Professional Documents
Culture Documents
Textbook Algorithms and Models For The Web Graph Anthony Bonato Ebook All Chapter PDF
Textbook Algorithms and Models For The Web Graph Anthony Bonato Ebook All Chapter PDF
https://textbookfull.com/product/algorithms-and-models-for-the-
web-graph-17th-international-workshop-waw-2020-warsaw-poland-
september-21-22-2020-proceedings-bogumil-kaminski/
https://textbookfull.com/product/graph-searching-games-and-
probabilistic-methods-1st-edition-bonato/
https://textbookfull.com/product/a-guide-to-graph-colouring-
algorithms-and-applications-lewis/
https://textbookfull.com/product/graph-algorithms-for-data-
science-with-examples-in-neo4j-1st-edition-tomaz-bratanic/
Hybrid System Identification: Theory and Algorithms for
Learning Switching Models Fabien Lauer
https://textbookfull.com/product/hybrid-system-identification-
theory-and-algorithms-for-learning-switching-models-fabien-lauer/
https://textbookfull.com/product/graph-algorithms-practical-
examples-in-apache-spark-and-neo4j-1st-edition-mark-needham/
https://textbookfull.com/product/discrete-mathematics-graph-
algorithms-algebraic-structures-coding-theory-and-
cryptography-1st-edition-sriraman-sridharan/
https://textbookfull.com/product/biological-network-analysis-
trends-approaches-graph-theory-and-algorithms-1st-edition-pietro-
hiram-guzzi/
https://textbookfull.com/product/genetic-algorithms-and-machine-
learning-for-programmers-create-ai-models-and-evolve-
solutions-1st-edition-frances-buontempo/
Anthony Bonato
Paweł Prałat
Andrei Raigorodskii (Eds.)
LNCS 10836
123
Lecture Notes in Computer Science 10836
Commenced Publication in 1973
Founding and Former Series Editors:
Gerhard Goos, Juris Hartmanis, and Jan van Leeuwen
Editorial Board
David Hutchison
Lancaster University, Lancaster, UK
Takeo Kanade
Carnegie Mellon University, Pittsburgh, PA, USA
Josef Kittler
University of Surrey, Guildford, UK
Jon M. Kleinberg
Cornell University, Ithaca, NY, USA
Friedemann Mattern
ETH Zurich, Zurich, Switzerland
John C. Mitchell
Stanford University, Stanford, CA, USA
Moni Naor
Weizmann Institute of Science, Rehovot, Israel
C. Pandu Rangan
Indian Institute of Technology Madras, Chennai, India
Bernhard Steffen
TU Dortmund University, Dortmund, Germany
Demetri Terzopoulos
University of California, Los Angeles, CA, USA
Doug Tygar
University of California, Berkeley, CA, USA
Gerhard Weikum
Max Planck Institute for Informatics, Saarbrücken, Germany
More information about this series at http://www.springer.com/series/7407
Anthony Bonato Paweł Prałat
•
123
Editors
Anthony Bonato Andrei Raigorodskii
Department of Mathematics Department of Discrete Mathematics
Ryerson University Moscow Institute of Physics and Technology
Toronto, ON Dolgoprudny
Canada Russia
Paweł Prałat
Department of Mathematics
Ryerson University
Toronto, ON
Canada
This Springer imprint is published by the registered company Springer International Publishing AG
part of Springer Nature
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
Preface
The 15th Workshop on Algorithms and Models for the Web Graph (WAW 2018) took
place at the Moscow Institute of Physics and Technology, Russia, May 17–18, 2018.
This is an annual meeting, which is traditionally co-located with another, related,
conference. WAW 2018 was co-located with the Workshop on Graphs, Networks, and
Their Applications. The co-location of the two workshops provided opportunities for
researchers in two different but interrelated areas to interact and to exchange research
ideas. It was an effective venue for the dissemination of new results and for fostering
research collaboration.
The World Wide Web has become part of our everyday life, and information
retrieval and data mining on the Web are now of enormous practical interest. The
algorithms supporting these activities combine the view of the Web as a text repository
and as a graph, induced in various ways by links among pages, hosts and users. The
aim of the workshop was to further the understanding of graphs that arise from the Web
and various user activities on the Web, and stimulate the development of
high-performance algorithms and applications that exploit these graphs. The workshop
gathered together researchers working on graph-theoretic and algorithmic aspects of
related complex networks, including social networks, citation networks, biological
networks, molecular networks, and other networks arising from the Internet.
This volume contains the papers presented during the workshop. Each submission
was reviewed by Program Committee members. Papers were submitted and reviewed
using the EasyChair online system. The committee members accepted 11 papers.
General Chairs
Andrei Z. Broder Google Research, USA
Fan Chung Graham University of California San Diego, USA
Organizing Committee
Anthony Bonato Ryerson University, Canada
Paweł Prałat Ryerson University, Canada
Andrei Raigorodskii MIPT, Russia
Program Committee
Konstantin Avratchenkov Inria, France
Paolo Boldi University of Milan, Italy
Anthony Bonato Ryerson University, Canada
Milan Bradonjic Bell, USA
Fan Chung Graham UC San Diego, USA
Collin Cooper King’s College London, UK
Andrzej Dudek Western Michigan University, USA
Alan Frieze Carnegie Mellon University, USA
Aristides Gionis Aalto University, Finland
David Gleich Purdue University, USA
Jeannette Janssen Dalhousie University, Canada
Bogumil Kaminski Warsaw School of Economics, Poland
Ravi Kumar Google Research, USA
Silvio Lattanzi Google Research, USA
Marc Lelarge Inria, France
Stefano Leonardi Sapienza University of Rome, Italy
Nelly Litvak University of Twente, The Netherlands
Michael Mahoney UC Berkeley, USA
Oliver Mason NUI Maynooth, Ireland
Dieter Mitsche Université de Nice Sophia-Antipolis, France
Peter Morters University of Bath, UK
Tobias Mueller Utrecht University, The Netherlands
Liudmila Ostroumova Yandex, Russia
Pan Peng TU Dortmund, Germany
Xavier Perez-Gimenez University of Nebraska-Lincoln, USA
Pawel Pralat Ryerson University, Canada
Yana Volkovich AppNexus, USA
Stephen Young Pacific Northwest National Laboratory, USA
VIII Organization
Sponsoring Institutions
1 Introduction
The induced subgraph isomorphism problem asks whether a large graph G con-
tains a connected graph H as an induced subgraph. When k is allowed to grow
with the graph size n, this problem is NP-hard in general. For example, k-
clique and k induced cycle, special cases of H, are known to be NP-hard [13,20].
For fixed k, this problem can be solved in polynomial time O(nk ) by search-
ing for H on all possible combinations of k vertices. Several randomized and
non-randomized algorithms exist to improve upon this trivial way of finding
H [14,25,27,29].
On real-world networks, many algorithms were observed to run much faster
than predicted by the worst-case running time of algorithms. This may be
ascribed to some of the properties that many real-world networks share [4],
such as the power-law degree distribution found in many networks [1,8,19,28].
One way of exploiting these power-law degree distributions is to design algo-
rithms that work well on random graphs with power-law degree distributions.
For example, finding the largest clique in a network is NP-complete for general
networks [20]. However, in random graph models such as the Erdős-Rényi ran-
dom graph and the inhomogeneous random graph, their specific structures can be
exploited to design fixed parameter tractable (FPT) algorithms that efficiently
find a clique of size k [10,12] or the largest independent set [15].
In this paper, we study algorithms that are designed to perform well for
the inhomogeneous random graph, a random graph model that can generate
graphs with a power-law degree distribution [2,3,5,6,24,26]. The inhomogeneous
random graph has a densely connected core containing many cliques, consisting
of vertices with degrees n log(n) and larger. In this densely connected core,
the probability of an edge being present is close to one, so that it contains
c Springer International Publishing AG, part of Springer Nature 2018
A. Bonato et al. (Eds.): WAW 2018, LNCS 10836, pp. 1–15, 2018.
https://doi.org/10.1007/978-3-319-92871-5_1
2 E. Cardinaels et al.
many complete graphs [18]. This observation was exploited in [11] to efficiently
determine whether a clique of size k occurs as a subgraph in an inhomogeneous
random graph. When searching for induced subgraphs however, some edges are
required not to be present. Therefore, searching for induced subgraphs in the
entire core is not efficient. We show that a connected subgraph H can be found
as an induced subgraph by scanning only vertices √ that are on the boundary of
the core: vertices with degrees proportional to n.
We present √ an algorithm that first selects the set of vertices with degrees
proportional to n, and then randomly searches for H as an induced subgraph on
a subset of k of those vertices. The first algorithm we present does not depend on
the specific structure of H. For general sparse graphs, the best known algorithms
to solve subgraph isomorphism on 3 or 4 vertices run in O(n1.41 ) or O(n1.51 ) time
with high probability [29]. For small values of k, our algorithm solves subgraph
isomorphism on k nodes in linear time with high probability on inhomogeneous
random graphs. However, the graph size needs to be very large for our algorithm
to perform well. We therefore present √a second algorithm that again selects the
vertices with degrees proportional to n, and then searches for induced subgraph
H in a more efficient way. This algorithm has the same performance guarantee
as our first algorithm, but performs much better in simulations.
We test our algorithm on large inhomogeneous random graphs, where it
indeed efficiently finds induced subgraphs. We also test our algorithm on real-
world network data with power-law degrees. There our algorithm does not per-
form well, probably due to the fact that the densely connected core of some
real-world
√ networks may not be the vertices of degrees at least proportional
to n. We then show that a slight modification of our algorithm that looks for
induced subgraphs on vertices of degrees proportional to nγ for some other value
of γ performs better on real-world networks, where the value of γ depends on
the specific network.
Notation. We say that a sequence of events (En )n≥1 happens with high prob-
ability (w.h.p.) if limn→∞ P (En ) = 1. Furthermore, we write f (n) = o(g(n)) if
limn→∞ f (n)/g(n) = 0, and f (n) = O(g(n)) if |f (n)|/g(n) is uniformly bounded,
where (g(n))n≥1 is nonnegative. Similarly, if lim supn→∞ |f (n)| /g(n) > 0, we
say that f (n) = Ω(g(n)) for nonnegative (g(n))n≥1 . We write f (n) = Θ(g(n)) if
f (n) = O(g(n)) as well as f (n) = Ω(g(n)).
1.1 Model
As a random graph null model, we use the inhomogeneous random graph or
hidden variable model [2,3,5,6,24,26]. Every vertex is equipped with a weight.
We assume that the weights are i.i.d. samples from the power-law distribution
P (wi > k) = Ck 1−τ (1.1)
for some constant C and for τ ∈ (2, 3). Two vertices with weights w and w are
connected with probability
ww
p(w, w ) = min ,1 , (1.2)
μn
Finding Induced Subgraphs in Scale-Free Inhomogeneous Random Graphs 3
where μ denotes the mean value of the power-law distribution (1.1). Choosing
the connection probability in this way ensures that the expected degree of a
vertex with weight w is w.
1.2 Algorithms
The following theorem gives a bound for the performance of Algorithm 1 for
small values of k.
Theorem 1. Choose f1 = f1 (n) ≥ 1/ log(n) and f1 < f2 < 1 and let k <
log1/3 (n). Then, with high probability, Algorithm 1 detects induced subgraph H
on k vertices in an inhomogeneous random graph with n vertices and weights
distributed as in (1.1) in time O(nk).
Thus, for small values of k, Algorithm 1 finds an instance of H in linear time.
4 E. Cardinaels et al.
The following theorem shows that indeed Algorithm 2 has similar perfor-
mance guarantees as Algorithm 1.
Theorem 2. Choose f1 = f1 (n) ≥ 1/ log(n) and f1 < f2 < 1. Choose s =
Ω(nα ) for some 0 < α < 1, such that s ≤ n/k. Then, Algorithm 2 detects
induced subgraph H on k < log1/3 (n) vertices on an inhomogeneous random
graph with n vertices and weights distributed as in (1.1) in time O(nk) with high
probability.
The proofs of Theorems 1 and 2 rely on the fact that for small k, any sub-
graph on k vertices is present in G with high probability. This means that after
the degree selection step of Algorithms 1 and 2, for small k, any motif finding
algorithm can be used to find motif H on the remaining graph G , such as the
Grochow-Kellis algorithm [14], the MAvisto algorithm [27] or the MODA algo-
rithm [25]. In the proofs of Theorems 1 and 2, we show that G has Θ(n(3−τ )/2 )
vertices with high probability. Thus, the degree selection step reduces the prob-
lem of finding a motif H on n vertices to finding a motif on a graph with
Θ(n(3−τ )/2 ) vertices, significantly reducing the running time of the algorithms.
We prove Theorem 1 using two lemmas. The first lemma relates the degrees of
the vertices to their weights. The connection probabilities in the inhomogeneous
random graph depend on the weights of the vertices. In Algorithm 1, we select
vertices based on their degrees instead of their unknown weights. The following
lemma shows that the weights of the vertices in V are close to their degrees.
6 E. Cardinaels et al.
√
Lemma
√ 1. Degrees and weights. Fix ε > 0, and define Jn = [(1−ε) f1 μn, (1+
ε) f2 μn]. Then, for some K > 0,
2
ε (1 − ε)
P (∃i ∈ V : wi ∈
/ Jn ) ≤ Kn exp − f1 μn . (2.1)
2(1 + ε)
Proof. Fix a vertex i ∈ V . Conditionally on the weight wi of vertex i, Di ∼
Poi(wi ) [5,16]. Then,
P D ∈ I | w < (1 − ε)√f μn
i n i 1
P wi < (1 − ε) f1 μn, Di ∈ In = √
P wi < (1 − ε) f1 μn
√ √
P Di > f1 μn | wi = (1 − ε) f1 μn
≤ √
1 − C((1 − ε) f1 μn)1−τ
≤ K1 P Di > f1 μn | wi = (1 − ε) f1 μn ,
(2.2)
for some K1 > 0. Here the first inequality follows because for Poisson random
variables P (Poi(λ1 ) > k) ≤ P (Poi(λ2 ) > k) for λ1 < λ2 . We use that by the
Chernoff bound for Poisson random variables
P (X > λ(1 + δ)) ≤ exp −h(δ)δ 2 λ/2 , (2.3)
where h(δ) = 2((1 + δ) ln(1 + δ) − δ)/δ 2 . Therefore, using that h(δ) ≥ 1/(1 + δ)
for δ ≥ 0 results in
2
ε (1 − ε)
P Di > f1 μn | wi = (1 − ε) f1 μn ≤ exp − f1 μn . (2.4)
2(1 + ε)
Combining this with (2.2) and taking the union bound over all vertices then
results in
2
ε (1 − ε)
P ∃i : Di ∈ In , wi < (1 − ε) f1 μn ≤ K1 n exp − f1 μn . (2.5)
2(1 + ε)
√
The bound for wi > (1 + ε) f2 μn follows similarly. Combining this with the
fact that f1 < f2 then proves the lemma.
The second lemma shows that after deleting all vertices with degrees outside
of In defined in Step 1 of Algorithm 1, still polynomially many vertices remain
with high probability.
Lemma 2. Polynomially many nodes remain. There exists γ > 0 such that
P |V | < γn(3−τ )/2 ≤ 2 exp −Θ(n(3−τ )/2 ) . (2.6)
Proof. Let E denote the event that all vertices i ∈ V satisfy wi ∈ Jn for some
ε > 0, with Jn as in Lemma 1. Let W be the set of vertices with weights in Jn .
Under the event E, |V | ≤ |W |. Then, by Lemma 1
2
ε (1 − ε)
P |V | < γn(3−τ )/2 ≤ P |W | < γn(3−τ )/2 + Kn exp − f1 μn .
2(1 + ε)
(2.7)
Finding Induced Subgraphs in Scale-Free Inhomogeneous Random Graphs 7
Furthermore,
√
P (wi ∈ Jn ) = C((1 − ε) f1 μn)1−τ − C((1 + ε) f2 μn)1−τ ≥ c1 ( μn)1−τ
(2.8)
for some constant c1 > 0 because f1 < f2 . Thus, each of the n vertices is in
√
set W independently with probability at least c1 ( μn)1−τ . Choose 0 < γ < c1 .
Applying the multiplicative Chernoff bound then shows that
(c1 − γ)2 (3−τ )/2
P |W | < γn (3−τ )/2
≤ exp − n , (2.9)
2c1
√
which proves the lemma together with (2.7) and the fact that f1 μn =
Ω(n(3−τ )/2 ) for τ ∈ (2, 3).
1
Now apply that k ≤ log 3 (n). Then
3−τ log 23 n
P (H not in the partitions) ≤ exp − dn 12 c3
log n
log 3 n (2.15)
3−τ
≤ exp −dn 2 −o(1) .
Hence, the inner expression grows polynomially such that the probability of not
finding H in one of the partitions is negligibly small. The running time of the
partial search is given by
|V | k n k 4
≤ ≤ nk ≤ nek , (2.16)
k 2 k 2
by following a random edge. The probability that vertex i is added can therefore
be bounded as
Di,G M log(n)
P (vertex i is added) = ≤ (2.18)
|V | |V |
s=1 Ds,G
for some constant M > 0 by the conditions on the degrees. Therefore, the prob-
ability that Sj does not overlap with one of the previously chose jk vertices can
be bounded from below by
kj M kj log(n) k−1
P (Sj does not overlap with previous sets) ≥ 1− 1− . (2.19)
|V | |V |
Thus, the probability that all j sets do not overlap can be bounded as
jk
M kj log(n)
P (Sj ∩ Sj−1 · · · ∩ S1 = ∅) ≥ 1− , (2.20)
|V |
which tends to one when jk = o(n(3−τ )/4 ). Let sdis denote the number of disjoint
sets out of the s sets constructed in Algorithm 2. Then, when s = Ω(nα ) for some
α > 0, sdis > nβ for some β > 0 with high probability, because k < log1/3 (n).
The probability that H is present as an induced subgraph is bounded sim-
ilarly as in Theorem 1. We already know that k − 1 edges are present. For all
other E − (k − 1) edges of H, and all k2 − E edges that are not present in H,
we can again use (2.10) and (2.11) to bound on the probability of edges being
present or not being present between vertices in V . Therefore, we can bound
the probability that H is not found similarly to (2.13) as
Because sdis > nβ for some β > 0, this term tends to zero exponentially. The
running time of the partial search can be bounded similarly to (2.16) as
k
s ≤ sk 2 = O(nk), (2.21)
2
3 Experimental Results
Fig. 1 shows the fraction of times Algorithm 1 succeeds to find a cycle of size
k in an inhomogeneous random graph on 107 vertices. Even though for large n
Algorithm 1 should find an instance of a cycle of size k in step 7 of the algorithm
with high probability, we see that Algorithm 1 never succeeds in finding one. This
is because of the finite size effects discussed before.
10 E. Cardinaels et al.
Fig. 1. The fraction of times step 7 in Algorithm 1 succeeds to find a cycle of length k
on an inhomogeneous random graph with n = 107 , averaged over 500 network samples
with f1 = 1/ log(n) and f2 = 0.9.
Figure 2a also plots the fraction of times Algorithm 2 succeeds to find a cycle.
We set the parameter s = 10000 so that the algorithm fails if the algorithm does
not succeed to detect motif H after executing step 13 of Algorithm 2 10000
times. Because s gives the number of attempts to find H, increasing s may
increase the success probability of Algorithm 2 at the cost of a higher running
time. However, in Fig. 2b we see that for small values of k, the mean number of
times Step 13 is executed when the algorithm succeeds is much lower than 10000,
so that increasing s in this experiment probably only has a small effect on the
success probability. We see that Algorithm 2 outperforms Algorithm 1. Figure 2b
also shows that the number of attempts needed to detect a cycle of length k is
small for k ≤ 6. For larger values of k the number of attempts increases. This
can again be ascribed to the finite size effects that cause the set V to be small,
so that large motifs may not be present on vertices in set V . We also plot the
success probability when using different values of the functions f1 and f2 . When
only the lower bound f1 on the vertex degrees is used, as in [11], the success
probability of the algorithm decreases. This is because the set V now contains
many high degree vertices that are much more likely to form clique motifs than
cycles or other connected motifs on k vertices. This makes f2 = ∞ a very efficient
bound for detecting clique motifs [11]. For the cycle motif however, we see in
Fig. 2b that more checks are needed before a cycle is detected, and in some cases
the cycle is not detected at all.
Setting f1 = 0 and f2 = ∞ is also less efficient, as Fig. 2a shows. In this
situation, the number of attempts needed to find a cycle of length k is larger
than for Algorithm 2 for k ≤ 6.
shows the fraction of runs where Algorithm 2 finds a cycle as an induced sub-
graph. We see that for the Wikipedia social network in Fig. 3a, Algorithm 2 is
more efficient than looking for cycles among all vertices in the network. For the
Baidu online encyclopedia in Fig. 3c however, we see that Algorithm 2 performs
much worse than looking for cycles among all possible vertices. In the other two
network data sets in Figs. 3b and d the performance on the reduced vertex set
and the original vertex set is almost the same. Figure 4 shows that in general,
Algorithm 2 indeed seems to finish in fewer steps than when using the full vertex
set. However, as Fig. 4c shows, for larger values of k the algorithm fails almost
always.
Table 1. Statistics of the data sets: the number of vertices n, the number of edges E,
and the power-law exponent τ fitted by the method of [7].
n E τ
Wikipedia 2,394,385 5,021,410 2.46
Gowalla 196,591 950,327 2.65
Baidu 2,141,300 17,794,839 2.29
AS-Skitter 1,696,415 11,095,298 2.35
Fig. 3. The fraction of times Algorithm 2 succeeds to find a cycle on four large network
data sets for detecting cycles of length k. The parameters are chosen as s = 10000,
f1 = 1/ log(n), f2 = 0.9. The black line uses Algorithm 2 on vertices of degrees in
In = [(μn)γ / log(n), (μn)γ ]. The values are averaged over 500 runs of Algorithm 2.
value of γ that works well. For the Gowalla, Wikipedia and Autonomous systems
network, this leads to a faster algorithm to detect cycles. Only for the Baidu net-
work other values of γ do not improve upon randomly selecting from all vertices.
This indicates that for most networks, cycles do appear mostly on degrees with
specific orders of magnitude, making it possible to sample these cycles faster.
Unfortunately, these orders of magnitude may be different for different networks.
Across all four networks, the best value of γ seems to be smaller than the value
of 0.5 that is optimal for the inhomogeneous random graph.
Finding Induced Subgraphs in Scale-Free Inhomogeneous Random Graphs 13
Fig. 4. The number of times step 12 of Algorithm 2 is invoked when the algorithm does
not fail on four large network data sets for detecting cycles of length k. The parameters
are chosen as s = 10000, f1 = 1/ log(n), f2 = 0.9. The black line uses Algorithm 2
on vertices of degrees in In = [(μn)γ / log(n), (μn)γ ]. The values are averaged over 500
runs of Algorithm 2.
4 Conclusion
We presented an algorithm which solves the induced subgraph problem on inho-
mogeneous random graphs with infinite variance power-law degrees in time
4
O(nek ) with high probability as n grows large. This algorithm is based on the
observation that for fixed k, any subgraph is present on k vertices with degrees
√
slightly smaller than μn with positive probability. Therefore, the algorithm
first selects vertices with those degrees, and then uses a random search method
to look for the induced subgraph on those vertices.
We show that this algorithm performs well on simulations of inhomogeneous
random graphs. Its performance on real-world data sets varies for different data
sets. This indicates that the degrees that contain the√ most induced subgraphs
of size k in real-world networks may not be close to n. We then show that on
these data sets, it may be more efficient to find induced subgraphs on degrees
proportional to nγ for some other value of γ. The value of γ may be different for
different networks.
14 E. Cardinaels et al.
√
Our algorithm exploits that induced subgraphs are likely formed among μn-
degree vertices. However, certain subgraphs may occur more frequently on ver-
tices of other degrees [17]. For example, star-shaped subgraphs on k vertices
√
appear more often on one vertex with degree much higher than μn corre-
sponding to the middle vertex of the star, and k − 1 lower-degree vertices cor-
responding to the leafs of the star [17]. An interesting open question is whether
there exist better degree-selection steps for specific subgraphs than the one used
in Algorithms 1 and 2.
Acknowledgements. The work of JvL and CS was supported by NWO TOP grant
613.001.451. The work of JvL was further supported by the NWO Gravitation Networks
grant 024.002.003, an NWO TOP-GO grant and by an ERC Starting Grant.
References
1. Albert, R., Jeong, H., Barabási, A.L.: Internet: diameter of the world-wide web.
Nature 401(6749), 130–131 (1999)
2. Boguñá, M., Pastor-Satorras, R.: Class of correlated random networks with hidden
variables. Phys. Rev. E 68, 036112 (2003)
3. Bollobás, B., Janson, S., Riordan, O.: The phase transition in inhomogeneous ran-
dom graphs. Random Struct. Algorithms 31(1), 3–122 (2007)
4. Brach, P., Cygan, M., L acki, J., Sankowski, P.: Algorithmic complexity of power law
networks. In: Proceedings of the Twenty-Seventh Annual ACM-SIAM Symposium
on Discrete Algorithms, SODA 2016, pp. 1306–1325. Society for Industrial and
Applied Mathematics, Philadelphia (2016)
5. Britton, T., Deijfen, M., Martin-Löf, A.: Generating simple random graphs with
prescribed degree distribution. J. Stat. Phys. 124(6), 1377–1397 (2006)
6. Chung, F., Lu, L.: The average distances in random graphs with given expected
degrees. Proc. Natl. Acad. Sci. USA 99(25), 15879–15882 (2002) (electronic)
7. Clauset, A., Shalizi, C.R., Newman, M.E.J.: Power-law distributions in empirical
data. SIAM Rev. 51(4), 661–703 (2009)
8. Faloutsos, M., Faloutsos, P., Faloutsos, C.: On power-law relationships of the inter-
net topology. ACM SIGCOMM Comput. Commun. Rev. 29, 251–262 (1999)
9. Fountoulakis, N., Friedrich, T., Hermelin, D.: On the average-case complexity of
parameterized clique. arXiv:1410.6400v1 (2014)
10. Fountoulakis, N., Friedrich, T., Hermelin, D.: On the average-case complexity of
parameterized clique. Theor. Comput. Sci. 576, 18–29 (2015)
11. Friedrich, T., Krohmer, A.: Cliques in hyperbolic random graphs. In: INFOCOM
Proceedings 2015, pp. 1544–1552. IEEE (2015)
12. Friedrich, T., Krohmer, A.: Parameterized clique on inhomogeneous random
graphs. Disc. Appl. Math. 184, 130–138 (2015)
13. Garey, M.R., Johnson, D.S., Garey, M.R.: Computers and Intractability: A Guide
to the Theory of NP-Completeness. W H FREEMAN & CO (2011)
14. Grochow, J.A., Kellis, M.: Network motif discovery using subgraph enumeration
and symmetry-breaking. In. RECOMB, pp. 92–106 (2007)
15. Heydari, H., Taheri, S.M.: Distributed maximal independent set on inhomogeneous
random graphs. In: 2017 2nd Conference on Swarm Intelligence and Evolutionary
Computation (CSIEC). IEEE, March 2017
Finding Induced Subgraphs in Scale-Free Inhomogeneous Random Graphs 15
16. van der Hofstad, R.: Random Graphs and Complex Networks, vol. 1. Cambridge
University Press, Cambridge (2017)
17. van der Hofstad, R., van Leeuwaarden, J.S.H., Stegehuis, C.: Optimal subgraph
structures in scale-free networks. arXiv:1709.03466 (2017)
18. Janson, S., L
uczak, T., Norros, I.: Large cliques in a power-law random graph. J.
Appl. Probab. 47(04), 1124–1135 (2010)
19. Jeong, H., Tombor, B., Albert, R., Oltvai, Z.N., Barabási, A.L.: The large-scale
organization of metabolic networks. Nature 407(6804), 651–654 (2000)
20. Karp, R.M.: Reducibility among combinatorial problems. In: Miller, R.E.,
Thatcher, J.W., Bohlinger, J.D. (eds.) Complexity of Computer Computations.
The IBM Research Symposia Series, pp. 85–103. Springer, Boston (1972). https://
doi.org/10.1007/978-1-4684-2001-2 9
21. Kashtan, N., Itzkovitz, S., Milo, R., Alon, U.: Efficient sampling algorithm for
estimating subgraph concentrations and detecting network motifs. Bioinformatics
20(11), 1746–1758 (2004)
22. Leskovec, J., Krevl, A.: SNAP Datasets: Stanford large network dataset collection
(2014). http://snap.stanford.edu/data. Accessed 14 Mar 2017
23. Niu, X., Sun, X., Wang, H., Rong, S., Qi, G., Yu, Y.: Zhishi.me - weaving chinese
linking open data. In: Aroyo, L., Welty, C., Alani, H., Taylor, J., Bernstein, A.,
Kagal, L., Noy, N., Blomqvist, E. (eds.) ISWC 2011. LNCS, vol. 7032, pp. 205–220.
Springer, Heidelberg (2011). https://doi.org/10.1007/978-3-642-25093-4 14
24. Norros, I., Reittu, H.: On a conditionally poissonian graph process. Adv. Appl.
Probab. 38(01), 59–75 (2006)
25. Omidi, S., Schreiber, F., Masoudi-Nejad, A.: MODA: an efficient algorithm for
network motif discovery in biological networks. Genes Genetic Syst. 84(5), 385–
395 (2009)
26. Park, J., Newman, M.E.J.: Statistical mechanics of networks. Phys. Rev. E 70,
066117 (2004)
27. Schreiber, F., Schwobbermeyer, H.: MAVisto: a tool for the exploration of network
motifs. Bioinformatics 21(17), 3572–3574 (2005)
28. Vázquez, A., Pastor-Satorras, R., Vespignani, A.: Large-scale topological and
dynamical properties of the internet. Phys. Rev. E 65, 066130 (2002)
29. Williams, V.V., Wang, J.R., Williams, R., Yu, H.: Finding four-node subgraphs in
triangle time. In: Proceedings of the Twenty-Sixth Annual ACM-SIAM Symposium
on Discrete Algorithms, SODA 2015, pp. 1671–1680. Society for Industrial and
Applied Mathematics, Philadelphia (2015)
The Asymptotic Normality of the Global
Clustering Coefficient in Sparse Random
Intersection Graphs
1 Introduction
The global clustering coefficient of a finite graph G is the ratio CG = 3NΔ /N∨ ,
where NΔ is the number of triangles and N∨ is the number of paths of length
2. Equivalently, CG represents the probability that a randomly selected path of
length 2 induces triangle in G. The global clustering coefficient is a commonly
used network characteristic, assessing the strength of the statistical association
between neighboring adjacency relations. For example, in a social network the
tendency of linking actors which have a common neighbor is reflected by a non-
negligible value of the global clustering coefficient.
Clustering in a social network can be explained by an auxiliary bipartite
structure: each actor is prescribed a collection of attributes and any two actors
sharing a common attribute have high chances of being adjacent, cf. [8]. The
respective random intersection graph (RIG) on the vertex set V = {v1 , . . . , vn }
and with the auxiliary attribute set W = {w1 , . . . , wm } defines adjacency rela-
tions with the help of a random bipartite graph H linking actors (=vertices) to
attributes: two actors are adjacent in RIG if they have a common neighbour in
H. We mention that RIG admits non-vanishing tunable global clustering coeffi-
cient, power-law degrees and short typical distances, see e.g., [4].
In this note we consider the uniform random intersection graph G(n, m, r),
where every vertex vi ∈ V is prescribed a random subset Si = S(vi ) ⊂ W of size r
and two vertices vi , vj are declared adjacent (denoted vi ∼ vj ) whenever Si ∩Sj =
∅. We assume that the sets S1 , . . . , Sn are independent. (The respective random
bipartite graph H is drawn uniformly at random from the class of bipartite
c Springer International Publishing AG, part of Springer Nature 2018
A. Bonato et al. (Eds.): WAW 2018, LNCS 10836, pp. 16–29, 2018.
https://doi.org/10.1007/978-3-319-92871-5_2
The Asymptotic Normality of the Global Clustering Coefficient 17
graphs with the property that each actor vi ∈ V has exactly r neighbours in
W .) The uniform random intersection graph has been widely studied in the
literature mainly as a model of secure wireless sensor network that uses random
predistribution of keys, see [5,14]. We denote for short G = G(n, m, r) and by G
we denote the instance (realization) of the random graph G.
We consider large random intersection graphs, where r2 = o(m) as
m, n → +∞. In this case the edge probability is, see (53),
Denote
N̄Δ = NΔ − ENΔ , N̄∨ = N∨ − EN∨ , σΔ
2 2
= EN̄Δ 2
, σ∨ = EN̄∨2 , σΔ∨ = E N̄Δ N̄∨ .
We start our analysis with an evaluation of the first and second moments of
the subgraph counts NΔ and N∨ .
The random variables gΔ1,2 , hΔ1,2,3 and g∨1,2 , h∨1,2,3 define the Hoeffding
decomposition of N̄Δ and N̄∨ , see (12). Their second moments entering (6),
(7), (8) are evaluated in (25), (26) and (31), (32) and (39), (40) respectively.
We note that (4) and (5) imply that the “theoretical clustering coefficient”
pΔ E(3NΔ ) 1
P Δi,j,k ∨ijk = = ≈ as n, m → +∞.
p∨ EN∨ r
Therefore, in order to have a non-vanishing global clustering coefficient we
need r to be bounded as n, m → +∞, cf. [3,13]. But we may still expect the
−1 −1
asymptotic normality of σΔ N̄Δ and σ∨ N̄∨ even for r → ∞ as n, m → +∞.
Indeed, assuming (2) we obtain from (4) for r3 = o(m) that
m r3
ENΔ ≈ 1 + → +∞ as n, m → +∞. (9)
6c3 r3 m
−1
Hence, for r3 = o(m) we can expect the asymptotic normality of σΔ N̄Δ . For
larger r such that m = O(r ) and r = o(m), the identity ENΔ = n3 pΔ
3 2
combined with (2) and the bound pΔ = O(r3 m−2 +r6 m−3 ) implies ENΔ = O(1).
−1
The latter bound rules out the asymptotic normality of σΔ N̄Δ . We refer to
Lemma 4 and the remark following it for various bounds on pΔ .
Our main result, Theorem 2 below gives sufficient conditions for the asymp-
totic normality of CG as n, m → +∞. We derive the asymptotic normality of CG
from a related asymptotic normality result for the bivariate vector of subgraph
counts (NΔ , N∨ ).
2 Proofs
∨ijk = I{vi ∼vj } I{vj ∼vk } , Δi,j,k = I{vi ∼vj } I{vj ∼vk } I{vk ∼vi } .
We note that g(Si , Sj ) := gi,j and h(Si , Sj , Sk ) := hi,j,k are symmetric functions
of their arguments Si , Sj and Si , Sj , Sk and they have the orthogonality property
(13) implies in particular that all distinct summands fi , gj1 ,j2 , hk1 ,k2 ,k3 are
uncorrelated whatever the indices i, j1 , j2 , k1 , k2 , k3 . A simple consequence of
(13) is the variance formula
2
n−1 n n
VarT = ET 2 = nEf12 + (n − 2)2 2
Eg1,2 + Eh21,2,3 . (14)
2 2 3
We construct decomposition (12) for T = N̄Δ and T = N̄∨ and use subscripts
Δ and ∨ to distinguish the respective terms ψΔ , fΔ j , gΔ i,j , hΔ i,j,k and ψ∨ , f∨ j ,
g∨ i,j , h∨ i,j,k .
Decomposition of N̄Δ . We put ψΔ (Si , Sj , Sk ) = Δi,j,k − pΔ and apply (12)
to T = N̄Δ . We shall show that for any j and k = j
fΔ j ≡ 0, (15)
r
gΔ j,k = I{s[j,k] =t} − p̄t pt . (16)
t=1
You may have seen a little leaf that has fallen into a stream and
been whirled along by the unresting current, torn and bruised and
helpless, then suddenly drift into a still and quiet pool and lie tranquil,
unvexed, while the stream, unable longer to clutch it, goes hurrying
by. So to Rose, after her troubled, changeful childhood, Farmdale
was the quiet pool, where she was to find a quiet, uneventful period.
Not that Rose ever thought of it as uneventful. To her school life
she brought an enthusiasm that never flagged; the school tests, the
class competitions, the school entertainments, the school games,
and even the school differences, she entered into them all heart and
soul. She studied hard, she took eager advantage of every
opportunity, and was none the less ready for every enjoyment with
the keen zest of her intense nature. Then outside the school was the
village with all its people and all their happenings, a little world of
itself. “Some of the girls call Farmdale dull and poky,” she repeated
wonderingly to Miss Silence. “I’m sure it isn’t dull to me—I don’t see
how they can think it is.”
The Blossom household quickly became home, and home folks to
Rose. But when Mrs. Blossom promised for her the same care she
would have given her own little Rachel, she included also, what she
would have expected of little Rachel had she lived, as she had of her
other daughters, the yielding of a ready, cheerful obedience. Mrs.
Blossom’s law was one Rose had known little of, the law of love, but
none the less was it law. Never in their girlhood, and hardly in their
maturer years, had Silence or Patience Blossom dreamed of acting
in opposition to their mother’s will—that reasonable, mild, but
inflexible will. And though Rose had not hesitated to face Mrs.
Hagood’s fury, yet when those clear, steadfast eyes looked into hers,
and that kindly but firm voice said, with its accent of decision, “Rose,
you cannot!” she instinctively realized that here was a force, the
force of moral strength, that impetuous willfulness would beat
powerless against. Nor was her affection for Mrs. Blossom any the
less sincere because of the obedient respect on which it was
founded.
Great-Uncle Samuel had been rightly informed that the Farmdale
high school was a good one, and the lessons Rose learned within its
walls were to her of value; but no less so was the unconscious
teaching of the pure and unselfish lives that were open before her
every day. Over an ardent young life, full of dreams and plans and
ambitions, all centered in self, a happier influence could not well
have fallen than that of these gentle, kindly women, whose spirit of
helpfulness and sympathy was always as ready and unfailing as the
flow of the fountain itself.
Was any one in distress, in perplexity, in trouble; there was no
counselor so wise, discreet, trustworthy, as Mrs. Blossom, who held
half the village secrets, and had served as a peacemaker times
without number. Was there a bride to be dressed; no one could do it
so well as Miss Silence or Mrs. Patience. Was any one sick; no
nurses were as tender and skillful and tireless as they. Did the
shadow of death rest over a home; no voices could speak words of
sweeter comfort to the dying, no other’s presence was so
unobtrusive, so helpful in the house of bereavement. Indeed, few
were the families in that little community to whom they were not
bound by the cords of a common sympathy in some hour of joy or
grief. And Rose was not the only one who often wondered how with
all the calls upon them they still managed to accomplish so much,
and with a manner so unhurried.
“I don’t see how you ever do it,” Rose exclaimed one day.
“It’s the busy people who find not only the most time but the most
happiness,” was Silence Blossom’s cheery answer.
And realizing, as she well did, how much more of real happiness
there was in the modest Blossom home than in the big Fifield house,
where no one ever thought of going to ask a service, and every life
was wholly self-centered, Rose could not but admit that this was
true.
“I don’t see what happiness you could find in sitting up all night
with Aunt Polly Brown,” she protested. “I’m sure I never want to go
where there’s sick people. I hope I’ll never be asked to.”
Already in that home where thoughtfulness for others was part of
the daily life, and interest in any who were suffering a matter of
course, it had come about naturally that Rose should be sent with a
handful of flowers, or some dainty for a sick neighbor, or was asked
to call at the door with a message of inquiry. So the next day she
took it as a matter of course when Miss Silence asked her to take a
bowl of chicken broth to Aunt Polly Brown.
“Take it right in to Aunt Polly,” said the young woman who opened
the door. “She’s in the bedroom right off the sitting-room.”
Rose hesitated. She would have refused if she had known exactly
how to do so. As it was, the bowl trembled a little as she walked
through into the bedroom, where on a high four-post bedstead,
under a “blazing star” quilt, Aunt Polly lay, a ruffled night cap
surrounding her shrunken face.
“Well, now,” as Rose told her errand, “it was reel kind of Silence
Blossom to send the broth. I was just thinkin’ that a taste o’ chicken
broth would relish. Sit down, won’t ye,” with a wistful accent, “and tell
me what’s goin’ on? Mary Jane never knows nothin’. Mebby I ain’t
goin’ to get well, but ’tany rate I like to know what folks is doin’.”
“I was standing on one foot wondering how quick I could get out,”
Rose said, relating it all to Miss Silence, on her return. “But when
she spoke that way I just thought that if I were old and sick I’d be
glad to have somebody come in; and I sat down and racked my brain
to tell everything I could think of. She seemed real cheered up when
I came away, and I promised her I’d come again.”
“I thought you never wanted to go where there were sick people,”
and Silence Blossom’s eyes twinkled.
“Well, it wasn’t so bad as I thought it was going to be, though her
hands are kind of skinny. And I don’t think I feel quite as I did about
sick folks now. Besides, it must be dreadful to lie in bed day after
day, and if I can make a little of the time pass, why I’m glad to.”
“There is where the gladness comes in,” said Mrs. Patience. “It is
making the hours of suffering a little brighter, a little easier. And now
you have learned this I think you will never forget it.”
“And I also remember that I promised to come down to Helen
Green’s to get out my Latin with her,” and gathering an armful of
books Rose hurried away.
“I am glad that Rose went in to see Aunt Polly; she is such a bit of
sunshine that she could not help but do her good. Besides, she has
always had such a morbid dread of a sick room,” Silence remarked
as she watched her away.
“I am glad, too,” agreed Mrs. Blossom, “for Rose can gain as well
as give. Of course I would not want her to go where there was any
danger, but her exuberant young nature will be made the deeper and
richer for being stirred and lifted out of itself.”
So among the threads of interest running from the Blossom home
Rose knit her threads. The people of Farmdale became her friends,
and because they were her friends she loved them, and so it was not
strange that she won love in return. With the Fifields her relations
through the years continued of the friendliest. On her part the
painfulness of being falsely accused had faded away; and on their
part the fact that it had been an unjust charge had not only made
them one and all feel that they owed her something in return, but had
awakened an interest in her that otherwise they might never have
felt. Miss Eudora regarded her in the light of a romance; Miss Jane
Fifield commended the fact that she was neither vain, nor, as she
was pleased to put it, “silly”; while Mr. Nathan, in his pride at Rose’s
persistence, and the quality he called her “grit,” went so far as to
freshen up the languages of his college days, that he might the more
help her.
At their time of life it was not to be expected that the Fifield nature
would greatly change; still their friendship for Rose, inexperienced
young girl though she was, brought a new and wholesome
atmosphere into the old house. Her flitting in and out, bright, breezy,
vivacious, was a welcome break in their old formality. A part of
Rose’s nature was her overflowing enthusiasm on the subject then in
mind; her studies, her school pleasures, whatever part was hers in
the life of the village, was all shared with her friends. So when she
came in beaming with excitement over the prettiness of the newest
Banby baby, Miss Fifield and Miss Eudora became conscious that
Mrs. Banby was a neighbor. Or if it were anxiety how little Mrs.
Mather, whose husband had just died and left her with five children,
was ever going to get through the winter; or rejoicing that Fanny
Barber, who had been so low with inflammatory rheumatism was
really improving, almost before they were aware, they would find
themselves becoming interested, an interest that could easily take
the form of a bundle of warm clothing for the widow, or a glass of
Miss Fifield’s famous quince jelly for the invalid. And so by the slight
touches of Rose’s hands they found themselves drawn gradually
from their cold isolation, and nearer to those about them.
CHAPTER XXIII
A VISIT FROM AN OLD FRIEND
It was the third May that Rose had been in Farmdale. The turf on
the open green was emerald velvet, the orchards were drifts of pink
and white, the lilacs by Mrs. Blossom’s gate were lifting spikes of
lavender, and shrubs and roses were heavy with the weight of bud or
bloom. In a swift rush Rose came down the walk, the white gate
clashed behind her, and she dashed into the house, rosy and
breathless with haste, waving a long envelope over her head.
“What do you think that is?” she cried.
Miss Silence glanced up from her sewing machine. “It looks to me
like an envelope.”
“And what do you think is inside it?” pursued Rose.
“A letter is usually inside an envelope,” answered Mrs. Patience.
“You won’t guess,” pouted Rose, “so I shall have to tell you, for I
couldn’t possibly keep it. This is my certificate that I have passed the
teachers’ examination I went to last week, and am duly qualified to
teach. Wish me joy!”
“But I thought thee went to the examination simply for the
practice,” said Grandmother Sweet.
“So I did. But all the same I wanted to pass, and was so afraid I
wouldn’t pass. That’s why I didn’t say more about it. And now that I
have a really, truly certificate to teach! I’m sure I’ve grown an inch
since I took it out of the post-office.”
“We are very glad you succeeded,” and Mrs. Patience held off a
hat to see if the bunch of flowers was in the right place.
“And that isn’t all,” Rose went on blithely. “You need sixteen points
to graduate from the high school, I have fourteen already, because
I’ve taken extra studies; to pass the teachers’ examination counts
two points, so now I can graduate this year.”
“But why do you want to graduate this year? I supposed of course
you were going one more,” and Silence looked her surprise.
“I want to get to teaching. I’m just crazy to begin.”
“Rose, Rose,” Mrs. Blossom in the next room had heard the
conversation, and now stepped to the doorway, “you are too young
to think of teaching; even if you are qualified you have not the self-
control a teacher needs.”
“Oh, don’t say that!” groaned Rose, “when I have struggled with
my temper, and prayed over it, and counted a hundred before I
spoke, and bitten my tongue till it bled, and did all the things I ever
heard of to hold on to myself.”
“And you have done very well,” commended Mrs. Blossom. “You
have overcome much, and learned some hard lessons in the bridling
of your quick tongue, and holding in check your temper. But you
have still more to learn, especially if you are going to teach. I know,
for I was a teacher myself, and while text-books and methods
change, boys and girls, as far as I can see, remain about the same.”
“All I ask is the chance to try some boys and girls.”
“Besides,” Mrs. Blossom’s voice was calmly even, “I do not think
you can teach, that any school board would hire a girl of seventeen.”
“But I know people who have taught when no older than that,”
persisted Rose.
“That might have been once but it is not now. Indeed I am quite
sure that a law has been passed in Ohio that a teacher cannot draw
pay unless she is over eighteen.”
“It is a mean old law,” scorned Rose.
“Another thing,” continued Mrs. Blossom, “your Uncle Samuel is
your guardian, and he did not expect, any more than we did, that you
would leave school till next year; and before taking such a step you
must consult him.”
“Great-Uncle Samuel won’t care,” urged Rose, “and I’ve set my
heart on getting through this year. Besides if I can’t teach I can go to
school another year, and take Latin and German, and review the
common branches.”
“You write to Mr. Jarvis first, and see what he says,” and Rose
knew further argument was useless.
Rose waited and fretted for two weeks before an answer to her
letter came, and when she read it she gave a gasp of surprise.
“What do you think?” she exclaimed. “Great-Uncle Samuel says I
have been a very prudent girl, while from my marks—you know I
have sent them to him every quarter—I seem to have made good
use of my opportunities; so if I will continue to be prudent he thinks
there will be money enough for me to go to college for four years.
This is what he writes: ‘Of course not to a big expensive college, that
would be quite beyond your means, the Fairville Woman’s College is
the one I have chosen for you. I am told that it is an excellent school,
that the location is healthy, and the moral tone excellent. That you
will make good use of its benefits I shall expect. Of course your Aunt
Sarah Hartly ought to have seen to this for you, but as long as she
wouldn’t I have done what seemed to me the best.’”
“Four years in college, will not that be fine?” Silence Blossom’s
own eyes were bright with pleasure.
“Yes, I suppose it will,” Rose spoke slowly. “But, you know, I never
had thought of such a thing as college being possible for me; I did
not think that there was money enough for that. Of course I shall like
it, the only thing is it will make me so old before I get to teaching.”
The older women looked at Rose’s face, that had never lost its
child expression, and laughed at her words.
“It may be though,” she went on, “that I can put in extra studies
and shorten the time.”
“No, no,” protested Mrs. Patience, “to do your best work you do
not want to hurry it.”
Grandmother Sweet stopped her knitting. “Rose, my husband
while a lad served five years as apprentice to a carpenter. His own
work was of the best, and he often said that time spent learning to
use one’s tools was time saved. Now, thee is planning to use books
as tools, and the better thee understands them the better work thee
will do.”
“Oh, of course,” Rose hastened to say, “now the chance has come
to me I wouldn’t miss it for anything. And I will make the best of it,
too. I’m going to send right away and get a prospectus of the college
to see what the entrance requirements are. I’m not going to be
conditioned, and I’d rather be a little ahead. I had planned anyway to
read Virgil this summer with Mr. Fifield, and I can study up whatever
else is needed.”
“I think if you are going to college this fall you will need to do some
sewing as well as studying,” suggested Miss Silence.
“Of course I shall. I know I can’t spend money for a great deal;
what I do have I want neat and in good shape. I’m so glad to know
about it now, for I can plan the dresses I will need when I graduate
from the high school so I can use them then.”
“How many will you need?” asked Silence Blossom.
“The other girls say three; a suit for the Baccalaureate sermon,
another for the senior reception, and the graduating dress.”
“That last will be white, and will answer for your best white dress
all the year, and if you get a pretty grey for your suit that will do for
fall wear.”
“That makes two new dresses,” reflected Rose. “I can’t afford any
more, and one other still to be evolved. I wish the waist wasn’t so
badly worn to the lavender and white striped silk Great-Aunt Sarah
sent in the last box; it would make a pretty dress, and I could mend
up the cream lace to trim it.”
Before Rose had ceased speaking Miss Silence was turning the
leaves of a fashion book. “There is a dress in this last number that I
believe we can copy, and use the purple silk she sent you once to
combine with it. The solid color will give it character, and the lace will
soften and keep it girlish.”
Rose was looking at the plate. “Yes, that will be pretty. You are the
very Wizard of Old Clothes. And if there are scraps enough of silk
and lace left I will make a little hat with purple violets for trimming to
wear with it.”
She paused and lifted an impressive finger. “But mind this, when I
get to earning for myself I will have some pretty dresses, and never
will I wear any more of Great-Aunt Sarah’s cast-offs!”
Mrs. Patience smiled indulgently. “You are young, Rose, it is only
natural you should feel so. But you know you are denying yourself
now so that day may come.”
“I know it,” Rose nodded. “When I have had to go without things I
wanted and that other girls did have, I’ve said, ‘Never mind, you are
having an education.’ I expect to have to say that pretty often when I
get to college—it’s hard to realize that I am going—but I’m not going
to forget that I’m working for a purpose.”
“And that’s better than fine clothes.”
Rose twisted her face. “I wouldn’t object to the fine clothes if I
could have them. But I suppose I shall need some dresses for
everyday wear; the blue dress I had last year will do for that, won’t
it?”
“Yes, and there is your green and red plaid. You can have some
separate waists, too. I’m sure, Rose, we can have your wardrobe in
shape, that if not fine, it will be neat and tasty.”
“What could I ever have done without you all?” Rose paused and
sighed. “I am glad that I can go to college. I shall be gladder the
longer I realize it. But I feel that it will just break my heart to leave
here. If I could only take you all with me or bring the college to
Farmdale.”
“We are glad that you can go to college, Rose,” Mrs. Blossom’s
voice had not quite its usual firmness, “but you may be sure of one
thing, we shall miss you more than you will us. But it is a long time till
September; we will not begin the parting yet.”
“And of course I shall come back in vacations; everybody goes
home then, and this is my home.”
“Do you think a college freshman will remember how to gather
eggs?” asked Mrs. Patience.
“This one will, you may be sure,” laughed Rose, “and how to make
omelet, and custard, and cake with them when they are gathered.
It’s a pity Great-Uncle Samuel never comes so I can show him how
you have taught me to cook.”
It was a busy summer for Rose; she went over all the studies in
which she would be examined for entrance to college, she sewed
and gathered and tucked and hemmed, and when the September
days came she packed her modest wardrobe in her new trunk with a
curious mingling of dread and delight; dread at leaving the life she
knew, the friends she had proved; delight in the new and wider world
opening before her.
There had been talk of Mrs. Patience going with Rose, but it had
not proved possible, so when one sunny September day the stage—
the same stage that had brought her to Farmdale, stopped at the
white gate, and her trunk was strapped on, with a mixture of tears
and smiles the good-bys were said, and Rose settled herself in the
same corner of the back seat she had occupied on that day which
now seemed so far, far in the past, no longer a forlorn little figure,
dingy, travel worn and friendless; but a trim young girl in a pretty grey
suit, leaning out and waving her handkerchief in answer to those
waved to her from nearly every house. For Rose’s friends included
almost every one in Farmdale, and all her friends were interested in
her start for college.
THE END