Topological Methods in Data Analysis and Visualization IV Theory Algorithms and Applications 1st Edition Hamish Carr 2024 Scribd Download
Topological Methods in Data Analysis and Visualization IV Theory Algorithms and Applications 1st Edition Hamish Carr 2024 Scribd Download
com
https://textbookfull.com/product/topological-
methods-in-data-analysis-and-visualization-iv-
theory-algorithms-and-applications-1st-edition-
hamish-carr/
https://textbookfull.com/product/topological-data-analysis-for-
scientific-visualization-1st-edition-julien-tierny-auth/
textbookfull.com
https://textbookfull.com/product/applications-of-topological-methods-
in-molecular-chemistry-1st-edition-remi-chauvin/
textbookfull.com
https://textbookfull.com/product/hybrid-intelligent-systems-ajith-
abraham/
textbookfull.com
Emergence of Pharmaceutical Industry Growth with
Industrial IoT Approach 1st Edition Valentina Emilia Balas
https://textbookfull.com/product/emergence-of-pharmaceutical-industry-
growth-with-industrial-iot-approach-1st-edition-valentina-emilia-
balas/
textbookfull.com
https://textbookfull.com/product/visual-c-for-kids-a-step-by-step-
computer-programming-tutorial-15th-edition-philip-conrod/
textbookfull.com
https://textbookfull.com/product/intersecting-film-music-and-
queerness-1st-edition-jack-curtis-dubowsky-auth/
textbookfull.com
https://textbookfull.com/product/becoming-michelangelo-apprenticing-
to-the-master-and-discovering-the-artist-through-his-drawings-alan-
pascuzzi/
textbookfull.com
Information Security Practices Emerging Threats and
Perspectives Awad Ahmed
https://textbookfull.com/product/information-security-practices-
emerging-threats-and-perspectives-awad-ahmed/
textbookfull.com
Mathematics and Visualization
Topological
Methods in Data
Analysis and
Visualization IV
Theory, Algorithms, and Applications
Mathematics and Visualization
Series Editors
Hans-Christian Hege
David Hoffman
Christopher R. Johnson
Konrad Polthier
Martin Rumpf
More information about this series at http://www.springer.com/series/4562
Hamish Carr • Christoph Garth • Tino Weinkauf
Editors
Topological Methods
in Data Analysis
and Visualization IV
Theory, Algorithms, and Applications
123
Editors
Hamish Carr Christoph Garth
University of Leeds Department of Computer Science
Leeds, United Kingdom Technical University of Kaiserslautern
Kaiserslautern, Germany
Tino Weinkauf
School of Computer Science
and Communication
KTH Royal Institute of Technology
Stockholm, Sweden
Cover illustration from Morse-Smale Analysis of Ion Diffusion in Ab Initio Battery Materials Simula-
tions by A. Gyulassy, A. Knoll, K. Chun Lau, B. Wang, P.-T. Bremer, M. E. Papka, L. A. Curtiss and
V. Pascucci. By courtesy of the authors.
Since Helman and Hesselink’s landmark paper on vector topology in 1989, topolog-
ical analysis has formed an increasingly important part of scientific visualization.
This is not only because it opens up novel forms of understanding but also because
as our data has increased past terascale, machine analysis necessarily substitutes
for laborious human inspection of visualizations. More and more, one can argue
that data analysis precedes rather than succeeds visualization and that topological
analysis is one of the key approaches given its strong mathematical underpinnings,
precise answers and verifiable outcomes.
From its early starts in vector field topology, topological visualization has
expanded to embrace analysis of scalar fields in the form of contour trees,
Reeb graphs and Morse-Smale complexes, analysis of abstract graphs and high-
dimensional data and, most recently, analysis of multivariate fields through Jacobi
sets, Reeb spaces and the joint contour net, linking with the mathematical field of
fibre topology in the process.
Topological visualization is, however, not concerned only with the topological
computation per se. One of the strongest features of the community is its focus on
the full range of theoretical understanding, algorithmic advances and application
work, all of which are represented in this volume.
Starting in 2005, biennial workshops have been held on topological visualization
in Budmerice (2005), Grimma (2007), Snowbird (2009), Zürich (2011), Davis
(2013) and Annweiler (2015), where informal discussions supplement formal
presentations and knit the community together. Notably, these workshops have
consistently resulted in quality publications under the Springer imprint which form
a significant part of the working knowledge in the area.
In the most recent workshop (2015), at Kurhaus Trifels in Annweiler, Germany,
bivariate analysis, Reeb spaces and fibre topology increased in importance, anchored
by keynotes from Professor Osamu Saeki (Kyushu), one of the leaders in fibre topol-
ogy, and Professor Kathrin Padberg-Gehle (Lüneburg), who works on computational
methods for nonlinear dynamical systems.
Of the 23 papers presented at TopoInVis 2015, 20 passed a second-round review
process for this volume. In addition, Professor Saeki contributed a survey of the
v
vi Preface
relevant fibre topology to this volume for the benefit of the community, which we
expect to shape approaches to data visualization in future years, and a further paper
was contributed directly to this volume.
We have grouped this paper in Part I with the two most closely related papers.
Of these, one deals with multi-modal analysis in a particular application domain
(atmospheric impacts of volcanic eruptions). The other deals with joint contour nets
(a quantized approximation of fibre topology) and their relation to analysis based on
Pareto set analysis.
We have then collected papers relating to high-dimensional data in Part II. Here,
the first paper applies scalar field topology to optimization problems, based on the
common description of optimization as a search landscape. In contrast, the second
paper discusses algorithms for computing and visualizing merge trees (one of the
principal forms of scalar analysis) in high-dimensional data. These are grouped with
a paper that considers the relative quality of different measures applied to reduce the
dimensionality of the data.
Part III then collects papers that use scalar topology in relatively low-dimensional
spaces (i.e. three-dimensional space). Here, the first paper compares similarity
between scalar fields, using histograms as summaries of geometric information
to supplement the underlying topological analysis. The second paper is more
applied in nature, as it addresses a practical domain problem—how to track
diffusion of ions into a battery material, using Morse-Smale analysis, to identify the
potential diffusion channels. Lastly, the third paper addresses the inverse problem
of (re-)constructing a scalar field from a known Morse-Smale complex.
Where Part III deals with scalar fields, Part IV considers vector and tensor
fields. Here, while the broad strokes of the analysis are well-understood, actual
computation of topological invariants has a number of practical problems. At the
heart of these is the tension between formal mathematical expression of continuous
models and practical numerical computation. The papers in this part therefore
primarily address issues of discontinuity and degeneracy in the analysis process.
Of these, the first paper deals with issues at the boundary of flow fields through
computation of escape maps, while the second computes similarity measures
between nearby integral curves to detect regions of shared behaviour. A third paper
extends existing ideas for decomposition of vector fields, in order to underpin
a future generation of algorithmic approaches, while a fourth paper extends
existing mathematical analysis of tensor fields as a preliminary to developing new
techniques.
Part V then considers a theme common to many of the newest approaches—
indirect detection of topological features to avoid the numerical problems of early
methods. Here, the goal is to detect coherent structures in a variety of contexts and
use them as the basis of the visualization. The best known techniques for this use
finite time Lyapunov exponents (FTLEs), and three of these papers extend these
techniques, while the fourth considers related computations.
In the first paper on FTLEs, they are used to detect regions of topological change
as a scalar field, which is then subjected to a second round of topological analysis
to detect ridge features. The second paper builds on the observation that not all
Preface vii
ix
x Contents
Osamu Saeki
1 Introduction
O. Saeki ()
Institute of Mathematics for Industry, Kyushu University, Motooka 744, Nishi-ku,
Fukuoka 819-0395, Japan
e-mail: [email protected]
visualization community in general, we will try to cover issues which might not be
so popular. Special attention is paid to Morse functions on manifolds with boundary,
since in many situations, data sets are given on a bounded domain in Euclidean
spaces which have boundary. We also describe the deformation of Morse functions,
from the view point of the simplification of the Reeb graph.
In Sect. 3, we review the singularity theory for generic multi-fields, which are
called stable maps in singularity theory. We will see that Morse theory can be
extended naturally to some dimension pairs, but not to all cases, at least theoretically.
In Sect. 4, the theory of singular fibers of differentiable maps is explained.
Mathematically, a fiber is a map around a given pre-image and it contains the
information of nearby pre-images. This is why it is important for grasping the
topological transitions of pre-images, which is essential in visualization.
In Sect. 5, we explain the concept of Reeb space of a given multi-field. This is
the straightforward generalization of Reeb graph for a scalar field. We will see that
several structure theorems are already known in singularity theory. In fact, once
you have a classification of singular fibers, a structure theorem then follows. Some
topological transitions of Reeb spaces are also presented with the simplification of
Reeb spaces in mind.
In Sect. 6, we will give several open problems related to singularity theory
and visualization of singular fibers. We will also explain how the visualization
techniques can be useful in singularity theory itself. We will give several examples
of ongoing projects in this direction as well. We end this paper by summarizing
the impact of such singularity theoretical results and techniques on computational
topology and visualization.
Throughout this paper, all manifolds and maps between them are differentiable
of class C1 unless otherwise indicated. A manifold is closed if it is compact and
has no boundary. The symbol Dk denotes the unit disk in Rk .
2 Morse Functions
@f @f @f
. p/ D . p/ D D . p/ D 0:
@x1 @x2 @xn
Theory of Singular Fibers and Reeb Spaces for Visualization 5
N R
Fig. 1 Example of level sets: one can observe a topological change of level sets as the value in R
passes through a critical value
By the implicit function theorem, we see easily that for a regular value r, the
pre-image f 1 .r/ is a smooth submanifold of dimension n 1 of N, as long as it
is non-empty (for example, see [26]). Based on this observation, we introduce the
following notion.
Definition 2.6 For a real number r 2 R, the set
f 1 .r/ D fp 2 N j f .p/ D rg
Fig. 2 Example of local level set changes for dimension 3: the top half corresponds to the case of
a critical point of index 1 or 2, while the bottom half corresponds to the case of index 0 or 3
for " 2 R with j"j sufficiently small. The other parts which sit outside of a
neighborhood of p do not change topologically in a sense similar to that in
Theorem 2.7. Examples for the case n D 3 are depicted in Fig. 2.
Definition 2.8 For a real number r 2 R, we call the set
Nr D fp 2 N j f .p/ rg
a sub-level set of f .
Note that if r is a regular value, then the corresponding sub-level set is a smooth
manifold whose boundary is the level set.
Suppose that a Morse function f is given. Starting from a real number r0 strictly
less than the minimum of f , let us consider the topological transition of the sub-
level sets Nr . Then, according to Theorem 2.7, its topological transition occurs near
a real number r only if r is a critical value of f . Furthermore, if p is a critical point
with value r, then by using the Morse Lemma, we can show that NrC" , with " >
0 sufficiently small, is obtained by attaching a -handle to Nr" , where is the
index of the critical point p. A -handle is an n-dimensional disk of the form D
Dn attached to @Nr" along @D Dn . In this way, we get a so-called handle
decomposition of the manifold N [24].
On the other hand, if we look at the transitions of the homotopy types of the
sub-level sets, then we get a decomposition of N as a CW complex [25].
For more details about handles, the reader is referred also to [12, Chap. 6]. An
application of handle decompositions for morphing 3D shapes has been explored in
[35].
Let us now define the Reeb graph of a Morse function.
Definition 2.9 Let f W N ! R be a Morse function on a closed manifold. Then,
each level set of f has finitely many connected components. Contracting each such
component to a point, we get a space Rf . More precisely, two points x; x0 2 N are
equivalent if they lie in the same component of a level set. This is an equivalence
relation, and the quotient space of N with respect to this equivalence relation is
8 O. Saeki
N R
qf f¯
Rf
Fig. 3 Reeb graph of the height function f on the torus: the original function is decomposed into
the composition of the quotient map qf and the function fN defined on the Reeb graph. The vertices
of Rf are the qf -images of the critical points of f
such that the differential dfp W Tp N ! R vanishes. We call such a point p a (usual)
critical point of f . Note that such a point p may lie in the interior as well as the
boundary of N in general. The other is a critical point of the restricted function
f@ D f j@N W @N ! R. Such a point is called a boundary critical point of f . Note that
a critical point of the second type necessarily lies on the boundary @N. Note also
that a usual critical point on @N is a boundary critical point, while the converse is
not true in general.
The following lemma is well-known.
Theorem 2.10 (Morse Lemma along Boundary) Let f W N ! R be a smooth
function defined on a manifold with boundary. If p 2 @N is not a usual critical point
of f , but is a non-degenerate critical point of f@ , then there exist local coordinates
.x1 ; x2 ; : : : ; xn / of N around p such that
(1) fxn 0g corresponds to N, and fxn D 0g corresponds to @N,
(2) f is locally expressed as
Fig. 4 Boundary critical points of functions defined on surfaces with boundary: the top half
represents x21 C x2 , while the bottom half represents x21 C x2
10 O. Saeki
outward a
Fig. 6 Birth-death of a usual critical point near the boundary: in the middle figures, the encircled
dots are usual critical points and are, at the same time, boundary critical points. The two functions
on the left have only a boundary critical point, while the two on the right have both a boundary
critical point and a usual critical point in the interior
(1) (2)
(3) (4)
(5)
Fig. 7 Possible transitions of Reeb graphs for generic 1-parameter deformations of Morse
functions on closed manifolds near a given parameter: these are local descriptions, and the part
lying outside of these graphs does not change during the deformation
to another Morse function whose Reeb graph is locally of the form as in the left hand
side of (2). However, for the resulting graph, the transition (4) from the right to left
cannot be applied. This is because after the birth of a pair of critical points, they
are so involved with each other that their values cannot be interchanged to make a
crossing of critical values.
We can also prove that the transition of Fig. 7 (1) is always possible. More
precisely, if the Reeb graph of a Morse function contains one of the two graphs
in the figure as a subgraph, then we can deform the given Morse function by passing
through a birth-death exactly once so that the resulting Morse function has the Reeb
graph obtained from the original one by replacing the subgraph with the other graph
in the figure. This gives a theoretical justification for simplification of a Reeb graph
for visualization purposes that uses the topological transition described in Fig. 7 (1).
The above theorem can be proved by using a result on local structures of Reeb
spaces of generic maps into the plane [19]. (For the definition of a Reeb space, refer
to Sect. 5.1.)
14 O. Saeki
3 Stable Maps
Although theorems like the Morse Lemma do not exist in general for multi-
fields, for some specific dimension pairs .n; m/, we do have such theorems. In the
following, if we say that a map is stable, then it means that it is C1 stable.
In the function case, the following is known.
Theorem 3.2 Let N be a closed n-dimensional manifold, n 1. Then a smooth
function f W N ! R is stable if and only if it is a Morse function. In particular, the
dimension pair .n; 1/ is always in the nice range.
The above theorem means that the notion of a stable map generalizes the notion
of a Morse function in a reasonable sense.
Let us introduce the following notion.
Definition 3.3 Let fi W Ni ! Rm , i D 0; 1, be smooth maps with dim N0 D
dim N1 D n. For singular points pi 2 Ni of fi , i D 0; 1, we define that they have
the same singularity type if for some open neighborhoods Ui of pi and Vi of fi .pi /
and diffeomorphisms W U0 ! U1 and W V0 ! V1 with .p0 / D p1 and
.f0 .p0 // D f1 .p1 / such that the following diagram is commutative:
Note that the Morse Lemma claims that a non-degenerate critical point of a
function has the same singularity type as the critical point of a quadratic function
˙x21 ˙ x22 ˙ ˙ x2n .
Let us now consider the case m D 2. Let f W N ! R2 be a smooth map of a
closed n-dimensional manifold, n 2.
Random documents with unrelated
content Scribd suggests to you:
by the Press which he had done nothing to conciliate, he
continued on his headlong course, and at the end of January,
1908, he signed a decree practically amounting to a suspension
of civil liberties."
Lisbon Correspondence,
London Times.
{507}
PORTUGAL: A. D. 1909.
Demonstration against the Religious Orders.
PORTUGAL: A. D. 1909.
Offer of Dom Miguel to renounce his Claim to the Throne.
Dom Miguel, son of the Dom Miguel who, from 1828 to 1833 held
the throne of Portugal in defiance of the rights of Maria da
Gloria, his elder brother’s daughter (see, in Volume IV.,
PORTUGAL: A. D. 1824-1889), had kept up his father’s
pretensions to the crown until the spring of 1909, when he
offered to renounce it, if permitted to live in Portugal as a
citizen. The permission was refused, for the reason that his
return, with that of a number of nobles of his party, "would
be regarded as a challenge to the rising tide of Liberalism."
PORTUGUESE AFRICA.
Edith Sellers,
Poor Relief in Vienna
(Contemporary Review, December, 1900).
{508}
The Act of the British Parliament, "to Provide for Old Age
Pensions" (August 1, 1908), declares in its first section that
"the receipt of an old age pension under this Act shall not
deprive the pensioner of any franchise, right, or privilege,
or subject him to any disability." The second section defines
the "statutory conditions for the receipt of an old age
pension by any person" to be: the person must have attained
the age of seventy; must satisfy the pension authorities that
he has been a British subject and resident in the United
Kingdom for at least twenty years; that his yearly means, as
calculated under the stipulations of the Act, do not exceed
thirty-one pounds ten shillings. But, notwithstanding the
fulfilment of these statutory conditions, a person is
disqualified while he is in receipt of any poor relief, other
than medical or surgical assistance on the recommendation of a
medical officer, or relief rendered by means of the
maintenance of a dependent in an asylum, infirmary, or
hospital, or any relief that by law is expressly declared not
to be a disqualification for any franchise, right, or
privilege. Furthermore, any person is disqualified for the
receipt of an old age pension "if, before he becomes entitled
to a pension, he has habitually failed to work according to
his ability, opportunity, and need, for the maintenance or
benefit of himself and those legally dependent upon him:
Provided that a person shall not be disqualified under this
paragraph if he has continuously for ten years up to attaining
the age of sixty, by means of payments to friendly, provident,
or other societies, or trade unions, or other approved steps,
made such provision against old age, sickness, infirmity, or
want or loss of employment as may be recognized as proper
provision for the purpose by regulations under this Act, and
any such provision, when made by the husband in the case of a
married couple living together, shall, as respects any right
of the wife to a pension, be treated as provision made by the
wife as well as by the husband."
Specific rules are given in the Act for "calculating the means
of a person" who seeks the pension; and the rate of weekly
pension to be paid is proportioned inversely to such
ascertained means, as follows: "Where the yearly means of the
pensioner as calculated under this Act:
Do not exceed £21.,—5s. 0d.;
exceed £21, but do not exceed £23, 12s. 6d.,—4s. 0d.;
exceed £23 12s. 6d., but do not exceed £26 5s.,—3s. 0d.;
exceed £26 5s., but do not exceed £28 17s. 6d.,—2s. 0c2.;
exceed £28 17s. 6d., but do not exceed £31 10s.,—1s. 0d.;
exceed £31 10s., no pension."
{509}
"What struck one in such cases was how the people had fought
against the horror of the Poor Law. There were 270,000 people
over 70 years of age in receipt of Poor Law relief. The
Old-Age Pensions Act had disclosed the presence in the
community of over 600,000 people the vast majority of whom
were living in circumstances of great poverty, and yet
disdained the charity of the Poor Law."
{510}
If the labourer died after subscribing for 200 weeks his wife
and children were entitled to receive what he had subscribed,
but nothing more.
"The lot of the widows and orphans was one of the black
features of the system. A married woman could not qualify for
an old-age pension. The amount of the weekly contribution was
fixed for ten years. In 1906 the receipts exceeded the
expenditure by £6,000,000; the cost of administration was only
£850,000. But that was only the minor part of the provision
made for elderly people in Germany. The main provision was
made under the head of infirmity or invalidity occurring
before the pension age—70. If the insurers, after having
subscribed for not less than four years, broke down and were
unable to earn wages, they were entitled to more generous
treatment. If curable they were cured in State sanatoriums and
received temporary sickness pensions. If incurable they
received a pension which was regulated by the number of years
they had subscribed, and varied from a minimum of £5
16s. in the lowest grade for four years’ subscriptions to £22
10s. in the highest grade for 50 years’ subscriptions. The
insurer began to pay his contributions at the age of 17, and
for an old-age pension he had to subscribe 50 weeks a year for
24 years—1,200 weeks in all. Though the system had not checked
Socialism or militant trade unionism, it had attained its real
purpose, for it had conferred an enormous boon upon the poor."
At the time when the remark quoted above, touching the
defective provision of the German law for widows and orphans,
was made, the Imperial Government was preparing to amend it.
The London Times of April 17, 1909, gave, in its
correspondence from Berlin, the account of a draft Bill, just
made public, which the Imperial Ministry of the Interior had
prepared for presentation to the Federal Council, the object
being to combine and coordinate "the seven compulsory
insurance laws of 1883 to 1899," together with certain
amendments and additions. "It is understood," wrote the
correspondent, "that the Bill will not reach the Reichstag
before the autumn of this year. Whereas many authorities …
have favored a thorough unification of the three systems of
invalidity and old age, accident, and sick insurance, the
immediate proposals of the Government would leave the three
systems separate and distinct, while codifying the law and the
regulations which are common to all branches of compulsory
insurance, and establishing a joint and threefold system of
higher administration." The main purpose of the bill was to
rectify that lack of proper provision for widows and orphans
which was noted above. "The need of solving this problem,"
said the correspondent, "is really the immediate occasion of
reform, and the proposed solution is the most important
feature of the reform scheme. An essential feature of the
tariff law of 1902 was the ear-marking—by the so-called Lex
Trimborn—for widows and orphans’ insurance of the surplus
revenue from the increased Customs duties on corn and cattle.
The Lex Trimborn takes effect on January 1, 1910, but the
surplus revenue is lacking. For the financial year 1906 there
was no surplus. For 1907 there was a surplus of about
£2,000,000. For the financial year 1908 there will be no
surplus, although £2,650,000 was estimated for. In these
circumstances the Government—while apparently still cherishing
the hope that, upon the average of a long period of years, the
revised tariff will do what was expected of it—proposes to
provide for widows and orphans insurance by a simple all-round
extension of the system of invalidity and old-age insurance.
That is to say, the ‘contributions’ of employers and employed
are to be raised, and an Imperial subsidy, of fixed amount,
without regard to the annual revenue from Customs, is to be
added to the contributions.
{512}
{513}
{514}
A Bill known as the "Right to Work" Bill came before the House
of Commons in April, 1909, with the endorsement of the trade
unions and the Labor Party. It was opposed by John Burns, the
former labor leader, but now speaking as President of the
Local Government Board and member of the Cabinet, who said:
"For three and a half years he had had intimate experience of
relief works, and he could not exaggerate the degradation of
the workmen, the demoralization of the honest labourer, the
extent to which money had been wasted and character impaired
by the relief works which he had had in the name of Parliament
to administer. Any member had only to take up the report of
any one of the distress committees to see that what the
minority report said had happened would increasingly happen so
long as these means of meeting unemployment were resorted to.
The amount of work would be disproportionate to the wages
paid, the wrong men would get the right work, and the best men
would be excluded, because modesty was a characteristic of
good workmanship and craftsmanship, and the worst men were
always in the front line when relief works were set on foot."