euler calculus & data robert ghrist university of pennsylvania depts. of mathematics & electrical/systems engineering machine learning summer school : june 2009 motivation tools euler calculus euler calculus χ = χ χ = Σk (-1)k # {k-cells} =2 Σk (-1)k rank Hk χ =7 χ =3 χ =2 χ =3 sheaves u geometry χ(AuB) = χ(A)+ χ(B) – χ(A B) blaschke hadwiger rota chen topology ∫ h dχ kashiwara macpherson schapira viro probability adler taylor networks integration consider the sheaf of constructible functions tools axiomatic approach to tameness in the work on o-minimal structures collections {Sn}n=1,2,... of boolean algebras of sets in Rn closed under projections, products,... elements of {Sn}n=1,2,... are called “definable” or “tame” sets results CF(X) = Z-valued functions whose level sets are locally finite and “tame” all definable sets are triangulable & have a well-defined euler characteristic all functions in CF(X) are of the form h = Σci1Ui for Ui definable all functions in CF(X) are integrable with respect to Euler characteristic ∫ h dχ = ∫ (Σ ci1Ui) dχ = Σ(∫ ci1Ui)dχ = Σci χ(Ui) euler integral explicit definition: integration [schapira, 1980’s; via kashiwara, macpherson, 1970’s] X the induced pushforward on sheaves of constructible functions is the correct way to understand dχ CF(X) in the case where Y is a point, CF(Y)=Z, and the pushforward is a homomorphism from CF(X) to Z which respects all the gluings implicit in sheaves... X CF(X) F F* Y CF(Y) pt ∫ F F* X CF(X) Y CF(Y) pt ∫ CF(pt)=Z dχ corollary: [schapira, viro; 1980’s] fubini theorem CF(pt)=Z dχ sheaf-theoretic constructions also give natural convolution operators, duality, integral transforms, ... problem a network of “minimal” sensors returns target counts without IDs how many targets are there? =0 =1 =2 =3 =4 problem counting let W = “target space” = space where finite # of targets live let X = “sensor space” = space which parameterizes sensors target i is detected on a target support Ui in X h:X→Z sensor field on X returns h(x) = #{ i : x lies in Ui } 2 theorem: [BG] assuming target supports with uniform χ(Ui)=N # targets = (1/N) ∫ X N≠0 h dχ trivial proof: ∫ h dχ = ∫ (Σ1Ui) dχ = Σ(∫ 1Ui dχ) = Σ χ(U ) = N # i i amazingly, one needs no convexity, no leray (“good cover”) condition, etc. this is a purely topological result. computation for h in CF(X), integrals with respect to dχ are computable via ∞ ∫ h dχ = s=0 Σ s χ({ h=s }) level set ∞ = s=0 Σ χ({ h>s })-χ({ h<-s }) = ΣV h(V)χ(v) “chambers” of h components of level sets upper excursion set weighted euler index example ∞ ∫ h dχ = s=0 Σ χ {h(x)>s} h>3 : χ = 2 h>2 : χ = 3 h>1 : χ = 3 h>0 : χ = -1 net integral = 2+3+3-1 = 7 example ∞ ∫ h dχ = s=0 Σ χ {h(x)>s} = Σ h(V)χ(V) V h=4 : Σ = 2 h=3 : Σ = 1 h=2 : Σ = 0 h=1 : Σ = -4 net integral = 4(1+1)+3(1+1-1)+2(1+1+1-1-1-1)+1(1-1-1-1-2) = 7 some applications in minimal sensing waves consider a sensor modality which counts each wavefronts and increments an internal counter: used to count # events 3 booms… whuh? 2 booms… the resulting target impacts are still nullhomotopic (no echoing) accurate event counts obtained via ad hoc network of acoustic sensors with no clocks, no synchronization, and no localization 17 wheels consider sensors which count passing vehicles and increment an internal counter acoustic sensors embedded in roads… such target impacts may not be contractible… theorem: [BG] if sensors read h = the total number of time intervals in which some vehicle is nearby, then # vehicles = ∫ h dχ wheels supports are the projected image of a contractible subset in space-time recall: F X CF(X) ∫ X F* Y pt CF(Y) Z ∫ dχ h(x) dχ(x) = ∫Y F*h(y) dχ(y) F*h(y) = ∫ -1 F (y) h(x) dχ(x) let X = domain x time ; let Y = domain ; let F = temporal projection map then F*h(y) = total # of (compact) time intervals on which some vehicle is at/near point w = sensor reading at y numerical integration ad hoc networks theorem: [BG] if the function h:R2→N is sampled over a network in a way that correctly samples the connectivity of upper and lower excursion sets, then the exact value of the euler integral of h is ∞ Σ( #comp{ h≥s } - #comp{ h<s } + 1) s=1 this is a simple application of alexander duality… ∞ ∞ χ{ h ≥ s } = Σ b0 {h ≥ s } – b1{h ≥ s } ∫ h dχ = Σ s=1 ∞ s=1 ~ =Σ b0{h ≥ s } – b0{h < s } s=1 ∞ = Σ b0{h ≥ s } – b0{h < s } + 1 s=1 this works in ad hoc setting : clustering gives fast computation bk χ = Σ (-1)k dim Hk k get real… real-valued integrands it’s helpful to have a well-defined integration theory for R-valued integrands: Def(X) = R-valued functions whose graphs are “tame” (definable in o-minimal) take a riemann-sum approach ∫ h dχ● = lim 1/n∫ floor(nh) dχ unfortunately, ∫ _ dχ ● & however, ∫ _ dχ ● & ∫ h dχ● = lim 1/n∫ ceil(nh) dχ ∫ _ dχ● are no longer homomorphisms Def(X)→R ∫ _ dχ● have an interpretation in o-minimal category lemma if h is affine on an open k-simplex, then ∫ h dχ● = (-1)k inf (h) ∫ h dχ● h = (-1)k sup (h) real-valued integrands intuition: the two measures correspond to the stratified morse indices of the graph of h in Def(X) with respect to two graph axis directions… I*, I* : Def(X)→CF(X) theorem: [BG] for h in Def(X) ∫ h dχ∙ = ∫ h I*h dχ ∫ h dχ∙ = ∫ h I*h dχ ∫ h dχ∙ = Σ (-1)n-μ(p) h(p) crit(h) μ = morse index corollary: [BG] if h : X → R is morse on an n-manifold, then ∫ h dχ∙ = Σ (-1)μ(p) h(p) crit(h) corollary: [BG] if h is univariate, then ∫ h dχ = totvar(h)/2 = - ∫ h dχ ∙ ∙ real-valued integrands Lebesgue ∫ h dχ● = ∫R χ{h≥s} - χ{h<-s} ds ∫ h dχ● = limε→0+∫R s χ{s ≤ h < s+ε} ds ∫ h dχ● = ∫R χ{h>s} - χ{h≤-s} ds ∫ h dχ● = limε→0+∫R s χ{s < h ≤ s+ε} ds Morse ∫ h dχ● = Σ (-1)n-μ(p) h(p) crit(h) ∫ h dχ● = Σ (-1)μ(p) h(p) crit(h) Duality ∫ h dχ● = - ∫ - h dχ● (Dh)(x) = limε→0+∫ h 1B(ε,x) dχ D(Dh) = h Fubini ∫X h dχ●(x) = ∫Y ∫ {F(x)=y} h(x) dχ●(x)dχ●(y) F:X→Y with h∙F=h incomplete data consider the following relative problem: D given h on the complement of a hole D, ∫ estimate h dχ over the entire domain theorem: [BG] for h:R2→Z a sum of indicator functions over homotopically trivial supports, none of which lies entirely within a contractible hole D, then ∫R h dχ ≤ ∫R h dχ ≤ ∫R h dχ 2 h = fill in D with maximum of h on ∂D 2 2 h = fill in D with minimum of h on ∂D reminder: f < g does not imply that ∫ f dχ < ∫ g dχ ...in this case the opposite occurs… incomplete data but what to choose in between upper and lower bounds? claim: a harmonic extension over a hole is a “best guess”... theorem: [BG] For h:R2→Z a sum of indicator functions over homotopically trivial supports, none of which lies entirely within a contractible hole D, then ∫R h dχ ≤ ∫R 2 2 f dχ ≤ ∫R h dχ 2 for f any “harmonic” extension of h over D (weighted average of h rel ∂D) the proof is surprisingly easy using morse theory: a “harmonic” extension has no local maxima or minima within D... # saddles in D - # maxima on ∂D = χ(D)=1 the integral over D is the heights of the maxima minus the heights of the saddles expected values in practice, harmonic extensions lead to non-integer target counts ∫ h dχ = 1+1-c this is an “expected” target count weights for the laplacian can be chosen based on confidence of data points toward a general theory of expected integrals integral transforms inversion X S W ∫ X h dχ = N ∫ W 1T dχ = N #T h = integral transform of 1T with kernel S fourier transform radon transform bessel transform open questions how to correct “side lobes” and energy loss in integral transforms? what is the appropriate integration theory for multi-modal and logical-valued data? how to efficiently compute integral transforms given discrete (sparse) data? …and, well, numerical analysis in general topological network topology closing credits… research sponsored by darpa (stomp program) national science foundation office of naval research primary collaborator yuliy baryshnikov, bell labs professional support university of pennsylvania a. mitchell java code david lipsky, uillinois, urbana a.j. friend, stanford naveen kasthuri, penn