1: %%%%%%%

2: %

3: %  This is a LaTeX file containing the paper

4: %

5: %    Affine and toric hyperplane arrangements

6: %

7: %  by Richard Ehrenborg, Margaret Readdy and Michael Slone

8: %

9: %

10: %	Last edited: May 16, 2008

11: %

12: %%%%%%%

13: 

14: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

15: %

16: % pdf settings

17: %

18: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

19: %

20: %\pdfpagewidth=8.5truein

21: %\pdfpageheight=11truein

22: %

23: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

24: 

25: 

26: 

27: 

28: \documentclass[11pt]{article}

29: \usepackage{latexsym}

30: \usepackage{amsmath,color}

31: \usepackage{epsfig}

32: %%\usepackage{mnsymbol}

33: 

34: %%%%% Here is the command to make the graphics package

35: %%%%% to work with pdfLaTeX

36: %\ifx\pdftexversion\undefined

37: %  \usepackage[dvips]{graphics}

38: %\else

39: %  \usepackage[pdftex]{graphics}

40: %\fi

41: %%%%%

42: 

43: 

44: 

45: \setlength{\topmargin}{ -1.5cm}

46: \setlength{\oddsidemargin}{ -0.5cm}

47: \textwidth 17cm

48: \textheight 22.4cm

49: 

50: 

51: 

52: 

53: \renewcommand{\baselinestretch}{1.0}

54: 

55: 

56: 

57: \font\german = eufm10 scaled\magstep1

58: \font\Cp = msbm10

59: 

60: \newcommand{\Rrr}{\hbox{\Cp R}}

61: \newcommand{\Zzz}{\hbox{\Cp Z}}

62: 

63: \newcommand{\bigcenter}[1]{\begin{center} {\Large\bf #1} \end{center}}

64: \newcommand{\qed}{\mbox{$\Box$}\vspace{\baselineskip}}

65: 

66: \newenvironment{proof}{\noindent {\bf Proof:}}{{\qed}}

67: \newenvironment{proof_special}{\noindent {\bf Proof:}}{}

68: \newenvironment{proof_}[1]{\noindent {\bf #1}}{{\qed}}

69: \newenvironment{proof_special_}[1]{\noindent {\bf #1}}{}

70: 

71: \newtheorem{theorem}{Theorem}[section]

72: \newtheorem{proposition}[theorem]{Proposition}

73: \newtheorem{lemma}[theorem]{Lemma}

74: \newtheorem{corollary}[theorem]{Corollary}

75: \newtheorem{conjecture}[theorem]{Conjecture}

76: \newtheorem{remark}[theorem]{Remark}

77: \newtheorem{example}[theorem]{Example}

78: \newtheorem{continuation}[theorem]{Continuation of Example}

79: \newtheorem{definition}[theorem]{Definition}

80: 

81: \renewcommand{\theequation}{\thesection.\arabic{equation}}

82: %%%% The next three lines make a new section reset the equation counter

83: \makeatletter

84: \@addtoreset{equation}{section}

85: \makeatother

86: 

87: 

88: \newcommand{\onethingatopanother}[2]

89: {\genfrac{}{}{0pt}{}{#1}{#2}}

90: 

91: \newcommand{\hz}{\hat{0}}

92: \newcommand{\ho}{\hat{1}}

93: 

94: \newcommand{\tensor}{\otimes}

95: 

96: \newcommand{\ab}{\av\bv}

97: \newcommand{\av}{{\bf a}}

98: \newcommand{\bv}{{\bf b}}

99: \newcommand{\cd}{\cv\dv}

100: \newcommand{\ctd}{\cv\mbox{-}2\dv}

101: \newcommand{\cv}{{\bf c}}

102: \newcommand{\dv}{{\bf d}}

103: 

104: \newcommand{\id}{\operatorname{id}}

105: \newcommand{\wt}{\operatorname{wt}}

106: \newcommand{\lcm}{\operatorname{lcm}}

107: 

108: 

109: \newcommand{\HH}{{\mathcal H}}

110: \newcommand{\LL}{{\mathcal L}}

111: \newcommand{\Lzero}{\LL \cup \{\hz\}}

112: \newcommand{\Lub}{\LL_{ub}}

113: \newcommand{\PP}{{\mathcal P}}

114: \newcommand{\Pzero}{\PP \cup \{\hz\}}

115: 

116: 

117: \newcommand{\zab}{\Zzz\langle\av,\bv\rangle}

118: \newcommand{\zcd}{\Zzz\langle\cv,\dv\rangle}

119: 

120: \newcommand{\tensordots}{\tensor \cdots \tensor}

121: \newcommand{\cupdots}{\cup \cdots \cup}

122: \newcommand{\capdots}{\cap \cdots \cap}

123: 

124: \newcommand{\pair}[2]{\left\langle#1 \: \vrule \: #2\right\rangle}

125: 

126: 

127: \newcommand{\zub}{z_{ub}}

128: \newcommand{\Zub}{Z_{ub}}

129: \newcommand{\zt}{z_{t}}

130: \newcommand{\Zt}{Z_{t}}

131: 

132: 

133: \newcommand{\ma}{\av}

134: \newcommand{\mb}{\bv}

135: \newcommand{\maa}{\av^{2}}

136: \newcommand{\mab}{\av\bv}

137: \newcommand{\mba}{\bv\av}

138: \newcommand{\mbb}{\bv^{2}}

139: \newcommand{\maaa}{\av^{3}}

140: \newcommand{\maab}{\av^{2}\bv}

141: \newcommand{\naab}{\av\av\bv}

142: \newcommand{\maba}{\av\bv\av}

143: \newcommand{\mabb}{\av\bv^{2}}

144: \newcommand{\nabb}{\av\bv\bv}

145: \newcommand{\mbaa}{\bv\av^{2}}

146: \newcommand{\nbaa}{\bv\av\av}

147: \newcommand{\mbab}{\bv\av\bv}

148: \newcommand{\mbba}{\bv^{2}\av}

149: \newcommand{\nbba}{\bv\bv\av}

150: \newcommand{\mbbb}{bv^{3}}

151: 

152: \newcommand{\mc}{\cv}

153: 

154: \newcommand{\mcc}{\cv^2}

155: \newcommand{\md}{\dv}

156: 

157: \newcommand{\mccc}{\cv^3}

158: \newcommand{\mcd}{\cv \dv}

159: \newcommand{\mdc}{\dv \cv}

160: 

161: 

162: 

163: 

164: \parskip=12pt

165: 

166: \begin{document}

167: 

168: 

169: 

170: \title{Affine and toric hyperplane arrangements}

171: \author{{\sc Richard EHRENBORG,}

172:         {\sc Margaret READDY}

173:           and

174:         {\sc Michael SLONE}}

175: \date{}

176: %%\date{\today}

177: \maketitle

178: 

179: 

180: 

181: 

182: 

183: 

184: 

185: 

186: 

187: \begin{abstract}

188: We extend the Billera--Ehrenborg--Readdy

189: map between the intersection lattice

190: and face lattice of

191: a central hyperplane arrangement to

192: affine and toric hyperplane arrangements.

193: For arrangements on the torus,

194: we also generalize Zaslavsky's fundamental

195: results on the number of regions.

196: \end{abstract}

197: 

198: 

199: \section{Introduction}

200: \label{section_introduction}

201: 

202: 

203: 

204: 

205: Traditionally combinatorialists have studied topological objects

206: that are spherical, such as polytopes, or which are homeomorphic to

207: a wedge of spheres, such as those obtained from shellable complexes.

208: In this paper we break from this practice and study hyperplane 

209: arrangements on the $n$-dimensional torus.

210: 

211: 

212: It is classical that the convex hull

213: of a finite collection of points in Euclidean space

214: is a polytope

215: and its boundary is a sphere.

216: The key ingredient in this construction is convexity.

217: At the moment there is no natural analogue of this process

218: to obtain a complex whose geometric realization is

219: a torus.

220: 

221: 

222: 

223: 

224: 

225: In this paper we are taking a zonotopal approach to 

226: working with arrangements on the torus.

227: Recall that a zonotope can be defined without the notion

228: of convexity, that is, it is a Minkowski sum of line segments.

229: Dually, a central hyperplane arrangement gives rise to

230: a spherical cell complex.

231: By considering an arrangement on the torus, we are able to

232: obtain a subdivision whose geometric realization is indeed the torus.

233: We will see later in

234: Section~\ref{section_toric}

235: that

236: this amounts to restricting ourselves to arrangements

237: whose subspaces

238: in the Euclidean space $\Rrr^{n}$

239: have coefficient matrices with rational entries.

240: Under the quotient map

241: $\Rrr^{n} \longrightarrow \Rrr^{n}/\Zzz^{n} = T^{n}$

242: these subspaces 

243: are sent to subtori of the $n$-dimensional torus $T^{n}$.

244: 

245: 

246: 

247: 

248: Zaslavsky

249: initiated the modern study of hyperplane arrangements in his

250: fundamental treatise~\cite{Zaslavsky}.

251: For early work in the field,

252: see the references given

253: in Gr\"unbaum's text~\cite[Chapter 18]{Grunbaum}.

254: Zaslavsky showed that evaluating the characteristic

255: polynomial of a central hyperplane arrangement at

256: $-1$ gives the number of regions

257: in the complement of the arrangement.

258: For central hyperplane arrangements,

259: Bayer and Sturmfels~\cite{Bayer_Sturmfels} proved the flag $f$-vector

260: of the arrangement can be determined from the intersection

261: lattice;

262: see Theorem~\ref{theorem_Bayer_Sturmfels}.

263: However, their result is stated as a sum

264: of chains in the intersection lattice and hence

265: it is hard to apply.

266: Billera, Ehrenborg and Readdy

267: improved the 

268: Bayer--Sturmfels result by showing that it is enough

269: to know the flag $f$-vector of the intersection lattice

270: to compute the flag $f$-vector of a central arrangement.

271: Recall that the $\cd$-index of a regular

272: cell complex is an efficient tool to encode its

273: flag $f$-vector without linear redundancies~\cite{Bayer_Klapper}.

274: The Billera--Ehrenborg--Readdy theorem gives an explicit way to compute

275: the $\cd$-index of the arrangement,

276: and hence its flag $f$-vector~\cite{Billera_Ehrenborg_Readdy_om}.

277: 

278: 

279: 

280: We generalize Zaslavsky's theorem on the number

281: of regions of a hyperplane arrangement to the toric case.

282: Although there is no intersection lattice per se,

283: one works with the intersection poset.

284: From the Zaslavsky result we obtain a toric

285: version of the Bayer--Sturmfels result for

286: hyperplane arrangements,

287: that is, there is a natural poset map from

288: the face poset to the intersection poset,

289: and furthermore, the cardinality

290: of the inverse image of a chain under this map

291: is described.

292: 

293: 

294: 

295: 

296: As in the case of a central hyperplane 

297: arrangement, our toric version of the Bayer--Sturmfels result determines

298: the flag $f$-vector of the face poset of

299: a toric arrangement in terms of its intersection poset.

300: However, this is far from being explicit.

301: Using the coalgebraic techniques

302: from~\cite{Ehrenborg_Readdy_c},

303: we are able to determine the flag $f$-vector

304: explicitly in terms of the flag $f$-vector

305: of the intersection poset. 

306: Moreover, the 

307: answer is given by a $\cd$ type of polynomial.

308: The flag $f$-vector of a regular spherical complex

309: is encoded by the $\cd$-index, a non-commutative polynomial

310: in the variables $\cv$ and~$\dv$,

311: whereas

312: the $n$-dimensional

313: toric analogue

314: is a $\cd$-polynomial plus the $\ab$-polynomial $(\av-\bv)^{n+1}$.

315: 

316: 

317: 

318: 

319: Zaslavsky also showed 

320: that evaluating the characteristic

321: polynomial of an affine arrangement at

322: $1$ gives the number of bounded regions

323: in the complement of the arrangement.

324: Thus we return to affine arrangements in

325: Euclidean space with the twist that

326: we study the {\em unbounded} regions.

327: The unbounded regions form a spherical

328: complex.

329: In the case of central arrangements,

330: this complex is exactly what was studied

331: previously  

332: by Billera, Ehrenborg and Readdy~\cite{Billera_Ehrenborg_Readdy_om}.

333: For non-central arrangements,

334: we determine the $\cd$-index of

335: this complex in terms of the lattice

336: of unbounded intersections of the arrangement.

337: 

338: 

339: Interestingly, the techniques for studying toric arrangements and

340: the unbounded complex of 

341: non-central arrangements are similar.

342: Hence, we present these

343: results

344: in the same paper.

345: For example, 

346: the toric and non-central

347: analogues of the Bayer--Sturmfels theorem

348: only differ

349: by which Zaslavsky invariant is used.

350: The coalgebraic translations of the

351: two analogues

352: involve exactly the same argument,

353: and

354: the resulting underlying maps

355: $\varphi_{t}$ (in the toric case) and 

356: $\varphi_{ub}$ (in the non-central case)

357: differ only slightly

358: in their definitions.

359: 

360: 

361: We end with many open questions about subdivisions of manifolds.

362: 

363: 

364: 

365: 

366: 

367: 

368: \section{Preliminaries}

369: \label{section_preliminaries}

370: 

371: 

372: 

373: All the posets we will work with are

374: graded,

375: that is,

376: posets having a unique minimal element

377: $\hz$, a

378: unique maximal element

379: $\ho$,

380: and rank function $\rho$.

381: For two elements $x$ and $z$ in a graded poset $P$ such that $x \leq z$,

382: let $[x,z]$ denote the interval

383: $\{y \in P \: : \: x \leq y \leq z\}$.

384: Observe that the interval $[x,z]$ is itself a graded poset.

385: Given a graded poset $P$ of rank $n+1$

386: and

387: $S \subseteq \{1, \ldots, n\}$,

388: the

389: {\em $S$-rank-selected poset $P(S)$}

390: is

391: the poset

392: consisting of the elements

393: $P(S) = \{x \in P \: : \: \rho(x) \in S\} \cup \{\hz, \ho\}$.

394: The partial order of

395: $[x,y]$ and $P(S)$ are each inherited from that

396: of $P$.

397: For a graded poset $P$ let $P^{*}$ denote the dual

398: poset, that is, the poset having

399: the same underlying set as $P$ but with the

400: order relation reversed:

401: $x <_{P^{*}} y$ if and only if $y <_{P} x$.

402: For standard poset terminology,

403: we refer the reader

404: to Stanley's work~\cite{Stanley_EC_I}.

405: 

406: 

407: The M\"obius function $\mu(x,y)$ on a poset $P$ is

408: defined recursively by

409: $\mu(x,x) = 1$

410: and for elements $x, y \in P$ with $x < y$ by 

411: $\mu(x,y) = - \sum_{x \leq z < y} \mu(x,z)$;

412: see Section~3.7 in~\cite{Stanley_EC_I}.

413: For a graded poset~$P$ 

414: with minimal element $\hz$ and maximal element $\ho$

415: we write $\mu(P) = \mu_{P}(\hz,\ho)$.

416: 

417: 

418: 

419: 

420: 

421: We now review important  results about

422: hyperplane arrangements, the $\cd$-index

423: and coalgebraic techniques.

424: All are essential

425: for proving the main results of this paper.

426: 

427: 

428: 

429: 

430: 

431: 

432: \subsection{Hyperplane arrangements}

433: 

434: 

435: Let $\HH = \{H_{1}, \ldots, H_{m}\}$

436: be a hyperplane arrangement in $\Rrr^{n}$,

437: that is,

438: a finite collection of affine hyperplanes in 

439: $n$-dimensional Euclidean space $\Rrr^{n}$.

440: For brevity, throughout this paper we will 

441: often refer to a

442: hyperplane arrangement as an arrangement.

443: We call an arrangement {\em essential} if

444: the normal vectors to the hyperplanes in $\HH$

445: span $\Rrr^{n}$.

446: An arrangement that is not essential can be made essential by

447: quotienting out by the subspace $V^{\perp}$ where

448: $V$ is the subspace orthogonal to 

449: all of the hyperplanes in $\HH$.

450: In this paper we are only interested

451: in essential arrangements.

452: 

453: 

454: 

455: Observe that the intersection 

456: $\bigcap_{i=1}^{m} H_{i}$

457: of all of the hyperplanes in an essential arrangement is either

458: the empty set $\emptyset$ or a singleton point.

459: We call an arrangement {\em central} if the

460: intersection of all the hyperplanes is one point.

461: We may assume that this point is the origin ${\mathbf 0}$

462: and hence all of the hyperplanes are codimension $1$

463: subspaces.

464: If the intersection is the empty set,

465: we call the arrangement {\em non-central}.

466: 

467: 

468: 

469: 

470: The {\em intersection lattice} $\LL$

471: is the lattice formed by ordering all the intersections

472: of hyperplanes in~$\HH$ by reverse inclusion.

473: If the intersection

474: of all the hyperplanes

475: in a given arrangement is empty,

476: then we include the empty set

477: $\emptyset$ as the 

478: the maximal element in the intersection lattice.

479: If the arrangement is central, the

480: maximal element

481: is  $\{{\mathbf 0}\}$.

482: In all cases, the minimal element

483: of $\LL$ will be all of

484: $\Rrr^{n}$.

485: 

486: 

487: 

488: 

489: 

490: 

491: For a hyperplane arrangement $\HH$

492: with intersection lattice $\LL$, the

493: {\em characteristic polynomial} is defined by

494: $$   \chi(\HH; t)

495:    =

496:      \sum_{\onethingatopanother{x \in \LL}{x \neq \emptyset}}

497:              \mu(\hz,x) \cdot t^{\dim(x)}  ,   $$

498: where $\mu$ denotes the

499: M\"obius function.

500: The characteristic polynomial is a combinatorial

501: invariant of the arrangement. 

502: The fundamental result of Zaslavsky~\cite{Zaslavsky}

503: is that this

504: invariant determines the number and type of

505: regions.

506: \begin{theorem}[Zaslavsky]

507: For a hyperplane arrangement $\HH$ in $\Rrr^{n}$

508: the number of

509: regions is given by $(-1)^{n} \cdot \chi(\HH; t=-1)$.

510: Furthermore,

511: the number of

512: bounded regions is given by $(-1)^{n} \cdot \chi(\HH; t=1)$.

513: \label{theorem_Zaslavsky}

514: \end{theorem}

515: For a graded poset $P$,

516: define the two Zaslavsky invariants $Z$ and $Z_{b}$

517: by

518: \begin{eqnarray*}

519:       Z(P) 

520:   & = &

521:       \sum_{\hz \leq x \leq \ho} (-1)^{\rho(x)} \cdot \mu(\hz,x)  ,\\

522:       Z_{b}(P) 

523:   & = &

524:       (-1)^{\rho(P)} \cdot \mu(P)    .

525: \end{eqnarray*}

526: In order to work with Zaslavsky's result, we need the following

527: reformulation of Theorem~\ref{theorem_Zaslavsky}.

528: \begin{theorem}

529: \begin{enumerate}

530: \item[(i)]

531: For a central hyperplane arrangement the number of

532: regions is given by $Z(\LL)$, where $\LL$

533: is the intersection lattice of the arrangement.

534: 

535: \item[(ii)]

536: For a non-central hyperplane arrangement the number of

537: regions is given by $Z(\LL) - Z_{b}(\LL)$,

538: where $\LL$

539: is the intersection lattice of the arrangement.

540: The number of bounded regions is given by~$Z_{b}(\LL)$.

541: \end{enumerate}

542: \label{theorem_Zaslavsky_poset}

543: \end{theorem}

544: 

545: 

546: Given a central hyperplane arrangement $\HH$

547: there are two associated lattices, namely,

548: the 

549: intersection lattice $\LL$

550: and the 

551: lattice $T$ of faces of the arrangement.

552: The minimal element of $T$ is the empty set $\emptyset$

553: and the maximal element is the whole space $\Rrr^{n}$.

554: The lattice of faces can be seen as the face poset

555: of the cell complex obtained by intersecting the arrangement

556: $\HH$ with a sphere of radius $R$

557: centered at the origin.

558: Each hyperplane corresponds to a great circle on the sphere.

559: An alternative way to view the lattice of faces $T$

560: is that the dual lattice $T^{*}$ is the 

561: face lattice of the zonotope corresponding to $\HH$.

562: 

563: 

564: 

565: 

566: Let $\Lzero$ denote the intersection lattice

567: with a new minimal element $\hz$ adjoined.

568: Define an order- and rank-preserving map $z$

569: from the dual lattice $T^{*}$ to 

570: the augmented lattice $\Lzero$

571: by sending a face of the arrangement, that is, a cone in

572: $\Rrr^{n}$, to its affine hull.

573: Note that under the map $z$ the minimal

574: element of $T^{*}$ is mapped to the 

575: minimal element of $\Lzero$.

576: Observe that $z$ maps chains to chains.

577: Hence we view $z$ as a map from the set

578: of chains of $T^{*}$

579: to the set of chains of $\Lzero$.

580: Bayer and Sturmfels~\cite{Bayer_Sturmfels}

581: proved the following result about the

582: inverse image of a chain under the map $z$.

583: \begin{theorem}[Bayer--Sturmfels]

584: Let $\HH$ be a central hyperplane arrangement

585: with intersection lattice $\LL$.

586: Let $c = \{\hz = x_{0} < x_{1} < \cdots < x_{k} = \ho\}$

587: be a chain in $\Lzero$. Then the cardinality

588: of the inverse image of the chain $c$ 

589: under the map $z : T^{*} \longrightarrow \Lzero$

590: is given by the product

591: $$      |z^{-1}(c)|

592:     =

593:         \prod_{i=2}^{k}

594:                Z([x_{i-1},x_{i}])   .   $$

595: \label{theorem_Bayer_Sturmfels}

596: \end{theorem}

597: 

598: 

599: \subsection{The $\cd$-index}

600: 

601: 

602: Let $P$ be a graded poset of rank $n+1$

603: with rank function $\rho$.

604: For $S = \{s_{1} < \cdots < s_{k-1}\}$

605: a subset of

606: $\{1, \ldots, n\}$ define $f_{S}$ to be

607: the number of chains

608: $c = \{\hz = x_{0} < x_{1} < \cdots < x_{k} = \ho\}$

609: that have elements with ranks in the set $S$,

610: that is, 

611: $$    f_{S}

612:    =

613:       |\{ c \:\: : \:\:

614:           \rho(x_{1}) = s_{1}, \ldots, 

615:           \rho(x_{k-1}) = s_{k-1} \}|   .  $$

616: Observe that $f_{S}$ is the number

617: of maximal chains in the rank-selected poset $P(S)$.

618: The flag $h$-vector is obtained by the relation

619: (here we also present its inverse)

620: $$       h_{S} 

621:       =

622:          \sum_{T \subseteq S}

623:              (-1)^{|S-T|} \cdot f_{T} 

624:  \:\:\:\: \mbox{ and } \:\:\:\:

625:          f_{S} 

626:       =

627:          \sum_{T \subseteq S}

628:              h_{T}               .  $$

629: Recall that by Philip Hall's theorem

630: the M\"obius function of the $S$-rank-selected poset

631: $P(S)$ is given by

632: $\mu(P(S)) = (-1)^{|S|-1} \cdot h_{S}$.

633: 

634: Let $\av$ and $\bv$ be two non-commutative variables

635: each having degree $1$.

636: For $S$ a subset of $\{1, \ldots, n\}$ let $u_{S}$ be the monomial

637: $u_{S} = u_{1} \cdots u_{n}$ where

638: $u_{i} = \bv$ if $i \in S$

639: and

640: $u_{i} = \av$ if $i \not\in S$.

641: Then the {\em $\ab$-index} is the noncommutative polynomial defined by

642: $$  \Psi(P) = \sum_{S} h_{S} \cdot u_{S} ,  $$

643: where the sum is over all subsets $S \subseteq \{1, \ldots, n\}$.

644: Observe that the $\ab$-index of a poset $P$ of rank $n+1$

645: is a homogeneous polynomial of degree $n$.

646: 

647: 

648: 

649: 

650: A poset $P$ is {\em Eulerian} if

651: every interval $[x,y]$, where $x < y$,

652: satisfies the Euler-Poincar\'e

653: relation, that is, there are  the same number of elements

654: of odd as even rank.

655: Equivalently, the M\"obius function of $P$ is given by

656: $\mu(x,y) = (-1)^{\rho(x,y)}$

657: for all $x \leq y$ in $P$.

658: The quintessential result is that

659: the $\ab$-index 

660: of Eulerian posets has the following

661: form.

662: \begin{theorem}

663: The $\ab$-index of an Eulerian poset $P$

664: can be expressed in terms of

665: the noncommutative variables

666: $\cv = \av + \bv$

667: and $\dv = \av \bv + \bv \av$.

668: \end{theorem}

669: This theorem

670: was originally proved for face lattices of convex polytopes

671: by Bayer and Klapper~\cite{Bayer_Klapper}.

672: Stanley provided a proof for all

673: Eulerian posets~\cite{Stanley_d}.

674: There are proofs which have both used and revealed  the underlying

675: algebraic structure.

676: See for instance~\cite{Ehrenborg_k-Eulerian,Ehrenborg_Readdy_homology}.

677: When the $\ab$-index $\Psi(P)$ is written in terms of

678: $\cv$ and $\dv$,

679: the resulting polynomial is called the {\em $\cd$-index}.

680: There are linear relations 

681: holding among the entries of the flag $f$-vector

682: of an Eulerian poset, known as the generalized

683: Dehn-Sommerville relations; see~\cite{Bayer_Billera}.

684: The importance of the $\cd$-index is that it removes all of these

685: linear redundancies among the flag $f$-vector entries.

686: 

687: 

688: 

689: Observe that the variables $\cv$ and $\dv$ have degrees

690: $1$ and $2$, respectively.

691: Thus the $\cd$-index of a poset of rank $n+1$ is

692: a homogeneous polynomial of degree $n$ in the

693: noncommutative variables $\cv$ and $\dv$.

694: Define the reverse of an $\ab$-monomial

695: $u = u_{1} u_{2} \cdots u_{n}$ to be

696: $u^{*} = u_{n} \cdots u_{2} u_{1}$

697: and extend by linearity to an involution on $\zab$.

698: Since $\cv^{*} = \cv$ and $\dv^{*} = \dv$,

699: this involution applied to a $\cd$-monomial

700: simply reverses the $\cd$-monomial.

701: Finally, for a graded poset $P$

702: we have  $\Psi(P)^{*} = \Psi(P^{*})$.

703: 

704: 

705: 

706: 

707: 

708: 

709: A direct approach to describe the $\ab$-index of a

710: poset $P$ is to give each chain a weight and

711: then sum over all chains.

712: For a chain

713: $c = \{\hz = x_{0} < x_{1} < \cdots < x_{k} = \ho\}$

714: in the poset $P$, define its {\em weight} to be

715: \begin{equation}

716:       \wt(c)

717:    =

718:       (\av-\bv)^{\rho(x_{0},x_{1})-1}

719:         \cdot

720:       \bv

721:         \cdot

722:       (\av-\bv)^{\rho(x_{1},x_{2})-1}

723:         \cdot

724:       \bv

725:         \cdots

726:       \bv

727:         \cdot

728:       (\av-\bv)^{\rho(x_{k-1},x_{k})-1}  ,

729: \label{equation_weight}

730: \end{equation}

731: where $\rho(x,y)$ denotes the rank difference

732: $\rho(y) - \rho(x)$.

733: Then $\ab$-index of $P$ is given by

734: $$

735:      \Psi(P) = \sum_{c} \wt(c),

736: $$

737: where the sum is over all chains $c$ in the poset $P$.

738: 

739: Finally, a third description of the $\ab$-index is

740: Stanley's recursion for the $\ab$-index of a graded

741: poset~\cite[Equation~(7)]{Stanley_d}.  It is:

742: \begin{equation}

743: \label{equation_Stanley_recursion}

744:  \Psi(P) = (\av-\bv)^{\rho(P)-1}

745:          + \sum_{\hz < x < \ho}

746:               (\av-\bv)^{\rho(x)-1} \cdot \bv \cdot \Psi([x,\ho]).

747: \end{equation}

748: The initial condition for

749:  this recursion is the unique poset of rank $1$,

750: $B_{1}$, where $\Psi(B_{1}) = 1$.

751: 

752: 

753: 

754: 

755: 

756: \subsection{Coalgebraic techniques}

757: 

758: A coproduct $\Delta$ on a free $\Zzz$-module $C$

759: is a linear map $\Delta : C \longrightarrow C \tensor C$.

760: In order to be explicit, we use the Sweedler notation~\cite{Sweedler}

761: for writing the coproduct. To explain this notation, notice that

762: $\Delta(w)$ is an element of $C \tensor C$ and thus has the form

763: $$   \Delta(w) = \sum_{i=1}^{k} w_{1}^{i} \tensor w_{2}^{i}   ,  $$

764: where $k$ is the number of terms and $w_{1}^{i}$ and $w_{2}^{i}$

765: belong to $C$. Since all the maps that are applied to $\Delta(w)$

766: treat each term the same, the Sweedler notation drops the index $i$

767: and one writes

768: $$   \Delta(w) = \sum_{w} w_{(1)} \tensor w_{(2)}   .  $$

769: Informally, this sum should be thought of as all the ways of breaking

770: the element $w$ in two pieces, where the

771: first piece is denoted by  $w_{(1)}$

772: and the second by $w_{(2)}$.

773: The Sweedler notation for the expression

774: $(\Delta \tensor \id) \circ \Delta$,

775: where $\id$ denotes the identity map,

776: is the following

777: $$    ((\Delta \tensor \id) \circ \Delta)(w)

778:    =

779:       \sum_{w} \sum_{w_{(1)}}

780:          w_{(1,1)} \tensor w_{(1,2)} \tensor w_{(2)}  .  $$

781: The right-hand side should be thought of as

782: first breaking $w$ into the two pieces $w_{(1)}$ and $w_{(2)}$

783: and then breaking

784: $w_{(1)}$ into the two pieces $w_{(1,1)}$ and $w_{(1,2)}$.

785: See Joni and Rota for a more detailed explanation~\cite{Joni_Rota}.

786: 

787: 

788: 

789: 

790: The coproduct $\Delta$ is coassociative

791: if

792: $(\Delta \tensor \id) \circ \Delta = 

793:  (\id \tensor \Delta) \circ \Delta$.

794: The Sweedler notation expresses coassociativity as

795: $$   \sum_{w} \sum_{w_{(1)}}

796:          w_{(1,1)} \tensor w_{(1,2)} \tensor w_{(2)}

797:    =

798:      \sum_{w} \sum_{w_{(2)}}

799:          w_{(1)} \tensor w_{(2,1)} \tensor w_{(2,2)}  .  $$

800: Informally coassociativity states that 

801: all the possible ways to break $w$ into two pieces and

802: then breaking the first piece

803: into the two pieces

804: is equivalent to

805: all the ways to break $w$ into two pieces and

806: then break the second piece

807: into two pieces.

808: Compare coassociativity with associativity of

809: a multiplication map $m: A \tensor A \longrightarrow A$

810: on an algebra $A$.

811: 

812: 

813: Assuming coassociativity, Sweedler notation

814: simplifies to

815: $$   \Delta^{2}(w) = \sum_{w} w_{(1)} \tensor w_{(2)} \tensor w_{(3)} , $$

816: where $\Delta^{2}$ is defined as

817: $(\Delta \tensor \id) \circ \Delta = (\id \tensor \Delta) \circ \Delta$,

818: and the three pieces have been renamed as

819: $w_{(1)}$, $w_{(2)}$ and $w_{(3)}$.

820: Coassociativity allows one to define

821: the $k$-ary coproduct

822: $\Delta^{k-1} : C \longrightarrow C^{\tensor k}$

823: by the recursion

824: $\Delta^{0} = \id$

825: and

826: $\Delta^{k} = (\Delta^{k-1} \tensor \id) \circ \Delta$.

827: The Sweedler notation for the $k$-ary coproduct is

828: $$   \Delta^{k-1}(w)

829:    = 

830:      \sum_{w}

831:          w_{(1)} \tensor w_{(2)} \tensordots w_{(k)}  .  $$

832: 

833: 

834: Let $\zab$ denote the polynomial ring in the non-commutative

835: variables $\av$ and $\bv$.

836: We define a coproduct~$\Delta$ on the algebra $\zab$

837: by letting $\Delta$ satisfy the following

838: identities:

839: $\Delta(1) = 0$, $\Delta(\av) = \Delta(\bv) = 1 \tensor 1$

840: and the Newtonian condition 

841: \begin{equation}

842:       \Delta(u \cdot v)

843:     =

844:       \sum_{u} u_{(1)} \tensor u_{(2)} \cdot v

845:     +

846:       \sum_{v} u \cdot v_{(1)} \tensor v_{(2)}    .  

847: \label{equation_Newtonian}

848: \end{equation}

849: For an $\ab$-monomial $u = u_{1} u_{2} \cdots u_{n}$

850: we have that

851: $$    \Delta(u)

852:     =

853:       \sum_{i=1}^{n}

854:                u_{1} \cdots u_{i-1}

855:             \tensor

856:                u_{i+1} \cdots u_{n}    .    $$

857: The fundamental result for this coproduct is that

858: the $\ab$-index is a coalgebra homomorphism~\cite{Ehrenborg_Readdy_c}.

859: We express this result as the following identity.

860: \begin{theorem}[Ehrenborg--Readdy]

861: For a graded poset $P$

862: with $\ab$-index $w = \Psi(P)$ and for any $k$-multilinear map~$M$ 

863: on $\zab$, 

864: the following coproduct identity holds:

865: $$   \sum_{c}

866:            M(\Psi([x_{0},x_{1}]),

867:              \Psi([x_{1},x_{2}]),

868:                 \ldots,

869:              \Psi([x_{k-1},x_{k}]))

870:   =

871:      \sum_{w}

872:            M(w_{(1)},

873:              w_{(2)},

874:                 \ldots,

875:              w_{(k)})    ,  $$

876: where the first sum is over all

877: chains $c = \{\hz = x_{0} < x_{1} < \cdots < x_{k} = \ho\}$

878: of length $k$ 

879: and the second sum is the Sweedler notation of the $k$-ary coproduct

880: of $w$, that is, $\Delta^{k-1}$.

881: \label{theorem_Ehrenborg_Readdy}

882: \end{theorem}

883: 

884: 

885: 

886: \subsection{The $\cd$-index of the face poset of a central arrangement}

887: 

888: 

889: We recall the definition of the 

890: omega map~\cite{Billera_Ehrenborg_Readdy_om}.

891: 

892: \begin{definition}

893: The linear map $\omega$ from $\zab$ to $\zcd$

894: is formed by 

895: first replacing every occurrence of $\av\bv$ 

896: in a given 

897: $\ab$-monomial 

898: by $2\dv$

899: and then replacing the remaining letters by $\cv$.

900: \end{definition}

901: For a central hyperplane arrangement~$\HH$

902: the $\cd$-index of the face poset is computed

903: as follows~\cite{Billera_Ehrenborg_Readdy_om}.

904: \begin{theorem}[Billera--Ehrenborg--Readdy]

905: Let $\HH$ be

906: a central hyperplane arrangement

907: with intersection lattice $\LL$

908: and face lattice $T$. Then the $\cd$-index

909: of the

910: face lattice $T$ is given by

911: $$  \Psi(T) = \omega(\av \cdot \Psi(\LL))^{*}  .  $$

912: \label{theorem_Billera_Ehrenborg_Readdy}

913: \end{theorem}

914: 

915: We review the basic ideas behind the proof of this theorem.

916: We will refer back to them when we

917: prove similar results for toric and affine arrangements

918: in Sections~\ref{section_toric}

919: and~\ref{section_affine}.

920: 

921: 

922: Define three linear operators $\kappa$, $\beta$ and $\eta$ on

923: $\zab$ by

924: $$

925: \kappa(v) = \begin{cases}

926: (\av-\bv)^{m} & \text{if $v=\av^{m}$ for some $m \geq 0$,} \\

927: 0 & \text{otherwise,}

928: \end{cases}

929: $$

930: $$

931: \beta(v) = \begin{cases}

932: (\av-\bv)^{m} & \text{if $v=\bv^{m}$ for some $m \geq 0$,} \\

933: 0 & \text{otherwise,}

934: \end{cases}

935: $$

936: and 

937: $$

938: \eta(v) = \begin{cases}

939: 2 \cdot (\av-\bv)^{m+k} & \text{if $v = \bv^{m} \av^{k}$ for some $m, k \geq 0$,} \\

940: 0 & \text{otherwise.}

941: \end{cases}

942: $$

943: Observe that $\kappa$ and $\beta$ are both algebra maps.

944: The following relations hold for a poset $P$;

945: see~\cite[Section 5]{Billera_Ehrenborg_Readdy_om}:

946: \begin{eqnarray}

947: \kappa(\Psi(P)) & = & (\av-\bv)^{\rho(P)-1} ,                  

948: \label{equation_kappa} \\

949: \beta(\Psi(P))  & = & Z_{b}(P) \cdot (\av-\bv)^{\rho(P)-1} ,   

950: \label{equation_beta} \\

951: \eta(\Psi(P))   & = & Z(P) \cdot (\av-\bv)^{\rho(P)-1} .

952: \label{equation_eta}

953: \end{eqnarray}

954: For $k \geq 1$ the operator $\varphi_{k}$

955: is defined by the coalgebra expression

956: $$   \varphi_{k}(v)

957:    =

958:      \sum_{v}

959:          \kappa(v_{(1)}) \cdot \bv \cdot

960:          \eta(v_{(2)}) \cdot \bv \cdots \bv \cdot

961:          \eta(v_{(k)})   ,  $$

962: where the coproduct splits $v$ into $k$ parts.

963: Finally $\varphi$ is defined as the sum

964: $$       \varphi(v) = \sum_{k \geq 1} \varphi_{k}(v)  .   $$

965: Note that 

966: in this expression

967: only a finite number of terms are non-zero.

968: The connection with hyperplane arrangements is given by

969: the following proposition.

970: \begin{proposition}

971: The $\ab$-index of the lattice of faces of a central hyperplane

972: arrangement is given by

973: $$   \Psi(T) = \varphi( \Psi( \Lzero ) )^{*}   .   $$

974: \label{proposition_varphi}

975: \end{proposition}

976: The function $\varphi$ satisfies the functional equation

977: $$     \varphi(v)

978:      = 

979:        \kappa(v)

980:      + 

981:        \sum_{v} \varphi(v_{(1)}) \cdot \bv \cdot \eta(v_{(2)}) .  $$

982: From this relation it follows that

983: the function $\varphi$ satisfies the initial conditions

984: $\varphi(1) = 1$ and $\varphi(\bv) = 2 \cdot \bv$

985: and the recursions:

986: \begin{eqnarray}

987: \varphi(v \cdot \av) & = & \varphi(v) \cdot \cv , 

988: \label{equation_varphi_a} \\

989: \varphi(v \cdot \bv \bv) & = & \varphi(v \cdot \bv) \cdot \cv , 

990: \label{equation_varphi_b_b} \\

991: \varphi(v \cdot \av \bv) & = & \varphi(v) \cdot 2 \dv ,

992: \label{equation_varphi_a_b}

993: \end{eqnarray}

994: for an $\ab$-monomial $v$;

995: see~\cite[Section~5]{Billera_Ehrenborg_Readdy_om}.

996: These recursions culminate in the following result.

997: \begin{proposition}

998: \label{proposition_culminate}

999: For an $\ab$-monomial $w$ that begins with $\av$,

1000: the two maps $\varphi$ and $\omega$ coincide,

1001: that is,

1002: $\varphi(w) = \omega(w)$.

1003: \end{proposition}

1004: Finally, Theorem~\ref{theorem_Billera_Ehrenborg_Readdy}

1005: follows by Proposition~\ref{proposition_culminate} and 

1006: from the fact that

1007: $\Psi(\Lzero) = \av \cdot \Psi(\LL)$.

1008: 

1009: 

1010: \subsection{Regular subdivisions of manifolds}

1011: 

1012: 

1013: The face poset $P(\Omega)$ of a cell complex $\Omega$ is

1014: the set of all cells in $\Omega$ together

1015: with a minimal element~$\hz$ and a maximal element $\ho$.

1016: One partially orders two

1017: cells $\tau$ and $\sigma$

1018: by requiring that

1019: $\tau < \sigma$ if the cell $\tau$ is contained in $\overline{\sigma}$,

1020: the closure of $\sigma$.

1021: In order to define a regular cell complex,

1022: consider the cell complex $\Omega$ embedded in 

1023: Euclidean space $\Rrr^{n}$.

1024: This condition is compatible with toric cell complexes

1025: since the $n$-dimensional torus can be embedded 

1026: in $2n$-dimensional Euclidean space.

1027: Let $B^{n}$ denote the ball

1028: $\{x \in \Rrr^{n} \: : \: x_{1}^{2} + \cdots + x_{n}^{2} \leq 1\}$

1029: and

1030: let $S^{n-1}$ denote the sphere

1031: $\{x \in \Rrr^{n} \: : \: x_{1}^{2} + \cdots + x_{n}^{2} = 1\}$.

1032: A cell complex $\Omega$ 

1033: is {\em regular} if 

1034: (i) $\Omega$ consists of a finite number of cells,

1035: (ii) for every cell $\sigma$ of $\Omega$

1036: the pair $(\overline{\sigma}, \overline{\sigma} - \sigma)$

1037: is homeomorphic

1038: to a pair $(B^{k},S^{k-1})$ for some integer $k$,

1039: and

1040: (iii)

1041: the boundary

1042: $\overline{\sigma} - \sigma$

1043: is the disjoint union of smaller cells in $\Omega$.

1044: See Section~3.8 in~\cite{Stanley_EC_I} for more details.

1045: For a discussion of regular cell complexes not embedded

1046: in $\Rrr^n$,

1047: see~\cite{Bjorner_topological_methods}.

1048: 

1049: 

1050: The face poset of a regular subdivision of the sphere

1051: is an Eulerian face poset

1052: and hence has a $\cd$-index.

1053: For regular subdivisions of compact manifolds,

1054: a similar result holds.

1055: This was independently observed by

1056: Ed~Swartz~\cite{Swartz}.

1057: \begin{theorem}

1058: Let $\Omega$ be a regular cell complex whose geometric realization

1059: is a compact $n$-dimensional manifold $M$. 

1060: Let $\chi(M)$ denote the Euler characteristic of $M$.

1061: Then the $\ab$-index

1062: of the face poset $P$ of $\Omega$ has the following form.

1063: \begin{enumerate}

1064: \item[(i)]

1065: If $n$ is odd then

1066: $P$ is an Eulerian poset and hence

1067: $\Psi(P)$ can written in terms of $\cv$ and $\dv$.

1068: 

1069: \item[(ii)]

1070: If $n$ is even then $\Psi(P)$ has the form

1071: $$      \Psi(P) 

1072:     = 

1073:         \left( 1-\frac{\chi(M)}{2}

1074:         \right) \cdot (\av-\bv)^{n+1}

1075:       + 

1076:         \frac{\chi(M)}{2} \cdot \cv^{n+1}

1077:       + 

1078:         \Phi   ,   $$

1079: where $\Phi$ is a homogeneous $\cd$-polynomial of degree $n+1$

1080: and $\Phi$ does not contain the term $\cv^{n+1}$.

1081: \end{enumerate}

1082: \label{theorem_manifold}

1083: \end{theorem}

1084: \begin{proof}

1085: Observe that the poset $P$ has rank $n+2$.

1086: By~\cite[Theorem~3.8.9]{Stanley_EC_I} we know that

1087: every interval~$[x,y]$ strictly contained in $P$

1088: is Eulerian. When the rank of $P$ is odd this implies

1089: that $P$ is also Eulerian; see~\cite[Exercise 69c]{Stanley_EC_I}.

1090: Hence in this case the $\ab$-index of $P$

1091: can be expressed as a $\cd$-index.

1092: When $n$ is even, we use~\cite[Theorem~4.2]{Ehrenborg_k-Eulerian}

1093: to conclude that

1094: the $\ab$-index of $P$ belongs to

1095: $\Rrr\langle\cv,\dv,(\av-\bv)^{n+1}\rangle$.

1096: Since

1097: $\Psi(P)$ has degree $n+1$,

1098: the $\ab$-index $\Psi(P)$ can be written in the form

1099: $$      \Psi(P) 

1100:     = 

1101:         c_{1} \cdot (\av-\bv)^{n+1}

1102:       + 

1103:         c_{2} \cdot \cv^{n+1}

1104:       + 

1105:         \Phi   ,   $$

1106: where $\Phi$ is a homogeneous $\cd$-polynomial of degree $n+1$

1107: that does not contain any

1108: $\cv^{n+1}$ terms. 

1109: By looking at the coefficients

1110: of $\av^{n+1}$ and $\bv^{n+1}$, we have

1111: $c_{1} + c_{2} = 1$ and

1112: $c_{2} - c_{1} = \mu(P) = \chi(M) - 1$,

1113: where the last identity is again~\cite[Theorem~3.8.9]{Stanley_EC_I}.

1114: Solving for $c_{1}$ and $c_{2}$ proves the result.

1115: \end{proof}

1116: 

1117: 

1118: \begin{corollary}

1119: Let $P$ be the face poset of a regular cell complex whose geometric realization

1120: is a compact $n$-dimensional manifold $M$. 

1121: If $n$ is odd then the flag $h$-vector of $P$ is symmetric, that is,

1122: $h_{S} = h_{\overline{S}}$.

1123: If $n$ is even the flag $h$-vector of $P$ satisfies

1124: $$     h_{S} - h_{\overline{S}}

1125:     =

1126:        (-1)^{|S|} \cdot \left( 2 - \chi(M) \right)  . $$

1127: \end{corollary}

1128: 

1129: 

1130: 

1131: For the $n$-dimensional torus Theorem~\ref{theorem_manifold}

1132: can be expressed as follows.

1133: \begin{corollary}

1134: Let $\Omega$ be a regular cell complex whose geometric realization

1135: is the $n$-dimensional torus~$T^{n}$. Then the $\ab$-index

1136: of the face poset $P$ of $\Omega$ has the following form:

1137: $$      \Psi(P) 

1138:     = 

1139:         (\av-\bv)^{n+1}

1140:       + 

1141:         \Phi   ,   $$

1142: where $\Phi$ is a homogeneous $\cd$-polynomial of degree $n+1$

1143: and $\Phi$ does not contain the term $\cv^{n+1}$.

1144: \end{corollary}

1145: \begin{proof}

1146: When $n$ is even this is Theorem~\ref{theorem_manifold}.

1147: When $n$ is odd this is Theorem~\ref{theorem_manifold}

1148: together with the two facts that

1149: $\chi(T^{n}) = 0$ and

1150: $(\av-\bv)^{n+1} = (\cv^{2} - 2 \dv)^{(n+1)/2}$.

1151: \end{proof}

1152: 

1153: 

1154: 

1155: 

1156: 

1157: 

1158: 

1159: 

1160: 

1161: \section{Toric arrangements}

1162: \label{section_toric}

1163: 

1164: 

1165: 

1166: \subsection{Toric subspaces and arrangements}

1167: 

1168: 

1169: \begin{figure}

1170: \setlength{\unitlength}{0.5mm}

1171: \begin{center}

1172: \begin{picture}(100,100)(-20,-20)

1173: 

1174: %% Coordinate axis

1175: \put(0,-20){\line(0,1){100}}

1176: \put(-20,0){\line(1,0){100}}

1177: 

1178: %%%% Arrow heads

1179: %%\put(0,80){\line(-1,-2){4}}

1180: %%\put(0,80){\line(1,-2){4}}

1181: 

1182: %% Complete the boundary of the box

1183: \put(0,60){\line(1,0){60}}

1184: \put(60,0){\line(0,1){60}}

1185: 

1186: %% The lines

1187: \put(-20,-10){\line(2,1){100}}

1188: \put(-10,-20){\line(1,2){50}}

1189: 

1190: \thicklines

1191: 

1192: %% Inside the box

1193: \put(0,0){\line(2,1){60}}

1194: \put(0,0){\line(1,2){30}}

1195: \put(0,30){\line(2,1){60}}

1196: \put(30,0){\line(1,2){30}}

1197: 

1198: %% Number the regions

1199: \put(28,28){{\small 1}}

1200: 

1201: \put(53,35){{\small 2}}

1202: \put(9,46){{\small 2}}

1203: \put(21,4){{\small 2}}

1204: 

1205: \put(4,21){{\small 3}}

1206: \put(46,9){{\small 3}}

1207: \put(35,53){{\small 3}}

1208: 

1209: 

1210: \end{picture}

1211: \hspace*{15 mm}

1212: \begin{picture}(100,100)(-5,-5)

1213: 

1214: \put(45,0){\circle*{3}}

1215: \multiput(30,30)(30,0){2}{\put(0,0){\circle*{3}}}

1216: \multiput(15,60)(30,0){3}{\put(0,0){\circle*{3}}}

1217: \put(45,90){\circle*{3}}

1218: 

1219: \put(45,0){\line(-1,2){15}}

1220: \put(45,0){\line(1,2){15}}

1221: 

1222: \put(30,30){\line(-1,2){15}}

1223: \put(30,30){\line(1,2){15}}

1224: \put(30,30){\line(3,2){45}}

1225: 

1226: \put(60,30){\line(-3,2){45}}

1227: \put(60,30){\line(-1,2){15}}

1228: \put(60,30){\line(1,2){15}}

1229: 

1230: \put(45,90){\line(-1,-1){30}}

1231: \put(45,90){\line(0,-1){30}}

1232: \put(45,90){\line(1,-1){30}}

1233: \end{picture}

1234: \end{center}

1235: \caption{A toric line arrangement

1236: which subdivides

1237: the torus $T^{2}$ into a non-regular cell complex

1238: and its intersection poset.}

1239: \label{figure_toric_one}

1240: \end{figure}

1241: 

1242: 

1243: 

1244: 

1245: 

1246: The $n$-dimensional torus $T^{n}$ is defined as the quotient

1247: $\Rrr^{n}/\Zzz^{n}$.

1248: Recall that the torus $T^{n}$ is an abelian group.

1249: When identifying the torus $T^{n}$ with the set $[0,1)^{n}$,

1250: the group structure is componentwise addition modulo $1$.

1251: \begin{lemma}

1252: Let $V$ be a $k$-dimensional affine subspace

1253: in $\Rrr^{n}$ with rational coefficients. That is, $V$ has the form

1254: $$     V 

1255:    =

1256:        \{ \vec{v} \in \Rrr^{n} \:\: : \:\: A \vec{v} = \vec{b}\} , $$

1257: where the matrix $A$ has rational entries and the vector $\vec{b}$

1258: is allowed to have real entries.

1259: Then the image of $V$ under the

1260: quotient map $\Rrr^{n} \to \Rrr^{n}/\Zzz^{n}$,

1261: denoted by $\overline{V}$,

1262: is a $k$-dimensional torus.

1263: \end{lemma}

1264: \begin{proof}

1265: By translating $V$, we may assume that

1266: the vector $\vec{b}$ is the zero vector,

1267: and therefore

1268: $V$ is a subspace.

1269: In this case, the intersection of $V$ with the integer lattice $\Zzz^{n}$

1270: is a subgroup of the free abelian group $\Zzz^{n}$.

1271: Since the matrix $A$ has all rational entries,

1272: the rank of this subgroup is $k$,

1273: that is, the subgroup is isomorphic to $\Zzz^{k}$.

1274: Hence the image $\overline{V}$

1275: is the quotient $V/(V \cap \Zzz^{n})$,

1276: which is isomorphic to

1277: the quotient $\Rrr^{k}/\Zzz^{k}$, that is,

1278: a $k$-dimensional torus.

1279: \end{proof}

1280: 

1281: We call the image $\overline{V}$ 

1282: a {\em toric subspace} of the torus $T^{n}$

1283: because it is homeomorphic to some $k$-dimensional torus.

1284: When we remove the condition that the matrix $A$ is rational, 

1285: the image is not necessarily homeomorphic to a torus.

1286: 

1287: 

1288: 

1289: The intersection of two toric subspaces is in general not a toric subspace,

1290: but instead is the disjoint union of a finite number of

1291: toric subspaces.

1292: For two affine subspaces $V$ and $W$ with rational coefficients,

1293: we have that $\overline{V \cap W} \subseteq \overline{V} \cap \overline{W}$.

1294: In general, this containment is strict.

1295: 

1296: Define the translate of a toric subspace $U$ by a point $x$

1297: on the torus to be the toric subspace $U + x = \{ u+x  : u \in U \}$.

1298: Alternatively, one may lift the toric subspace to an affine subspace

1299: in Euclidean space, translate it and then map back to the torus.

1300: Then for two toric subspaces $V$ and $W$, their intersection

1301: has the form

1302: $$     V \cap W = \bigcup_{p=1}^{r} (U + x_{p})  , $$

1303: where $U$ is a toric subspace, $r$ is a non-negative integer

1304: and $x_{1}, \ldots, x_{r}$ are points on the torus $T^{n}$.

1305: 

1306: 

1307: 

1308: 

1309: \begin{figure}

1310: \setlength{\unitlength}{0.5mm}

1311: \begin{center}

1312: \begin{picture}(100,100)(-20,-20)

1313: 

1314: %% Coordinate axis

1315: \put(0,-20){\line(0,1){100}}

1316: \put(-20,0){\line(1,0){100}}

1317: 

1318: %% Complete the boundary of the box

1319: \put(0,60){\line(1,0){60}}

1320: \put(60,0){\line(0,1){60}}

1321: 

1322: %% The lines

1323: \put(-20,-10){\line(2,1){100}}

1324: \put(-6,-18){\line(1,3){32}}

1325: \put(-20,12){\line(1,0){100}}

1326: 

1327: \thicklines

1328: 

1329: %% Inside the box

1330: \put(0,0){\line(2,1){60}}

1331: \put(0,30){\line(2,1){60}}

1332: \put(0,0){\line(1,3){20}}

1333: \put(20,0){\line(1,3){20}}

1334: \put(40,0){\line(1,3){20}}

1335: \put(0,12){\line(1,0){60}}

1336: \end{picture}

1337: \hspace*{15 mm}

1338: \begin{picture}(100,100)(-5,-5)

1339: 

1340: \put(45,0){\circle*{3}}

1341: \multiput(15,30)(30,0){3}{\put(0,0){\circle*{3}}}

1342: \multiput(0,60)(15,0){7}{\put(0,0){\circle*{3}}}

1343: \put(45,90){\circle*{3}}

1344: 

1345: \put(45,0){\line(-1,1){30}}

1346: \put(45,0){\line(0,1){30}}

1347: \put(45,0){\line(1,1){30}}

1348: 

1349: \put(15,30){\line(-1,2){15}}

1350: \put(15,30){\line(0,1){30}}

1351: \put(15,30){\line(1,2){15}}

1352: \put(15,30){\line(1,1){30}}

1353: \put(15,30){\line(3,2){45}}

1354: 

1355: \put(45,30){\line(-3,2){45}}

1356: \put(45,30){\line(-1,1){30}}

1357: \put(45,30){\line(-1,2){15}}

1358: \put(45,30){\line(0,1){30}}

1359: \put(45,30){\line(1,2){15}}

1360: \put(45,30){\line(1,1){30}}

1361: \put(45,30){\line(3,2){45}}

1362: 

1363: \put(75,30){\line(-1,2){15}}

1364: \put(75,30){\line(0,1){30}}

1365: \put(75,30){\line(1,2){15}}

1366: 

1367: 

1368: \put(45,90){\line(-3,-2){45}}

1369: \put(45,90){\line(-1,-1){30}}

1370: \put(45,90){\line(-1,-2){15}}

1371: \put(45,90){\line(0,-1){30}}

1372: \put(45,90){\line(1,-2){15}}

1373: \put(45,90){\line(1,-1){30}}

1374: \put(45,90){\line(3,-2){45}}

1375: \end{picture}

1376: \end{center}

1377: \caption{A toric line arrangement

1378: and its intersection poset.}

1379: \label{figure_toric_two}

1380: \end{figure}

1381: 

1382: 

1383: A {\em toric hyperplane arrangement} $\HH = \{H_{1}, \ldots, H_{m}\}$

1384: is a finite collection of toric hyperplanes.

1385: Define the {\em intersection poset} $\PP$

1386: of a toric arrangement

1387: to be the set of all connected components

1388: arising from

1389: all possible intersections of the toric hyperplanes,

1390: that is, 

1391: all connected components of

1392: $\bigcap_{i \in S} H_{i}$

1393: where $S \subseteq \{1, \ldots, m\}$,

1394: together with the empty set.

1395: Order the elements of

1396: the intersection poset $\PP$ by reverse inclusion,

1397: that is, the torus $T^{n}$ is the minimal element of $\PP$

1398: corresponding to the empty intersection,

1399: and the empty set is the maximal element.

1400: A toric subspace $V$ is contained in the intersection

1401: poset $\PP$ if there are

1402: toric hyperplanes $H_{i_{1}}, \ldots, H_{i_{k}}$

1403: in the arrangement such that

1404: $V \subseteq H_{i_{1}} \cap \cdots \cap H_{i_{k}}$

1405: and there is no toric subspace $W$

1406: satisfying

1407: $V \subset W \subseteq H_{i_{1}} \cap \cdots \cap H_{i_{k}}$.

1408: In other words,

1409: $V$ has to be a maximal toric subspace in some intersection

1410: of toric hyperplanes from the arrangement.

1411: 

1412: 

1413: The notion of using the intersection poset can be found

1414: in

1415: work of Zaslavsky, where he considers

1416: topological dissections~\cite{Zaslavsky_paper}.

1417: In this setting there is not an intersection

1418: lattice, but rather an intersection poset.

1419: 

1420: 

1421: 

1422: 

1423: 

1424: 

1425: To every toric hyperplane arrangement $\HH = \{H_{1}, \ldots, H_{m}\}$

1426: there is an associated periodic hyperplane arrangement $\widetilde{\HH}$

1427: in the Euclidean space $\Rrr^{n}$. Namely, the inverse

1428: image of the toric hyperplane~$H_{i}$ under the quotient map

1429: $\Rrr^{n} \to \Rrr^{n}/\Zzz^{n}$ is the union of parallel integer

1430: translates of a real hyperplane.

1431: Let~$\widetilde{\HH}$

1432: be the collection of all these integer translates.

1433: Observe that every face of the toric arrangement~$\HH$

1434: can be lifted to a parallel class of faces in

1435: the periodic real arrangement $\widetilde{\HH}$.

1436: 

1437: 

1438: 

1439: 

1440: 

1441: 

1442: For a toric hyperplane arrangement $\HH$

1443: define the {\em toric characteristic polynomial}

1444: to be

1445: $$   \chi(\HH; t)

1446:    =

1447:      \sum_{\onethingatopanother{x \in \PP}{x \neq \emptyset}}

1448:              \mu(\hz,x) \cdot t^{\dim(x)}  .   $$

1449: Also for a toric hyperplane arrangement $\HH$ define

1450: $T_{t}$ to be the face poset of the induced subdivision

1451: of the torus $T^{n}$. Note that $T_{t}$ is a graded poset

1452: of rank $n+2$: the minimal element $\hz$ is the empty face,

1453: the maximal element $\ho$ is the torus and the rank of

1454: face $x$ is given by $\dim(x) + 1$.

1455: 

1456: 

1457: \begin{example}

1458: {\rm

1459: Consider the line arrangement consisting

1460: of the two lines $y = 2 \cdot x$

1461: and $x = 2 \cdot y$ in the plane $\Rrr^{2}$.

1462: In $\Rrr^{2}$ they intersect in one point,

1463: namely

1464: the origin,

1465: whereas on the torus $T^{2}$ they intersect

1466: in three points,

1467: namely

1468: $(0,0)$, $(2/3,1/3)$ and $(1/3,2/3)$.

1469: The characteristic polynomial

1470: is given by $\chi(\HH;t) = t^{2} - 2 \cdot t + 3$.

1471: However, this arrangement is not

1472: regular, since the induced subdivision of $T^{2}$

1473: is not regular. The boundary of each region

1474: is a wedge of two circles.

1475: See Figure~\ref{figure_toric_one}.

1476: }

1477: \label{example_toric_one}

1478: \end{example}

1479: 

1480: 

1481: 

1482: \begin{example}

1483: {\rm

1484: Consider the line arrangement consisting

1485: of the three lines $y = 3 \cdot x$,

1486: $x = 2 \cdot y$ and

1487: $y = 1/5$.

1488: It subdivides the torus into a regular cell complex.

1489: The subdivision and the associated

1490: intersection poset are shown in 

1491: Figure~\ref{figure_toric_two}.

1492: The characteristic polynomial

1493: is given by $\chi(\HH;t) = t^{2} - 3 \cdot t + 8$.

1494: Furthermore, the $\ab$-index of the

1495: subdivision of the torus is given by

1496: $\Psi(T_{t}) = (\av-\bv)^{3} + 7 \cdot \mdc + 8 \cdot \mcd$,

1497: as the following calculation shows.

1498: \newcommand{\ts}{\:\:\:\:\:\:}

1499: $$

1500:   \begin{array}{c r r c r c c}

1501:  S & f_{S} & h_{S} & u_{S} & (\av-\bv)^{3} 

1502:    & 7 \cdot \mdc & 8 \cdot \mcd \\ \hline

1503: \emptyset &  1 &  1 & \av\av\av &  1 \ts & 0 & 0 \\

1504: \{1\}     &  7 &  6 & \bv\av\av & -1 \ts & 7 & 0 \\

1505: \{2\}     & 15 & 14 & \av\bv\av & -1 \ts & 7 & 8 \\

1506: \{3\}     &  8 &  7 & \av\av\bv & -1 \ts & 0 & 8 \\

1507: \{1,2\}   & 30 &  9 & \bv\bv\av &  1 \ts & 0 & 8 \\

1508: \{1,3\}   & 30 & 16 & \bv\av\bv &  1 \ts & 7 & 8 \\

1509: \{2,3\}   & 30 &  8 & \av\bv\bv &  1 \ts & 7 & 0 \\

1510: \{1,2,3\} & 60 & -1 & \bv\bv\bv & -1 \ts & 0 & 0 \\

1511:   \end{array} $$

1512: Recall that

1513: $\mdc = \maba + \nabb + \nbaa + \mbab$

1514: and

1515: $\mcd = \naab + \maba + \mbab + \nbba$.

1516: Here in the last three columns we indicate

1517: the contribution of a given term to each $\ab$-monomial.

1518: Observe that the sum of the last three columns gives

1519: the flag $h$-vector entries.

1520: }

1521: \label{example_toric_two}

1522: \end{example}

1523: 

1524: 

1525: 

1526: 

1527: 

1528: 

1529: 

1530: We now give a natural interpretation of the toric characteristic polynomial.

1531: Recall that the intersection of toric subspaces

1532: is the disjoint union of toric subspaces that are translates of each other.

1533: Let~$G$ be the collection of finite intersections of toric

1534: subspaces of the $n$-dimensional torus $T^{n}$,

1535: that is, $G$ consists of sets of the form

1536: $V = W_{1} \capdots W_{q}$, where $W_{1}, \ldots, W_{q}$ are

1537: toric subspaces.

1538: Such a set~$V$ can be written as a union,

1539: more precisely,

1540: $V = \bigcup_{p=1}^{r} (U + x_{p})$,

1541: where $U$ is a toric subspace, $r$ a non-negative integer,

1542: and $x_{1}, \ldots, x_{r}$ are points on the torus.

1543: Observe that the empty set $\emptyset$

1544: and the torus~$T^{n}$ belong to $G$.

1545: Furthermore, $G$ is closed under finite intersections.

1546: Let $L$ be the distributive lattice consisting of all

1547: subsets of the torus $T^{n}$ that are obtained from

1548: the collection~$G$ by finite intersections, finite unions

1549: and complements. The set $G$ is the generating set for the lattice~$L$.

1550: A {\em valuation}~$v$ is a function on the lattice $L$

1551: to an abelian group

1552: satisfying

1553: $v(\emptyset) = 0$ and $v(A) + v(B) = v(A \cap B) + v(A \cup B)$

1554: for all sets $A, B \in L$.

1555: 

1556: 

1557: The next theorem is analogous to Theorem~2.1

1558: in~\cite{Ehrenborg_Readdy_valuation_1}.

1559: The proof here is more involved

1560: due to the fact that the collection of

1561: toric subspaces is not closed under intersections.

1562: \begin{theorem}

1563: There is a valuation $v$ on the distributive lattice~$L$

1564: to $\Zzz[t]$

1565: such that 

1566: for a $k$-dimensional toric subspace $V$

1567: its valuation is $v(V) = t^{k}$.

1568: \end{theorem}

1569: \begin{proof}

1570: Define the function $v$ on the generating set $G$ by

1571: $$     v\left(\bigcup_{p=1}^{r} (U + x_{p})\right)

1572:     =

1573:        r \cdot t^{k}   ,  $$

1574: where we assume that $U$ is a $k$-dimensional toric subspace

1575: and the $r$ translates

1576: $U + x_{1}, \ldots, U + x_{r}$ are pairwise disjoint.

1577: Observe that the function $v$ is additive with respect to

1578: disjoint unions, that is, for elements

1579: $V_{1}, \ldots, V_{m}$ in $G$

1580: which are pairwise disjoint

1581: and 

1582: $V_{1} \cupdots V_{m} \in G$.

1583: In this case, each $V_{i}$ is a disjoint union of translates of

1584: the same affine subspace $U$ and both sides of the identity

1585: $v(V_{1}) + \cdots + v(V_{m})

1586:     =

1587:  v(V_{1} \cupdots V_{m})$

1588: count the number of translates of $U$ 

1589: times $t^{\dim(U)}$.

1590: 

1591: 

1592: Groemer's integral theorem~\cite{Groemer}

1593: (see also~\cite[Theorem~2.2.1]{Klain_Rota})

1594: states that a function $v$

1595: defined on a generating set $G$ extends to a valuation

1596: on the distributive lattice generated by $G$ if

1597: for all $V_{1}, \ldots, V_{m}$ in $G$ such that

1598: $V_{1} \cupdots V_{m} \in G$, the inclusion-exclusion

1599: formula holds:

1600: \begin{equation}

1601: v(V_{1} \cupdots V_{m})

1602:   =

1603: \sum_{i} v(V_{i})

1604:   -

1605: \sum_{i<j} v(V_{i} \cap V_{j})

1606:   +

1607: \cdots .

1608: \label{equation_inclusion_exclusion}

1609: \end{equation}

1610: To verify this for our generating set $G$,

1611: first consider the case when 

1612: the union $V_{1} \cupdots V_{m}$

1613: is a toric subspace.

1614: This case implies that

1615: $V_{1} \cupdots V_{m} = V_{i}$ for some index $i$.

1616: It then follows that the inclusion-exclusion

1617: formula~(\ref{equation_inclusion_exclusion}) holds

1618: trivially.

1619: 

1620: Before considering the general case,

1621: we introduce some notation.

1622: For $S$ a non-empty subset of the index set $\{1, \ldots, m\}$,

1623: let $V_{S} = \bigcap_{i \in S} V_{i}$.

1624: Equation~(\ref{equation_inclusion_exclusion})

1625: can then be written as

1626: $$

1627: v(V_{1} \cupdots V_{m})

1628:   =

1629: \sum_{S} (-1)^{|S|-1} \cdot v(V_{S})  ,

1630: $$

1631: where the sum ranges over non-empty subsets $S$ of $\{1, \ldots, m\}$.

1632: Now assume that 

1633: $V_{1} \cupdots V_{m}$ is the disjoint union

1634: $(U + x_{1}) \cupdots (U + x_{r})$.

1635: Let $V_{S,p}$ denote the intersection

1636: $V_{S} \cap (U + x_{p})$.

1637: Observe that

1638: $U + x_{p} = \bigcup_{i=1}^{m} V_{\{i\},p}$

1639: and since $U + x_{p}$ is itself a toric subspace,

1640: we have already proved that

1641: the inclusion-exclusion formula~(\ref{equation_inclusion_exclusion})

1642: holds for this union.

1643: Hence we have

1644: \begin{eqnarray*}

1645: v(V_{1} \cupdots V_{m})

1646:   & = &

1647: \sum_{p=1}^{r} v(U + x_{p}) \\

1648:   & = &

1649: \sum_{p=1}^{r} \sum_{S} (-1)^{|S|-1} \cdot v(V_{S,p}) \\

1650:   & = &

1651: \sum_{S} (-1)^{|S|-1} \cdot \sum_{p=1}^{r} v(V_{S,p}) \\

1652:   & = &

1653: \sum_{S} (-1)^{|S|-1} \cdot v(V_{S}) ,

1654: \end{eqnarray*}

1655: where $S$ ranges over all non-empty subsets of $\{1, \ldots, m\}$.

1656: The last step follows since the

1657: union $V_{S} = \bigcup_{p=1}^{r} V_{S,p}$

1658: is pairwise disjoint.

1659: \end{proof}

1660: 

1661: 

1662: By M\"obius inversion we directly have the following theorem.

1663: The proof is standard.  See the 

1664: references~\cite{Athanasiadis,Chen,Ehrenborg_Readdy_valuation_1,Jozefiak_Sagan}.

1665: \begin{theorem}

1666: The characteristic polynomial of a toric arrangement is given by

1667: $$  \chi(\HH) = v\left( T^{n} - \bigcup_{i=1}^{m} H_{i} \right) .  $$

1668: \label{theorem_characteristic}

1669: \end{theorem}

1670: 

1671: 

1672: When each region is an open ball we can now determine the number of

1673: regions in a toric arrangement.  The proof is analogous to the proofs

1674: in~\cite{Ehrenborg_Readdy_valuation_1,Ehrenborg_Readdy_valuation_2}.

1675: 

1676: 

1677: 

1678: \begin{theorem}

1679: Let $\HH$ be a toric hyperplane arrangement on the $n$-dimensional

1680: torus $T^{n}$ that subdivides the torus into 

1681: regions that are open $n$-dimensional balls.

1682: Then the number of regions in the complement of the arrangement 

1683: is given by $(-1)^{n} \cdot \chi(\HH; t=0)$.

1684: \label{theorem_toric_Zaslavsky_version_1}

1685: \end{theorem}

1686: \begin{proof}

1687: Observe that the Euler valuation $\varepsilon$

1688: of a $k$-dimensional torus

1689: is given by the Kronecker delta~$\delta_{k,0}$.

1690: Hence the Euler valuation of a toric subspace $V$

1691: of the $n$-dimensional torus $T^{n}$ is given

1692: by setting $t=0$ in the valuation $v(V)$, that

1693: is, $\varepsilon(V) = v(V)|_{t = 0}$.

1694: Since the two valuations $\varepsilon$ and $v|_{t=0}$

1695: are additive with respect to disjoint unions,

1696: they agree for any member of the generating set $G$.

1697: Hence they also agree for any member in the distributive lattice $L$.

1698: In particular, we have that

1699: \begin{equation}

1700:     \varepsilon\left( T^{n} - \bigcup_{i=1}^{m} H_{i} \right) 

1701:   =

1702:     \left. v\left( T^{n} - \bigcup_{i=1}^{m} H_{i} \right)\right|_{t = 0} .

1703: \label{equation_Euler_and_v}

1704: \end{equation}

1705: Since the Euler valuation of an open ball

1706: is $(-1)^{n}$ and

1707: $T^{n} - \bigcup_{i=1}^{m} H_{i}$

1708: is a disjoint union of open balls,

1709: the left-hand side 

1710: of~(\ref{equation_Euler_and_v})

1711: is $(-1)^{n}$ times the number of regions.

1712: The right-hand side is $\chi(\HH; t=0)$ by 

1713: Theorem~\ref{theorem_characteristic}.

1714: \end{proof}

1715: 

1716: 

1717: \addtocounter{theorem}{-5}

1718: \begin{continuation}

1719: {\rm

1720: Setting $t=0$ in the characteristic polynomial

1721: in Example~\ref{example_toric_one}

1722: we obtain $3$, which is indeed

1723: is the number of regions of this arrangement.

1724: }

1725: \end{continuation}

1726: \addtocounter{theorem}{4}

1727: 

1728: 

1729: 

1730: 

1731: 

1732: 

1733: We call a toric hyperplane arrangement $\HH = \{H_{1}, \ldots, H_{m}\}$

1734: {\em rational}

1735: if each hyperplane $H_{i}$

1736: is of the form

1737: $\vec{a}_{i} \cdot \vec{x} = b_{i}$

1738: where the vector $\vec{a}_{i}$ has integer entries and

1739: $b_{i}$ is an integer for $1 \leq i \leq m$.

1740: This is equivalent to assuming every constant $b_{i}$ is rational

1741: since every vector $\vec{a}_{i}$ was already assumed to be rational.

1742: In what follows it will be convenient to

1743: assume every coefficient is integral in a given rational

1744: arrangement.

1745: 

1746: 

1747: 

1748: Define

1749: $\lcm(\HH)$ to be

1750: the least common multiple of all the

1751: $n \times n$ minors of the $n \times m$ matrix

1752: $(\vec{a}_{1}, \ldots, \vec{a}_{m})$.

1753: We can now give a different interpretation

1754: of the toric chromatic polynomial

1755: by counting lattice points.

1756: \begin{theorem}

1757: For a rational hyperplane arrangement $\HH$

1758: there exists a constant $k$ such that

1759: for every $q > k$ where $q$ is a multiple of $\lcm(\HH)$,

1760: the toric characteristic polynomial

1761: evaluated at $q$ is given by

1762: the number of lattice points in

1763: $\left( \frac{1}{q} \Zzz \right)^{n}/\Zzz^{n}$

1764: that do not lie on any of the toric hyperplanes $H_{i}$, that is,

1765: $$

1766:     \chi(\HH; t=q)

1767:   =

1768:     \left|

1769:       \left( \frac{1}{q} \Zzz \right)^{n}/\Zzz^{n}

1770:         -

1771:       \bigcup_{i=1}^{m} H_{i}

1772:     \right|    . 

1773: $$

1774: \label{theorem_lattice_points}

1775: \end{theorem}

1776: The condition that $q$ is a multiple of $\lcm(\HH)$

1777: implies that every subspace $x$

1778: in the intersection poset~$\PP$

1779: intersects the toric lattice

1780: $\left(\frac{1}{q} \Zzz \right)^{n}/\Zzz^{n}$

1781: in exactly $q^{\dim(x)}$ points.

1782: Theorem~\ref{theorem_lattice_points}

1783: now follows by M\"obius inversion.

1784: This theorem

1785: is the toric analogue of

1786: Athanasiadis'

1787: finite field method.

1788: See especially \cite[Theorem~2.1]{Athanasiadis_II}.

1789: 

1790: 

1791: In the case when $\lcm(\HH) = 1$, the toric arrangement 

1792: $\HH$ is

1793: called {\em unimodular}. 

1794: Novik, Postnikov and Sturmfels~\cite{Novik_Postnikov_Sturmfels}

1795: state Theorem~\ref{theorem_toric_Zaslavsky_version_1}

1796: in the special case of unimodular arrangements.

1797: Their first proof is based upon Zaslavsky's result on the number

1798: of bounded regions in an affine arrangement.

1799: The second proof, due to Vic Reiner, is equivalent to our

1800: proof for arbitrary toric arrangements.

1801: See also the paper~\cite{Zaslavsky_paper}

1802: by Zaslavsky, where more general arrangement

1803: are considered.

1804: 

1805: 

1806: 

1807: \subsection{Graphical arrangements}

1808: 

1809: We digress in this subsection to discuss an application

1810: to graphical arrangements.

1811: For a graph $G$ on the vertex set $\{1, \ldots, n\}$

1812: define the graphical arrangement $\HH_{G}$ to

1813: be the collection of hyperplanes of the form $x_{i} = x_{j}$

1814: for each edge $ij$ in the graph $G$.

1815: \begin{corollary}

1816: For a connected graph $G$ on $n$ vertices the regions

1817: in the complement

1818: of the graphical arrangement $\HH_{G}$ on the torus $T^{n}$

1819: are each homotopy equivalent to the $1$-dimensional torus $T^{1}$.

1820: Furthermore, the number of regions is given by

1821: $(-1)^{n-1}$ times the linear coefficient of the chromatic

1822: polynomial of $G$.

1823: \end{corollary}

1824: \begin{proof}

1825: Recall the chromatic polynomial of the graph $G$ is equal to

1826: the characteristic polynomial of the graphical arrangement

1827: $\HH_{G}$.

1828: The intersection lattice

1829: of the real arrangement $\HH_{G}$ is the same

1830: as the intersection poset of the toric arrangement $\HH_{G}$.

1831: Translating the graphical arrangement in the

1832: direction $(1, \ldots, 1)$ leaves the

1833: arrangement on the torus invariant.

1834: Since $G$ is connected this is the only direction

1835: that leaves the arrangement invariant.

1836: Hence each region is

1837: homotopy equivalent to $T^{1}$.

1838: By adding the hyperplane $x_{1} = 0$ to the arrangement

1839: we obtain a new arrangement $\HH^{\prime}$

1840: with the same number of regions, 

1841: but with each region homeomorphic to a ball.

1842: Since the intersection lattice of $\HH^{\prime}$

1843: is just the Cartesian product of the two-element poset

1844: with the  intersection lattice of $\HH_{G}$, we have

1845: $$   \chi(\HH^{\prime}, t)

1846:    = 

1847:      (t-1) \cdot \chi(\HH_{G}, t)/t .  $$

1848: The number of regions is obtained by setting $t=0$ in this equality.

1849: \end{proof}

1850: 

1851: A similar statement holds for graphs that are disconnected.

1852: The result follows from the fact that

1853: the complement of the graphical arrangement

1854: is the product of the complements

1855: of each connected component.

1856: \begin{corollary}

1857: For a graph $G$ on $n$ vertices 

1858: consisting of $k$ components,

1859: the regions

1860: in the complement

1861: of the graphical arrangement $\HH_{G}$ on the torus $T^{n}$

1862: are each homotopy equivalent to the $k$-dimensional torus $T^{k}$.

1863: The number of regions is given by

1864: $(-1)^{n-k}$ times the coefficient of $t^{k}$ in the chromatic

1865: polynomial of $G$.

1866: \end{corollary}

1867: 

1868: Stanley~\cite{Stanley_acyclic} proved the celebrated result

1869: that the chromatic polynomial of a graph

1870: evaluated at $t=-1$

1871: is $(-1)^{n}$ times the number of acyclic orientations of the

1872: graph.

1873: A similar interpretation for the linear coefficient

1874: of the chromatic polynomial is due to

1875: Greene and Zaslavsky~\cite{Greene_Zaslavsky}:

1876: \begin{theorem}[Greene--Zaslavsky]

1877: Let $G$ be a connected graph and $v$ a given vertex of the graph.

1878: The linear coefficient of the chromatic polynomial

1879: is $(-1)^{n-1}$ times the number of acyclic orientations

1880: of the graph such that the only sink is the vertex $v$.

1881: \label{theorem_Greene_Zaslavsky}

1882: \end{theorem}

1883: \begin{proof}

1884: It is enough to give a bijection between

1885: regions in the complement of the graphical arrangement

1886: on the torus $T^{n}$

1887: and acyclic orientations with the vertex $v$ as the unique sink.

1888: For a region $R$ of the arrangement intersect it with

1889: the hyperplane $x_{v} = 0$ to obtain the face $S$.

1890: Let $\HH^{\prime}$ be the arrangement~$\HH_{G}$

1891: together with the hyperplane $x_{v} = 0$.

1892: Lift $S$ to a face $\widetilde{S}$

1893: in the periodic arrangement~$\widetilde{\HH^{\prime}}$

1894: in~$\Rrr^{n}$.

1895: Observe that $\widetilde{S}$ is the interior of a polytope.

1896: When minimizing the linear functional

1897: $L(x) = x_{1} + \cdots + x_{n}$

1898: on the closure of the face $\widetilde{S}$, the optimum

1899: is a lattice point $k = (k_{1}, \ldots, k_{n})$.

1900: Pick a point $x = (x_{1}, \ldots, x_{n})$ 

1901: in $\widetilde{S}$ close to the optimum,

1902: that is, each coordinate $x_{i}$ lies in the interval

1903: $[k_{i},k_{i}+\epsilon)$ for some small $\epsilon > 0$.

1904: 

1905: 

1906: Let $y = (y_{1}, \ldots, y_{n})$ be the image of the point

1907: $x$ on the torus $T^{n}$, that is,

1908: $y_{i} = x_{i} \bmod 1$. Note that

1909: each entry $y_{i}$ lies in the half open interval $[0,1)$

1910: and that $y_{v} = 0$.

1911: Construct an orientation of the graph $G$

1912: by letting the edge $ij$ be oriented $i \rightarrow j$ if $y_{i} > y_{j}$.

1913: Note that this orientation is acyclic and has the vertex $v$

1914: as a sink. 

1915: 

1916: To show that the vertex $v$ is the unique sink,

1917: assume that the vertex $i$ is also a sink, where $i \neq v$.

1918: In other words, for all neighbors $j$ of the vertex $i$ we have that

1919: $y_{i} < y_{j}$. We can continuously move the point

1920: $x$ in $\widetilde{S}$

1921: by decreasing the value of the $i$th coordinate $x_{i}$.

1922: Observe that there is no hyperplane in the periodic

1923: arrangement blocking the coordinate $x_{i}$ from passing through

1924: the integer value $k_{i}$ and continuing down to $k_{i}-1+\epsilon$.

1925: This contradicts the fact

1926: that we chose the original point~$x$ close to the optimum

1927: of the linear functional $L$. Hence the vertex $i$ cannot be a sink.

1928: 

1929: 

1930: It is straightforward to verify that

1931: this map from regions to the set of acyclic orientations

1932: with the unique sink at $v$ is a bijection.

1933: \end{proof}

1934: 

1935: The technique of assigning a point to every region

1936: of a toric arrangement using a linear functional

1937: was used by

1938: Novik, Postnikov and Sturmfels

1939: in their paper~\cite{Novik_Postnikov_Sturmfels}.

1940: See their first proof of the number of regions

1941: of a toric arrangement.

1942: 

1943: 

1944: 

1945: 

1946: 

1947: 

1948: \subsection{The toric Bayer--Sturmfels result}

1949: 

1950: 

1951: Define the {\em toric Zaslavsky invariant} of a graded poset $P$

1952: having $\hz$ and $\ho$ by

1953: $$    \Zt(P) 

1954:    = 

1955:       \sum_{x \mbox{ {\rm \tiny coatom of} } P}

1956:       (-1)^{\rho(\hz,x)} \cdot \mu(\hz,x)    

1957:    = 

1958:       (-1)^{\rho(P) - 1} \cdot

1959:       \sum_{x \mbox{ {\rm \tiny coatom of} } P}

1960:       \mu(\hz,x)    .  $$

1961: 

1962: We reformulate 

1963: Theorem~\ref{theorem_toric_Zaslavsky_version_1}

1964: as follows.

1965: \begin{theorem}

1966: For a toric hyperplane arrangement $\HH$ 

1967: on the torus $T^n$

1968: that subdivides the torus into 

1969: open $n$-dimensional balls,

1970: the number of

1971: regions is given by $\Zt(\PP)$, where $\PP$

1972: is the intersection poset of the arrangement~$\HH$.

1973: \label{theorem_toric_Zaslavsky_version_2}

1974: \end{theorem}

1975: 

1976: 

1977: 

1978: As a corollary of Theorem~\ref{theorem_toric_Zaslavsky_version_2},

1979: we can describe the $f$-vector of the subdivision $T_{t}$ of the torus.

1980: For similar results for more general manifolds

1981: see~\cite[Section~3]{Zaslavsky_paper}.

1982: \begin{corollary}

1983: The number of $i$-dimensional regions in the subdivision $T_{t}$

1984: of the $n$-dimensional torus is given by the sum

1985: $$    f_{i+1}(T_{t})

1986:     =

1987:       (-1)^{i}

1988:     \cdot

1989:       \sum_{\onethingatopanother{x \leq y}

1990:             {\onethingatopanother{\dim(x) = i}{\dim(y) = 0}}}

1991:               \mu(x,y)  ,

1992: $$

1993: where $\mu(x,y)$ denotes the M\"obius function

1994: of the interval $[x,y]$

1995: in the intersection poset $\PP$.

1996: \label{corollary_f_vector_I}

1997: \end{corollary}

1998: \begin{proof}

1999: Each $i$-dimensional region is contained in a unique 

2000: $i$-dimensional subspace $x$. By restricting the arrangement

2001: to the subspace $x$ and applying

2002: Theorem~\ref{theorem_toric_Zaslavsky_version_1}, we have that

2003: the number of $i$-dimensional regions in $x$ is given by

2004: $(-1)^{i} \cdot

2005:  \sum_{x \leq y, \dim(y) = 0}   \mu(x,y)$.

2006: %%%% \sum_{\onethingatopanother{x \leq y}{\dim(y) = 0}}   \mu(x,y)$.

2007: Summing over all $x$, the result follows.

2008: \end{proof}

2009: 

2010: 

2011: 

2012: 

2013: 

2014: 

2015: 

2016: 

2017: For the remainder of this section we will assume that

2018: the induced subdivision of the torus is a regular cell complex.

2019: Let $T_{t}$ be the face poset of 

2020: the subdivision of the torus induced by the

2021: toric arrangement.

2022: Define the map

2023: $\zt : T_{t}^{*} \longrightarrow \Pzero$

2024: by sending each face to the

2025: smallest toric subspace in the intersection poset that contains

2026: the face

2027: and

2028: sending

2029: the minimal element

2030: in $T_{t}^{*}$ to $\hz$.

2031: Observe

2032: that the map $\zt$ is order- and rank-preserving,

2033: as well as being surjective.

2034: As in the central hyperplane arrangement case,

2035: we view the map $\zt$ as a map from the set of chains

2036: of $T_{t}^{*}$ to the set of chains of $\Pzero$.

2037: 

2038: 

2039: 

2040: 

2041: Let $x$ be an element in the intersection poset $\PP$

2042: of a toric hyperplane arrangement $\HH$.

2043: Then the interval $[x,\ho]$ is the intersection poset

2044: of a toric arrangement in the toric subspace $x$.

2045: The atoms of the interval $[x,\ho]$ are the toric hyperplanes

2046: in this smaller toric arrangement.

2047: 

2048: More interesting is the geometric interpretation

2049: of the interval $[\hz,x]$. It is the intersection lattice

2050: of a central hyperplane arrangement

2051: in $\Rrr^{n - \dim(x)}$.

2052: Without loss of generality we may assume that

2053: $x$ contains the zero point $(0, \ldots, 0)$,

2054: that is, when we lift the toric subspace $x$ to an affine subspace $V$

2055: in~$\Rrr^{n}$

2056: we may assume that $V$ is a subspace of $\Rrr^{n}$.

2057: Any toric subspace $y$ in the interval $[\hz,x]$,

2058: that is, a toric subspace containing $x$,

2059: can be lifted to a subspace $W$ containing the subspace $V$.

2060: In particular, the toric hyperplanes in $[\hz,x]$

2061: lift to hyperplanes in $\Rrr^{n}$ containing $V$.

2062: This lifting is a poset isomorphism and

2063: we obtain an essential central arrangement

2064: of dimension $n - \dim(x)$

2065: by quotienting out by the subspace~$V$.

2066: We conclude by noticing that an interval $[x,y]$ in

2067: $\PP$, where $y < \ho$, is the intersection lattice

2068: of a central hyperplane arrangement.

2069: 

2070: 

2071: 

2072: The toric analogue of Theorem~\ref{theorem_Bayer_Sturmfels}

2073: is as follows.

2074: \begin{theorem}

2075: Let $\PP$ be the intersection poset of a toric

2076: hyperplane arrangement

2077: whose induced subdivision is regular.

2078: Let $c = \{\hz = x_{0} < x_{1} < \cdots < x_{k} = \ho\}$

2079: be a chain in $\Pzero$

2080: with $k \geq 2$.

2081: Then the cardinality

2082: of the inverse image of the chain $c$ is given by the product

2083: $$      |\zt^{-1}(c)|

2084:     =

2085:         \prod_{i=2}^{k-1}

2086:                Z([x_{i-1},x_{i}])  

2087:     \cdot

2088:                \Zt([x_{k-1},x_{k}])     .  $$

2089: \label{theorem_Bayer_Sturmfels_toric}

2090: \end{theorem}

2091: \begin{proof}

2092: We need to count the number of ways we can select a chain

2093: $d = \{\hz = y_0 < y_1 < \dots < y_k = \ho\}$ in $T_{t}^{*}$

2094: such that $\zt(y_{i}) = x_{i}$.

2095: The number of ways to select

2096: the element $y_{k-1}$ in $T_{t}^{*}$

2097: is the number of regions

2098: in the arrangement restricted to the toric subspace $x_{k-1}$.

2099: By Theorem~\ref{theorem_toric_Zaslavsky_version_2}

2100: this can be done in $\Zt([x_{k-1},x_{k}])$ number of ways.

2101: Observe now that

2102: all other elements in the chain $d$

2103: contain the face~$y_{k-1}$.

2104: 

2105: To count the number of ways

2106: to select the element $y_{k-2}$, we follow the original argument

2107: of Bayer--Sturmfels. 

2108: We would like to pick the face $y_{k-2}$

2109: such that it contains the face $y_{k-1}$ and 

2110: it is a region in the toric subspace $x_{k-2}$. 

2111: This is equal to the number of regions in

2112: the central arrangement having the intersection lattice $[x_{k-2},x_{k-1}]$,

2113: which is given by $Z([x_{k-2},x_{k-1}])$.

2114: By iterating this procedure

2115: until we reach the element $y_{1}$, the result follows.

2116: \end{proof}

2117: 

2118: 

2119: 

2120: 

2121: \begin{corollary}

2122: The flag $f$-vector entry

2123: $f_{S}(T_{t})$

2124: of the face poset

2125: $T_{t}$ of a toric arrangement

2126: whose induced subdivision is regular subdivision of $T^{n}$

2127: is divisible by $2^{|S|-1}$

2128: for $S \subseteq \{1, \ldots, n+1\}$

2129: with

2130: $S \neq \emptyset$.

2131: \label{corollary_evenness}

2132: \end{corollary}

2133: \begin{proof}

2134: The proof follows from the fact

2135: that the Zaslavsky invariant $Z$ is an even integer

2136: and that a given flag $f$-vector entry is the appropriate

2137: sum of products appearing in

2138: Theorem~\ref{theorem_Bayer_Sturmfels_toric}.

2139: \end{proof}

2140: 

2141: 

2142: 

2143: 

2144: 

2145: 

2146: \subsection{The connection between posets and coalgebras}

2147: 

2148: 

2149: For an $\ab$-monomial $v$ 

2150: define the linear map $\lambda_{t}$ 

2151: by letting

2152: $$

2153:     \lambda_{t}(v)

2154:   =

2155:      \begin{cases}

2156:         (\av-\bv)^{m}   & \text{if $v=\bv^{m}$    for some $m \geq 0$,} \\

2157:         (\av-\bv)^{m+1} & \text{if $v=\bv^{m}\av$ for some $m \geq 0$,} \\

2158:         0               & \text{otherwise.}

2159: \end{cases}

2160: $$

2161: 

2162: 

2163: Define the linear operator $H^{\prime}$ on $\zab$

2164: to be the one which removes 

2165: the last letter in each $\ab$-monomial, that is,

2166: $H^{\prime}(w \cdot \av) = H^{\prime}(w \cdot \bv) = w$

2167: and $H^{\prime}(1) = 0$.

2168: We use the prime in the notation to distinguish it from the~$H$ map

2169: defined in~\cite[Section~8]{Billera_Ehrenborg_Readdy_om}

2170: which instead removes the first letter in each $\ab$-monomial.

2171: From~\cite{Billera_Ehrenborg_Readdy_om} we have the

2172: following lemma.

2173: \begin{lemma}

2174: For a graded poset $P$ with $\ho$ of rank greater than or equal to $2$,

2175: the following identity holds:

2176: $$

2177: H^{\prime}(\Psi(P))

2178:       =

2179: \sum_{x \mbox{ {\rm \tiny coatom of} } P} \Psi([\hz,x]) .

2180: $$

2181: \end{lemma}

2182: 

2183: 

2184: The next lemma gives the relation between

2185: the toric Zaslavsky invariant $\Zt$ and

2186: the map~$\lambda_{t}$.

2187: \begin{lemma}

2188: For a graded poset $P$ with $\ho$ of rank greater than or equal to $1$,

2189: the following identity holds:

2190: $$

2191:      \lambda_{t}(\Psi(P)) = \Zt(P) \cdot (\av-\bv)^{\rho(P)-1}.

2192: $$

2193: \end{lemma}

2194: \begin{proof}

2195: When $P$ has rank $1$, both sides are equal to $1$.

2196: For an $\ab$-monomial $v$ different from $1$, we have that

2197: $\lambda_{t}(v) = \beta(H^{\prime}(v)) \cdot (\av-\bv)$.

2198: Hence

2199: \begin{eqnarray*}

2200: \lambda_{t}(\Psi(P))

2201:   & = &

2202: \beta(H^{\prime}(\Psi(P))) \cdot (\av-\bv) \\

2203:   & = &

2204: \sum_{x \mbox{ {\rm \tiny coatom of} } P}

2205:   \beta(\Psi([\hz,x])) \cdot (\av-\bv) \\

2206:   & = &

2207: (-1)^{\rho(P)} \cdot

2208: \sum_{x \mbox{ {\rm \tiny coatom of} } P}

2209:   \mu(\hz,x) \cdot (\av-\bv)^{\rho(P)-1} ,

2210: \end{eqnarray*}

2211: which concludes the proof.

2212: \end{proof}

2213: 

2214: 

2215: Define a sequence of functions

2216: $\varphi_{t,k}\colon\zab\to\zab$ by $\varphi_{t,1}=\kappa$,

2217: and for $k \geq 2$,

2218: $$

2219: \varphi_{t,k}(v)

2220:   =

2221: \sum_{v}

2222:      \kappa(v_{(1)}) \cdot

2223:      \bv \cdot \eta(v_{(2)}) \cdot

2224:      \bv \cdot \eta(v_{(3)}) \cdot \bv \cdots

2225:      \bv \cdot \eta(v_{(k-1)}) \cdot

2226:      \bv \cdot \lambda_{t}(v_{(k)}).                      

2227: $$

2228: Finally, let $\varphi_{t}(v)$ be the

2229: sum $\varphi_{t}(v) = \sum_{k\geq 1}\varphi_{t,k}(v)$.

2230: 

2231: 

2232: 

2233: 

2234: 

2235: 

2236: \begin{theorem}

2237: The $\ab$-index of the face poset $T_{t}$ 

2238: of a toric arrangement is given by

2239: $$

2240:      \Psi(T_{t})^{*} = \varphi_{t}(\Psi(\Pzero)).

2241: $$

2242: \label{theorem_poset_varphi_toric}

2243: \end{theorem}

2244: \begin{proof}

2245: The $\ab$-index of the poset $T_{t}$

2246: is given by the sum 

2247: $\Psi(T_{t}) = \sum_{c} |\zt^{-1}(c)| \cdot \wt(c)$.

2248: Fix $k \geq 2$ and sum over all chains

2249: $c = \{\hz = x_{0} < x_{1} < \cdots < x_{k} = \ho\}$

2250: of length $k$.  We then have

2251: \begin{eqnarray*}

2252:   &   &

2253: \sum_{c} |\zt^{-1}(c)| \cdot \wt(c) \\

2254:   & = &

2255: \sum_{c} \prod_{i=2}^{k-1} Z([x_{i-1}, x_{i}]) 

2256:          \cdot \Zt([x_{k-1}, x_{k}])

2257:   \cdot

2258:       (\av-\bv)^{\rho(x_{0},x_{1})-1}

2259:         \cdot

2260:       \bv

2261:         \cdots

2262:       \bv

2263:         \cdot

2264:       (\av-\bv)^{\rho(x_{k-1},x_{k})-1} \\

2265:   & = &

2266: \sum_{c} 

2267:    \kappa(\Psi([x_{0},x_{1}]))

2268:        \cdot 

2269:    \prod_{i=2}^{k-1} \left(\bv \cdot \eta(\Psi([x_{i-1}, x_{i}]))\right)

2270:        \cdot 

2271:    \bv

2272:        \cdot 

2273:    \lambda_{t}(\Psi([x_{k-1},x_{k}])) \\

2274:   & = &

2275: \sum_{w}

2276:    \kappa(w_{(1)})

2277:        \cdot 

2278:    \prod_{i=2}^{k-1} \left(\bv \cdot \eta(w_{(i)})\right)

2279:        \cdot 

2280:    \bv

2281:        \cdot 

2282:    \lambda_{t}(w_{(k)}) \\

2283:   & = &

2284: \varphi_{t,k}(w)  ,  

2285: \end{eqnarray*}

2286: where we let $w$ denote the $\ab$-index of 

2287: the augmented intersection poset $\Pzero$.

2288: For $k=1$ we have that

2289: $(\av-\bv)^{\rho(T_{t})-1} = \varphi_{t,1}(\Psi(\Pzero))$.

2290: Summing over all $k \geq 1$, we obtain the result.

2291: \end{proof}

2292: 

2293: 

2294: 

2295: 

2296: 

2297: 

2298: \subsection{Evaluating the function $\varphi_{t}$}

2299: 

2300: \begin{proposition}

2301: For an $\ab$-monomial $v$,

2302: the following identity holds:

2303: $$ 

2304:         \varphi_{t}(v) 

2305:     =

2306:         \kappa(v)

2307:       +

2308:         \sum_{v} \varphi(v_{(1)}) \cdot \bv \cdot \lambda_{t}(v_{(2)}) .

2309: $$

2310: \label{proposition_varphi_t}

2311: \end{proposition}

2312: \begin{proof}

2313: Using the coassociative identity

2314: $\Delta^{k-1} = (\Delta^{k-2} \tensor \id) \circ \Delta$,

2315: for $k \geq 2$ we have that

2316: \begin{eqnarray*}

2317: \varphi_{t,k}(v)

2318:   & = &

2319: \sum_{v}              \kappa(v_{(1)}) \cdot

2320:                       \bv \cdot \eta(v_{(2)}) \cdot \bv \cdots

2321:                       \bv \cdot \eta(v_{(k-1)}) \cdot

2322:                       \bv \cdot \lambda_{t}(v_{(k)}) \\             

2323:   & = &

2324: \sum_{v}

2325: \sum_{v_{(1)}}        \kappa(v_{(1,1)}) \cdot

2326:                       \bv \cdot \eta(v_{(1,2)}) \cdot \bv \cdots

2327:                       \bv \cdot \eta(v_{(1,k-1)}) \cdot

2328:                       \bv \cdot \lambda_{t}(v_{(2)}) \\             

2329:   & = &

2330: \sum_{v}

2331:                       \varphi_{k-1}(v_{(1)}) \cdot

2332:                       \bv \cdot \lambda_{t}(v_{(2)}) .

2333: \end{eqnarray*}

2334: By  summing over all $k \geq 1$, the result follows.

2335: \end{proof}

2336: 

2337: 

2338: 

2339: \begin{lemma}

2340: Let $v$ be an $\ab$-monomial that begins with $\av$

2341: and let $x$ be either $\av$ or $\bv$. Then

2342: $$     \varphi_{t}(v \cdot \av \cdot x)

2343:    = 

2344:        \kappa(v \cdot \av \cdot x)

2345:      + 

2346:        {1}/{2} \cdot \omega(v \cdot \av\bv)  .  $$

2347: \label{lemma_toric_varphi_I}

2348: \end{lemma}

2349: \begin{proof}

2350: Using Proposition~\ref{proposition_varphi_t}

2351: we have

2352: \begin{eqnarray*}

2353:        \varphi_{t}(v \cdot \av \cdot x)

2354:   & = &

2355:        \kappa(v \cdot \av \cdot x)

2356:      +

2357:        \varphi(v \cdot \av) \cdot \bv \cdot \lambda_{t}(1)

2358:      +

2359:        \varphi(v) \cdot \bv \cdot \lambda_{t}(x)  \\

2360:   &   &

2361:      + \sum_{v}

2362:        \varphi(v_{(1)}) \cdot \bv \cdot

2363:                  \lambda_{t}(v_{(2)} \cdot \bv \cdot x)  \\

2364:   & = &

2365:        \kappa(v \cdot \av \cdot x)

2366:      +

2367:        \varphi(v) \cdot \cv \cdot \bv

2368:      +

2369:        \varphi(v) \cdot \bv \cdot (\av-\bv) \\

2370:   & = &

2371:        \kappa(v \cdot \av \cdot x)

2372:      +

2373:        \omega(v) \cdot \dv \\

2374:   & = &

2375:        \kappa(v \cdot \av \cdot x)

2376:      +

2377:        1/2 \cdot \omega(v \cdot \av\bv) ,

2378: \end{eqnarray*}

2379: since $\lambda_{t}(v_{(2)} \cdot \bv \cdot x) = 0$.

2380: \end{proof}

2381: 

2382: \begin{lemma}

2383: Let $v$ be an $\ab$-monomial that begins with $\av$,

2384: let $k$ be a positive integer 

2385: and let $x$ be either $\av$ or $\bv$.

2386: Then the following evaluation holds:

2387: $$     \varphi_{t}(v \cdot \av \bv^{k} \cdot x)

2388:    = 

2389:        \kappa(v \cdot \av \bv^{k} \cdot x)

2390:      + 

2391:        {1}/{2} \cdot \omega(v \cdot \av\bv^{k+1})  .  $$

2392: \label{lemma_toric_varphi_II}

2393: \end{lemma}

2394: \begin{proof_special}

2395: Using Proposition~\ref{proposition_varphi_t}

2396: we have

2397: \begin{eqnarray}

2398:        \varphi_{t}(v \cdot \av \bv^{k}\cdot x)

2399:      -

2400:        \kappa(v \cdot \av \bv^{k} \cdot x)

2401:   & = &

2402:        \varphi(v \cdot \av \bv^{k}) \cdot \bv \cdot \lambda_{t}(1)

2403:      +

2404:        \varphi(v \cdot \av) \cdot \bv \cdot \lambda_{t}(\bv^{k-1} \cdot x)

2405: \nonumber \\

2406:   &   &

2407:      +

2408:        \varphi(v) \cdot \bv \cdot \lambda_{t}(\bv^{k} \cdot x)

2409:      +

2410:        \sum_{i+j=k-2}

2411:            \varphi(v \cdot \av \bv^{i+1}) \cdot \bv \cdot \lambda_{t}(\bv^{j} \cdot x)

2412: \nonumber \\

2413:   & = &

2414:        \varphi(v)

2415:      \cdot

2416:        \left(

2417:        \vphantom{\sum_{i+j=k-2}}

2418: %% The \vphantom command is to make the left parenthesis the right size.

2419:        2 \dv \cv^{k-1} \cdot \bv

2420:      +

2421:        \cv \cdot \bv \cdot (\av-\bv)^{k}

2422:      +

2423:        \bv \cdot (\av-\bv)^{k+1} 

2424:        \right.

2425: \nonumber \\

2426:   &   &

2427:        \left.

2428:      +

2429:        \sum_{i+j=k-2}

2430:            2 \dv \cv^{i} \cdot \bv \cdot (\av-\bv)^{j+1}

2431:        \right) .

2432: \label{equation_messy}

2433: \end{eqnarray}

2434: In order to simplify this expression, consider the butterfly poset

2435: of rank $k$. 

2436: This is the poset consisting

2437: of two rank $i$ elements, for $i = 1, \ldots, k-1$,

2438: adjoined with a minimal and maximal element.

2439: Each of the rank $i$ elements covers

2440: the rank $i-1$ element(s) for $i = 1, \ldots, k-1$.

2441: The butterfly poset

2442: is the unique poset having the $\cd$-index $\cv^{k-1}$.

2443: It is also Eulerian.

2444: Applying~(\ref{equation_Stanley_recursion})

2445: to the butterfly poset, we have

2446: $$      \cv^{k-1}

2447:     =

2448:         (\av-\bv)^{k-1}

2449:       +

2450:         2 \cdot \sum_{i+j=k-2} \cv^{i} \cdot \bv \cdot (\av-\bv)^{j}

2451: .  $$

2452: Using this relation to simplify

2453: equation~(\ref{equation_messy}), we obtain

2454: \begin{eqnarray*}

2455: \hspace*{35 mm}

2456:        \varphi_{t}(v \cdot \av \bv^{k}\cdot x)

2457:      -

2458:        \kappa(v \cdot \av \bv^{k} \cdot x)

2459:   & = &

2460:        \varphi(v)

2461:      \cdot

2462:        \dv \cdot \cv^{k} \\

2463:   & = &

2464:        1/2

2465:      \cdot

2466:        \omega(v \cdot \av \bv^{k+1}) .

2467: \hspace*{35 mm} \qed

2468: \end{eqnarray*}

2469: \end{proof_special}

2470: 

2471: By combining Lemmas~\ref{lemma_toric_varphi_I}

2472: and~\ref{lemma_toric_varphi_II}, we have

2473: the following proposition.

2474: \begin{proposition}

2475: For an $\ab$-monomial $v$ that begins with the letter $\av$,

2476: the following holds:

2477: $$     \varphi_{t}(v)

2478:     =

2479:        \kappa(v)

2480:     +

2481:        1/2

2482:     \cdot

2483:        \omega(H^{\prime}(v) \cdot \bv)  .  $$

2484: \label{proposition_toric_varphi}

2485: \end{proposition}

2486: 

2487: 

2488: We now obtain the main result

2489: for computing  the $\ab$-index

2490: of the face poset of a toric arrangement.

2491: \begin{theorem}

2492: Let $\HH$ be a toric hyperplane arrangement on

2493: the $n$-dimensional torus $T^{n}$

2494: that subdivides the torus into a regular cell complex.

2495: Then the $\ab$-index of the face poset $T_{t}$

2496: can be computed from the 

2497: $\ab$-index of the intersection poset $\PP$

2498: as follows:

2499: $$     \Psi(T_{t})

2500:     =

2501:         (\av-\bv)^{n+1}

2502:     +

2503:        \frac{1}{2}

2504:      \cdot

2505:        \omega(\av \cdot H^{\prime}(\Psi(\PP)) \cdot \bv)^{*}  .  $$

2506: \label{theorem_toric}

2507: \end{theorem}

2508: 

2509: Observe that in

2510: Lemmas~\ref{lemma_toric_varphi_I} and~\ref{lemma_toric_varphi_II},

2511: Proposition~\ref{proposition_toric_varphi}

2512: and

2513: Theorem~\ref{theorem_toric}

2514: no

2515: rational coefficients were introduced.

2516: Only the $\ab$-monomial $\av^{n}$ is mapped

2517: to a $\cd$-polynomial with an odd coefficient,

2518: hence $1/2 \cdot \omega(v \cdot \bv)$

2519: has all integer coefficients.

2520: 

2521: 

2522: 

2523: \addtocounter{theorem}{-20}

2524: \begin{continuation}

2525: {\rm

2526: The flag $f$-vector of the intersection poset $\PP$ in

2527: Example~\ref{example_toric_two}

2528: is given by

2529: $(f_{\emptyset},f_{1},f_{2},f_{12}) = (1,3,7,15)$,

2530: the flag $h$-vector by

2531: $(h_{\emptyset},h_{1},h_{2},h_{12}) = (1,2,6,6)$,

2532: and so the  $\ab$-index is 

2533: $\Psi(P) = \maa + 2 \cdot \mba + 6 \cdot \mab + 6 \cdot \mbb$.

2534: Thus 

2535: \begin{eqnarray*}

2536:        \Psi(T_{t})

2537:   & = &

2538:         (\av-\bv)^{3}

2539:     +

2540:        1/2

2541:      \cdot

2542:        \omega(\av \cdot 

2543:               H^{\prime}(\maa + 2 \cdot \mba + 6 \cdot \mab + 6 \cdot \mbb)

2544:               \cdot \bv)^{*}  \\

2545:   & = &

2546:         (\av-\bv)^{3}

2547:     +

2548:        1/2

2549:      \cdot

2550:        \omega(\av \cdot (7 \cdot \ma + 8 \cdot \mb)

2551:                 \cdot \bv)^{*}  \\

2552:   & = &

2553:         (\av-\bv)^{3}

2554:     +

2555:        1/2

2556:      \cdot

2557:        \omega(7 \cdot \maab + 8 \cdot \mabb)^{*}  \\

2558:   & = &

2559:         (\av-\bv)^{3}

2560:     +

2561:        7 \cdot \mdc + 8 \cdot \mcd ,

2562: \end{eqnarray*}

2563: which agrees with the

2564: calculation in Example~\ref{example_toric_two}.

2565: }

2566: \end{continuation}

2567: \addtocounter{theorem}{19}

2568: 

2569: 

2570: 

2571: 

2572: Theorem~\ref{theorem_toric}

2573: gives a different approach 

2574: from

2575: Corollary~\ref{corollary_f_vector_I}

2576: for determining 

2577: the $f$-vector of $T_{t}$.

2578: For notational ease, for positive integers $i$ and $j$,

2579: let $[i,j] = \{ i, i+1, \ldots, j\}$

2580: and

2581: $[j] = \{1, \ldots, j\}$.

2582: 

2583: \begin{corollary}

2584: The number of $i$-dimensional regions in the subdivision $T_{t}$

2585: of the $n$-dimensional torus is given by the following

2586: sum

2587: of  flag $h$-vector entries from

2588: the intersection poset $\PP$:

2589: $$

2590:       f_{i+1}(T_{t})

2591:   =

2592:       h_{[n-i,n]}(\PP)

2593:    +

2594:       h_{[n-i,n-1]}(\PP) 

2595:    +

2596:       h_{[n-i+1,n]}(\PP)

2597:    +

2598:       h_{[n-i+1,n-1]}(\PP)     ,

2599: $$

2600: for $1 \leq i \leq n-1$.

2601: The number

2602: of vertices is given by

2603: $f_{1}(T_{t}) = 1 + h_{n}(\PP)$

2604: and the number of

2605: maximal regions

2606: by

2607: $f_{n+1}(T_{t})

2608:  = h_{[n-1]}(\PP)

2609:  + h_{[n]}(\PP)$.

2610: \label{corollary_f_vector_II}

2611: \end{corollary}

2612: \begin{proof}

2613: Let $\pair{\cdot}{\cdot}$ denote the inner product on $\zab$

2614: defined by $\pair{u}{v} = \delta_{u,v}$

2615: for two $\ab$-monomials $u$ and $v$.

2616: For $1 \leq i \leq n-1$ we have

2617: \begin{eqnarray*}

2618:          f_{i+1}(T_{t})

2619:   & = &

2620:          1 + h_{i+1}(T_{t}) \\

2621:   & = &

2622:          1 + \pair{\av^{i} \bv \av^{n-i}}{\Psi(T_{t})} \\

2623:   & = &

2624:        \frac{1}{2}

2625:      \cdot

2626:        \pair{\av^{i} \bv \av^{n-i}}

2627:        {\omega(\av \cdot H^{\prime}(\Psi(\PP)) \cdot \bv)^{*}}  \\

2628:   & = &

2629:        \frac{1}{2}

2630:      \cdot

2631:        [\cv^{i-1} \dv \cv^{n-i}]

2632:        \omega(\av \cdot H^{\prime}(\Psi(\PP)) \cdot \bv)^{*}  

2633:     +

2634:        \frac{1}{2}

2635:      \cdot

2636:        [\cv^{i} \dv \cv^{n-i-1}]

2637:        \omega(\av \cdot H^{\prime}(\Psi(\PP)) \cdot \bv)^{*}  \\

2638:   & = &

2639:        \pair{\av^{n-i} \cdot \av\bv \cdot \bv^{i-1}

2640:               +

2641:              \av^{n-i-1} \cdot \av\bv \cdot \bv^{i}}

2642:             {\av \cdot H^{\prime}(\Psi(\PP)) \cdot \bv} \\

2643:   & = &

2644:        \pair{\av^{n-i-1} \cdot (\av+\bv) \cdot \bv^{i-1}}

2645:             {H^{\prime}(\Psi(\PP))} \\

2646:   & = &

2647:        \pair{\av^{n-i-1} \cdot (\av+\bv) \cdot \bv^{i-1} \cdot (\av+\bv)}

2648:             {\Psi(\PP)} .

2649: \end{eqnarray*}

2650: Expanding in terms of the flag $h$-vector the result follows.

2651: The expressions for $f_{1}$ and $f_{n+1}$ are

2652: obtained by similar calculations.

2653: \end{proof}

2654: 

2655: 

2656: The fact that Corollaries~\ref{corollary_f_vector_I}

2657: and~\ref{corollary_f_vector_II}

2658: are equivalent follows from

2659: the coalgebra techniques

2660: in Theorem~\ref{theorem_Ehrenborg_Readdy}.

2661: 

2662: 

2663: 

2664: 

2665: 

2666: 

2667: \section{The complex of unbounded regions}

2668: \label{section_affine}

2669: 

2670: 

2671: \subsection{Zaslavsky and Bayer--Sturmfels}

2672: 

2673: The {\em unbounded Zaslavsky invariant} is defined by

2674: $$    \Zub(P)

2675:     =

2676:       Z(P) - 2 \cdot Z_{b}(P)  .  $$

2677: As the name suggests, the number of unbounded regions in 

2678: a non-central

2679: arrangement is given by this invariant.

2680: By taking the difference of the

2681: two statements in Theorem~\ref{theorem_Zaslavsky_poset} part (ii),

2682: we immediately have the following result.

2683: \begin{lemma}

2684: For a non-central hyperplane arrangement $\HH$ the number of

2685: unbounded regions is given by $\Zub(\LL)$,

2686: where $\LL$

2687: is the intersection lattice of the arrangement $\HH$.

2688: \label{lemma_Zaslavsky_unbounded}

2689: \end{lemma}

2690: 

2691: 

2692: 

2693: 

2694: 

2695: 

2696: \begin{figure}

2697: \begin{center}

2698: \scalebox{.7}{\epsfig{file=ekazonotope.eps}}

2699: \end{center}

2700: \caption{The non-central arrangement $x,y,z=0,1$.}

2701: \label{figure_affine}

2702: \end{figure}

2703: 

2704: 

2705: 

2706: 

2707: 

2708: 

2709: 

2710: 

2711: Let $\HH$ be a non-central hyperplane arrangement 

2712: in $\Rrr^n$

2713: with intersection

2714: lattice $\LL$ having the empty set $\emptyset$

2715: as the maximal element.

2716: Let $\Lub$ denote 

2717: the {\em unbounded intersection lattice},

2718: that is,

2719: the subposet of the intersection lattice

2720: consisting of all affine subspaces 

2721: with the points (dimension zero affine subspaces) omitted

2722: but with the empty set $\emptyset$ continuing to be the maximal element.

2723: Equivalently, the poset

2724: $\Lub$ is the rank-selected poset

2725: $\LL([1, n-1])$,

2726: that is,

2727: the poset

2728: $\LL$ with the coatoms removed.

2729: 

2730: 

2731: Let $T$ be the face lattice of the arrangement

2732: $\HH$ with the minimal element $\hz$ denoting the empty face

2733: and the maximal element denoted by $\ho$.

2734: Similarly, let $T_{ub}$ denote the set of all faces

2735: in the face lattice~$T$ which are not bounded.

2736: Observe that $T_{ub}$ includes the minimal and maximal elements of $T$

2737: and that $T_{ub}$ is the face poset of

2738: an $(n-1)$-dimensional sphere. 

2739: Pick~$R$ large enough

2740: so that all of the bounded faces

2741: are strictly inside a ball of radius~$R$.

2742: Intersect

2743: the arrangement~$\HH$ with a sphere of radius $R$.

2744: The resulting cell complex has face poset $T_{ub}$.

2745: Our goal is to compute the $\cd$-index of $T_{ub}$ in terms

2746: of the $\ab$-index of $\Lub$.

2747: 

2748: 

2749: 

2750: 

2751: 

2752: 

2753: 

2754: The collection of unbounded

2755: faces of the arrangement $\HH$

2756: forms a lower order ideal in the poset $T^{*}$.

2757: Let~$Q$ be the subposet of $T^{*}$

2758: consisting of this ideal with a  maximal element $\ho$

2759: adjoined.

2760: We define the rank of an element in $Q$ to be its rank

2761: in the original poset $T^{*}$, that is, for $x \in Q$

2762: let $\rho_{Q}(x) = \rho_{T^{*}}(x)$.

2763: This rank convention will simplify the later arguments.

2764: As posets $T_{ub}^{*}$ and $Q$ are isomorphic.

2765: However, since their rank functions differ, their $\ab$-indexes 

2766: satisfy

2767: $\Psi(T_{ub})^{*} \cdot (\av-\bv) = \Psi(Q)$.

2768: 

2769: 

2770: 

2771: Restrict the zero map

2772: $z : T^{*} \longrightarrow \Lzero$

2773: to 

2774: form the map $\zub : Q \longrightarrow \Lzero$.

2775: Note that the map $\zub$ is order- and rank-preserving,

2776: but not necessarily surjective.

2777: As before we view the map $\zub$ as a map from the set of chains

2778: of $Q$ to the set of chains of $\Lzero$.

2779: We have the following theorem, which is analogous

2780: to the Bayer--Sturmfels result

2781: (Theorem~\ref{theorem_Bayer_Sturmfels}).

2782: \begin{theorem}

2783: Let $\HH$ be a non-central hyperplane arrangement

2784: with intersection lattice $\LL$.

2785: Let $c=\{\hz = x_0 < x_1 < \dots < x_k = \ho\}$ be a chain in

2786: $\Lzero$ with $k \geq 2$.

2787: Then the cardinality of the inverse image of the chain $c$

2788: under $\zub$ is given by

2789: $$      |\zub^{-1}(c)|

2790:      =

2791:         \prod_{i=2}^{k-1}

2792:              Z([x_{i-1},x_{i}])

2793:        \cdot

2794:              \Zub([x_{k-1},x_{k}]) . $$

2795: \label{theorem_Bayer_Sturmfels_unbounded}

2796: \end{theorem}

2797: \begin{proof}

2798: We need to count the number of ways we can select a chain

2799: $d = \{\hz = y_0 < y_1 < \dots < y_k = \ho\}$ in 

2800: the poset of unbounded regions $Q$

2801: such that $\zub(y_{i}) = x_{i}$.

2802: The number of ways to select

2803: the element~$y_{k-1}$ in $Q$

2804: is the number of unbounded regions

2805: in the arrangement restricted to the subspace $x_{k-1}$.

2806: By Lemma~\ref{lemma_Zaslavsky_unbounded}

2807: this can be done in $\Zub([x_{k-1},x_{k}])$ number of ways.

2808: Since $y_{k-1}$ is an unbounded face of the arrangement

2809: and all other elements in the chain $d$

2810: contain the face $y_{k-1}$, the other elements must be unbounded.

2811: 

2812: The remainder of the proof is the same as

2813: that of Theorem~\ref{theorem_Bayer_Sturmfels_toric}.

2814: \end{proof}

2815: 

2816: 

2817: 

2818: 

2819: \begin{corollary}

2820: The flag $f$-vector entry

2821: $f_{S}(T_{ub})$

2822: is divisible by $2^{|S|}$

2823: for any index set $S \subseteq \{1, \ldots, n\}$.

2824: \end{corollary}

2825: \begin{proof}

2826: The proof is the same as 

2827: Corollary~\ref{corollary_evenness}

2828: with the extra observation that

2829: the Zaslavsky invariant $\Zub$ is even.

2830: \end{proof}

2831: 

2832: 

2833: 

2834: 

2835: 

2836: \subsection{The connection between posets and coalgebras}

2837: 

2838: Define $\lambda_{ub}$ by

2839: $\lambda_{ub} = \eta - 2 \cdot \beta$.

2840: By equations~(\ref{equation_beta}) 

2841: and~(\ref{equation_eta}),

2842: for a graded poset $P$ we have

2843: $$  \lambda_{ub}(\Psi(P)) = \Zub(P) \cdot (\av-\bv)^{\rho(P)-1}  .  $$

2844: Define a sequence of functions

2845: $\varphi_{ub,k}\colon\zab\to\zab$ by $\varphi_{ub,1}=\kappa$ and for

2846: $k>1$,

2847: $$

2848: \varphi_{ub,k}(v)

2849:      =

2850: \sum_{v}

2851:        \kappa(v_{(1)}) \cdot

2852:        \bv \cdot \eta(v_{(2)}) \cdot

2853:        \bv \cdot \eta(v_{(3)}) \cdot \bv \cdots

2854:        \bv \cdot \eta(v_{(k-1)}) \cdot

2855:        \bv \cdot \lambda_{ub}(v_{(k)}).                      

2856: $$

2857: Finally, let $\varphi_{ub}(v)$ be the

2858: sum $\varphi_{ub}(v) = \sum_{k\geq 1}\varphi_{ub,k}(v)$.

2859: 

2860: 

2861: Similar to Theorem~\ref{theorem_poset_varphi_toric} we

2862: have the next result.

2863: The proof only differs in replacing

2864: the map

2865: $\zt : T_{t}^{*} \longrightarrow \Pzero$

2866: with 

2867: $\zub : Q \longrightarrow \Lzero$

2868: and the invariant $\Zt$ by $\Zub$.

2869: \begin{theorem}

2870: The $\ab$-index of the poset $Q$ 

2871: of unbounded regions of a

2872: non-central 

2873: arrangement is given by

2874: $$

2875:    \Psi(Q) = \varphi_{ub}(\Psi(\Lzero)).

2876: $$

2877: \label{theorem_poset_varphi_unbounded}

2878: \end{theorem}

2879: 

2880: 

2881: \subsection{Evaluating the function $\varphi_{ub}$}

2882: 

2883: 

2884: In this subsection we

2885: analyze the behavior of $\varphi_{ub}$.

2886: \begin{lemma}

2887: For any $\ab$-monomial $v$,

2888: $$       \varphi_{ub}(v)

2889:      =

2890:          \varphi(v)

2891:        -

2892:          2 \cdot 

2893:          \sum_v \varphi(v_{(1)}) \cdot \bv \cdot \beta(v_{(2)}) . $$

2894: \label{lemma_phi_ub_recursion}

2895: \end{lemma}

2896: \begin{proof}

2897: Using the coassociative identity

2898: $\Delta^{k-1} = (\Delta^{k-2} \tensor \id) \circ \Delta$,

2899: we have for $k \geq 2$

2900: \begin{eqnarray*}

2901: \varphi_{ub,k}(v)

2902:   & = &

2903: \varphi_{k}(v)

2904:   -

2905: 2 \cdot 

2906: \sum_{v}            \kappa(v_{(1)}) \cdot

2907:                     \bv \cdot \eta(v_{(2)}) \cdot

2908:                     \bv \cdots

2909:                     \bv \cdot \eta(v_{(k-1)}) \cdot

2910:                     \bv \cdot \beta(v_{(k)}) \\

2911:   & = &

2912: \varphi_{k}(v)

2913:   -

2914: 2 \cdot 

2915: \sum_{v}

2916: \sum_{v_{(1)}}      \kappa(v_{(1,1)}) \cdot

2917:                     \bv \cdot \eta(v_{(1,2)}) \cdot

2918:                     \bv \cdots

2919:                     \bv \cdot \eta(v_{(1,k-1)}) \cdot

2920:                     \bv \cdot \beta(v_{(2)}) \\

2921:   & = &

2922: \varphi_{k}(v)

2923:   -

2924: 2 \cdot 

2925: \sum_{v}

2926: \varphi_{k-1}(v_{(1)}) \cdot \bv \cdot \beta(v_{(2)}) .

2927: \end{eqnarray*}

2928: The result then follows by

2929: summing over all $k \geq 2$ and adding

2930: $\varphi_{ub,1}(v) = \kappa(v) = \varphi_{1}(v)$.

2931: \end{proof}

2932: 

2933: 

2934: \begin{lemma}

2935: Let $v$ be an $\ab$-monomial.  Then 

2936: $$   \varphi_{ub}(v \cdot \av)

2937:    =

2938:       \varphi(v) \cdot (\av - \bv) .   $$

2939: \label{lemma_varphi_ub_I}

2940: \end{lemma}

2941: \begin{proof}

2942: By Lemma~\ref{lemma_phi_ub_recursion}

2943: and the Newtonian relation~(\ref{equation_Newtonian}) we have

2944: $$   \varphi_{ub}(v \cdot \av)

2945:  = 

2946:      \varphi(v \cdot \av) 

2947:    -

2948:      2 \cdot

2949:      \varphi(v) \cdot \bv \cdot \beta(1)

2950:    -

2951:      2 \cdot

2952:      \sum_v \varphi(v_{(1)}) \cdot \bv \cdot \beta(v_{(2)} \cdot \av) .  $$

2953: By equation~(\ref{equation_varphi_a})

2954: $\varphi(v \cdot \av) = \varphi(v) \cdot \cv$.

2955: The summation above is

2956: zero because $\beta(v_{(2)} \cdot \av)$ is always zero.  Hence

2957: $\varphi_{ub}(v \cdot \av) 

2958:    =

2959:   \varphi(v) \cdot (\cv - 2\bv)

2960:    =

2961:   \varphi(v) \cdot (\av - \bv)$.

2962: \end{proof}

2963: 

2964: \begin{lemma}

2965: Let $v$ be an $\ab$-monomial.  Then

2966: \[

2967: \varphi_{ub}(v \cdot \bv\bv) = \varphi_{ub}(v \cdot \bv) \cdot (\av - \bv).

2968: \]

2969: \label{lemma_varphi_ub_II}

2970: \end{lemma}

2971: \begin{proof}

2972: Let $u = v \cdot \bv$. 

2973: Applying

2974: Lemma~\ref{lemma_phi_ub_recursion}

2975: and the Newtonian relation~(\ref{equation_Newtonian})

2976: to $u$ gives

2977: \begin{eqnarray*}

2978: \varphi_{ub}(u \cdot \bv) 

2979:   & = &

2980: \varphi(u \cdot \bv)

2981:   - 

2982:     2 \cdot

2983:     \varphi(u) \cdot \bv \cdot \beta(1)

2984:   - 

2985:     2 \cdot

2986:     \sum_{u} \varphi(u_{(1)}) \cdot \bv \cdot \beta(u_{(2)} \cdot \bv) \\

2987:   & = &

2988: \varphi(u) \cdot (\cv - 2\bv)

2989:   - 

2990:     2 \cdot

2991:     \sum_{u} \varphi(u_{(1)}) \cdot \bv \cdot \beta(u_{(2)}) \cdot (\av - \bv) \\

2992:   & = &

2993: \left(\varphi(u) 

2994:       - 

2995:       2 \cdot

2996:       \sum_{u} \varphi(u_{(1)}) \cdot \bv \cdot 

2997:                  \beta(u_{(2)})

2998: \right) \cdot (\av - \bv) \\

2999:   & = &

3000: \varphi_{ub}(u) \cdot (\av - \bv).

3001: \end{eqnarray*}

3002: Here we have used the two facts

3003: $\varphi(u \cdot \bv) = \varphi(u) \cdot \cv$

3004: and 

3005: $\beta(u_{(2)} \cdot \bv) = \beta(u_{(2)}) \cdot (\av-\bv)$.

3006: \end{proof}

3007: 

3008: \begin{lemma}

3009: Let $v$ be an $\ab$-monomial.

3010: Then $\varphi_{ub}(v \cdot \ab) = 0$.

3011: \label{lemma_varphi_ub_III}

3012: \end{lemma}

3013: \begin{proof}

3014: Directly we have

3015: \begin{eqnarray*}

3016: \varphi_{ub}(v \cdot \ab)

3017:   & = &

3018:   \varphi(v \cdot \ab)

3019:    - 

3020:      2 \cdot

3021:      \varphi(v) \cdot \bv \cdot \beta(\bv)

3022:    - 

3023:      2 \cdot

3024:      \varphi(v \cdot \av) \cdot \bv \cdot \beta(1) 

3025:    - 

3026:      2 \cdot

3027:      \sum_v \varphi(v_{(1)}) \cdot \bv \cdot \beta(v_{(2)} \cdot \av \bv) \\

3028:   & = &

3029:   \varphi(v) \cdot 2\dv

3030: - 

3031:   2 \cdot \varphi(v) \cdot \bv \cdot (\av - \bv)

3032: - 

3033:   2 \cdot \varphi(v) \cdot \cv \bv \\

3034:   & = &

3035:   2 \cdot \varphi(v) \cdot (\dv - \bv(\av - \bv) - \cv\bv) \\

3036:   & = &

3037: 0 ,

3038: \end{eqnarray*}

3039: where we have used the facts

3040: $\varphi(v \cdot \av\bv) = \varphi(v) \cdot 2\dv$

3041: and 

3042: $\beta(v_{(2)} \cdot \av\bv) = 0$.

3043: \end{proof}

3044: 

3045: 

3046: The previous three lemmas enable us to determine $\varphi_{ub}$.

3047: In order to obtain more compact notation, 

3048: define a map $r\colon\zab\to\zab$ by $r(1)=0$,

3049: $r(v \cdot \av)=v$,

3050: and

3051: $r(v \cdot \bv)=0$.

3052: By using the chain definition of the $\ab$-index,

3053: it is straightforward to see that

3054: $\Psi(\Lub) = r(\Psi(\LL))$.

3055: 

3056: 

3057: \begin{proposition}

3058: Let $w$ be an $\ab$-polynomial homogeneous of degree greater than zero.

3059: Then

3060: $$    \varphi_{ub}(\av \cdot w)

3061:     =

3062:       \omega(\av \cdot r(w)) \cdot (\av - \bv)   .  $$

3063: \end{proposition}

3064: \begin{proof}

3065: The case

3066: $w = v \cdot \av$ 

3067: follows from Lemma~\ref{lemma_varphi_ub_I}.

3068: The remaining  case is $w = v \cdot \bv$.

3069: Note that $\av \cdot v \cdot \bv$ can be factored as

3070: $u \cdot \av\bv \cdot \bv^{k}$ for a monomial $u$.

3071: Hence

3072: $\varphi_{ub}(u \cdot \av\bv \cdot \bv^{k})

3073:   =

3074:  \varphi_{ub}(u \cdot \av\bv) \cdot (\av-\bv)^{k}

3075:   =

3076:  0$

3077: by Lemmas~\ref{lemma_varphi_ub_II}

3078: and~\ref{lemma_varphi_ub_III}.

3079: \end{proof}

3080: 

3081: 

3082: 

3083: 

3084: 

3085: 

3086: 

3087: We combine all of these results to 

3088: conclude that the $\cd$-index of

3089: the poset of unbounded regions~$T_{ub}$ can be computed

3090: in terms of the $\ab$-index of the unbounded intersection lattice

3091: $\Lub$.

3092: 

3093: 

3094: 

3095: \begin{theorem}

3096: Let $\HH$ be a non-central hyperplane arrangement with

3097: the unbounded intersection

3098: lattice $\Lub$ and poset of unbounded regions $T_{ub}$.

3099: Then the $\ab$-index of $T_{ub}$ is given by

3100: $$    \Psi(T_{ub})

3101:     =

3102:       \omega(\av \cdot \Psi(\Lub))^{*} . $$

3103: \label{theorem_unbounded}

3104: \end{theorem}

3105: \begin{proof}

3106: We have that

3107: \begin{eqnarray*}

3108: \Psi(T_{ub})^{*} \cdot (\av-\bv)

3109:   & = &

3110: \Psi(Q) \\

3111:   & = &

3112: \varphi_{ub}(\av \cdot \Psi(\LL)) \\

3113:   & = &

3114: \omega(\av \cdot r(\Psi(\LL))) \cdot (\av-\bv) \\

3115:   & = &

3116: \omega(\av \cdot \Psi(\Lub)) \cdot (\av-\bv) .

3117: \end{eqnarray*}

3118: By cancelling $\av-\bv$ on both sides of this identity,

3119: the result follows.

3120: \end{proof}

3121: 

3122: 

3123: 

3124: \begin{figure}

3125: \begin{center}

3126: \scalebox{.3}{\epsfig{file=spherical.eps}}

3127: \end{center}

3128: \caption{The spherical subdivision

3129: obtained from the non-central arrangement $x,y,z=0,1$.}

3130: \label{figure_spherical}

3131: \end{figure}

3132: 

3133: 

3134: 

3135: 

3136: \begin{example}

3137: {\rm

3138: Consider the non-central hyperplane arrangement

3139: consisting of the six hyperplanes

3140: $x = 0,1$, 

3141: $y = 0,1$ and

3142: $z = 0,1$.

3143: See Figure~\ref{figure_affine}.

3144: After intersecting this arrangement

3145: with a sphere of large enough radius

3146: we obtain the cell complex in

3147: Figure~\ref{figure_spherical}.

3148: The polytopal realization of this complex

3149: is known as the rhombicuboctahedron.

3150: The dual of the face lattice of this spherical complex is

3151: not realized by a zonotope. However, one can view

3152: the dual lattice as the face lattice

3153: of a $2 \times 2 \times 2$ pile of cubes.

3154: 

3155: The intersection lattice $\LL$ is the

3156: face lattice of the three-dimensional crosspolytope,

3157: in other words,

3158: the octahedron.

3159: Hence the lattice of unbounded intersections

3160: $\Lub$

3161: has the flag $f$-vector

3162: $(f_{\emptyset},f_{1},f_{2},f_{12}) = (1,6,12,24)$

3163: and the flag $h$-vector 

3164: $(h_{\emptyset},h_{1},h_{2},h_{12}) = (1,5,11,7)$.

3165: The $\ab$-index is given by

3166: $\Psi(\Lub) = \maa + 5 \cdot \mba + 11 \cdot \mab + 7 \cdot \mbb$.

3167: Hence the $\cd$-index of $T_{ub}$ is

3168: \begin{eqnarray*}

3169: \Psi(T_{ub})

3170:   & = &

3171:       \omega(\maaa + 5 \cdot \maba + 11 \cdot \maab + 7 \cdot \mabb)^{*} \\

3172:   & = &

3173:       \mccc + 22 \cdot \mdc + 24 \cdot \mcd  .

3174: \end{eqnarray*}

3175: }

3176: \end{example}

3177: 

3178: 

3179: 

3180: 

3181: 

3182: 

3183: 

3184: \section{Concluding remarks}

3185: 

3186: 

3187: 

3188: 

3189: For regular subdivisions of manifolds there is now

3190: a plethora of questions to ask.

3191: \begin{itemize}

3192: \item[(i)] What is the right analogue of a regular subdivision

3193: in order that it  be polytopal? 

3194: Can flag $f$-vectors be classified for

3195: polytopal subdivisions?

3196: \item[(ii)] Is there a Kalai convolution for manifolds

3197: that will generate more inequalities for flag $f$-vectors?

3198: \cite{Kalai}

3199: \item[(iii)] Is there a lifting technique that will yield

3200:              more inequalities for higher dimensional manifolds?

3201: \cite{Ehrenborg_lifting}

3202: \item[(iv)] Are there minimization inequalities for the

3203: $\cd$-coefficients in the polynomial $\Psi$?

3204: As a first step, can one prove the non-negativity of $\Psi$?

3205: \cite{Billera_Ehrenborg,Ehrenborg_Karu}

3206: \item[(v)] Is there an extension of the toric $g$-inequalities

3207: to manifolds?

3208: \cite{Bayer_Ehrenborg,Kalai_g,Karu,Stanley_h}

3209: \item[(vi)] Can the coefficients for $\Psi$ be minimized for regular toric

3210: arrangements as was done in the case of central hyperplane arrangements?

3211: \cite{Billera_Ehrenborg_Readdy_om}

3212: \end{itemize}

3213: 

3214: 

3215: 

3216: 

3217: 

3218: 

3219: 

3220: The most straightforward manifold to study is

3221: $n$-dimensional projective space $P^{n}$. We offer the following

3222: result in obtaining the $\ab$-index of subdivisions of $P^{n}$.

3223: \begin{theorem}

3224: Let $\Omega$ be a centrally symmetric regular subdivision of

3225: the $n$-dimensional sphere $S^{n}$. Assume that when 

3226: antipodal points of the sphere are identified,

3227: a regular subdivision

3228: $\Omega^{\prime}$ of the projective space $P^{n}$

3229: is obtained.

3230: Then the $\ab$-index of $\Omega^{\prime}$ is given by

3231: $$    \Psi(\Omega^{\prime})

3232:    =

3233:       \frac{\cv^{n+1} + (\av-\bv)^{n+1}}{2}

3234:    +

3235:       \frac{\Phi}{2}   ,   $$

3236: where the $\cd$-index of $\Omega$ is

3237: $\Psi(\Omega) = \cv^{n+1} + \Phi$.

3238: \end{theorem}

3239: \begin{proof}

3240: Each chain $c = \{\hz = x_{0} < x_{1} < \cdots < x_{k} = \ho\}$ with

3241: $k \geq 2$ in $\Omega^{\prime}$ corresponds to two chains in~$\Omega$

3242: with the same weight $\wt(c)$. The chain

3243: $c = \{\hz = x_{0} < x_{1} = \ho\}$ corresponds to exactly

3244: one chain in~$\Omega$ and has weight $(\av-\bv)^{n+1}$.

3245: Hence

3246: $\Psi(\Omega) = 2 \cdot \Psi(\Omega^{\prime}) - (\av-\bv)^{n+1}$,

3247: proving the result.

3248: \end{proof}

3249: 

3250: 

3251: The results in this paper have been stated

3252: for hyperplane arrangements. In true generality

3253: one could

3254: work with the underlying oriented matroid,

3255: especially  since there are nonrealizable ones 

3256: such as the non-Pappus oriented matroid.

3257: All of these can be  represented as pseudo-hyperplane

3258: arrangements. However, we have chosen to work with

3259: hyperplane arrangements in order not to lose

3260: the geometric intuition.

3261: 

3262: 

3263: 

3264: 

3265: Poset transformations related to the $\omega$ map appear

3266: in~\cite{Ehrenborg_r_Birkhoff,Ehrenborg_Readdy_Tchebyshev,Hsiao}.

3267: Are there toric or affine analogues of these poset transforms?

3268: 

3269: 

3270: Another way to encode the flag $f$-vector data of a poset

3271: is to use the quasisymmetric function of a

3272: poset~\cite{Ehrenborg_Hopf}.

3273: In this language the $\omega$ map is translated to

3274: Stembridge's $\vartheta$ map;

3275: see~\cite{Billera_Hsiao_van_Willigenburg,Stembridge}.

3276: Would the results of

3277: Theorems~\ref{theorem_toric}

3278: and~\ref{theorem_unbounded}

3279: be appealing in the quasisymmetric function viewpoint?

3280: 

3281: 

3282: 

3283: 

3284: 

3285: Richard Stanley has asked if the coefficients of the toric

3286: characteristic polynomial are alternating. If so, is there any

3287: combinatorial interpretation of the absolute values of the

3288: coefficients.

3289: 

3290: 

3291: A far reaching generalization of Zaslavsky's results for hyperplane

3292: arrangements is by Goresky and MacPherson~\cite{Goresky_MacPherson}.

3293: Their results determine the cohomology groups of the

3294: complement of a complex hyperplane arrangement. 

3295: For a toric analogue of the Goresky--MacPherson results,

3296: see work of De~Concini and Procesi~\cite{De_Concini_Procesi}.

3297: For algebraic considerations of toric arrangements,

3298: see~\cite{Douglass,Macmeikan_I, Macmeikan_II, Macmeikan_III}.

3299: 

3300: 

3301: 

3302: Greene and Zaslavsky~\cite{Greene_Zaslavsky}

3303: also give an interpretation of the derivative of the chromatic

3304: polynomial of a graph evaluated at $1$. Given an edge $ij$ of a graph $G$,

3305: the number of acyclic orientations of the graph

3306: with unique source at $i$ and

3307: unique sink at $j$ is given by

3308: $(-1)^{n} \cdot \left.\frac{d}{dt} \chi(G;t)\right|_{t=1}$.

3309: Is there a geometric proof of this fact

3310: analogous to the proof of 

3311: Theorem~\ref{theorem_Greene_Zaslavsky}?

3312: 

3313: 

3314: 

3315: 

3316: In Section~\ref{section_toric} we restricted ourselves

3317: to studying arrangements that cut the torus into

3318: regular cell complexes. In a future paper~\cite{Ehrenborg_Slone},

3319: two of the authors are developing the notion

3320: of a $\cd$-index for non-regular cell complexes.

3321: 

3322: 

3323: 

3324: \section*{Acknowledgements}

3325: 

3326: The authors thank

3327: the MIT Mathematics Department where

3328: this research was carried out.

3329: We also thank the

3330: referee for his careful comments

3331: and for bringing the 1977 paper of Zaslavsky

3332: to our attention,

3333: Andrew Klapper for his suggestions to improve the exposition,

3334: and Tricia Hersh for suggesting that we look at

3335: graphical arrangements on the torus.

3336: The first and third authors were partially

3337: supported by

3338: National Security Agency grant H98230-06-1-0072.

3339: 

3340: 

3341: 

3342: 

3343: 

3344: 

3345: 

3346: \newcommand{\journal}[6]{{\sc #1,} #2, {\it #3} {\bf #4} (#5), #6.}

3347: \newcommand{\book}[4]{{\sc #1,} ``#2,'' #3, #4.}

3348: \newcommand{\bookf}[5]{{\sc #1,} ``#2,'' #3, #4, #5.}

3349: \newcommand{\books}[6]{{\sc #1,} ``#2,'' #3, #4, #5, #6.}

3350: \newcommand{\collection}[6]{{\sc #1,}  #2, #3, in {\it #4}, #5, #6.}

3351: \newcommand{\thesis}[4]{{\sc #1,} ``#2,'' Doctoral dissertation, #3, #4.}

3352: \newcommand{\springer}[4]{{\sc #1,} ``#2,'' Lecture Notes in Math.,

3353:                           Vol.\ #3, Springer-Verlag, Berlin, #4.}

3354: \newcommand{\preprint}[3]{{\sc #1,} #2, preprint #3.}

3355: \newcommand{\preparation}[2]{{\sc #1,} #2, in preparation.}

3356: \newcommand{\appear}[3]{{\sc #1,} #2, to appear in {\it #3}}

3357: \newcommand{\submitted}[3]{{\sc #1,} #2, submitted to {\it #3}}

3358: \newcommand{\JCTA}{J.\ Combin.\ Theory Ser.\ A}

3359: \newcommand{\AdvancesinMathematics}{Adv.\ Math.}

3360: \newcommand{\JournalofAlgebraicCombinatorics}{J.\ Algebraic Combin.}

3361: 

3362: \newcommand{\communication}[1]{{\sc #1,} personal communication.}

3363: 

3364: % \newpage

3365: 

3366: {\small

3367: 

3368: \begin{thebibliography}{99}

3369: 

3370: \bibitem{Athanasiadis}

3371: \journal{C.\ A.\ Athanasiadis}

3372:         {Characteristic polynomials of subspace arrangements

3373:         and finite fields}

3374:         {\AdvancesinMathematics}

3375:         {122}{1996}{193--233}

3376: 

3377: \bibitem{Athanasiadis_II}

3378: \journal{C.\ A.\ Athanasiadis}

3379:         {Extended Linial hyperplane arrangements for root systems

3380:          and a conjecture of Postnikov and Stanley}

3381:         {J.\ Algebraic Combin.}

3382:         {10}{1999}{207--225}

3383: 

3384: \bibitem{Bayer_Billera}

3385: \journal{M.\ Bayer and L.\ Billera}

3386:         {Generalized Dehn-Sommerville relations for polytopes,

3387:          spheres and Eulerian partially ordered sets}

3388:         {Invent.\ Math.}

3389:         {79}{1985}{143--157}

3390: 

3391: \bibitem{Bayer_Ehrenborg}

3392: \journal{M.\ Bayer and R.\ Ehrenborg}

3393:         {The toric $h$-vectors of partially ordered sets}

3394:         {Trans.\ Amer.\ Math.\ Soc.}

3395:         {352}{2000}{4515--4531}

3396: 

3397: \bibitem{Bayer_Klapper}

3398: \journal{M.\ Bayer and A.\ Klapper}

3399:         {A new index for polytopes}

3400:         {Discrete Comput.\ Geom.}

3401:         {6}{1991}{33--47}

3402: 

3403: \bibitem{Bayer_Sturmfels}

3404: \journal{M.\ Bayer and B.\ Sturmfels}

3405:         {Lawrence polytopes}

3406:         {Canad.\ J.\ Math.}

3407:         {42}{1990}{62--79}

3408: 

3409: \bibitem{Billera_Ehrenborg}

3410: \journal{L.\ J.\ Billera and R.\ Ehrenborg}

3411:         {Monotonicity of the cd-index for polytopes}

3412:         {Math.\ Z.}

3413:         {233}{2000}{421--441}

3414: 

3415: \bibitem{Billera_Ehrenborg_Readdy_om}

3416: \journal{L.\ J.\ Billera, R.\ Ehrenborg, and M.\ Readdy}

3417:         {The {\ctd}-index of oriented matroids}

3418:         {J.\ Combin.\ Theory Ser.\ A}

3419:         {80}{1997}{79--105}

3420: 

3421: \bibitem{Billera_Hsiao_van_Willigenburg}

3422: \journal{L.\ J.\ Billera, S.\ K.\ Hsiao and S.\ van Willigenburg}

3423:         {Peak quasisymmetric functions and Eulerian enumeration}

3424:         {\AdvancesinMathematics}

3425:         {176}{2003}{248--276}

3426: 

3427: \bibitem{Bjorner_topological_methods}

3428: {\sc A.\ Bj\"orner,}

3429: Topological methods.

3430: Handbook of combinatorics,

3431: Vol.\ 2, 1819--1872,

3432: Elsevier, Amsterdam, 1995. 

3433: 

3434: \bibitem{Chen}

3435: \journal{B.\ Chen}

3436:         {On characteristic polynomials of

3437:          subspace arrangements}

3438:         {\JCTA}

3439:         {90}{2000}{347--352}

3440: 

3441: \bibitem{De_Concini_Procesi}

3442: \journal{C.\ De Concini and C.\ Procesi}

3443:         {On the geometry of toric arrangements}

3444:         {Transform.\ Groups}

3445:         {10}{2005}{387--422}

3446: 

3447: \bibitem{Douglass}

3448: \journal{J.\ M.\ Douglass}

3449:         {Toral arrangements and hyperplane arrangements}

3450:         {Rocky Mountain J.\ Math.}

3451:         {28}{1998}{939--956}

3452: 

3453: 

3454: 

3455: \bibitem{Ehrenborg_Hopf}

3456: \journal{R.\ Ehrenborg}

3457:         {On posets and Hopf algebras}

3458:         {\AdvancesinMathematics}

3459:         {119}{1996}{1--25}

3460: 

3461: \bibitem{Ehrenborg_k-Eulerian}

3462: \journal{R.\ Ehrenborg}

3463:         {$k$-Eulerian posets}

3464:         {Order.}

3465:         {18}{2001}{227--236}

3466: 

3467: \bibitem{Ehrenborg_r_Birkhoff}

3468: \submitted{R.\ Ehrenborg}

3469:           {The $r$-signed Birkhoff transform}

3470:           {Trans.\ Amer.\ Math.\ Soc.}

3471: 

3472: \bibitem{Ehrenborg_lifting}

3473: \journal{R.\ Ehrenborg}

3474:         {Lifting inequalities for polytopes}

3475:         {Adv.\ Math.}

3476:         {193}{2005}{205--222}

3477: 

3478: \bibitem{Ehrenborg_Karu}

3479: \journal{R.\ Ehrenborg and K.\ Karu}

3480:         {Decomposition theorem for the $\cd$-index of Gorenstein* posets}

3481:         {J.\ Algebraic Combin.}

3482:         {26}{2007}{225--251}

3483: 

3484: \bibitem{Ehrenborg_Readdy_c}

3485: \journal{R.\ Ehrenborg and M.\ Readdy}

3486:         {Coproducts and the $\cd$-index}

3487:         {\JournalofAlgebraicCombinatorics}

3488:         {8}{1998}{273--299}

3489: 

3490: \bibitem{Ehrenborg_Readdy_valuation_1}

3491: \journal{R.\ Ehrenborg and M.\ Readdy}

3492:         {On valuations, the characteristic polynomial and

3493:          complex subspace arrangements}

3494:         {Adv.\ Math.}

3495:         {134}{1998}{32--42}

3496: 

3497: \bibitem{Ehrenborg_Readdy_valuation_2}

3498: \journal{R.\ Ehrenborg and M.\ Readdy}

3499:         {The Dowling transform of subspace arrangements}

3500:         {\JCTA}

3501:         {91}{2000}{322--333}

3502: 

3503: \bibitem{Ehrenborg_Readdy_homology}

3504: \journal{R.\ Ehrenborg and M.\ Readdy}

3505:         {Homology of Newtonian coalgebras}

3506:         {European J.\ Combin.}

3507:         {23}{2002}{919--927}

3508: 

3509: 

3510: \bibitem{Ehrenborg_Readdy_Tchebyshev}

3511: \appear{R.\ Ehrenborg and M.\ Readdy}

3512:        {The Tchebyshev transforms of the first and second kind}

3513:        {Ann.\ Comb.}

3514: 

3515: 

3516: \bibitem{Ehrenborg_Slone}

3517: \preparation{R.\ Ehrenborg and M.\ Slone}

3518:             {The $\cd$-index of non-regular cell complexes}

3519: 

3520: 

3521: 

3522: \bibitem{Goresky_MacPherson}

3523:   {\sc M.\ Goresky and R.\ MacPherson,}

3524:   ``Stratified Morse theory''

3525:   Ergebnisse der Mathematik und ihrer Grenzgebiete (3)

3526:   [Results in Mathematics and Related Areas (3)],

3527:   14.\ Springer-Verlag, Berlin, 1988.

3528: 

3529: 

3530: \bibitem{Greene_Zaslavsky}

3531: \journal{C.\ Greene and T.\ Zaslavsky}

3532:         {On the interpretation of Whitney numbers

3533:          through arrangements of hyperplanes, zonotopes,

3534:          non-Radon partitions, and orientations of graphs}

3535:         {Trans.\ Amer.\ Math.\ Soc.}

3536:         {280}{1983}{97--126}

3537: 

3538: 

3539: \bibitem{Groemer}

3540: \journal{H.\ Groemer}

3541:         {On the extension of additive functionals on classes of convex sets}

3542:         {Pacific J.\ Math.}

3543:         {75}{1978}{397--410}

3544: 

3545: 

3546: \bibitem{Grunbaum}

3547: \books{B.\ Gr\"unbaum}

3548:       {Convex polytopes}

3549:       {second edition}

3550:       {Springer-Verlag}{New York}{2003}

3551: 

3552: 

3553: \bibitem{Hsiao}

3554: \journal{S.\ K.\ Hsiao}

3555:         {A signed analog of the Birkhoff transform}

3556:         {\JCTA}

3557:         {113}{2006}{251--272}

3558: 

3559: 

3560: \bibitem{Joni_Rota}

3561: \journal{S.\ A.\ Joni and G.-C.\ Rota}

3562:         {Coalgebras and bialgebras in combinatorics}

3563:         {Stud.\ Appl.\ Math.}

3564:         {61}{1979}{93--139}

3565: 

3566: 

3567: \bibitem{Jozefiak_Sagan}

3568: \journal{T.\ J\'ozefiak and B.\ Sagan}

3569:         {Basic derivations for subarrangements of Coxeter arrangements}

3570:         {J.\ Algebraic Combin.}

3571:         {2}{1993}{291--320}

3572: 

3573: \bibitem{Kalai_g}

3574: \journal{G.\ Kalai}

3575:         {Rigidity and the lower bound theorem. I}

3576:         {Invent.\ Math.}

3577:         {88}{1987}{125--151}

3578: 

3579: \bibitem{Kalai}

3580: \journal{G.\ Kalai}

3581: 	{A new basis of polytopes}

3582: 	{\JCTA}

3583: 	{49}{1988}{191--209}

3584: 

3585: \bibitem{Karu}

3586: \journal{K.\ Karu}

3587:         {Hard Lefschetz theorem for nonrational polytopes}

3588:         {Invent.\ Math.}

3589:         {157}{2004}{419--447}

3590: 

3591: \bibitem{Klain_Rota}

3592: \book{D.\ Klain and G.-C.\ Rota}

3593:      {Introduction to geometric probability}

3594:      {Lezioni Lincee, Cambridge University Press, Cambridge}

3595:      {1997}

3596: 

3597: 

3598: 

3599: \bibitem{Macmeikan_I}

3600: \journal{C.\ Macmeikan}

3601:         {The Poincar\'e polynomial of an mp arrangement}

3602:         {Proc.\ Amer.\ Math.\ Soc.}

3603:         {132}{2004}{1575--1580}

3604: 

3605: \bibitem{Macmeikan_II}

3606: \journal{C.\ Macmeikan}

3607:         {Modules of derivations for toral arrangements}

3608:         {Indag.\ Math. (N.S.)}

3609:         {15}{2004}{257--267}

3610: 

3611: \bibitem{Macmeikan_III}

3612: \collection{C.\ Macmeikan}

3613:         {Toral arrangements.  The COE Seminar on

3614:          Mathematical Sciences 2004}

3615:          {37--54}

3616:         {Sem.\ Math.\ Sci., 31}

3617:         {Keio Univ., Yokohama}

3618:         {2004}

3619: 

3620: 

3621: \bibitem{Novik_Postnikov_Sturmfels}

3622: \journal{I.\ Novik, A.\ Postnikov and B.\ Sturmfels}

3623:         {Syzygies of oriented matroids}

3624:         {Duke Math.\ J.}

3625:         {111}{2002}{287--317}

3626: 

3627: 

3628: \bibitem{Stanley_acyclic}

3629: \journal{R.\ Stanley}

3630:         {Acyclic orientations of graphs}

3631:         {Discrete Math.}

3632:         {5}{1973}{171--178}

3633: 

3634: 

3635: \bibitem{Stanley_EC_I}

3636: \book{R.\ P.\ Stanley}

3637:      {Enumerative Combinatorics, Vol. I}

3638:      {Wadsworth and Brooks/Cole, Pacific Grove}

3639:      {1986}

3640: 

3641: 

3642: \bibitem{Stanley_h}

3643: {\sc R.\ P.\ Stanley,}

3644: Generalized $h$-vectors, intersection cohomology of toric varieties,

3645: and related results.

3646: Commutative algebra and combinatorics (Kyoto, 1985),

3647: 187--213, Adv. Stud. Pure Math., 11, North-Holland, Amsterdam, 1987.

3648: 

3649: 

3650: \bibitem{Stanley_d}

3651: \journal{R.\ P.\ Stanley}

3652:         {Flag $f$-vectors and the $cd$-index}

3653:         {Math.\ Z.}

3654:         {216}{1994}{483--499}

3655: 

3656: \bibitem{Stembridge}

3657: \journal{J.\ Stembridge}

3658:         {Enriched $P$-partitions}

3659:         {Trans.\ Amer.\ Math.\ Soc.}

3660:         {349}{1997}{763--788}

3661: 

3662: \bibitem{Swartz}

3663: \preprint{E.\ Swartz}

3664:          {Face enumeration -- from spheres to manifolds}

3665:          {2007}

3666: 

3667: \bibitem{Sweedler}

3668: \book{M.\ Sweedler}

3669:      {Hopf Algebras}

3670:      {Benjamin, New York}

3671:      {1969}

3672: 

3673: \bibitem{Zaslavsky}

3674:         {\sc T.\ Zaslavsky,}

3675: 	{Facing up to arrangements:  face count formulas for

3676: 	partitions of space by hyperplanes,}

3677: 	{\it Mem.\ Amer.\ Math.\ Soc.}

3678: 	{\bf 154}

3679:         {(1975).}

3680: 

3681: 

3682: \bibitem{Zaslavsky_paper}

3683: \journal{T.\ Zaslavsky}

3684:         {A combinatorial analysis of topological dissections}

3685:         {\AdvancesinMathematics}

3686:         {25}{1977}{267--285}

3687: 

3688: 

3689: \end{thebibliography}

3690: 

3691: }

3692: 

3693: 

3694: 

3695: 

3696: \bigskip

3697: 

3698: {\em R.\ Ehrenborg, M.\ Readdy and M.\ Slone,

3699: Department of Mathematics,

3700: University of Kentucky,

3701: Lexington, KY 40506-0027,}

3702: \{{\tt jrge},{\tt readdy},{\tt mslone}\}{\tt @ms.uky.edu}

3703: 

3704: 

3705: \end{document}

3706: 

3707: 

3708: 

3709: 

3710: 

3711: 

3712: