Real and Complex Singularities XI International

Workshop on Real and Complex Singularities July

26 30 2010 Instituto De Ciencias Matematicas E
De De Sao Victor Goryunov pdf download

https://ebookgate.com/product/real-and-complex-singularities-xi-international-workshop-on-real-and-
complex-singularities-july-26-30-2010-instituto-de-ciencias-matematicas-e-de-de-sao-victor-goryunov/

★★★★★ 4.7/5.0 (33 reviews) ✓ 191 downloads ■ TOP RATED

"Great resource, downloaded instantly. Thank you!" - Lisa K.

DOWNLOAD EBOOK
 Real and Complex Singularities XI International Workshop on
Real and Complex Singularities July 26 30 2010 Instituto De
Ciencias Matematicas E De De Sao Victor Goryunov pdf
download

TEXTBOOK EBOOK EBOOK GATE

Available Formats

■ PDF eBook Study Guide TextBook

EXCLUSIVE 2025 EDUCATIONAL COLLECTION - LIMITED TIME

INSTANT DOWNLOAD VIEW LIBRARY

Instant digital products (PDF, ePub, MOBI) available
Download now and explore formats that suit you...

The Japanese Australian Workshop on Real and Complex

Singularities JARCS III The University of Sydney Sydney 15
18 September 2009 1st Edition Toshizumi Fukui
https://ebookgate.com/product/the-japanese-australian-workshop-on-
real-and-complex-singularities-jarcs-iii-the-university-of-sydney-
sydney-15-18-september-2009-1st-edition-toshizumi-fukui/
ebookgate.com

Reflections on Groups and Organizations On the Couch With

Manfred Kets de Vries 1st Edition Kets De Vries

https://ebookgate.com/product/reflections-on-groups-and-organizations-
on-the-couch-with-manfred-kets-de-vries-1st-edition-kets-de-vries/

ebookgate.com

The Bourbon Restoration Guillaume De Bertier De Sauvigny

https://ebookgate.com/product/the-bourbon-restoration-guillaume-de-
bertier-de-sauvigny/

ebookgate.com

de Lahunta s Veterinary Neuroanatomy and Clinical

Neurology 5th Edition De Lahunta

https://ebookgate.com/product/de-lahunta-s-veterinary-neuroanatomy-
and-clinical-neurology-5th-edition-de-lahunta/

ebookgate.com
Portal Hypertension IV Proceedings of the 4th Baveno
International Consensus Workshop 1st Edition Roberto De
Franchis
https://ebookgate.com/product/portal-hypertension-iv-proceedings-of-
the-4th-baveno-international-consensus-workshop-1st-edition-roberto-
de-franchis/
ebookgate.com

String theory and the real world from particle physics to

astrophysics Ecole d ete de physique des Houches session
LXXXVII 2 July 27 July 2007 Ecole thematique du CNRS 1st
ed Edition C. Bachas
https://ebookgate.com/product/string-theory-and-the-real-world-from-
particle-physics-to-astrophysics-ecole-d-ete-de-physique-des-houches-
session-lxxxvii-2-july-27-july-2007-ecole-thematique-du-cnrs-1st-ed-
edition-c-bachas/
ebookgate.com

Functional Analytic and Complex Methods Proceedings of the

International Graz Workshop 1st Edition Wolfgang Tutschke

https://ebookgate.com/product/functional-analytic-and-complex-methods-
proceedings-of-the-international-graz-workshop-1st-edition-wolfgang-
tutschke/
ebookgate.com

Singularities in Physics and Engineering 2nd Edition

Paramasivam Senthilkumaran

https://ebookgate.com/product/singularities-in-physics-and-
engineering-2nd-edition-paramasivam-senthilkumaran/

ebookgate.com

Singularities Formation Structure and Propagation 1st

Edition Jens Eggers

https://ebookgate.com/product/singularities-formation-structure-and-
propagation-1st-edition-jens-eggers/

ebookgate.com
 569

Real and Complex Singularities

XI International Workshop on
Real and Complex Singularities
July 26–30, 2010
Instituto de Ciências Matemáticas e de Computação
Universidade de São Paulo
São Carlos, SP, Brazil

Victor Goryunov
Kevin Houston
Roberta Wik-Atique
Editors

American Mathematical Society

Real and Complex Singularities
XI International Workshop on
Real and Complex Singularities
July 26–30, 2010
Instituto de Ciências Matemáticas e de Computação
Universidade de São Paulo
São Carlos, SP, Brazil

Victor Goryunov
Kevin Houston
Roberta Wik-Atique
Editors
 569

Real and Complex Singularities

XI International Workshop on
Real and Complex Singularities
July 26–30, 2010
Instituto de Ciências Matemáticas e de Computação
Universidade de São Paulo
São Carlos, SP, Brazil

Victor Goryunov
Kevin Houston
Roberta Wik-Atique
Editors

American Mathematical Society

Providence, Rhode Island
 EDITORIAL COMMITTEE
Dennis DeTurck, managing editor
George Andrews Abel Klein Martin J. Strauss

2010 Mathematics Subject Classification. Primary 58Kxx, 57Rxx, 57Qxx, 32Sxx,

14Pxx, 37Cxx.

Library of Congress Cataloging-in-Publication Data

International Workshop on Real and Complex Singularities (11th : 2010 : Universidade de São
Paulo) Real and complex singularities : XI International Workshop on Real and Complex Singu-
larities, July 26-30, 2010, Universidade de São Paulo, São Carlos, SP Brazil / Victor Goryunov,
Kevin Houston, Roberta Wik-Atique, editors.
p. cm. — (Contemporary mathematics ; v. 569)
Includes bibliographical references.
ISBN 978-0-8218-5359-7 (alk. paper)
1. Singularities (Mathematics)–Congresses. I. Goryunov, Victor, 1955- II. Houston, Kevin,
1968- III. Wik-Atique, Roberta, 1964- IV. Title.

QA614.58.1527 2010
514.746—dc23 2011052531

Copying and reprinting. Material in this book may be reproduced by any means for edu-
cational and scientific purposes without fee or permission with the exception of reproduction by
services that collect fees for delivery of documents and provided that the customary acknowledg-
ment of the source is given. This consent does not extend to other kinds of copying for general
distribution, for advertising or promotional purposes, or for resale. Requests for permission for
commercial use of material should be addressed to the Acquisitions Department, American Math-
ematical Society, 201 Charles Street, Providence, Rhode Island 02904-2294, USA. Requests can
also be made by e-mail to [email protected].
Excluded from these provisions is material in articles for which the author holds copyright. In
such cases, requests for permission to use or reprint should be addressed directly to the author(s).
(Copyright ownership is indicated in the notice in the lower right-hand corner of the first page of
each article.)


c 2012 by the American Mathematical Society. All rights reserved.
The American Mathematical Society retains all rights
except those granted to the United States Government.
Printed in the United States of America.

∞ The paper used in this book is acid-free and falls within the guidelines
established to ensure permanence and durability.
Visit the AMS home page at http://www.ams.org/
10 9 8 7 6 5 4 3 2 1 17 16 15 14 13 12
This volume is dedicated to David Mond
on the occasion of his 60th birthday

v
 Contents

Preface ix
David Mond
K. Houston xi
Milnor fibrations and the concept of d-regularity for analytic map germs
J.L. Cisneros-Molina, J. Seade, and J. Snoussi 1
Bi-Lipschitz G-triviality and Newton polyhedra, G = R, C, K, RV , CV , KV
J.C.F. Costa, M.J. Saia, and C.H. Soares Júnior 29
Symplectic Sμ singularities
W. Domitrz and Z. Trȩbska 45
Topology of the real Milnor fiber for isolated singularities
R. Araújo dos Santos, D. Dreibelbis, and N. Dutertre 67
Compact 3-manifolds supporting some R -actions
2

C. Maquera and W.T. Huaraca 77

Timelike canal hypersurfaces of spacelike submanifolds in a de Sitter space
M. Kasedou 87
Residues in K-theory
D. Lehmann 101
Multicusps
Y. Mizota and T. Nishimura 115
Small growth vectors of the compactifications of the contact
systems on J r (1, 1)
P. Mormul 123
Vassiliev type invariants for generic mappings, revisited
T. Ohmoto 143
Sections of Analytic Variety
B. Oréfice and J.N. Tomazella 161
The Artin-Greenberg function of a plane curve singularity
S. Saleh 177
Singularities with critical locus a complete intersection
and transversal type A1
M. Shubladze 193

vii
 Preface

This volume is a collection of papers presented at the 11th Workshop on Real

and Complex Singularities, São Carlos, Brazil, July 26-30, 2010. The meeting was
a part of a highly successful series of biennial conferences organized by the Sin-
gularity Theory group at São Carlos, University of São Paulo, Brazil. It is the
longest running sequence of international workshops in singularities, which have
been providing an exceptional opportunity for both young researchers and recog-
nised leaders working in the field to meet together in a very productive scientific
atmosphere. In 2010 a total of 160 participants from 13 countries (Brazil, Canada,
Denmark, France, Germany, Japan, Mexico, Norway, Poland, Portugal, Spain, UK,
USA) came to the Workshop.
The meeting in 2010 had a special dedication – to David Mond’s 60th birthday.
David Mond, professor at the University of Warwick, United Kingdom, is one of the
leading experts in the area and has inspired many young mathematicians. Many
years ago he fell in love with South America. Nowadays David has very strong
connections with the Singularity group at Sao Carlos.
The main subject of singularity theory is the geometry and topology of spaces
and maps defined by polynomials or analytic equations which are not regular. The
theory uses techniques from several branches of mathematics and contributes to
the development of rather distant fields, including algebraic geometry, knot theory,
optics, computer vision, and many others. The possibility of application in a large
number of different areas is one of the reasons of the success of singularity theory.
This book reflects the high level of the conference. It discusses the most recent
results and applications of singularity theory and shows promising directions for
future research in the field. Therefore, it will be an excellent reference for expe-
rienced researchers and an ideal introduction for younger, such as PhD students
and post-docs, not only for current results but also for the variety of methods and
techniques used in singularity theory.
The volume covers pure singularity theory (invariants, classification theory,
Milnor fibres) and applications to other areas (singularities in topology and dif-
ferential geometry, algebraic geometry and bifurcation theory). In particular, the
reader will find here papers on plane curve singularities, metric theory of singulari-
ties, symplectic singularities, cobordisms of maps, Goursat distributions, sections of
analytic varieties, Vassiliev type invariants, projections of hypersurfaces, properties
of the Jacobian ideal.
We thank members of the Scientific Committee: Ragnar Buchweitz, Francisco
Castro Jiménez, James Damon, Washington L. Marar, Walter Neumann, Juan
José Nuño Ballesteros, Marcio Soares and Duco van Straten. We are also thankful
to members of the Organizing Committee: Alexandre Fernandes, Nivaldo Grulha Jr,
ix
x PREFACE

Regilene Oliveira, Marcelo Saia, João Tomazella and Catiana Casonatto. We would
like to express our gratitude to all others who work hard for the success of the
meeting.
The workshop was funded by Brazilian funding bodies Fapesp, CNPq, CAPES,
USP and INCTMat and the Japanese funding body JSPS, whose support we grate-
fully acknowledge.
We thank the referees for their diligent work in refereeing all the papers in
this volume. We thank the staff members of the American Mathematical Society
involved with the preparation of this book, and all those who have contributed in
whatever way to these proceedings.

Victor Goryunov
Kevin Houston
Roberta Wik-Atique
 David Mond

To celebrate the 60th birthday of David Mond the 2010 Real and Complex Sin-
gularities conference, a meeting held biennually in São Carlos, Brazil, was dedicated
to him. David has been an organizer and proceedings editor for earlier Real and
Complex Singularities conferences. Furthermore, having two of his PhD students
based in São Carlos means that David has strong connections with the singularity
group there and that the conference was a great opportunity to honour him.
David had not originally intended to become an academic. After completing his
undergraduate degree in Mathematics and Philosophy at St. Catherine’s College,
Oxford, 1971, he was inspired by the work of Dutch furniture maker Gerrit Rietveld
and initially pursued the ambition of making furniture himself. In 1973 with an
Oxford friend he visited South America intending to travel through Venezuela and
Colombia to Peru. Their journey came to a halt in Bogotá, Colombia, where David
remained in the end for eight years.
It was in Colombia that David returned to mathematics, taking a Master’s
degree in Mathematics in the Universidad Nacional de Colombia, and later working
as a mathematics instructor at Universidad de los Andes, and Universidad Nacional,
both in Bogotá. His first five, essentially expository, papers appeared in Colombian
mathematical journals and hence were written in Spanish – a language in which,
along with Portuguese, he is fluent. After finishing the MSc in Mathematics at the
Universidad Nacional, he moved in 1979 to the UK to begin a PhD at the University
of Liverpool under the supervision of C.T.C. Wall FRS. His PhD, submitted in 1982,
was entitled The Classification of Germs of Maps from Surfaces to 3-space, with
Applications to the Differential Geometry of Immersions. From the late 1960s there
had been much progress in the classification of singularities, particularly for right
equivalence following V.I. Arnold’s classifications. What was novel and significant
about David’s thesis was that it concerned the A-equivalence (also known as right-
left equivalence) for germs from R2 to R3 . The stable maps in this case were
classified by Whitney back in the 50s, David was classifying simple maps and those
with Ae -codimension less than or equal to 4. In the late 70s and early 80s such a
classification required doing by hand long, tedious calculations involving tangent
spaces, although David insists that it is often through long tedious calculations that
one builds up an understanding of what is going on. Aspects of the classification
were applied in differential geometry, for example to study the tangent developable
of a space curve.
After his PhD, David had a brief research visit at the IHES in France and
returned to Colombia to work again at the Universidad Nacional. In 1985 he
moved to the University of Warwick in the UK. Of particular note during this
time is the publication Some remarks on the geometry and classification of germs

xi
xii DAVID MOND

of maps from surfaces to 3-space in Topology 26 (1987) 361-383. It is here that

David starts a thread that passes through much of his work by studying complex
analytic maps, in this case investigating how invariants of map germs from C2
to C3 control the determinacy of the germ. It is in this paper that the (still
unresolved) Mond conjecture first begins to take form: the Ae -codimension of a map
germ from (Cn , 0) to (Cn+1 , 0) is less than or equal to the rank of the vanishing
homology for the image of a local stabilization of the map, with equality if the
map is quasihomogeneous. Various Mond students - including this author - have
attempted to prove this statement during their PhD studies.
Also at this time David had his first PhD student, Washington ‘Ton’ Marar from
Brazil, with whom he further developed the ideas in the Topology paper to show
the multiple point spaces for finitely A-determined corank 1 maps from Cn to Cp ,
n < p, were in fact isolated complete intersection singularities (or zero-dimensional).
His interest in classification continued with his second student, Diana Ratcliffe,
producing a thesis on classifying multi-germs from R2 and R3 . She later became
his post-doctoral assistant and produced a computer program that relieved the
drudgery from the extensive calculations involved in classifications. David remarked
in one seminar that her program could do a calculation in seconds that he had spent
weeks doing during his postgraduate days. A visit to Holland at this time led to a
collaboration with Ruud Pellikaan, and later to a series of papers with Duco van
Straten.
Other students followed in the early 90s with Thomas Cooper and this author
both completing in 1994. By this time David had collaborated with other notable
Singularity Theorists: Terry Gaffney, on the germs of maps from the plane to the
plane; Jim Damon (in their paper for Inventiones Mathematicae) on a result that
related Ae −codimension and the rank of the vanishing homology for maps with
n ≥ p (which added evidence to the Mond conjecture); Victor Goryunov, on a
powerful spectral sequence that allowed one to calculate the rational cohomology
of the image of a finite map.
David continued his association with São Carlos, co-supervising Roberta Wik-
Atique with Maria Ruas, again on classification of singularities. However, following
the importance of free divisors in the paper co-authored with Damon, he began
to study these objects in more detail. A hypersurface D is a free divisor if the
module of ambient vector fields which are tangent to D at its smooth points is
freely generated. This property has significant consequences for the topology of
the singularity, and, as the Inventiones paper showed, for the topology of its non-
linear sections, because it implies the conservation of a certain singular multiplic-
ity. This interest eventually produced a string of papers singly and co-authored
with people such as Francisco Castro-Jiménez, Luis Narváez-Macarro, Martin Hol-
land, Francisco Calderón Moreno, Ragnar-Olaf Buchweitz, Michel Granger, Alicia
Nieto-Reyes, Mathias Schulze, Ignacio de Gregorio (another PhD student and later
postdoctoral assistant) and Christian Sevenheck. In this string of papers, the im-
portant concept of linear free divisor is introduced. (This is a free divisor where
the basis is formed by linear vector fields.) The importance partly stems from the
connection to quiver representations as shown in the paper with Buchweitz. They
also have connection with Frobenius manifolds since for a large class of examples,
the Gauss-Manin system associated to a generic linear section of the Milnor fibre
of a linear free divisor is a Frobenius manifold.
 DAVID MOND xiii

Recent students have been Paul Cadman (2010) and Ayşe Altintas (2011). The
latter was one of David’s students present at the conference in his honour. The
conference was a great success and during the outing to a nearby lake the attendees
were treated to a performance of David playing the flute. At the conference dinner
friends, colleagues and ex-students took turns to pay tribute and sing the praises
of a man they are proud to call their friend and teacher.

Kevin Houston
Leeds, UK, 2011.
Contemporary Mathematics
Volume 569, 2012
http://dx.doi.org/10.1090/conm/569/11241

Milnor Fibrations and the Concept of d-regularity for

Analytic Map Germs

José Luis Cisneros-Molina, José Seade, and Jawad Snoussi

Abstract. In this expository article we review Milnor’s proof of his celebrated
fibration theorem and the way how it extends to the case of real analytic map
germs. We highlight the similarities and the differences between the complex
analytic and the real analytic settings, and we indicate the key geometric
aspects in Milnor’s proof that make it work in the complex case, unlike the
theorem in the real setting which is more stringent. We explain too how, from
the viewpoint of Milnor type fibrations, the difference between the real and
the complex settings naturally leads to the concept of d-regularity.

Introduction
Milnor’s fibration theorem for holomorphic maps is a key-stone in singularity
theory. This theorem says that given a holomorphic map-germ f : ( n , 0) → ( , 0)
with a critical point at the origin 0 ∈ n , one has two locally trivial fibrations
associated to it, and these fibrations are equivalent. The first is Milnor’s fibration:

(1) ε \ K −→ 1 ,
φ :=
f
|f |
:

where K is the link of f at 0, that is K = f −1 (0) ∩ ε with ε being a sufficiently

small sphere around 0.
The second is nowadays called the Milnor-Lê fibration; this is a fibration on
  
a “Milnor tube” N (ε, δ) = ε ∩ f −1 (∂ δ ) where ε is a ball bounded by ε , 

∂ δ is the sphere bounding the ball of radius δ centred at the origin of . When
0 < δ  ε  1 the map f induces a fibration
(2) f : N (ε, δ) → ∂ δ .
When 0 is an isolated critical point of f , the map φ actually defines an open-
book decomposition of the sphere, with binding K.

2010 Mathematics Subject Classification. Primary 32S05, 32S55, 57Q45; Secondary 58K65.
Key words and phrases. Real and complex singularities, Milnor fibration, canonical pencil,
d-regularity.
This research was partially supported by CONACYT (Mexico) grants G-36357-E, J-49048-
F, U-55084, by DGAPA-UNAM-PAPIIT (Mexico) grants IN105806, IN102208, IN108111, by the
CNRS, France and by ECOS–ANUIES grant M06-M02, France–Mexico.
The first author is a Regular Associate of the Abdus Salam International Centre for Theo-
retical Physics, Trieste, Italy.

2012
c American Mathematical Society

1
2 JOSÉ LUIS CISNEROS-MOLINA, JOSÉ SEADE, AND JAWAD SNOUSSI

It is natural to ask whether there is anything similar for real analytic map-
 f

germs ( n , 0) → ( p , 0) with n ≥ p. This question goes back to Milnor’s book,
and actually to his unpublished preprint [20], where he proves that if f has an
isolated critical point, then one always has a fibration in a tube, of type (2), now

over a small (p − 1)-sphere in p , and this tube can always be inflated to the
 ψ

sphere, defining an equivalent fibration n−1 \ K −→ p−1 , which restricted to a
neighbourhood of K is the obvious projection f /|f |. Yet, this theorem has several
weaknesses, as for instance the following (the first two of these were pointed out by
Milnor in his book):
i) The condition of having an isolated critical point is very stringent. When
this is satisfied we say that the corresponding map-germ satisfies the Milnor
condition. In [19, 9] the authors classify the pairs (n, p) for which there
exist analytic map-germs satisfying Milnor’s condition. There are many other
related works by various authors, as for instance N. A’Campo [1, 2], B. Perron
[25], L. Rudolph [30, 15] and many others. In particular one has that when
p = 2, even though the Milnor condition is indeed stringent, there are plenty
of map-germs satisfying it, for all n ≥ 2. More generally, one can study
Milnor fibrations associated to the larger class of real analytic mappings with
an isolated critical value. This was begun in [27] and continued in [11, 12].
ii) We cannot always take the map ψ to be globally the canonical projection φ =
f /f . When the equality holds we say that the map germ satisfies the strong
Milnor condition. Examples of singularities with the strong Milnor condition
have been given, for instance, in [17, 31, 29, 10, 28, 23].
iii) If one has that the two types of fibrations exist, one on a Milnor tube and
another on a sphere minus the link, a priori we do not know whether or not
the two fibrations are necessarily equivalent. This point was first noticed by R.
N. Araujo dos Santos (see for instance [3]), and we fully answered it in [12].
We refer to our recent article [13] for a survey on the topic of Milnor fibra-
tions for real singularities, where we explore carefully the above points and give an
overview of the developments about them that one has in the literature.
Here we focus on explaining the main ideas and results in [11, 12], which give
an essentially complete answer to the problem of existence and equivalence of the
two types of Milnor fibrations for maps with an isolated critical value.
The key-idea is a certain regularity condition that we introduced in [11, 12],
called d-regularity, which involves the family of varieties associated to every map-
 f
 
germ ( n , 0) → ( p , 0) as follows: for each line L through the origin in p set:


XL := {x ∈ n  f (x) ∈ L } .
The family {XL } is the canonical pencil of f . When f has an isolated critical value,
these analytic varieties are all smooth away from V , the axis of the pencil. The
d-regularity property means that there exists ε > 0 such that every sphere of radius
≤ ε intersects transversally each manifold XL \ V .
We show that d-regularity is precisely the condition required in order to inflate
the fibration on a Milnor tube conserving the value ff  , until we get the Milnor
f
fibration on the sphere with projection map f  . The theorem says:
 
Theorem. Let f : ( n , 0) → ( p , 0), n > p, be a real analytic map-germ,
which is surjective over a neighbourhood of 0 and with dim f −1 (0) > 0. Assume
 MILNOR FIBRATIONS AND THE CONCEPT OF d-REGULARITY 3

further that f has the Thom property, so one has a Milnor-Lê fibration
f : N (ε, δ) −→ ∂ δ ∼= p−1 .
Then one has also a locally trivial Milnor fibration:

φ=
f
f 
: ε \ (ε ∩ f −1 (0)) −→ p−1 ,
if and only if f is d-regular, and in this case these two fibrations are equivalent.
The concept of d-regularity is actually implicit in Milnor’s proof of his fibration
theorem. This concept is also used in [11] to give an alternative way for looking at
Milnor’s fibration theorem for holomorphic mappings and its generalisation by Lê
Dũng Tráng to functions on singular spaces. This concept also appears implicitly in
[27, Lemma 5.7] to prove the equivalence of the two types of Milnor fibrations, as
in (1) and (2), for functions of the type f ḡ. This regularity condition also appears
implicitly in [5] for map-germs with an isolated critical point.
The name “d-regularity” comes from the facts that this concept is naturally
associated to a function “distance to the critical point”, and also that every map-
germ (with an isolated singularity) which is (c)-regular in K. Bekka’s sense, is
d-regular in our sense (see [28] for details and [6] for definitions).
The purpose of this expository article is to give an introduction to d-regularity
and explain the theorem above. We begin the article with a section on real and
complex examples where it is easy to describe the two types of Milnor fibrations,
and where d-regularity appears implicitly to give the equivalence of the two types of
fibrations. Then we go back to the holomorphic setting, because it is actually from
Milnor’s work that this concept springs. We review Milnor’s original proof of the
fibration theorem from a perspective that leads towards the concept of d-regularity.
Finally, in Section 3 we explain how Thom’s af condition relates to these fibration
theorems, we speak about d-regularity and explain how this concept allows to prove
that whenever the two types of fibration exist, they are equivalent.

1. An example: weighted homogeneous singularities

In this section we look at Milnor fibrations for real and complex weighted homo-
geneous singularities. We start with a quick look at the celebrated Pham-Brieskorn
singularities, which provide the paradigm of Milnor fibrations. Throughout this
 
work, given a map germ n → p , we denote by 0 the origin in the source, and by
0 that in the target.

1.1. Milnor fibrations for the Pham-Brieskorn singularities. Consider

a Pham-Brieskorn polynomial f : n → :
f
(z1 , ..., zn ) → z1a1 + ... + znan , ai ≥ 2 .
It is clear that the origin 0 ∈ n is its only critical point, so V := f −1 (0) is a
hypersurface with an isolated singularity at 0.
Notice that this polynomial is weighted homogeneous. In fact, let d be the
lowest common multiple of the ai and for each i = 1, · · · , n set di = d/ai . Then for
every non-zero complex number λ ∈ ∗ one has a ∗ -action on n determined by
λ · (z1 , · · · , zn ) → (λd1 z1 , · · · , λdn zn ) .
4 JOSÉ LUIS CISNEROS-MOLINA, JOSÉ SEADE, AND JAWAD SNOUSSI

Observe one has:

f (λd1 z1 , · · · , λdn zn ) = λd f (z1 , · · · , zn ) .
We thus get a holomorphic action of ∗ which has the origin 0 as its only fixed
point and V is an invariant set, union of orbits. It is clear that this ∗ -action has
the following four properties:
• Property 1. If we restrict the action to the positive real numbers t ∈ + , 
then each such orbit is a curve that converges to 0 as t tends to 0, and it goes to
infinity as t tends to ∞. Furthermore, a direct computation shows that for the
points in each such orbit, the distance to 0 is a function with no critical point, so
each orbit is transversal to all spheres centred at 0. Hence the variety V intersects

transversally every (2n − 1)-sphere r around the origin and the intersection Kr :=

V ∩ r is a real codimension 2 smooth submanifold of the sphere r . Also, this 
real flow determines a 1-parameter group of diffeomorphisms and given arbitrary
 
spheres r , r centred at 0, we get a diffeomorphism from r into r taking Kr  
into Kr . Thence the diffeomorphism type of Kr is independent of the choice of

the sphere r . We denote this manifold simply by K and call it the link of the
singularity (see [21, 14]).
• Property 2. The above real analytic flow also has the property that for
points in n \ V , the argument of the complex number f (z) is constant on each
orbit, i.e., f (z)/|f (z)| = f (tz)/|f (tz)| for all t ∈ + . 
• Property 3. If we restrict the ∗ -action to the complex numbers of unit

length we get an 1 -action that leaves invariant every sphere around 0, and for each
eiθ and each z = (z1 , · · · , zn ) we have f (eiθ · (z1 , · · · , zn )) = eiθd · f (z1 , · · · , zn ).
Hence for each ζ ∈ we have that the multiplication by eiθ in n transports the
fibre f −1 (ζ) into the fibre f −1 (eiθd · ζ). Therefore, if for each real number δ > 0

we consider the “tube” N (δ) = f −1 (∂ δ ), where ∂ δ ∼  
= 1 is the boundary of the
disc in of radius δ, centred at 0, then 1

acts on N (δ). A direct computation
shows that the orbits of this action are transverse to the fibers of f . So the action
determines a locally trivial fibration (a fibre bundle):
f : N (δ) → ∂ δ .
Together with Property 2, this implies that the fibres f −1 (λ) are all diffeomorphic.
Now observe that for each line Lθ through the origin in , we may consider
the set
Xθ := {z ∈ n | f (z) ∈ Lθ } .
Then each Xθ is the zero-set of f followed by an orthogonal projection. Hence
each Xθ is a real analytic hypersurface with an isolated singularity at 0, their union
fills the entire space n and their intersection is V . By property 1, each of these
hypersurfaces is transversal to all the spheres, and by Property 3, the 1 -action 
permutes these hypersurfaces: multiplication by eit carries each Xθ into another
one of these hypersurfaces. In other words one has:
• Property 4. These varieties define a pencil in n , a sort of open-book where
the binding is now the singular variety V , and each of these varieties is transverse
to every sphere around 0. If we remove V from n , for every ball around 0 we get
a canonical projection

(3) Φ=
f
|f |
: 2n \ V −→ 1 ,
 MILNOR FIBRATIONS AND THE CONCEPT OF d-REGULARITY 5

which actually is a fibre bundle; the fibre over a point eiθ is the corresponding
connected component of Xθ \ V . (The other component is the fibre over e−iθ .)
We now focus our attention near the origin, say restricted to the unit ball
 2n

in n . Since each Xθ meets transversally the sphere 2n−1 = ∂ 2n , the 
intersection is a smooth codimension 1 submanifold of the sphere, containing the
  
link K = V ∩ 2n−1 . And since the orbits of the 1 -action preserve the sphere 2n−1 ,

it follows that the restriction of Φ to 2n−1 defines a locally trivial fibration:

(4) φ=
f

|f |
: 2n−1 \ K −→ 1 
This is the classical Milnor fibration for the map f . It is worth saying that for
these maps, the Pham-Brieskorn polynomials, the fact that (4) is a fibre bundle
was first proved by Pham and used by Brieskorn to prove important results about
the topology of the corresponding links (for more about this, see [32, Chapter 1]
and the bibliography in it).
Remark 1.1. Since f has a unique critical point at 0, the implicit function
theorem implies that at each point x ∈ V ∗ := V \ {0} we can find local coordinates
so that f is locally a linear projection n → . Hence the fibers of f are locally
(in a neighbourhood of x) parallel complex discs of complex dimension n − 1. This
implies the following statement, that we do not need here, but which is the key
for generalising this discussion to the case of (real and complex) analytic mappings
with non-isolated critical points: every smooth map-germ with an isolated critical
point has the Thom af -property. We refer to 3.1 for a precise definition; for more
about this important concept see for instance [11] and the bibliography in it.
In our case, that of the Pham-Brieskorn singularities, Remark 1.1 tells us that,

since V is transverse to the unit sphere 2n−1 , the fibre of f passing through each
point in the sphere sufficiently near V , is also transverse to the sphere. Since

K = V ∩ 2n−1 is a compact set, it follows that there exists a real number δ > 0

such that all fibres f −1 (λ) with |λ| = δ, are transverse to 2n−1 . In other words,

the tube N (δ) in Property 3 is everywhere transverse to 2n−1 . Therefore, setting

N (1, δ) := N (δ) ∩ 2n , where the 1 means that the ball has radius 1, we have that
the fibre bundle described by Property 3 determines a fibre bundle:
(5) 
f : N (1, δ) → ∂ δ ∼
= 1. 
This is the second classical version of Milnor’s fibration for the map f . Following
the modern literature, we call this the Milnor-Lê fibration of f .

Notice that by Property 2, each + -orbit is everywhere transverse to the tube

N (δ) and transverse to the sphere 2n−1 , and the complex numbers f (z) have
constant argument along each orbit. Thence the integral lines of this action deter-

mine a diffeomorphism between N (1, d) and 2n−1 minus the part of the sphere

◦
−1
contained inside the open solid tube f ( δ ). This determines the classical equiv-
alence between the Milnor fibration in the sphere (4) and the Milnor-Lê fibration
in the tube (2).
We now remark that everything we said above works in exactly the same way
for all weighted homogeneous complex singularities, i.e., for all complex polyno-
mials f for which there is a ∗ action on n as above, for some positive integers
{d; d1 , · · · , dn },
λ · (z1 , · · · , zn ) = (λd11 z1 , · · · , λdnn zn ),
6 JOSÉ LUIS CISNEROS-MOLINA, JOSÉ SEADE, AND JAWAD SNOUSSI

∗
satisfying that for all λ ∈ and for all z ∈ n
one has:
f (λ · z) = λd · f (z) .
This follows because if we look carefully at the above discussion we see that Prop-
erties 1 to 4 are intrinsically given by the weights of the action and the fact that f
brings out scalars to the power d. All the statements above concerning the fibration
theorems follow from these properties. More precisely:
i) Property 1 implies there is a flow on V \ {0} given by the + -action, which 
is transverse to all spheres around the origin 0. Since f has an isolated critical

point at 0, this implies that given a sphere ε , all fibers f −1 (t) sufficiently near

V are also transverse to ε . Property 3 then implies that given a tube N (ε, δ) :=
  
f −1 ( δ ) ∩ ε ), the flow defined by the 1 -action permutes the fibers in this tube
and we have a Milnor-Lê fibration
(6) f : N (ε, δ) → ∂ δ ∼= 1 .
ii) Properties 2 and 4 imply that one has a decomposition of all of n as an
“open-book” with binding the singular variety V and pages the real hypersurfaces
Xθ \ V .
iii) Then Property 2 implies that the Xθ are transverse to every sphere ε , and 
therefore the above open book decomposition of n determines the Milnor fibration

(7) φ=
f
|f |
: 2n−1
ε \ K −→ 1 .

iv) Finally, properties 1 and 2 imply that the flow given by the + -action 
determines an equivalence between the two fibrations of f , the Milnor fibration and
the Milnor-Lê-fibration.
As we will see in the sequel, there are several interesting families of singularities
which can be equipped with flows that satisfy properties analogous to (i)-(iv), and
whenever we have these properties, we have the two types of fibrations and they
are equivalent.

1.2. Polar weighted singularities. Let us consider now polynomial maps

  
f : 2n → 2 , n ≥ 1 that carry the origin 0 ∈ 2n into the origin 0 ∈ 2 . We 

identify the plane 2 with the complex line and equip ∗ with polar coordinates
{r eiθ | r > 0 ; θ ∈ [0, 2π)}; notice that as a manifold ∗ is the cylinder 1 × + .  

We also identify 2n with n .
The following concept is introduced in [10].
Definition 1.2. A map f as above is a polar weighted homogeneous polynomial
 
map if there exists an action of 1 × + on n of the form
(λ, r) · (z1 , . . . , zn ) = (λ r z1 , . . . , λdn r pn zn ) ,
d1 p1
λ∈ 1 , r ∈ + ,
where the dj , pj are positive integers such that gcd(d1 , . . . , dn ) = 1 = gcd(p1 , . . . , pn ),
and one has:
f ((λ, r) · (z1 , . . . , zn )) = λd r p f (z1 , . . . , zn ) ,
for some positive integers d, p.
Notice this includes the singularities given by weighted homogeneous complex
polynomials; in that case the degrees d and p coincide.
 MILNOR FIBRATIONS AND THE CONCEPT OF d-REGULARITY 7

The paradigm of polar weighted homogeneous maps is given by the Twisted

Pham-Brieskorn singularities studied in [29] (see also Chapter VII in [32]). These
are the singularities defined by maps of the following type:
(z1 , ..., zn ) → z1a1 z̄σ(1) + ... + znan z̄σ(n) , ai ≥ 2 ,
where σ is a permutation of the set {1, ..., n}. The explicit weights are easily
computable from the exponents ai and the permutation σ (see Lemmas 4.3 and 4.4
in Chapter VII of [32]). The name comes from their similarity with the classical
Pham-Brieskorn polynomials and the fact proved in [29, 24], that if σ is the identity,
then these singularities are equivalent to the usual Pham-Brieskorn singularities:
(z1 , ..., zn ) → z1a1 −1 + ... + znan −1 .
When σ is not the identity, these singularities are not equivalent to complex singu-
larities, generally speaking (by [26]).
Using the polar action given in Definition 1.2 one can show (see [10]) that
polar weighted homogeneous polynomials have 0 ∈ as the only critical value, and
assuming that the dimension of V = f −1 (0) is more than 0, the above Properties 1-4
continue to hold in this setting. Therefore every polar weighted singularity has the
two types of Milnor fibrations that complex singularities have: one on the sphere
given as in equation (4) and another in the tube as in equation (6). Furthermore,
just as in the complex case, these two fibrations are equivalent, by exactly the same
arguments as above.
In fact one has an associated pencil as in Property 4. The elements Xθ in
this pencil are real algebraic hypersurfaces in n which are smooth away from
V := f −1 (0), they fill out the whole ambient space and meet exactly at V . The + 
action leaves invariant every element of the pencil, which therefore is transverse to

all spheres around 0. And the orbits of the 1 -action are tangent to all the spheres
around 0 and permute the hypersurfaces Xθ . Therefore one has a global fibration
as in equation (3), which restricts to the Milnor fibration (4) on each sphere, and
by Property 2 this is equivalent to the fibration on a tube as in equation (6).

1.3. Milnor fibrations for weighted homogeneous real polynomial

mappings. Consider now a weighted homogeneous polynomial map germ
f := (f1 , · · · , fp ) : ( n, 0) → (p, 0) ,
with an isolated critical point at 0 and dim V > 0, where V = f −1 (0). We now
have a weighted action of  
on n of the type
t · (x1 , · · · , xn ) → (td1 x1 , · · · , tdn xn ), di ≥ 1
such that for each i = 1, · · · , p one has:
fi (t · (x1 , · · · , xn )) = td fi (x1 , · · · , xn ) .
Observe that in this setting we have a real analytic flow that clearly has Property
1 as well as the following property, analogous to Property 2:

• Property 2 . The line through the origin in p determined by a point f (x)

is the same for all x in the same + -orbit.

Regarding Property 3, now there is no 1 action, so Property 3 makes no sense.

◦
Recall that this property was used only to show that the tube N (δ) = f −1 (∂ δ )
was a fiber bundle over the boundary of the δ-disc, with projection map f . We
8 JOSÉ LUIS CISNEROS-MOLINA, JOSÉ SEADE, AND JAWAD SNOUSSI

claim (following [21, Chapter 10]) that the mere fact that f has an isolated critical
point implies that it satisfies:
• Property 3 . If we let N  
 (ε, δ) be the tube f −1 ( δ \ 0) ∩ ε , where δ is 

now a ball in p , centred at 0, of radius δ > 0, then for ε δ > 0 sufficiently small

one has that N (ε, δ) is everywhere transverse to the sphere ε . 
Then, just as in Milnor’s work (see for instance [20, p. 6]), one has a locally
trivial fibration:
f: N (ε, δ) −→ δ \ 0,
 
whose restriction to N (ε, δ) = f −1 (∂ δ ) ∩ ε is also a fibration over ∂ δ ∼
= p−1 .  
This is just a version of Ehresmann’s fibration theorem relative to the boundary of
the tube.
It is now an exercise to show that properties 1 and 2 imply that we can carry

the tube N (ε, δ) into the sphere ε and get a Milnor fibration on every sphere
centred at 0:

φ :=
f
f 
 
: ε \ (V ∩ ε ) −→ p−1 . 
Regarding Property 4, for each line L through the origin in p consider the
set:
XL = {x = (x1 , · · · , xn ) | f (x) ∈ L } .
Each of these is a real algebraic variety of dimension n − p + 1, smooth away from

V , and invariant under the -action. Hence they meet transversally every sphere
around 0. It is now an exercise to show that Properties (1), (2 ) and (3 ) imply
that one has Property (4) and there is a locally trivial fiber bundle:

Φ := f /f  : n \ V → p−1 ,
whose restriction to the boundary sphere is the Milnor fibration.
So, in this latter example, Property 4 was a consequence of the previous prop-
erties (1), (2 ) and (3 ). Yet, we will see in Section 3 that given a real analytic
 
map germ n → p with an isolated critical point and dimf −1 (0) > 0, having an
analogous “Property (4)” is a necessary and sufficient condition for having a Milnor
fibration on the spheres, and in this case this is equivalent to the fibration on a
Milnor tube, which all such maps have. “Property (4)” is precisely the d-regularity
condition mentioned in the title of this article. This will be discussed in Section 3.

2. On the geometry in Milnor’s proof of the fibration theorem

As we mentioned in the introduction, Milnor proved that given any complex
polynomial f : ( n , 0) → ( , 0), the map φ = |ff | : 2n−1
ε  
\K → 1 is a locally trivial
fibration. Instead of using a ∗
-action restricted to 1
 
and to + as in Section 1,
Milnor constructs two suitable vector fields which do the same job (essentially
properties 1 to 4).
The aim of this section is to describe the geometry behind Milnor’s original
proof of his fibration theorem for holomorphic maps. We shall describe the geo-
metric ideas without giving all the technical details for which we refer to Milnor’s
book [21, §4].
 MILNOR FIBRATIONS AND THE CONCEPT OF d-REGULARITY 9

2.1. Complex and real gradients. Let  ,  be the standard Hermitian

inner product on n given by

n
v, w = vj wj , for all v = (v1 , . . . , vn ), w = (w1 , . . . , wn ) ∈ n
.
j=1

The Hermitian vector space can also be thought of as a Euclidian vector space
of dimension 2n over the real numbers, defining the Euclidean inner product  , Ê
to be the real part of the Hermitian inner product. That is,

v, wÊ = v, w, for all v, w ∈ n

.

Let f : n
→ be a complex analytic function and define its gradient by
 ∂f ∂f 
grad f = ,..., .
∂z1 ∂zn
With this definition the chain rule for the derivative of f along a path z = p(t) has
the form
df   dp
(8) p(t) = , grad f .
dt dt
In other words, the directional derivative of f along a vector v at the point z is
equal to the inner product v, grad f .
Let f : ( n , 0) → ( , 0) be an analytic function with a critical value at 0 and
let V = f −1 (0). By the Bertini-Sard Theorem [33, Thm. (3.3)] there exists a ball
 ε of radius ε centred at 0 such that 0 ∈ is the only critical value of f restricted

to ε . Writing f as

(9) f (z) = |f (z)|eiθ(z)

we can associate to f the real analytic functions

θ: ε \ V → ,
log|f | : ε \ V → .

where θ is locally well defined.

From (9) we have that these functions are related to the function log f as
follows:
log f = log|f | + iθ.
Thus,

θ = −i log f + i log|f |,
log|f | = log f − iθ.

Since i log|f | and iθ are pure imaginary, we have that

(10) θ = (−i log f ) ,

(11) log|f | = (log f ).
10 JOSÉ LUIS CISNEROS-MOLINA, JOSÉ SEADE, AND JAWAD SNOUSSI

Differentiating (10) and (11) along a curve z = p(t) we obtain

dθ(p(t))  d(−i log f )  
= p(t)
dt dt
dp  
= , grad(−i log f p(t)
dt
dp  
(12) = , i grad log f p(t)
dt
dp  
(13) = , i grad log f p(t) ,
dt Ê
and
d log|f (p(t))|  d(log f )  
= p(t)
dt dt
dp  
(14) = , grad log f p(t)
dt
dp  
(15) = , grad log f p(t) .
dt Ê
Hence, by (8), (12) and (14) we have that
dθ(p(t))  dp     d log f (p(t)) 
(16) =  −i , grad log f p(t) = ,
dt dt dt
d log|f (p(t))| dp    d log f (p(t)) 
(17) = , grad log f p(t) =  .
dt dt dt
That is:
Proposition 2.1. The directional derivative of the function θ(z) (respectively
of log|f (z)|) in the direction dp
dt is the imaginary (respectively the real) part of the
directional derivative of the function log f in the direction dp
dt . Therefore (by (13)
and (15)) their corresponding real gradients are given by
(18) gradÊ θ = i grad log f,
(19) gradÊ log|f | = grad log f,
and these are of course normal to the respective level hypersurfaces.
For the case of the function θ(z), the level hypersurface corresponding to a
constant angle θ0 is given by
θ −1 (θ0 ) = { z ∈ε \ V | θ(z) = θ0 }
= { z ∈ ε \ V | f (z) ∈ L+
θ } 0

= f (Lθ ) ∩ ε ,
−1 +
0

where L+
is the open real ray in with angle θ0 with respect to the positive real
θ0
axis. We shall denote
Eθ0 = f −1 (L+
θ0 ) ∩ ε . 

Since f is a submersion on ε outside V we have that Eθ0 is a codimension 1
submanifold of n . Hence by (13) we have the following lemma:
Lemma 2.2. The vector i grad log f (z) is normal to Eθ(z) at the point z ∈ ε \V .
 MILNOR FIBRATIONS AND THE CONCEPT OF d-REGULARITY 11

Figure 1. The vectors z, grad log f (z) and i grad log f (z).

For the case of the function log|f (z)|, the level hypersurface corresponding to
a positive constant value log δ is given by
log|f |−1 (log δ) = { z ∈ε | log|f (z)| = log δ }
= { z ∈ ε | |f (z)| = δ }
= f −1 (∂ δ ) ∩ ε ,

where ∂ δ is the boundary of the disc of radius δ in . In other words, the level
hypersurface of log|f (z)| corresponding to the value log δ is the tube
N (ε, δ) = f −1 (∂ δ ) ∩ ε.
Again, since f is a submersion outside V we have that N (ε, δ) is a codimension 1
submanifold of n . Hence by (15) we have the following lemma:
Lemma 2.3. The vector grad log f (z) is normal to the tube N (ε, |f (z)|) at the

point z ∈ ε \ V .
Figure 1 ilustrates Lemmas 2.2 and 2.3 showing the vectors grad log f (z) and
i grad log f (z) and the fact that they are orthogonal.
Remark 2.4. Note that from the definition of the gradient we have that
grad f (z)
(20) grad log f (z) =
f¯(z)
so it is well defined everywhere, even though log f is only locally defined as a single-
valued function.
2.2. Milnor Fibration. Now that we have described the real gradients of the
real functions θ and log|f | in terms of log f the aim is to prove
Milnor’s Fibration Theorem. Let f : ( n
, 0) → ( , 0) be a holomorphic
map-germ. Then the map

φ :=
f
|f |
: ε \ K −→ 1
12 JOSÉ LUIS CISNEROS-MOLINA, JOSÉ SEADE, AND JAWAD SNOUSSI

is the projection of a smooth fibre bundle, where ε is a sufficiently small sphere


centred at 0 and K = f −1 (0) ∩ ε .
The proof consists of two main steps:
(1) To prove that φ is a submersion.
(2) To prove that φ is locally trivial.
To show that φ is a submersion, Milnor characterises its possible critical points
[21, Lemma 4.1]. In order to do this, consider the map

Φ :=
f
|f |
: ε \ V → 1 ,

where as before, ε is a ball of radius ε centred at 0 such that the restriction of f

to ε has 0 ∈ as its only critical value.
Lemma 2.5. The map Φ is a submersion.
Proof. The map Φ can be seen as the composition of the restriction of f to
ε \ V 
and the projection π : \ {0} → 1 given by π(x) = |x|
x
. The lemma follows
since both maps are submersions. 
Remark 2.6. The fibres of Φ are precisely the Eθ since given eiθ ∈ 1
 
Φ−1 (eiθ ) = f −1 π −1 (eiθ ) ∩ ε 
 
= f −1 L+θ ∩ ε 
= Eθ .
Therefore, given z ∈ Eθ we have that ker dz Φ = Tz Eθ .
Now we can give Milnor’s characterisation of the critical points of the map
 
φ := |ff | : ε \ K → 1 .

Lemma 2.7 ([21, Lemma 4.1]). The critical points of the map φ : ε \ K → 1  

are precisely those points z ∈ ε \ K for which the vector i grad log f (z) is a real
multiple of the vector z.

Proof. The map φ is the restriction of the map Φ to ε \ K. Thus, a point

z ∈ ε \ K is a critical point of φ if and only if the tangent space to ε \ K 

at z coincides with the kernel of dz Φ, i.e., Tz ( ε \ K) = ker dz Φ = Tz Eθ , since
 
ε \ K and Eθ have both codimension 1. But Tz ( ε \ K) = Tz Eθ if and only if the

normal vector to Eθ at z is a real multiple of the normal vector to ε \ K at z. By
Lemma 2.2 the normal vector to Eθ at z is i grad log f (z) and the normal vector to
ε \ K at z is z itself. 
Remark 2.8. Observe that just as in Section 1, for each line Lθ through the
origin in we may consider the set
Xθ := {z ∈ n
| f (z) ∈ Lθ } .
Then each Xθ is a real analytic hypersurface with singular set equal to the singular
set of V . Notice that Xθ \ V = Eθ ∪ Eθ+π , where the latter manifolds are those in
Lemma 2.2. The family {Xθ } of all these real hypersurfaces is the canonical pencil
of f (see Section 3). Then Lemma 2.7, together with Lemma 2.2, yields to the
following proposition.
 MILNOR FIBRATIONS AND THE CONCEPT OF d-REGULARITY 13

Proposition 2.9. The critical points of the map φ are the points in ε \ K
where the elements Xθ \ V of the canonical pencil are tangent to the sphere ε .
 
Using Lemma 2.7, to show that φ := |ff | : ε \K → 1 is a submersion for every
sufficiently small ε, we only need to prove the following lemma.
Lemma 2.10 ([21, Lemma 4.2]). There exists ε0 > 0 such that for every z ∈
n
\ V with z ≤ ε0 , the two vectors z and i grad log f (z) are linearly independent
over . 
Remark 2.11. Notice that Lemma 2.10 is equivalent to saying that for every

z ∈ n \ V with z < ε0 , the manifolds z \ V and Eθ(z) are transverse at z,
  
where z is the sphere of radius z. In other words, in ε0 all the spheres ε \ V
with ε ≤ ε0 and all the Eθ are transverse. In other words, this will show that every
complex valued holomorphic map is d-regular.
To prove Lemma 2.10 Milnor proves a slightly stronger statement when f is a
polynomial, which is the heart of the proof of his fibration theorem.
Lemma 2.12 ([21, Lemma 4.3]). Given any polynomial f which vanishes at the
origin, there exists a number ε0 > 0 so that, for all z ∈ n \ V with z ≤ ε0 , the
two vectors z and i grad log f (z) are either linearly independent over the complex
numbers or else
grad log f (z) = λz
where λ is a non-zero complex number whose argument has absolute value less than
say π4 .
It is easy to see that Lemma 2.12 implies Lemma 2.10: in the first case, if the
vectors z and i grad log f (z) are linearly independent over the complex numbers,
then they are linearly independent over the real numbers; in the second case, λ lies
in the open quadrant of the complex plane which is centred about the positive real
axis. Thus (λ) > 0 and therefore λ cannot be pure imaginary. This implies that
z and i grad log f (z) cannot be linearly dependent over . 
Idea of the proof of Lemma 2.12. The proof is a nice application of the
Curve Selection Lemma [21, Lemma 3.1] and it is done by contradiction: suppose
that there were points z ∈ n \ V arbitrarily close to the origin with
(21) grad log f (z) = λz = 0,

 with |arg
π
and  λ| > 4. In other words, assume
 that
 λ lies in the open half-plane
 (1 + i)λ < 0 or the open half-plane  (1 − i)λ < 0.
Then consider the set W of points z ∈ n for which the vectors grad f and z are
linearly dependent. We have that W is an algebraic set. Using (20) is easy to see
that z ∈ n \ V is in W if and only if equation (21) holds for some complex number
λ(z). Let U+ (respectively U− ) be the
 open set consistingof all z satisfying
 the real
polynomial inequality  (1 + i)λ (z) < 0 (respectively  (1 − i)λ (z) < 0), where
λ (z) is some real positive multiple of λ(z) defined by a real polynomial function,
and thus λ(z) and λ (z) have the same argument. The original supposition implies
that there exists points z arbitrarily close to the origin with z ∈ W ∩ (U+ ∪ U− ).
Then the Curve Selection Lemma (see [21, Lemma 3.1] for the precise statement)
asserts that there exists a real analytic curve p : [0, ε) → n with p(0) = 0 and
p(t) ∈ W ∩ (U+ ∪ U− ) for all t > 0. This proves Lemma 2.12 since the existence of
such a curve contradicts Lemma 2.13. 
14 JOSÉ LUIS CISNEROS-MOLINA, JOSÉ SEADE, AND JAWAD SNOUSSI

Lemma 2.13 ([21, Lemma 4.4]). Let p : [0, ε) →

n
 be a real analytic path with
p(0) = 0 such
 that,
 for each t > 0, the number f p(t) is non-zero and the vector
grad log f p(t) is a complex multiple λ(t)p(t). Then the argument of the complex
number λ(t) tends to zero as t → 0.
Idea of the proof. By Remark 2.4 we have that
(22) grad f (p(t)) = λ(t)p(t)f (p(t)).
Consider the Taylor expansions of p(t), f (p(t)) and grad f (p(t)) denoting their
corresponding non-zero leading coefficients by a, b and c, and their corresponding
leading exponents by α, β and γ, which are integers with α ≥ 1, β ≥ 1 and γ ≥ 0.
Substituting these Taylor expansions in (22) one can prove that λ(t) has a Taylor
expansion of the form
(23) λ(t) = λ0 tγ−α−β (1 + k1 t + k2 t2 + . . . ),
and the leading coefficients satisfy the equation
(24) c = λ0 ab̄.
Substituting (24) in the power series expansion of the identity
df dp
= , grad f ,
dt dt
and comparing leading coefficients we obtain
β = αa2 λ̄0 ,
which proves that λ0 is a positive real number. Therefore
λ(t) λ0 tγ−α−β (1 + k1 t + k2 t2 + . . . )
lim =
t→0 |λ(t)| λ0 tγ−α−β 1 + k1 t + k2 t2 + . . .
1 + k1 t + k2 t2 + . . .
= lim = 1.
t→0 1 + k1 t + k2 t2 + . . .

Hence arg λ(t) → 0 as t → 0. 

Now by Lemma 2.7 and Lemma 2.10 we conclude the first main step of the
proof of the Fibration Theorem.
Corollary 2.14 ([21, Cor. 4.5]). If ε ≤ ε0 then the map

φ=
f
|f |
: ε \ K → 1
is a submersion.
Remark 2.15. Recall that by Remark 2.6 the fibres of the map Φ are given by

the Eθ . Since φ is the restriction of Φ to ε \ K, we have that the fibres of φ are
given by
Fθ := Eθ ∩ ε . 
By Remark 2.11 this intersection is transverse and therefore Fθ is a smooth (2n−2)-
dimensional manifold.
 MILNOR FIBRATIONS AND THE CONCEPT OF d-REGULARITY 15

To prove that φ is the projection of a locally trivial fibration it is enough to


give a complete vector field w on ε \ K which projects under φ to the unit vector

field u tangent to 1 given by u(eiθ ) = ieiθ , so it is transverse to the fibres of φ. Let
p(t) be an integral curve of the vector field w. The function p(t) depends smoothly
on both t and the initial value z0 = p(0). We denote this dependence by
p(t) = ht (z0 ).
Then each ht is a diffeomorphism mapping ε \ K to itself, which sends the fibre
Fθ onto the fibre Fθ+t .

Proof of the Fibration Theorem [21, Thm. 4.8]. Given eiθ ∈ 1 let U
be a small neighbourhood of eiθ . Then the correspondence
U × Fθ → φ−1 (U )
(ei(t+θ) , z) → ht (z),
for |t| < constant, and z ∈ Fθ , is a diffeomorphism, proving the local triviality of
φ. 

Since φ is a submersion, by Corollary 2.14, we can take w to be a lifting of

 
the vector field u on 1 to ε \ K. The main difficulty is to guarantee that the

vector field w is complete, since ε \ K is non-compact we need to insure that p(t)
cannot tend towards
 K  as t tends toward some finite limit t0 . This is equivalent to
guarantee that f p(t) cannot tend to zero or that log|f (p(t))| cannot tend to −∞
as t tends to a finite value t0 . One way to do this is to make log|f (p(t))| to increase
or decrease “slowly” by keeping its derivative small in absolute value (compare with
[21, Lemma 4.7]). Suppose that
 
 d log|f (p(t))| 
  < 1,
 dt 

then we would have

   t0
d log|f (p(t))| 

 
log|f (p(t0 ))| − log|f (p(0))| =  dt
0 dt

t0  

≤  d log|f (p(t))| dt
 dt 
0
t0
< dt = t0 .
0

Then log|f (p(t))| cannot tend to −∞ as t tends to any finite limit t0 . Hence by
(14) we need the vector field w to satisfy
     
 
 w p(t) , grad log f p(t)  < 1.

As we mentioned above, the other property that the vector field w needs to

have is to project under φ to the unit tangent vector field u of 1 . In other words,
we need that the integral curves p(t) of w project under φ to the path on 1 which 
winds around the unit circle in the positive direction with unit velocity, this is the
same as
 
θ p(t) = t + constant.
16 JOSÉ LUIS CISNEROS-MOLINA, JOSÉ SEADE, AND JAWAD SNOUSSI

In turn, by (12) this is equivalent to have

dθ(p(t)) dp  
= , i grad log f p(t) = 1.
dt dt

In summary, we need a vector field w on ε \ K with the following properties:
(1) w(z),
 
z = 0 (tangent to ε ),
 
(2)  w(z), grad log f (z)  < 1 (complete),
(3)  w(z), i grad log f (z) = 1 (projects onto u).
Milnor solves the problem in a beautiful way, using Lemma 2.12 to construct a

vector field w on ε0 \ V with these properties as follows.
Lemma 2.16 ([21, Lemma 4.6]). If ε ≤ ε0 then there exists a smooth tangential
 
vector field v(z) on ε0 \ V such that v(z) is tangent to ε \ K and for each

z ∈ ε \ K, the complex inner product
v(z), i grad log f (z)
π
is non-zero, and the absolute value of its argument is less than 4.

The condition v(z), i grad log f (z) = 0 guarantees that v(z) is transverse to
the fibres of φ. In fact, suppose that v(z), i grad log f (z) = 0, in particular we have
that v(z), i grad log f (z) = 0 so v(z) would be tangent to Eθ(z) (see Lemma 2.2).

Since v(z) is also tangent to ε then we would have that v(z) is tangent to Fθ(Z) =
ε ∩ Eθ(z) , but we want v(z) to be transverse to the fibres of φ.
We shall follow Milnor’s proof literally and we shall explain the geometric
meaning of each step.
Proof. It suffices to construct such a vector field locally, in the neighbourhood
of some given point z ∈ n \ V .
Case 1: If the vectors z and grad log f (z) are linearly independent over , then
the linear equations
(25) v, z = 0,
(26) v, i grad log f (z) = 1,
have a simultaneous solution v. The first equation guarantees that v, z = 0, so

that v is tangent to ε at z.
If the vectors z and grad log f (z) are linearly independent over , then z
does not lie in the complex line generated by grad log f (z) (where it also lies
i grad log f (z)). Also it implies that z and grad log f (z) are linearly independent
 
over which by Lemma 2.3 implies that the sphere z and the tube N (ε0 , |f (z)|)

are transverse at z. The intersection Tz z ∩ Tz N (ε0 , |f (z)|) is precisely the vector
space orthogonal (with respect to the real Euclidean inner product) to the real plane

generated by z and grad log f (z). The intersection Tz z ∩ Tz N (ε0 , |f (z)|) has real

dimension 2n−2 and notice that it does not coincide with Tz Fθ(z) = Tz z ∩Tz Eθ(z)
because in that case, it would be also orthogonal to i grad log f (z) contradicting the
fact that z does not lie in the complex line generated by grad log f (z).

The idea now is to take a vector v(z) in Tz z ∩ Tz N (ε0 , |f (z)|) such that
it satisfies property (3). In this way, it satisfies property (1) by construction and
also property (2) since the directional derivative of log|f (z)| along a vector tangent
to N (ε0 , |f (z)|) is zero because N (ε0 , |f (z)|) is the level hypersurface of log|f (z)|.
 MILNOR FIBRATIONS AND THE CONCEPT OF d-REGULARITY 17

The solution to Milnor’s linear equations (25) and (26) is a vector of this kind:
(25) obviously implies property (1) and Milnor kills two birds with one stone with
equation (26) obtaining properties (2) and (3) at the same time since

(v, grad log f (z)) = (v, i grad log f (z)) = 0,

(v, i grad log f (z)) = 1.

Case 2: If grad log f (z) is equal to a multiple λz, then set v = iz. Clearly

(27) iz, z = 0;

and by Lemma 2.12 the number

(28) iz, i grad log f (z) = λ̄z2 = 0

has argument with absolute value less than π4 .


In this case Tz z ∩ Tz N (ε0 , |f (z)|) coincides with Tz Fθ(z) , so we cannot take
v there. Taking iz instead (27) obviously implies property (1), on the other hand,
iz is transverse to Fθ(z) , if it was tangent to Fθ(z) , then iz would be orthogonal
to i grad log f (z) which implies that iz is a real multiple of grad log f (z) which
is equivalent to z being a real multiple of i grad log f (z) which is impossible by
Lemma 2.10. In this case we do not get directly properties (2) and (3), but we
obtain them after a normalisation as we shall see.
In either case one can chose a local tangential vector field v(z) which takes the
constructed value v at z. The condition
π
|argv(z), i grad log f (z)| <
4
will hold throughout a neighbourhood of z. Using a partition of unity we obtain a
global vector field v(z) having the same property. 

Normalising by setting

v(z)
w(z) =
v(z), i grad log f (z)

we obtain a smooth vector field w on ε 0

\ V which satisfies
1. Property (1), since v satisfies it and this does not change multiplying by a real
scalar.
2. Property (3), since

v(z), i grad log f (z)

w(z), i grad log f (z) = = 1.
v(z), i grad log f (z)

3. Property (2), since

w(z), i grad log f (z)
π
has argument with absolute value less than 4 and its real part is equal to 1.
In this way we obtain the vector field w used in the proof of the fibration theorem.
18 JOSÉ LUIS CISNEROS-MOLINA, JOSÉ SEADE, AND JAWAD SNOUSSI

2.3. Milnor-Lê Fibration. Let 0 < δ  ε ≤ ε0 and consider the Milnor

 
tube N (ε, δ) = f −1 (∂ δ ), where δ is the disc of radius δ in with centre at 0.
Then the restriction
f : N (ε, δ) → ∂ δ 
is a locally trivial fibration. In the case when f has an isolated critical point, this
was proved by Milnor (see [20, Thm. 2],[21, Thm. 11.2]) using the fact that the

fibres in N (ε, δ) are transverse to the sphere ε (compare with Remark 1.1 and
Example 3.2) and a version of Ehresmann Fibration Theorem for manifolds with
boundary. In the general case it was proved by Lê in [18, Thm. (1,1)], using a
theorem of Hironaka [16, §5, Cor. 1] which asserts that every holomorphic function
f : n → satisfies the Thom af -property, and therefore the fibres in N (ε, δ) are

transverse to the sphere ε (compare with Proposition 3.3 in Section 3).
In order to prove that the fibration on the Milnor tube is equivalent to the one
on the sphere, we need a vector field v whose integral curves are transverse to the

tube N (ε, δ) and to the sphere ε and such that f (z) has constant argument along
each orbit. That is, we want a smooth vector field with the Properties (1) and (2)
in Section 1. Once we have such a vector field, given an integral curve p(t) of v we
have that f (p(t)) lies on the ray Lθ with θ = arg f (z). Let z be a point in N (ε, δ)
and let p(t) be the integral curve of v through z. Following p(t) we travel “away”
form the origin, transversally to the tubes and to the spheres until we reach a point

z  on ε \ K. Since f (p(t)) has constant argument along p(t) we have that
f (z) f (z  )
= .
|f (z)| |f (z  )|
Thus, the correspondence z → z  is a diffeomorphism which gives the equivalence
between the Milnor Fibration and the Milnor-Lê Fibration.
Milnor constructed such a vector field in the following lemma. Originally Milnor
used this vector field just to prove that the fibre in the tube is diffeomorphic to the
fibre in the sphere, since he did not have the fibration on the tube in the general
case.
Lemma 2.17 ([21, Lemma 5.9]). Let ε ≤ ε0 . There exists a smooth vector field

v on ε \ V so that the inner product
v(z), grad log f (z)
is real and positive, for all z ∈ ε \ V , and so that the inner product v(z), z has
positive real part.
dp
Let p(t) be an integral curve of v, that is dt = v(p(t)). By (8) we have that
d log f (p(t)) dp
=  , grad log f (p(t)).
dt dt
The condition that  dp
dt , grad log f (p(t)) = c with c real and positive, by (17) and
(16) implies that
d log|f (p(t))| dp
(29) = , grad log f (p(t)) = c > 0,
dt dt
dθ(p(t)) dp
(30) = , grad log f (p(t)) = 0.
dt dt
 MILNOR FIBRATIONS AND THE CONCEPT OF d-REGULARITY 19

From (29) we have that

log|f (p(t))| = ct + k, with k constant,
hence
|f (p(t))| = ect+k ,
whose derivative with respect to t does not vanish. Therefore the integral lines p(t)
of v are transverse to the Milnor tubes. From (30) we have that
θ(p(t)) = θ0 , for a constant value θ0 ,
in other words, f (p(t)) has constant argument as required. On the other hand, the
condition that v(z), z > 0 implies that
dp(t)2 dp
= 2 , p(t) > 0,
dt dt
therefore the integral lines p(t) are transverse to all the spheres.
Proof. The proof is analogous to the proof of Lemma 2.16. It suffices to
construct such a vector field locally, in the neighbourhood of some given point

z ∈ ε\V.
Case 1: If the vectors z and grad log f (z) are linearly independent over , then
the linear equations
v, grad log f (z) = 1,
v, z = k, with k ∈ and (c) > 0,
have a simultaneous solution v.
Case 2: If grad log f (z) is equal to a multiple λz, set v = grad log f (z). We have
that
grad log f (z), grad log f (z) = λz, λz = λz2 ∈ + , 
since z = 0 and λ = 0. On the other hand,
grad log f (z), z = λz, z = λz2 ,
since |arg λ| < π4 , we have that |λz2 | < π4 , that is, its real part is positive.
In either case one can choose a local tangential vector field v(z) which takes the
constructed value v at z. Using a partition of unity we obtain a global vector field
with the desired properties. 
Remark 2.18. The existence of the vector field constructed in Lemma 2.17 is
 
equivalent to saying that in ε0 all the spheres ε \ V with ε ≤ ε0 and all the Eθ
are transverse (compare with Remark 2.11).

3. d-regularity for real analytic map-germs

We now consider a real analytic map-germ
f: n → p
x → (f1 (x), . . . , fp (x))
−1
with n > p and set V = f (0).
First we want to know under what conditions one can have a “Milnor-Lê”
fibration:
(31) f : N (ε, δ) → ∂ δ ,
20 JOSÉ LUIS CISNEROS-MOLINA, JOSÉ SEADE, AND JAWAD SNOUSSI

 
where 0 < δ  ε ≤ ε0 and N (ε, δ) = f −1 (∂ δ ) ∩ ε is the so-called Milnor tube,
 δ being the ball of radius δ in p
 
and ε the ball of radius ε in n , both with
centre at the origin. We have just seen that this is always the case when f is
holomorphic.
f

For simplicity we require all along this section that the map N (ε, δ) → ∂ δ , is
surjective. This means that f has to be surjective over some neighborhood of the

origin in p . In this case we will call it locally surjective.

Milnor proved in [21] that every real analytic map f : n → p with n > p 
and with an isolated critical point induces a fibration on tubes
f : N (ε, δ) = ε ∩ f −1(∂ δ ) → ∂ δ .
In order to do that, he used an Ehresmann fibration theorem for manifolds with
boundary, which in turn requires the fibers of the induced map to be transverse to
the sphere. This last condition is always fulfilled when f has an isolated critical
point, and this is essentially Example 3.2 and Proposition 3.3.
D.T. Lê proved in [18] that every holomorphic function f : X → induces
a topological locally trivial fibration on tubes N (ε, δ). In order to apply Milnor’s
argument he first pointed out that any holomorphic function satisfies Thom’s af -
property ([16]) and hence all the fibers in the tube are transverse to the sphere.
So it became clear that if one needs to prove a fibration theorem in the tube it is
sufficient to prove transversality of the fibers with the spheres. Let us explain the
relation between this transversality and Thom’s condition.
 
Let f : ( n , 0) → ( p , 0) , n > p be a real analytic map with an isolated
critical value at 0. Endow V = f −1 (0) with a stratification V = Vα such that in
α
a neighborhood U of the origin the partition

U = (U \ V ) ∪ Vα
α
is a Whitney regular stratification of U .
Definition 3.1. The map f has the Thom af -property if there exists a stratifi-
cation as above, satisfying that for every sequence of points xn ∈ U \ V converging
to a point x in some Vα ⊂ V , such that the sequence of spaces Txn f −1 (f (xn )),
tangent to the fibers of f , converges to some linear space T , one has Tx Vα ⊂ T .
That is, the space T contains the tangent space to Vα at x .
 
Example 3.2. Every smooth map-germ ( n , 0) → ( p , 0), n > p, with an
isolated critical point at 0 has the Thom af -property. This is an immediate conse-
quence of the Implicit Function Theorem.
Assume now that the map-germ f has Thom’s af -property. Assume also that
0 is a stratum and that the neighbourhood U is sufficiently small so that every
other stratum contains 0 in its closure. We know that for sufficiently small ε, every
stratum Vα , other than 0 itself, intersects transversally the sphere n−1  . Hence,

ε
by Thom’s af -property, for each point x ∈ Vα ∩ n−1 there exists a neighborhood

ε
Ux such that for every y ∈ Ux ∩ n−1 the fiber f −1 (f (y)) intersects transversally

ε
the sphere n−1 . Since the sphere is compact and the map f is continuous, there

ε
exists 0 < δ  ε such that for every ξ ∈ p such that ξ = δ the fiber f −1 (ξ)

is transversal to the sphere n−1ε at each point of intersection. In other words,
sufficiently small spheres are transverse to fibers sufficiently close to f −1 (0). When
 MILNOR FIBRATIONS AND THE CONCEPT OF d-REGULARITY 21

f −1 (0) is reduced to the origin, this intersection is vacuous and hence transversal.
Thus we have the following proposition whose complete proof can be read in [12,
Proposition 5.1]:
 
Proposition 3.3. Let f : n → p be a real analytic map with isolated
critical value satisfying Thom property at 0. There exists ε0 > 0 such that for any

0 < ε ≤ ε0 there exists 0 < δ  ε and for any ξ ∈ p with ξ = δ the fiber f −1 (ξ)
is transverse to the sphere n−1
ε 
.
 
Now we observe that N (ε, δ) = f −1 (∂ δ ) ∩ ε is a compact manifold with
boundary, which allows us to apply the Ehresmann fibration theorem. Of course
the same remarks apply if we consider the solid tube N 
 (ε, δ) = f −1 ( δ \ {0}) ∩ ε , 

just as in Property 3 in Section 1. In fact, around each point x ∈ δ \ {0} we 
  
can chose a closed ball x ⊂ δ \ {0}. The restriction of f to ε ∩ f −1 ( x ) is 
a proper map whose fibers are transverse to the boundary sphere, so it induces a
locally trivial fibration.
So we have the following theorem, which is proved by an easy extension of
Milnor’s arguments in the last chapter of his book (see [27, Theorem 1.3], also [12,
Remark 5.7]).
 
Theorem 3.4. Let f : ( n , 0) → ( p , 0)) be an analytic map-germ with an
isolated critical value at 0. Assume further that n > p, and for every ε > 0
sufficiently small one has that there exists δ > 0 such that every fiber f −1 (t) with

t ≤ δ meets transversally the sphere ε . Then one has locally trivial fibrations:
f :N 
 (ε, δ) → δ \ {0} and f : N (ε, δ) → ∂ δ ∼= p−1 .  
Now, what about fibrations from the sphere minus the link into the sphere?
In fact it is easy to adapt Milnor’s methods to show that the above tube N (ε, δ)

can always be inflated to the sphere ε using appropriate vector fields. This is
similar to what we did in Section 1 using properties (1)-(3). One gets a locally
trivial fibration:

φ̄ : n−1
ε \ V → p−1 , 

which restricted to a neighbourhood of the link K = V ∩ n−1 is given by the
ε ∩ f −1 (δ
◦ ◦
f
projection f  (this is Theorem 1.3 in [27]). If we write T (ε, δ) =
◦
\{0}), then by Theorem 3.4, the restriction of f to T (ε, δ) is a locally trivial
fibration. If we compose it with the radial projection we get a locally trivial fibration

(32) φT =
f ◦
f 
:T (ε, δ) → p−1 . 
We are interested in finding conditions ensuring that the above map φ̄ can be
taken as ff  globally. That is, we want to know when

φ=
f
f 
: n−1
ε \ V → p−1 
is a locally trivial fibration, equivalent to the fibration in Theorem 3.4. Here is
where we come to d-regularity.
We use again an idea from J. Milnor in [20, 21]. Consider the map f : N (ε, δ) →
 
∂ δ and compose it with the projection to the unit sphere in p to obtain the map:

φN =
f
f 
: N (ε, δ) → p−1 . 
22 JOSÉ LUIS CISNEROS-MOLINA, JOSÉ SEADE, AND JAWAD SNOUSSI

If we have a fibration in the tube then this last map φN is also a fibration. We
want to transform its fibers into those of φ. In order to do that we will need a flow
that moves a point in the tube diffeomophically into a point in the sphere preserving
the value ff  . This flow can be given by a vector field transverse to the tubes and
the spheres and tangent to the fibers of φ̂ = f
f  : ε \ V → p−1 . In particular the
spheres need to be transverse to the fibers of φ̂. This is the d-regularity condition.
Let us explain it carefully.
Consider the two-fold covering of p−1
given by
π:  p−1
→ p−1 .
We have the following commutative diagramm:
φ̂= ff 
ε \ VJJ / p−1
JJJ
JJ
ψ JJ$
π

p−1
where the map ψ is given by the homogeneous coordinates (f1 : . . . : fp ).
Let l ∈  p−1
and call θ − and θ + the corresponding antipodal points in the
sphere  p−1
. The fiber ψ −1 (l) is the union of the fibers φ−1 (θ − ) and φ−1 (θ + ). If we

call L the line in p corresponding to the projective point l, then the fiber ψ −1 (l)

is the intersection (f −1 (L) \ V ) ∩ ε .
 
Definition 3.5. Let f = (f1 , . . . , fp ) : U ⊂ n → p be a locally surjective
real analytic map defined in a neighborhood of the origin and L ⊂ p be a line 
containing the origin. We define XL to be the inverse image of the line L by the
map f . The family of the spaces XL when L varies in p−1

is called the canonical
pencil associated to f .
This family of real analytic spaces has been introduced in [31, 29, 28], and
later in [11, 12, 5]. By abuse of notation, we will make no difference between a

line L containing the origin in p and the projective point corresponding to it in
 p−1
.
Note that for any L and L in 
p−1
the corresponding pre-images satisfy
X L ∩ X L = V
and the whole pencil covers the neighborhood where the map f is defined, i.e.,

U= XL .
L∈ ÊÈ p−1

From now on we will suppose that the map f has an isolated critical value at
0 and is locally surjective. Then each XL has dimension n − p + 1. In this case XL
is non-singular outside V .

Each line L intersects the sphere p−1 in two antipodal points θ − and θ + .
We can then decompose the line L into two open half lines L− and L+ containing
respectively the points θ − and θ + , so that we can express
L = L− ∪ {0} ∪ L+ .
If we define Eθ− , respectively Eθ+ , to be the inverse image f −1 (L− ), respec-
tively f − (L+ ), then we can express each element of the canonical pencil as the
 MILNOR FIBRATIONS AND THE CONCEPT OF d-REGULARITY 23

following union:
(33) XL = Eθ − ∪ V ∪ Eθ + .
Using this notation we can describe the fibers of the map φ : n−1
ε → p−1 as
follows:
φ−1 (θ − ) = Eθ− ∩ n−1
ε 
and
φ−1 (θ + ) = Eθ+ ∩ n−1
ε
If we denote by K the link of f in n−1
ε , then we can write:
XL ∩ n−1
ε = (Eθ − ∩ ε ) ∪ K ∪ (Eθ ∩ n−1
n−1
+
ε ) = φ−1 (θ − ) ∪ K ∪ φ−1 (θ + ).
We can now give the precise definition of d-regularity.

Definition 3.6. Let f : U → p be a locally surjective real analytic map with
an isolated critical value at 0. We say that f is d-regular if there exists ε0 > 0 such

that for every ε ≤ ε0 and for every line L through the origin in p , the sphere n−1 ε 
and the manifold XL \ V are transverse.
Remark 3.7. Recalling that each XL can be decomposed as in equation (33)
above, the d-regularity condition states that the Eθ ’s are transverse to the spheres.
Since the family of the Eθ ’s depends on (p − 1)-parameters, the existence of such an
ε0 is not guaranteed when p > 1 (compare with the second example in 3.9 below).
On the other hand, it is well known that we have transversality (in the stratified
sense) of V with all sufficiently small spheres [21, 8, 33]. Hence if f is d-regular,
then the whole family XL is transverse to the spheres. One can show further that,
since the domain of f is smooth at 0, the transversality of the XL with the spheres
implies that the pencil has a local uniform conical structure as in [11].
Now, how does one verify whether or not a map is d-regular?
One can use the following procedure, which is implicit in [5, p. 179]: Consider
the (((p(p − 1)/2) + 1) × n)-matrix M whose first p(p − 1)/2 rows are given by
fi grad fj − fj grad fi
for 1 ≤ i < j ≤ n, and the last row is the vector
(X1 , . . . Xn ).
Proposition 3.8. The map f = (f1 , . . . , fp ) is d-regular if and only if there

exists no arbitrary close point to the origin in n outside V where the matrix M
has rank strictly less than p.
Proof. Let x ∈ U be a point such that f (x) = 0. The element of the canonical
pencil XL that contains x is the inverse image by f of the line generated by the
vector (f1 (x), . . . , fp (x)). So the space XL is defined by
XL = {y ∈ n \ V, (f1(x) : · · · : fp (x)) = (f1(y) : · · · : fp (y))}.
This space is then defined by the equations
fi (x)fj − fj (x)fi = 0, 1 ≤ i < j ≤ p.
The tangent space to XL at x is the kernel of the matrix given by the rows
fi (x) gradx fj − fj (x) gradx fi , 1 ≤ i < p.
24 JOSÉ LUIS CISNEROS-MOLINA, JOSÉ SEADE, AND JAWAD SNOUSSI

The space XL and the sphere n−1 

x are not transverse if and only if the dimen-
sion of the intersection of their corresponding tangent spaces is bigger than n − p.
This is equivalent to saying that the matrix M has rank strictly smaller than p. 
Examples 3.9. Consider the real analytic map:

f: 3 → 2
(x, y, z) → (xz + y 3 , x)
One can check that:
V := f −1 (0) = {(0, 0, z), z ∈ }
and V coincides with the critical locus of f . So it has an isolated critical value at

the origin of 2 .
Now we apply Proposition 3.8. So the points where the XL ’s are not transverse
to the corresponding spheres are the points where the matrix
 
−y 3 3xy 2 x2
x y z
has rank strictly less than 2. This is achieved only on V . So this map is d-regular.
Consider now the slightly different map

g: 3 → 2
(x, y, z) → (x2 z + y 3 , x)
this map still has isolated critical value. However, all along the line {(x, 0, x), x ∈
}, the spaces XL are not transverse to the corresponding spheres, so it is not
d-regular.
Remark 3.10.  
• When f : 2n → 2 is given by the real and imagi-
nary part of a holomorphic map, then it has an isolated critical value. By
Remark 2.11 every such map is d-regular.
• From the examples in Section 1 we see that polar weighted homogeneous
polynomials and real weighted homogeneous maps with an isolated critical
value, are d-regular maps (see [10, 22]).
• In [27] it is proved that when f and g are holomorphic maps 2 → such
that the product f ḡ has an isolated critical value at the origin, then the
map f ḡ is d-regular. The same statement holds when f, g are defined in
n > 2 complex variables provided the meromorphic map f /g is semi-tame
(see [7]).

• In [28] it is proved that every real analytic map n+2 → 2 with an
isolated critical point such that the family of hypersurfaces Xt (defined
as above) satisfy that for any t the strata (Xt \ {0}, {0}) is (c)-regular
with respect to the function distance to the origin, is also d-regular. Pre-
sumably the same statement holds for real analytic maps with an isolated
critical value.
• The strongly non-degenerate mixed functions in [23] are all d-regular.
Notice that the projection map of a locally trivial (differentiable) fibration is
a fortiori a submersion. Hence, if we expect φ to be a fibration it needs to be
 MILNOR FIBRATIONS AND THE CONCEPT OF d-REGULARITY 25

a submersion. The following statement is proved in [12, Proposition 3.2]. An

equivalent result is proved in [5] for maps with an isolated critical point.
Proposition 3.11. The map f is d-regular if and only if there exists ε0 such
that for any ε ≤ ε0 the map φ induced on n−1
ε 
is a submersion.
Hence, given a locally surjective real analytic map f with an isolated critical
value and with the Thom af -property, d-regularity is necessary for the associated
map φ = ff  : n−1
ε  
→ p−1 to be a locally trivial fibration. We now explain why
and how d-regularity is also a sufficient condition for φ to be the projection map of
a locally trivial fibration, [12, Theorem 5.3]. The method we expose here has the
advantage that we get for free that whenever the two types of fibrations exist, one
on a Milnor tube, the other on a small sphere minus the link, these are equivalent.
For this we use the following:
Definition 3.12. Consider the map
F: U \ V → p
f (x)
x → x .
f (x)
We will call it the spherification map associated to f .
This map has the peculiarity that its Milnor tubes are the spheres (minus
 
the link), for it carries spheres in n into spheres in p (preserving the radius).
Also, each of its fibers is contained in a single element of the canonical pencil of
f . In [12, Proposition 3.2] we proved that the map f is d-regular if and only if
the spherification map is a submersion in a sufficiently small neighborhood of the
origin.
This will allow us to lift correctly some particular vector fields.
 
From now on we assume f : n → p has an isolated critical value, it is locally
surjective, d-regular, and with the Thom af -property.

Let v be the radial vector field in p , v(y1 , · · · , yp ) = (y1 , · · · , yp ). Since f and
F have no critical point outside V , we can lift v to obtain two vector fields vf and
vF using respectively f and F.
The vector vf is transverse to all sufficiently small tubes N (ε, δ) and tangent
to all XL ’s. Meanwhile the vector field vF is transverse to all spheres and tangent
to each XL . One can easily prove that sufficiently near the origin these vector fields
cannot point in opposite directions. So their sum w = vf + vF is a never vanishing
smooth vector field, which is tangent to the XL ’s and transverse to the spheres and

to f −1 (∂ δ ). The details of this proof are done in [12], Proposition 5.2. One has:
 
Lemma 3.13. Let an analytic map-germ f : ( n , 0) → ( p , 0) have an isolated
critical value at 0. The map f is d-regular if and only if there exists a smooth vector

field w on ε \ V which has the following properties:

(i) It is radial, i.e., it is transverse to all spheres in ε centred at 0.
(ii) It is tangent to each XL \ V .

(iii) It is transverse to all the tubes f −1 (∂ δ ).
The flow generated by the vector field w allows us to deform the fibration in
◦
the tube into the desired one in the sphere. In fact, if as before we call T (ε, δ) to
n−1 ∩ f −1 (δ \{0}). The space T (ε, δ) should be thought of as being the “cap”
◦ ◦
ε
26 JOSÉ LUIS CISNEROS-MOLINA, JOSÉ SEADE, AND JAWAD SNOUSSI

   
◦ ◦ ◦
of the tube f −1 ( δ \{0}) ∩ ε , since T (ε, δ) is the set where f −1 ( δ \{0}) ∩ ε
meets the boundary sphere e . 
Any point in N (ε, δ) can be moved following the flow generated by the vector
n−1
◦
field w until it reaches (in a unique way) a point in ε \ T (ε, δ). We obtain then
a fibration:
n−1 \ T (ε, δ) → ∂ δ ∼
= p−1 .
◦
φ1 : ε
Since the flow is tangent to the XL ’s, any point moves along the same XL . So
the map φ1 coincides with f /f .
On the other hand we have already seen in Equation (32), that we have a
fibration on the complement

p−1 .
◦
φT : T (ε, δ) →
These two fibrations glue together giving the desired one
φ: n−1
ε \ K → p−1 .
Let us summarize the previous discussion as follows. We use the same notation.
 
We consider an analytic map-germ f : ( n , 0) → ( p , 0) with an isolated critical
value at 0.
• If f has the Thom af -property then given ε > 0 sufficiently small, there exists
δ > 0 such that all fibers f −1 (t) with t ≤ δ intersect transversally the sphere ε . 
• If given ε > 0 sufficiently small, there exists δ > 0 such that all fibers

f −1 (t) with t ≤ δ intersect transversally the sphere ε , then one has Milnor-Lê
fibrations:
 (ε, δ) →
f :N δ \ {0} ,  (ε, δ) = f −1 (
where N δ \ {0}) ∩ ε ,
and
f : N (ε, δ) → ∂ δ ∼= p−1 ,  (ε, δ) = f −1 (∂
where N δ ) ∩ ε .
• If f is d-regular, then there exists a smooth vector field w as in Lemma
3.13. The integral lines of this vector field are transverse to all spheres around 0,
 
transverse to all Milnor tubes f −1 (∂ η ) ∩ ε , and tangent to each element XL of
the canonical pencil.
• If f is d-regular and it has the above fibration f : N (ε, δ) → δ \ {0}, then 
the vector field w allows us to identify its boundary N (ε, δ) with the complement
◦
of the cap T (ε, δ) in the sphere. One thus gets that f has a Milnor fibration

φ=
f
f 
: n−1
ε \V → p−1 ,
and this fibration is equivalent to the one on the tube N (ε, δ).
We remark that the above construction can also be used conversely: Given f
with isolated critical value and d-regular, we have a vector field w as in Lemma
3.13. If we know that f has a Milnor fibration n−1 ε 
\ V −→ p−1 , and f has a
f /f 

Milnor-Lê fibration on the tube N (ε, δ), then we can use the integral lines of w to
identify the fibers of φ on the sphere, with those of f on the tube. Thus we arrive
to the following theorem, which is implicit in [12] and completes previous results
by various authors in [29, 3, 4, 27, 11, 10], including the classical holomorphic
case by Milnor, which is the paradigm of this whole discussion.
 MILNOR FIBRATIONS AND THE CONCEPT OF d-REGULARITY 27


Theorem 3.14. Let an analytic map-germ f : ( n , 0) → ( p , 0) have an 
isolated critical value at 0. If the two fibrations exist (one on a Milnor tube, another
on the sphere minus the link), then these fibrations are smoothly equivalent. That
is, there exists a diffeomorphism between their corresponding total spaces, carrying
fibers into fibers.

References
[1] N. A’Campo. Le nombre de Lefschetz d’une monodromie. Nederl. Akad. Wetensch. Proc.
Ser. A 76, Indag. Math., 35:113–118, 1973. MR0320364 (47:8903)
[2] N. A’Campo. Monodromy of real isolated singularities. Topology, 42(6):1229–1240, 2003.
MR1981355 (2004c:14011)
[3] R. N. Araujo dos Santos. Uniform (m)-condition and strong Milnor fibrations. In Singularities
II, volume 475 of Contemp. Math., pages 189–198. Amer. Math. Soc., Providence, RI, 2008.
MR2454367 (2010c:58046)
[4] R. N. Araujo dos Santos. Equivalence of real Milnor’s fibrations for quasi homogeneous sin-
gularities. Singularities II, volume 475 of Contemp. Math., pp. 189–198. Amer. Math. Soc.,
Providence, RI, 2008. MR2454367 (2010c:58046)
[5] R. N. Araujo dos Santos and M. Tibăr. Real map germs and higher open book structures.
Geom. Dedicata, 147, 177-185 (2010). MR2660575 (2011g:58062)
[6] K. Bekka. C-régularité et trivialité topologique In Singularity theory and its applications,
Part I (Coventry, 1988/1989), Lecture Notes in Math. (1462), 42–62, 1991. MR1129023
(92h:58012)
[7] A. Bodin and A. Pichon. Meromorphic functions, bifurcation sets and fibred links. Math.
Res. Lett., 14(3):413–422, 2007. MR2318645 (2008f:32038)
[8] D. Burghelea and A. Verona. Local homological properties of analytic sets. Manuscripta
Math., 7:55–66, 1972. MR0310285 (46:9386)
[9] P. T. Church and K. Lamotke. Non-trivial polynomial isolated singularities. Nederl. Akad.
Wetensch. Proc. Ser. A 78=Indag. Math., 37:149–154, 1975. MR0365592 (51:1844)
[10] J. L. Cisneros-Molina. Join theorem for polar weighted homogeneous singularities. In Singu-
larities II, volume 475 of Contemp. Math., pages 43–59. Amer. Math. Soc., Providence, RI,
2008. MR2454370 (2010e:32030)
[11] J. L. Cisneros-Molina, J. Seade, and J. Snoussi. Refinements of Milnor’s fibration theorem
for complex singularities. Adv. Math., 222(3):937–970, 2009. MR2553374 (2010k:32042)
[12] J. L. Cisneros-Molina, J. Seade, and J. Snoussi. Milnor Fibrations and d-regularity for real
analytic Singularities. International Journal of Mathematics, vol. 21, No. 4, 419-434 (2010).
MR2647448 (2011e:58063)
[13] J. L. Cisneros-Molina, J. Seade, and J. Snoussi. Milnor Fibrations for Real and Complex
Singularities In Topology of Algebraic Varieties and Singularities, volume 538 of Contemp.
Math., pages 345–362. Amer. Math. Soc., Providence, RI, 2011. MR2768672
[14] A. H. Durfee. Neighborhoods of algebraic sets. Trans. Amer. Math. Soc., 276(2):517–530,
1983. MR688959 (84m:32013)
[15] M. Hirasawa and L. Rudolph. Constructions of Morse maps for knots and links, and up-
per bounds on the Morse-Novikov number. Preprint 2003, math.GT/0311134. To appear in
Journal of Knot Theory and its ramifications.
[16] H. Hironaka. Stratification and flatness. In Real and complex singularities (Proc. Ninth Nordic
Summer School/NAVF Sympos. Math., Oslo, 1976), pages 199–265. Sijthoff and Noordhoff,
Alphen aan den Rijn, 1977. MR0499286 (58:17187)
[17] A. Jacquemard. Fibrations de Milnor pour des applications réelles. Boll. Un. Mat. Ital. B
(7), 3(3):591–600, 1989. MR1010525 (91a:55019)
[18] D. T. Lê. Some remarks on relative monodromy. In Real and complex singularities (Proc.
Ninth Nordic Summer School/NAVF Sympos. Math., Oslo, 1976), pages 397–403. Sijthoff
and Noordhoff, Alphen aan den Rijn, 1977. MR0476739 (57:16296)
[19] E. Looijenga. A note on polynomial isolated singularities. Nederl. Akad. Wetensch. Proc.
Ser. A 74=Indag. Math., 33:418–421, 1971. MR0303557 (46:2694)
[20] J. Milnor. On isolated singularities of hypersurfaces. Preprint, June 1966.
28 JOSÉ LUIS CISNEROS-MOLINA, JOSÉ SEADE, AND JAWAD SNOUSSI

[21] J. Milnor. Singular points of complex hypersurfaces. Annals of Mathematics Studies, No. 61.
Princeton University Press, Princeton, N.J., 1968. MR0239612 (39:969)
[22] M. Oka. Topology of polar weighted homogeneous hypersurfaces. Kodai Math. J., 31(2):163–
182, 2008. MR2435890 (2009g:32061)
[23] M. Oka. Non-degenerate mixed functions. Kodai Math J., 33(1):1–62, 2010. MR2732230
(2012a:32032)
[24] M. Oka. On Mixed Brieskorn Variety. Preprint, arXiv:0909.4605, 2009.
[25] B. Perron. Le nœud “huit” est algébrique réel. Invent. Math., 65(3):441–451, 1982. MR643562
(84d:57005)
[26] A. Pichon and J. Seade. Real singularities and open-book decompositions of the 3-sphere.
Ann. Fac. Sci. Toulouse Math. (6), 12(2):245–265, 2003. MR2123256 (2005m:57031)
[27] A. Pichon and J. Seade. Fibred multilinks and singularities f g. Math. Ann., 342(3):487–514,
2008. MR2430986 (2009g:32062)
[28] M. A. S. Ruas and R. N. Araujo dos Santos. Real Milnor fibrations and (c)-regularity.
Manuscripta Math., 117(2):207–218, 2005. MR2150482 (2006g:32047)
[29] M. A. S. Ruas, J. Seade, and A. Verjovsky. On real singularities with a Milnor fibration. In
Trends in singularities, Trends Math., pages 191–213. Birkhäuser, Basel, 2002. MR1900787
(2003c:32031)
[30] L. Rudolph. Isolated critical points of mappings from R4 to R2 and a natural splitting of the
Milnor number of a classical fibered link. I. Basic theory; examples. Comment. Math. Helv.,
62(4):630–645, 1987. MR920062 (88k:57009)
[31] J. Seade. Open book decompositions associated to holomorphic vector fields. Bol. Soc. Mat.
Mexicana (3), 3(2):323–335, 1997. MR1679312 (2000c:32085)
[32] J. Seade. On the topology of isolated singularities in analytic spaces, volume 241 of Progress
in Mathematics. Birkhäuser Verlag, Basel, 2006. MR2186327 (2007a:32031)
[33] J.-L. Verdier. Stratifications de Whitney et théorème de Bertini-Sard. Invent. Math., 36:295–
312, 1976. MR0481096 (58:1242)

Instituto de Matemáticas, Unidad Cuernavaca, Universidad Nacional Autónoma de

México, Avenida Universidad s/n, Colonia Lomas de Chamilpa, Cuernavaca, Morelos,
México.
E-mail address: [email protected]

Instituto de Matemáticas, Unidad Cuernavaca, Universidad Nacional Autónoma de

México, Avenida Universidad s/n, Colonia Lomas de Chamilpa, Cuernavaca, Morelos,
México.
E-mail address: [email protected]

Instituto de Matemáticas, Unidad Cuernavaca, Universidad Nacional Autónoma de

México, Avenida Universidad s/n, Colonia Lomas de Chamilpa, Cuernavaca, Morelos,
México
E-mail address: [email protected]
Contemporary Mathematics
Volume 569, 2012
http://dx.doi.org/10.1090/conm/569/11248

Bi-Lipschitz G-triviality and Newton polyhedra, G = R, C, K,

R V , CV , K V

J. C. F. Costa, M. J. Saia, and C. H. Soares Júnior

Abstract. In this work we provide estimates for the bi-Lipschitz G-triviality,
G = C or K, for a family of map germs satisfying a Lojasiewicz condition. We
work with two cases: the class of weighted homogeneous map germs and the
class of non-degenerate map germs with respect to some Newton polyhedron.
We also consider the bi-Lipschitz triviality for families of map germs defined
on an analytic variety V . We give estimates for the bi-Lipschitz GV -triviality
where G = R, C or K in the weighted homogeneous case. Here we assume that
the map germ and the analytic variety are both weighted homogeneous with
respect to the same weights. The method applied in this paper is based in the
construction of controlled vector fields in the presence of a suitable Lojasiewicz
condition. In the last section of this work we compare our results with other
results related to this work showing tables with all estimates that we know,
including ours.

1. Introduction
A basic problem in Singularity Theory is the local classification of mappings
up to diffeomorphisms. However, this problem is too rigid. Therefore, it seems
natural to investigate classification of mappings by morphisms weaker than dif-
feomorphisms. In this paper we are interested in the study of mappings up to
bi-Lipschitz maps.
A mapping φ : U ⊂ R → Rs is called Lipschitz if there exists a constant c > 0
such that
 φ(x) − φ(y)  ≤ c  x − y , ∀x, y ∈ U.
When  = s and φ has a Lipschitz inverse, we say that φ is bi-Lipschitz.
Equivalence relations defined in terms of bi-Lipschitz maps are important tools
in the study of equisingularity of mappings and sets from the metric point of view.
Metric classification refers to classification up to bi-Lipschitz maps.
We can define Lipschitz versions of the equivalence relations involving the clas-
sical Mather’s groups R, C and K as follows:
i) Two map germs f, g : (Rn , 0) → (Rp , 0) are bi-Lipschitz R-equivalent if there
exists a germ of a bi-Lipschitz homeomorphism h : (Rn , 0) → (Rn , 0) such that
g = f ◦ h−1 .

2010 Mathematics Subject Classification. Primary 58K40,58K60.

Key words and phrases. bi-Lipschitz determinacy, Newton filtration, controlled vector fields.
This work is partially supported by CAPES, CNPq, FAPESP and FUNCAP.

2012
c American Mathematical Society

29
30 J.C.F. COSTA, M. J. SAIA, AND C. H. SOARES JÚNIOR

ii) Two map germs f, g : (Rn , 0) → (Rp , 0) are bi-Lipschitz K-equivalent if there
exists a pair of bi-Lipschitz homeomorphisms (h, H), with h : (Rn , 0) → (Rn , 0),
H = (H1 , H2 ) : (Rn × Rp , 0) → (Rn × Rp , 0) satisfying the conditions: H1 = h,
H(Rn × {0}) = Rn × {0} and H ◦ (Id, f ) = (Id, g) ◦ h, where Id denotes the germ
of the identity in Rn . When h = Id we call the bi-Lipschitz K-equivalence by
bi-Lipschitz C-equivalence.
In this article we consider the bi-Lipschitz G-triviality along of a family of map
germs ft with f0 = f satisfying a Lojasiewicz condition.
The bi-Lipschitz R-triviality of families of function germs was studied by Fer-
nandes and Ruas in [10] for the weighted homogeneous case and by Fernandes and
Soares Jr. in [11] for the Newton non-degenerate case. However the study of bi-
Lipschitz triviality with respect to the groups C and K has a gap in the literature,
which is the main subject of this article.
With respect to the class of weighted homogeneous map germs, our main results
are Theorems 2.3 and 2.6 in which we obtain sufficient conditions for the bi-Lipschitz
K and bi-Lipschitz C-triviality of families, respectively. These results are given in
terms of the weights and degrees of weighted homogeneity. For the class of non-
degenerate map germs with respect to some Newton polyhedron, the main results
are Theorems 2.14 and 2.15 in which we also obtain sufficient conditions for triviality
of families, similar to the weighted homogeneous case. We also consider the bi-
Lipschitz GV -triviality for families of map germs defined on an analytic variety V ,
where G = R, C or K. In this case our main results are Theorems 4.1, 4.3 and 4.4,
respectively.
It is worth mentioning here the difference between the classification questions
considered in this paper. For example, the bi-Lipschitz R-equivalence admits mod-
uli (cf. [12]) while the bi-Lipschitz K-equivalence does not have moduli, this is a
recent result due to Valette and Ruas [18]. The results presented here and in the
work of Ruas and Valette on the finiteness theorems can be useful for the the study
of Lipschitz classification of map germs.
The method applied to obtain our results is an application of a Thom-Levine
type theorem with the construction of controlled vector fields in the presence of
a suitable Lojasiewicz condition. Consider a deformation ft (x) of f , then the bi-
Lipschitz G-triviality of ft is proven by solving localized equations of type −ρ ∂f∂t =
t

ξ(f ) where ρ is a control function and ξ is a germ of smooth vector field in the G-
tangent space of f . If the deformation ft satisfies an appropriate filtration condition,
we obtain the bi-Lipschitz property of the vector field ρ−1 ξ.
This method was used by several authors to obtain estimates for the C -G-
triviality of map germs, 0 ≤  < ∞. For instance, the works of Damon [9, 8] and
Kuo [13], among others, treated the C 0 -G-triviality of map germs.
More recently Abderrahmane [1] studies the C 0 -R-triviality for function germs
which satisfy a Newton non-degeneracy condition. Estimates for the C -R-triviality
in families of real functions germs of class at least C +1 appear in the work of
Bromberg and Lopes de Medrano [3], for the weighted homogeneous case. The C -
G-triviality, with 0 ≤  < ∞ and G = R, C or K appear in [15] for homogeneous
map germs, in [16] for weighted homogeneous map germs and in [19] for map germs
which satisfy a Newton non-degeneracy condition.
Concerning the GV -triviality for families of map germs defined on an analytic
variety V , we see estimates for the C -GV -triviality with  > 1 and G = R, C or K
and

most

always horns

Numbers

and
are

arboreal and the

carnivorous consideration

is hungry it

The

other

see DOGS

kinds on

These so eats
this

or a however

strength

the to

concerned The

red extremely

by

suckled

is OUTHERN of
badgers

are BOOK suppose

believe jet

through

is of

case

eyes species

felt

amusing

well by
groves not

his Capuchins I

Volume

more

fortunate not bearing

feet

is African
Ewart

sharp eat from

steppe thus look

are

colour

of

the the

fully
in

is

instance had

it

and are

sugar mud P

and

wounded throw
loss does produced

in carries

struggle of RAT

burrow ancient large

of
BY in LACK

regularly they

Dr Zebras

never was

of sleep

ruin majestic

naturally very mules

most could who

home

refuge variation

boats

are

Hope vessels

domestication

Gardens

a Parson W
in

are Barbary here

Malacca colour Fox

Hon rush smaller

the white might

and caused vicious

all special
entirely Barrier

remarkable from Rhodope

suspending of

here sometimes

are

wolf

when

would

companions forced and

the
Old

three

this

is vessels

has Far in

a is

in betake
coolie came F

289 cat Palace

are 23

markings to the

RINCE bone

THER horse

line bull

Baker has

Cape me

like
passing

fact

ATS it yet

This Lake

a attention the

shoulders hunts From

induced was bacon

outside would and

of northern
a

far insects its

the half

fur as

Northern in remarkable

animals charge

the

is all

this through
trained

were

slow a

varieties

and

in E

it size

caused
on

him always

travellers done hand

which unique

CATTLE

with same

flesh

the

been
his the right

This Brown

our has

eyes the tortoiseshell

summer

Maltese

Indian comes

white into of

of bounds
their snap

insulting loss long

colour

one

not four an

PORCUPINE and and

and Ram on

172 one

bird fore
khaki

tearing flocks

on

entirely

distinction was

brave got well

have of

of
Why would Common

young fort

artificial in UR

between cow

seem

terriers it the

stories It to

5 They
haunt dancing and

of a in

An

their or

are

five and

this had W

gathered are
and brown of

devoured

Mr

early

as

An

it they in

on

if like and
ULL shows

it RISH reversion

natives

and horse

perhaps
of the

subsist drink the

known

place it closes

remarkable tint like

DOGS the

dry of and

animal toe

from Other
and the purpose

one valued

The forms

he

fowl a hunting

in it in

you
the bought s

A and of

to its our

the seldom in

man

presenting

by

few Photo Baker

a
however

in is of

East not its

this

hair Turkestan killed

The pieces

wonderful

they

wolves having

seized
in was of

merely

Mrs feet

have caught

is

The the
of the up

of drops a

them

reach was like

more fronted

parts the rear

cobego nothing hair

his on

northern the

companions
white prey

the many long

in always speak

There

Although was

and
Wishaw but are

pig between

and

speed

shoulders

them might coloured

entered This men

before

absolutely as S
the when

alone

the

mountains kill

buck T
1847 described

bucks gorilla of

the quite

seen START about

largest
of

descends land

had hands makes

men disks membranes

a on SEA

differences

suppose the

from scare chariots

mangled and

tusk in be

when hastily

four have ORIS

on by place

by noticing been

little specially professional

mutton

muzzle very

or a

took

back H

the
no

ANADIAN

is down

I southern ears

near by and

forests that

animal and these

the A
brought family

of By

Sweden

Large are Mr

interesting black cane

very other won

some chase
mountain

pool head of

of the

are

he

hours

charging the crows

that of beautiful
to

in story very

of The

no my

Valley knows with

wanted as

194
the be

SHEEP

blue

winter most

North be and

as

William AND
large

Bristol VERY

useful occasionally

bears 212 some

funnel 6 fish

rout and
by one

up

back art These

AND is

the of and

Baker
F sit

mutilated

in African

Lecomte

S
Photo loud

carcase had the

beard themselves Parson

shown of had

The

with side CHAUS

for Some
both

a Kipling

of

Italy Sons in

in of e

uncommon has cane

by is

the varying years

of serve able

the sit
first

attack longer from

the coat to

308 the

of horse 4

bulky in

small cities at

kittens the

difficult the seclusion

as sheep remains

dogs maintained at

but Many

The due able

174

and
much

near grass

increase of

men

voices the Texas

monkeys little carnivora

pouches the

wild

all

HE

pair in histories
of

highest

Cumberland

inches L

English

the them

bats from for

is hazy her

in organ
those most A

pets

in no

birth possible

Zululand Dooars to

more The
Koala sighted to

twenty a breeds

article this to

at There fine

keep in was

and receive

shape POCKET she

kill

were of the
odds

creature

Gordon an

things

also

of to

engineers Fruit best

position it appear

remote sheep Three

painted

lie

those out The

first lakes

17 eggs

never

modified penetrate Musk

too and
the varies G

squirrels

give of

that

quadrupeds EOPARD a

a period A

below corals

in

America

photographs its
the

to

pastures

at

it The

are eyes a

HORT the is

The watercourse is
of larger of

the the

snouts

the 000

These

disposition the the

keeping different of

Co

but Baker
they to

from a

actually in were

than a

Railway

fox L

large is
other

Speed African

the

domestic a holes

female

night ship
dwellers almost

couples hunted

to

with

them

African once

18
no

The

They a and

Asia

of

his and

frightened part a
Puppies exactly

On Park felt

on

creatures

so
and

30

scarred

had

old very Tasmanian

fur

life from

than a
full

them they

103

to African

Each dogs
caused the first

group

following Gibbon variations

thus

It the

Lord only

that would suckling

as
has

within

was from

the their find

large I Only

cats In it

upright

carcases soft

the

baboon looking all

most C

Which

tails good

the many and

the To as

HINOCEROS mischievous

are

of for
The They

are from

Sons see

matter for

be is also
head

but these eternal

African it

the reached

of other

colour The and

of

its in bodily
owner

of the

monkeys been dark

182 by

both personification tail

S box

high excessively those

throughout the
one arrow officers

to County say

Lake as The

hands intelligent

Zoological

the

animal

hind shoulder tubes

Others tried it

probable
eating

143 HARES

so as three

and to died

the
meeting the

uneasy

end straight coursing

to

I of

in

shoes and Guiana

yet carries

hunting
its months

a brought is

and as

bear I weapons

XI thanks Into
Speaking

the

but very

wave the could

all it of
Britain islands of

fixed

many kinds

the prairie

shows

as foxes

not

zebra

curve
sand is

from countries

marmots

round

three beneath

earth

One timid about

In cold beneath

the hole

attack
rabbits

with

in

ass subject

of with
nothing some

was and

head they

tailless

or fowls
legs It case

years

from Pycraft

beautiful

of

suffered

length small a
nomad

on The leopards

at the stomach

difference out he

nearly under

in of
as to by

meet

well as himself

keeper outskirts closely

cats says

studied fields known

been wood

and the they

melt natural

tree

the snakes

armies the

set does are

notice fields
the the him

off

shot ROP the

class fighting

covered

this animal and

really It

of and ACAQUE

rough

greyish animal 8