how DuckDB can access NVMe SSDs directly. By default, DuckDB relies on the POSIX file interface. In contrast, we rely on the xNVMe library and explore how it can be leveraged in DuckDB. We leverage the block-based nature of the DuckDB buffer manager to bypass the synchronous POSIX I/O interface, the file system and the block manager. Instead, we directly issue asynchronous I/Os against the SSD logical block address space. Our preliminary experimental study compares different ways to manage asynchronous I/Os atop xNVMe. The speed-up we observe over the DuckDB baseline is significant, even for the simplest scan query over a TPC-H table. As expected, the speed-up increases with the scale factor, and the Linux NVMe passthru improves performance. Future work includes a more thorough experimental study, a flexible solution that combines raw NVMe access and legacy POSIX file interface as well the co-design of DuckDB and SSDs.

cs.DB↗

arXiv2022

Predicting Bearings' Degradation Stages for Predictive Maintenance in the Pharmaceutical Industry

In the pharmaceutical industry, the maintenance of production machines must be audited by the regulator. In this context, the problem of predictive maintenance is not when to maintain a machine, but what parts to maintain at a given point in time. The focus shifts from the entire machine to its component parts and prediction becomes a classification problem. In this paper, we focus on rolling-elements bearings and we propose a framework for predicting their degradation stages automatically. Our main contribution is a k-means bearing lifetime segmentation method based on high-frequency bearing vibration signal embedded in a latent low-dimensional subspace using an AutoEncoder. Given high-frequency vibration data, our framework generates a labeled dataset that is used to train a supervised model for bearing degradation stage detection. Our experimental results, based on the FEMTO Bearing dataset, show that our framework is scalable and that it provides reliable and actionable predictions for a range of different bearings.

stat.ML↗

arXiv2022

VLDB 2021: Designing a Hybrid Conference

In 2020, while main database conferences one by one had to adopt a virtual format as a result of the ongoing COVID-19 pandemic, we decided to hold VLDB 2021 in hybrid format. This paper describes how we defined the hybrid format for VLDB 2021 going through the key design decisions. In addition, we list the lessons learned from running such a conference. Our goal is to share this knowledge with fellow conference organizers who target a hybrid conference format as well, which is on its way to becoming the norm rather than the exception. For readers who are more interested in the highlights rather than details, a short version of this report appears in SIGMOD Record.

cs.DB↗

arXiv2020

Unionized Data Governance in Virtual Power Plants

Flexible electricity networks continuously coordinate and optimize operations through ICT systems. An overlay data grid conveys information about the state of the electricity grid, as well as the status of demand and production of electricity in households and industry. Data is thus the basis for decisions that affect electricity costs and availability of assets on the electricity grid. It is crucial that these decisions are formed and monitored according to a well-defined governance model. No such data governance model exists today. In this paper, we focus on the central role of virtual power plants in flexible electricity networks. We define the problem of data governance in a virtual power plant, insisting on the issues linked to the inherent asymmetry of this system. We propose unionization as a framing device to reason about this issue. The central contribution of this paper is thus principles for a unionized data governance model for virtual power plants.

cs.CY↗

arXiv2015

Garbage Collection Techniques for Flash-Resident Page-Mapping FTLs

Storage devices based on flash memory have replaced hard disk drives (HDDs) due to their superior performance, increasing density, and lower power consumption. Unfortunately, flash memory is subject to challenging idiosyncrasies like erase-before-write and limited block lifetime. These constraints are handled by a flash translation layer (FTL), which performs out-of-place updates, wear-leveling and garbage-collection behind the scene, while offering the application a virtualization of the physical address space. A class of relevant FTLs employ a flash-resident page-associative mapping table from logical to physical addresses, with a smaller RAM-resident cache for frequently mapped entries. In this paper, we address the problem of performing garbage-collection under such FTLs. We observe two problems. Firstly, maintaining the metadata needed to perform garbage-collection under these schemes is problematic, because at write-time we do not necessarily know the physical address of the before-image. Secondly, the size of this metadata must remain small, because it makes RAM unavailable for caching frequently accessed entries. We propose two complementary techniques, called Lazy Gecko and Logarithmic Gecko, which address these issues. Lazy Gecko works well when RAM is plentiful enough to store the GC metadata. Logarithmic Gecko works well when RAM isn't plentiful and efficiently stores the GC metadata in flash. Thus, these techniques are applicable to a wide range of flash devices with varying amounts of embedded RAM.

cs.DB↗

arXiv2015

Modelling and Managing SSD Write-amplification

How stable is the performance of your flash-based Solid State Drives (SSDs)? This question is central for database designers and administrators, cloud service providers, and SSD constructors. The answer depends on write-amplification, i.e., garbage collection overhead. More specifically, the answer depends on how write-amplification evolves in time. How then can one model and manage write-amplification, especially when application workloads change? This is the focus of this paper. Managing write-amplification boils down to managing the surplus physical space, called over-provisioned space. Modern SSDs essentially separate the physical space into several partitions, based on the update frequency of the pages they contain, and divide the over-provisioned space among the groups so as to minimize write-amplification. We introduce Wolf, a block manager that allocates over-provisioned space to SSD partitions using a near-optimal closed-form expression, based on the sizes and update frequencies of groups of pages. Our evaluation shows that Wolf is robust to workloads change, with an improvement factor of 2 with respect to the state-of-the-art. We also show that Wolf performs comparably and even slightly better than the state of the art with stable workloads (over 20% improvement with a TPC-C workload).

cs.DB↗

arXiv2014

EagleTree: Exploring the Design Space of SSD-Based Algorithms

Solid State Drives (SSDs) are a moving target for system designers: they are black boxes, their internals are undocumented, and their performance characteristics vary across models. There is no appropriate analytical model and experimenting with commercial SSDs is cumbersome, as it requires a careful experimental methodology to ensure repeatability. Worse, performance results obtained on a given SSD cannot be generalized. Overall, it is impossible to explore how a given algorithm, say a hash join or LSM-tree insertions, leverages the intrinsic parallelism of a modern SSD, or how a slight change in the internals of an SSD would impact its overall performance. In this paper, we propose a new SSD simulation framework, named EagleTree, which addresses these problems, and enables a principled study of SSD-Based algorithms. The demonstration scenario illustrates the design space for algorithms based on an SSD-based IO stack, and shows how researchers and practitioners can use EagleTree to perform tractable explorations of this complex design space.

cs.DB↗

arXiv2009

uFLIP: Understanding Flash IO Patterns

Does the advent of flash devices constitute a radical change for secondary storage? How should database systems adapt to this new form of secondary storage? Before we can answer these questions, we need to fully understand the performance characteristics of flash devices. More specifically, we want to establish what kind of IOs should be favored (or avoided) when designing algorithms and architectures for flash-based systems. In this paper, we focus on flash IO patterns, that capture relevant distribution of IOs in time and space, and our goal is to quantify their performance. We define uFLIP, a benchmark for measuring the response time of flash IO patterns. We also present a benchmarking methodology which takes into account the particular characteristics of flash devices. Finally, we present the results obtained by measuring eleven flash devices, and derive a set of design hints that should drive the development of flash-based systems on current devices.

cs.PF↗

arXiv2008

Relations between the leading terms of a polynomial automorphism

Let $I$ be the ideal of relations between the leading terms of the polynomials defining an automorphism of $K^n$. In this paper, we prove the existence of a locally nilpotent derivation which preserves $I$. Moreover, if $I$ is principal, i.e. $I=(R)$, we compute an upper bound for $°_2(R)$ for some degree function $°_2$ defined by the automorphism. As applications, we determine all the principal ideals of relations for automorphisms of $K^3$ and deduce two elementary proofs of the Jung-van der Kulk Theorem about the tameness of automorphisms of $K^{2}$.

math.AG↗

arXiv2007

On algebraic automorphisms and their rational invariants

Let X be an affine irreducible variety over an algebraically closed field k of characteristic zero. Given an automorphism F, we denote by k(X)^F its field of invariants, i.e. the set of rational functions f on X such that f(F)=f. Let n(F) be the transcendence degree of k(X)^F over k. In this paper, we study the class of automorphisms F of X for which n(F)= dim X - 1. More precisely, we show that under some conditions on X, every such automorphism is of the form F=A_g, where A is an algebraic action of a linear algebraic group G of dimension 1 on X, and where g belongs to G. As an application, we determine the conjugacy classes of automorphisms of the plane for which n(F)=1.

math.AG↗

arXiv2006

Group actions, $k$-derivations and finite morphisms

Let $G$ be an affine algebraic group over an algebraically closed field $k$ of characteristic zero. In this paper, we consider finite $G$-equivariant morphisms $F:X\to Y$ of irreducible affine $G$-varieties. First we determine under which conditions on $Y$ the induced map $F^G:X//G\to Y//G$ of quotient varieties is also finite. This result is reformulated in terms of kernels of derivations on $k$-algebras $A\subset B$ such that $B$ is integral over $A$. Second we construct explicitly two examples of finite $G$-equivariant maps $F$. In the first one, $F^G$ is quasifinite but not finite. In the second one, $F^G$ is not even quasifinite.

math.AG↗

arXiv2006

Invariant hypersurfaces for derivations in positive characteristic

Let $A$ be an integral $k$-algebra of finite type over an algebraically closed field $k$ of characteristic $p>0$. Given a collection ${\cal{D}}$ of $k$-derivations on $A$, that we interpret as algebraic vector fields on $X=Spec(A)$, we study the group spanned by the hypersurfaces $V(f)$ of $X$ invariant for ${\cal{D}}$ modulo the rational first integrals of ${\cal{D}}$. We prove that this group is always a finite $\mathbb{Z}/p$-vector space, and we give an estimate for its dimension. This is to be related to the results of Jouanolou and others on the number of hypersurfaces invariant for a foliation of codimension 1. As an application, given a $k$-algebra $B$ between $A^p$ and $A$, we show that the kernel of the pull-back morphism $Pic(B)\rightarrow Pic(A)$ is a finite $\mathbb{Z}/p$-vector space. In particular, if $A$ is a UFD, then the Picard group of $B$ is finite.

math.AG↗

arXiv2006

Minimal invariant varieties and first integrals for algebraic foliations

Let $X$ be an irreducible algebraic variety over $\mathbb{C}$, endowed with an algebraic foliation ${\cal{F}}$. In this paper, we introduce the notion of minimal invariant variety $V({\cal{F}},Y)$ with respect to $({\cal{F}},Y)$, where $Y$ is a subvariety of $X$. If $Y=\{x\}$ is a smooth point where the foliation is regular, its minimal invariant variety is simply the Zariski closure of the leaf passing through $x$. First we prove that for very generic $x$, the varieties $V({\cal{F}},x)$ have the same dimension $p$. Second we generalize a result due to X. Gomez-Mont. More precisely, we prove the existence of a dominant rational map $F:X\to Z$, where $Z$ has dimension $(n-p)$, such that for every very generic $x$, the Zariski closure of $F^{-1}(F(x))$ is one and only one minimal invariant variety of a point. We end up with an example illustrating both results.

math.AG↗

arXiv2006

Cohomology of regular differential forms for affine curves

Let $C$ be a complex affine reduced curve, and denote by $H^1(C)$ its first truncated cohomology group, i.e. the quotient of all regular differential 1-forms by exact 1-forms. First we introduce a nonnegative invariant $μ'(C,x)$ that measures the complexity of the singularity of $C$ at the point $x$. Then, if $H_1(C)$ denotes the first singular homology group of $C$ with complex coefficients, we establish the following formula: $$ dim H^1(C)=dim H_1(C) + \sum_{x\in C} μ'(C,x) $$ Second we consider a family of curves given by the fibres of a dominant morphism $f:X\to \mathbb{C}$, where $X$ is an irreducible complex affine surface. We analyze the behaviour of the function $y\mapsto dim H^1(f^{-1}(y))$. More precisely, we show that it is constant on a Zariski open set, and that it is lower semi-continuous in general.

math.AG↗

arXiv2006

Relative cohomology of polynomial mappings

Let $F$ be a polynomial mappping from $\mathbb{C}^n$ to $\mathbb{C}^q$ with $n>q$. We study the De Rham cohomology of its fibres and its relative cohomology groups, by introducing a special fibre $F^{-1}(\infty)$ "at infinity" and its cohomology. Let us fix a weighted homogeneous degree on $\mathbb{C}[x_1,...,x_n]$ with strictly positive weights. The fibre at infinity is the zero set of the leading terms of the coordinate functions of $F$. We introduce the cohomology groups $H^k(F^{-1}(\infty))$ of $F$ at infinity. These groups enable us to compute all the other cohomology groups of $F$. For instance, if the fibre at infinity has an isolated singularity at the origin, we prove that every weighted homogeneous basis of $H^{n-q}(F^{-1}(\infty))$ is a basis of all the groups $H^{n-q}(F^{-1}(y))$ and also a basis a the $(n-q)^{th}$ relative cohomology group of $F$. Moreover the dimension of $H^{n-q}(F^{-1}(\infty))$ is given by a global Milnor number of $F$, which only depends on the leading terms of the coordinate functions of $F$.

math.AG↗

arXiv2006

Relative exactness modulo a polynomial map and algebraic $(\mathbb{C}^p,+)$-actions

Let $F=(f_1,...,f_q)$ be a polynomial dominating map from $\mathbb{C}^n$ to $\mathbb{C}^q$. We study the quotient ${\cal{T}}^1(F)$ of polynomial 1-forms that are exact along the fibres of $F$, by 1-forms of type $dR+\sum a_idf_i$, where $R,a_1,...,a_q$ are polynomials. We prove that ${\cal{T}}^1(F)$ is always a torsion $\mathbb{C}[t_1,...,t_q]$-module. The we determine under which conditions on $F$ we have ${\cal{T}}^1(F)=0$. As an application, we study the behaviour of a class of algebraic $(\mathbb{C}^p,+)$-actions on $\mathbb{C}^n$, and determine in particular when these actions are trivial.

math.AG↗

arXiv2006

Surjectivity of quotient maps for algebraic $(\mathbb{C},+)$-actions and polynomial maps with contractible fibres

In this paper, we establish two results concerning algebraic $(\mathbb{C},+)$-actions on $\mathbb{C}^n$. First let $ϕ$ be an algebraic $(\mathbb{C},+)$-action on $\mathbb{C}^3$. By a result of Miyanishi, its ring of invariants is isomorphic to $\mathbb{C}[t_1,t_2]$. If $f_1,f_2$ generate this ring, the quotient map of $ϕ$ is the map $F:\mathbb{C}^3\to \mathbb{C}^2$, $x\mapsto (f_1(x),f_2(x))$. By using some topological arguments, we prove that $F$ is always surjective. Second, we are interested in dominant polynomial maps $F:\mathbb{C}^n\to \mathbb{C}^{n-1}$ whose connected components of their connected fibres are contractible. For such maps, we prove the existence of an algebraic $(\mathbb{C},+)$-action $ϕ$ on $\mathbb{C}^n$ for which $F$ is invariant. Moreover we give some conditions so that $F^*(\mathbb{C}[t_1,...,t_{n-1}])$ is the ring of invariants of $ϕ$.

math.AG↗

arXiv2006

Families of k-derivations on k-algebras

Let $A$ be an integral $k$-algebra of finite type over a field $k$ of characteristic zero. Let ${\cal{F}}$ be a family of $k$-derivations on $A$ and $M_{\cal{F}}$ the $A$-module spanned by ${\cal{F}}$. In this paper, we generalize a result due to A. Nowicki and construct an element $\partial$ of $M_{\cal{F}}$ such that $\ker \partial=\cap_{d\in {\cal{F}}} \ker d$. Such a derivation is called ${\cal{F}}$-minimal. Then we establish a density theorem for ${\cal{F}}$-minimal derivations in $M_{\cal{F}}$.

math.AC↗