Dans ce travail nous étudions les variétés determinantales essentiellement isolées (EIDS). Ce type de singularité est une généralization de la notion de singularité isolée. La variété determinantale générique $M_{m,n}^t$ est un sous-ensemble des matrices, mxn, tels que le rang est inférieur que t, où t≤m≤n. Une variété X est determinantal si X est définie comme la pré-image d'une fonction holomorphe, $F:\mathbb{C}^N \to M$, sur la variété determinantale générique avec la condition $codim X=codim M_{m,n}^t$.Certains travaux précédents ont étudié les variétés determinantales avec singularité isolée et ils ont défini le nombre de Milnor d'une surface determinantale et la caractéristique évanescente d'Euler.Nous étudions l'ensemble des hyperplans limites d'hyperplans tangents à une surface determinantale en $\mathbb{C}^4$ et 3-variété en $\mathbb{C}^5$ pour donner une caractérisation de ces hyperplans, par le fait que le nombre de Milnor de leur section avec la surface dans le premier cas ou la 3- variété dans le deuxième cas n'est pas minimum.Nous montrons également que, si X est une EIDS, de dimension d et H et H' sont des hyperplans fortement généraux, si $P \subset H$ et $P'\subset H'$ sont des plans de codimension d-2, les nombres de Milnor des surfaces genériques sont égaux.Nous étudions aussi la modification de Nash d'une EIDS et donnons des conditions suffisantes pour que cette transformation soit lisse.Un autre objectif de notre travail est l'étude de l'obstruction d'Euler d Nous obtenons des formules inductives qui relient l'obstruction d'Euler de X à la caractéristique d'Euler évanescente du lissage essentiel de leurs sections génériques. / In this work, we study the essentially isolated determinantal singularities (EIDS). This type of singularities is a natural generalization of isolated ones. A generic determinantal variety $M_{m,n}^t$ is a subset of the space of mxn matrices, given by matrices of rank less than t, where t≤m≤n. A variety X is determinantal if X is defined as the pre-image of $M_{m,n}^t$ by a holomorphic function $F:\mathbb{C}^N \to M$ with the condition $codim X=codim M_{m,n}^t$.Several recent works investigate determinantal variety with isolated singularities and they are difened the Milnor number and the vanishing Euler characteristic.In this work we study the set of limits of tangent hyperplanes to surface in $\mathbb{C}^4$ and 3-variety in $\mathbb{C}^5$ to give a characterization of this set by the fact that the Milnor number of its section with the surface in the first case or the 3-dimensional determinantal variety in the second case is not minimum. We also prove that if X is a d- dimensional EIDS and H and H' are strongly general hyperplans, if $P \subset H$ and $P'\subset H'$ are d-2 linear plans, the Milnor number of the generic surfaces are equal.We study the Nash transformation of an EIDS and give sufficient conditions for this transformation to be smooth.Another aim of our study is the Euler obstruction of essentially isolated determinantal singularities. We obtain inductive formulas associating the Euler obstruction with the vanishing Euler characteristic of the essencial smoothing of their generic sections.

Submitted by Ronildo Prado (ronisp@ufscar.br) on 2017-08-09T18:34:01Z No. of bitstreams: 1 TeseDAHA.pdf: 605987 bytes, checksum: 218da6f6f0b14c9296bc76440e616467 (MD5) / Approved for entry into archive by Ronildo Prado (ronisp@ufscar.br) on 2017-08-09T18:34:10Z (GMT) No. of bitstreams: 1 TeseDAHA.pdf: 605987 bytes, checksum: 218da6f6f0b14c9296bc76440e616467 (MD5) / Approved for entry into archive by Ronildo Prado (ronisp@ufscar.br) on 2017-08-09T18:34:17Z (GMT) No. of bitstreams: 1 TeseDAHA.pdf: 605987 bytes, checksum: 218da6f6f0b14c9296bc76440e616467 (MD5) / Made available in DSpace on 2017-08-09T18:34:26Z (GMT). No. of bitstreams: 1 TeseDAHA.pdf: 605987 bytes, checksum: 218da6f6f0b14c9296bc76440e616467 (MD5) Previous issue date: 2017-04-19 / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) / In this work, we show relations between invariants of map germs. First, we consider an analytic function germ f : (X, 0) —(C, 0) on an isolated determinantal singularity and we present a relation between the Euler obstruction of f and the determinantal Milnor number of f. In the particular case where (X, 0) is an isolated complete intersection singularity, we obtain a simple way to calculate the Euler obstruction of f as the difference between the dimension of two algebras. After, we work with map germs f : (X, 0) —— (C2, 0), where (X, 0) is a plane curve with isolated singularity. We introduce the image Milnor number to these map germs and we present a positive answer to the Mond’s conjecture in this context. The Mond’s conjecture proposes an inequality between two other invariants, the A^-codimension and the image Milnor number, in the case of map germs f : (Cn, 0) —(Cn+1, 0) when the dimensions (n,n + 1) is in Mather’s nice dimensions. The conjecture is true for n = 1, 2, and for the cases n > 3 is an open problem. / Neste trabalho, mostramos relações entre invariantes de germes de aplicações. Primeiro, consideramos um germe de funçao analítica f : (X, 0)^(C, 0) sobre uma singularidade determinantal isolada e apresentamos uma relaçao entre a obstrução de Euler de f e o número de Milnor determinantal de f. No caso particular em que (X, 0) e uma interseçao completa com singularidade isolada, obtemos um modo simples de calcular a obstrucao de Euler de f como a diferenca entre dimensães de duas algebras. Depois, trabalhamos com germes de aplicacoes f : (X, 0)^(C2, 0), onde (X, 0) e uma curva plana com singularidade isolada. Introduzimos o número de Milnor da imagem para estes germes de aplicacães e apresentamos uma resposta positiva para a conjectura de Mond neste contexto. A conjectura de Mond propoe uma desigualdade entre outros dois invariantes, a A^-codimensao e o numero de Milnor da imagem, para o caso de germes de aplicacoes f : (Cn, 0)^(Cn+1,0) quando as dimensoes (n,n + 1) estao nas boas dimensoes de Mather. A conjectura e verdadeira para n = 1, 2, e para os casos n > 3 e um problema em aberto.