ebook img

Bernstein polynomials PDF

13 Pages·2000·0.082 MB·English
Save to my drive
Quick download
Download
Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.

Preview Bernstein polynomials

On-LineGeometricModelingNotes BERNSTEIN POLYNOMIALS KennethI.Joy VisualizationandGraphicsResearchGroup DepartmentofComputerScience UniversityofCalifornia,Davis Overview Polynomials are incredibly useful mathematical tools as they are simply defined, can be calculated quickly on computer systems and represent a tremendous variety of functions. They can be differentiated andintegratedeasily,andcanbepiecedtogethertoformsplinecurvesthatcanapproximateanyfunctionto anyaccuracydesired. Moststudentsareintroductedtopolynomialsataveryearlystageintheirstudiesof mathematics,andwouldprobablyrecallthemintheformbelow: p(t) = a tn+a tn−1+···+a t+a n n−1 1 0 whichrepresentsapolynomialasalinearcombinationofcertainelementarypolynomials(cid:8)(1,t,t2,...,tn(cid:9). In general, any polynomial function that has degree less than or equal to n, can be written in this way, andthereasonsaresimply • The set of polynomials of degree less than or equal to n forms a vector space: polynomials can be addedtogether,canbemultipliedbyascalar,andallthevectorspacepropertieshold. • The set of functions (cid:8)1,t,t2,...,tn(cid:9) form a basis for this vector space – that is, any polynomial of degreelessthanorequaltoncanbeuniquelywrittenasalinearcombinationsofthesefunctions. This basis, commonly called the power basis, is only one of an infinite number of bases for the space of polynomials. Inthesenoteswediscussanotherofthecommonlyusedbasesforthespaceofpolynomials,theBernstein basis,anddiscussitsmanyusefulproperties. BernsteinPolynomials TheBernsteinpolynomialsofdegreenaredefinedby (cid:18) (cid:19) n B (t) = ti(1−t)n−i i,n i fori = 0,1,...,n,where (cid:18) (cid:19) n n! = i i!(n−i)! Therearen+1nth-degreeBernsteinpolynomials. Formathematicalconvenience,weusuallysetB = 0, i,n ifi < 0ori > n. These polynomials are quite easy to write down: the coefficients (cid:0)n(cid:1) can be obtained from Pascal’s i triangle; the exponents on the t term increase by one as i increases; and the exponents on the (1−t) term decreasebyoneasiincreases. Inthesimplecases,weobtain • TheBernsteinpolynomialsofdegree1are B (t) = 1−t 0,1 B (t) = t 1,1 andcanbeplottedfor0 ≤ t ≤ 1as (cid:3)(cid:5)(cid:4)(cid:7)(cid:6)(cid:8)(cid:10)(cid:9)(cid:12)(cid:11)(cid:14)(cid:13) (cid:15)(cid:17)(cid:16)(cid:19)(cid:18)(cid:16)(cid:10)(cid:20)(cid:12)(cid:21)(cid:14)(cid:22) (cid:2) (cid:0) (cid:1) 2 • TheBernsteinpolynomialsofdegree2are B (t) = (1−t)2 0,2 B (t) = 2t(1−t) 1,2 B (t) = t2 2,2 andcanbeplottedfor0 ≤ t ≤ 1as (cid:3)(cid:5)(cid:4)(cid:7)(cid:6)(cid:8)(cid:10)(cid:9)(cid:12)(cid:11)(cid:14)(cid:13) (cid:24)(cid:26)(cid:25)(cid:28)(cid:27)(cid:25)(cid:10)(cid:29)(cid:12)(cid:30)(cid:14)(cid:31) (cid:15)(cid:17)(cid:16)(cid:19)(cid:18)(cid:20)(cid:10)(cid:21)(cid:12)(cid:22)(cid:14)(cid:23) (cid:2) (cid:0) (cid:1) • TheBernsteinpolynomialsofdegree3are B (t) = (1−t)3 0,3 B (t) = 3t(1−t)2 1,3 B (t) = 3t2(1−t) 2,3 B (t) = t3 3,3 andcanbeplottedfor0 ≤ t ≤ 1as 3 (cid:3)(cid:5)(cid:4)(cid:7)(cid:6)(cid:8)(cid:10)(cid:9)(cid:12)(cid:11)(cid:14)(cid:13) !(cid:26)"(cid:28)#"(cid:10)$(cid:12)%(cid:14)& (cid:15)(cid:17)(cid:16)(cid:19)(cid:18)(cid:20)(cid:10)(cid:21)(cid:12)(cid:22)(cid:14)(cid:23) (cid:24)(cid:26)(cid:25)(cid:28)(cid:27)(cid:29)(cid:10)(cid:30)(cid:12)(cid:31)(cid:14) (cid:2) (cid:0) (cid:1) 4 ARecursiveDefinitionoftheBernsteinPolynomials TheBernsteinpolynomialsofdegreencanbedefinedbyblendingtogethertwoBernsteinpolynomials ofdegreen−1. Thatis,thekthnth-degreeBernsteinpolynomialcanbewrittenas B (t) = (1−t)B (t)+tB (t) k,n k,n−1 k−1,n−1 Toshowthis,weneedonlyusethedefinitionoftheBernsteinpolynomialsandsomesimplealgebra: (cid:18) (cid:19) (cid:18) (cid:19) n−1 n−1 (1−t)B (t)+tB (t) = (1−t) tk(1−t)n−1−k +t tk−1(1−t)n−1−(k−1) k,n−1 k−1,n−1 k k−1 (cid:18) (cid:19) (cid:18) (cid:19) n−1 n−1 = tk(1−t)n−k + tk(1−t)n−k k k−1 (cid:20)(cid:18) (cid:19) (cid:18) (cid:19)(cid:21) n−1 n−1 = + tk(1−t)n−k k k−1 (cid:18) (cid:19) n = tk(1−t)n−k k = B (t) k,n TheBernsteinPolynomialsareAllNon-Negative A function f(t) is non-negative over an interval [a,b] if f(t) ≥ 0 for t ∈ [a,b]. In the case of the Bernstein polynomials of degree n, each is non-negative over the interval [0,1]. To show this we use the recursivedefinitionpropertyaboveandmathematicalinduction. ItiseasilyseenthatthefunctionsB (t) = 1−tandB (t) = tarebothnon-negativefor0 ≤ t ≤ 1. If 0,1 1,1 weassumethatallBernsteinpolynomialsofdegreelessthankarenon-negative,thenbyusingtherecursive definitionoftheBernsteinpolynomial,wecanwrite B (t) = (1−t)B (t)+tB (t) i,k i,k−1 i−1,k−1 and argue that B (t) is also non-negative for 0 ≤ t ≤ 1, since all components on the right-hand side i,k of the equation are non-negative components for 0 ≤ t ≤ 1. By induction, all Bernstein polynomials are non-negativefor0 ≤ t ≤ 1. Inthisprocess,wehavealsoshownthateachoftheBernsteinpolynomialsispositivewhen0 < t < 1. 5 TheBernsteinPolynomialsformaPartitionofUnity Asetoffunctionsf (t)issaidtopartitionunityiftheysumtooneforallvaluesoft. Thek+1Bernstein i polynomialsofdegreek formapartitionofunityinthattheyallsumtoone. To show that this is true, it is easiest to first show a slightly different fact: for each k, the sum of the k+1BernsteinpolynomialsofdegreekisequaltothesumofthekBernsteinpolynomialsofdegreek−1. Thatis, k k−1 X X B (t) = B (t) i,k i,k−1 i=0 i=0 Thiscalculationisstraightforward,usingtherecursivedefinitionandcleverlyrearrangingthesums: k k X X B (t) = [(1−t)B (t)+tB (t)] i,k i,k−1 i−1,k−1 i=0 i=0 "k−1 # " k # X X = (1−t) B (t)+B (t) +t B (t)+B (t) i,k−1 k,k−1 i−1,k−1 −1,k−1 i=0 i=1 k−1 k X X = (1−t) B (t)+t B (t) i,k−1 i−1,k−1 i=0 i=1 k−1 k−1 X X = (1−t) B (t)+t B (t) i,k−1 i,k−1 i=0 i=0 k−1 X = B (t) i,k−1 i=0 (wherewehaveutilizedB (t) = B (t) = 0). k,k−1 −1,k−1 Oncewehaveestablishedthisequality,itissimpletowrite n n−1 n−2 1 X X X X B (t) = B (t) = B (t) = ··· = B (t) = (1−t)+t = 1 i,n i,n−1 i,n−2 i,1 i=0 i=0 i=0 i=0 The partition of unity is a very important property when utilizing Bernstein polynomials in geometric modeling and computer graphics. In particular, for any set of points P , P , ..., P , in three-dimensional 0 1 n space,andforanyt,theexpression P(t) = P B (t)+P B (t)+···+P B (t) 0 0,n 1 1,n n n,n 6 isanaffinecombinationofthesetofpointsP ,P ,...,P andif0 ≤ t ≤ 1, itisaconvexcombinationof 0 1 n thepoints. DegreeRaising Anyofthelower-degreeBernsteinpolynomials(degree<n)canbeexpressedasalinearcombination ofBernsteinpolynomialsofdegreen. Inparticular,anyBernsteinpolynomialofdegreen−1canbewritten asalinearcombinationofBernsteinpolynomialsofdegreen. Wefirstnotethat (cid:18) (cid:19) n tB (t) = ti+1(1−t)n−i i,n i (cid:18) (cid:19) n = ti+1(1−t)(n+1)−(i+1) i (cid:0)n(cid:1) = i B (t) (cid:0)n+1(cid:1) i+1,n+1 i+i i+1 = B (t) i+1,n+1 n+1 and (cid:18) (cid:19) n (1−t)B (t) = ti(1−t)n+1−i i,n i (cid:0)n(cid:1) = i B (t) (cid:0)n+1(cid:1) i,n+1 i n−i+1 = B (t) i,n+1 n+1 andfinally 1 1 B (t)+ B (t) = ti(1−t)n−i+ti+1(1−t)n−(i+1) (cid:0)n(cid:1) i,n (cid:0) n (cid:1) i+1,n i i+1 = ti(1−t)n−i−1((1−t)+t) = ti(1−t)n−i−1 1 = B (t) (cid:0)n−1(cid:1) i,n−1 i Usingthisfinalequation,wecanwriteanarbitraryBernsteinpolynomialintermsofBernsteinpolynomials 7 ofhigherdegree. Thatis, " # (cid:18) (cid:19) n−1 1 1 B (t) = B (t)+ B (t) i,n−1 i (cid:0)n(cid:1) i,n (cid:0) n (cid:1) i+1,n i i+1 (cid:18) (cid:19) (cid:18) (cid:19) n−i i+1 = B (t)+ B (t) i,n i+1,n n n which expresses a Bernstein polynomial of degree n − 1 in terms of a linear combination of Bernstein polynomials of degree n. We can easily extend this to show that any Bernstein polynomial of degree k (lessthann)canbewrittenasalinearcombinationofBernsteinpolynomialsofdegreen–e.g.,aBernstein polynomialofdegreen−2canbeexpressedasalinearcombinationoftwoBernsteinpolynomialsofdegree n−1, each of which can be expressed as a linear combination of two Bernstein polynomials of degree n, etc. ConvertingfromtheBernsteinBasistothePowerBasis Sincethepowerbasis{1,t,t2,...,tn}formsabasisforthespaceofpolynomialsofdegreelessthanor equal to n, any Bernstein polynomial of degree n can be written in terms of the power basis. This can be directlycalculatedusingthedefinitionoftheBernsteinpolynomialsandthebinomialtheorem,asfollows: (cid:18) (cid:19) n B (t) = tk(1−t)n−k k,n k (cid:18) (cid:19) n−k (cid:18) (cid:19) n X n−k = tk (−1)i ti k i i=0 n−k (cid:18) (cid:19)(cid:18) (cid:19) X n n−k = (−1)i ti+k k i i=0 n (cid:18) (cid:19)(cid:18) (cid:19) X n n−k = (−1)i−k ti k i−k i=k n (cid:18) (cid:19)(cid:18) (cid:19) X n i = (−1)i−k ti i k i=k wherewehaveusedthebinomialtheoremtoexpand(1−t)n−k. ToshowthateachpowerbasiselementcanbewrittenasalinearcombinationofBernsteinPolynomials, 8 weusethedegreeelevationformulasandinductiontocalculate: tk = t(tk−1) n (cid:0) i (cid:1) X = t k−1 B (t) (cid:0) n (cid:1) i,n−1 i=k−1 k−1 n (cid:0)i−1(cid:1) X = k−1 tB (t) (cid:0)n−1(cid:1) i−1,n−1 i=k k−1 n−1 (cid:0) i (cid:1) X i = k−1 B (t) (cid:0) n (cid:1)n i,n i=k−1 k−1 n−1 (cid:0)i(cid:1) X = k B (t), (cid:0)n(cid:1) i,n i=k−1 k wheretheinductionhypothesiswasusedinthesecondstep. Derivatives DerivativesofthenthdegreeBernsteinpolynomialsarepolynomialsofdegreen−1. Usingthedefini- tionoftheBernsteinpolynomialwecanshowthatthisderivativecanbewrittenasalinearcombinationof Bernsteinpolynomials. Inparticular d B (t) = n(B (t)−B (t)) k,n k−1,n−1 k,n−1 dt for0 ≤ k ≤ n. Thiscanbeshownbydirectdifferentiation (cid:18) (cid:19) d d n B (t) = tk(1−t)n−k k,n dt dt k kn! (n−k)n! = tk−1(1−t)n−k + tk(1−t)n−k−1 k!(n−k)! k!(n−k)! n(n−1)! n(n−1)! = tk−1(1−t)n−k + tk(1−t)n−k−1 (k−1)!(n−k)! k!(n−k−1)! (cid:18) (cid:19) (n−1)! (n−1)! = n tk−1(1−t)n−k + tk(1−t)n−k−1 (k−1)!(n−k)! k!(n−k−1)! = n(B (t)−B (t)) k−1,n−1 k,n−1 9 Thatis,thederivativeofaBernsteinpolynomialcanbeexpressedasthedegreeofthepolynomial,multiplied bythedifferenceoftwoBernsteinpolynomialsofdegreen−1. TheBernsteinPolynomialsasaBasis Why do the Bernstein polynomials of order n form a basis for the space of polynomials of degree less thanorequalton? 1. Theyspanthespaceofpolynomials–anypolynomialofdegreelessthanorequaltoncanbewritten asalinearcombinationoftheBernsteinpolynomials. ThisiseasilyseenifonerealizesthatThepowerbasisspansthespaceofpolynomialsandanymember ofthepowerbasiscanbewrittenasalinearcombinationofBernsteinpolynomials. 2. Theyarelinearlyindependent–thatis,ifthereexistconstantsc ,c ,...,c sothattheidentity 0 1 n 0 = c B (t)+c B (t)+···+c B (t) 0 0,n 1 1,n n n,n holdsforallt,thenallthec ’smustbezero. i Ifthisweretrue,thenwecouldwrite 0 = c B (t)+c B (t)+···+c B (t) 0 0,n 1 1,n n n,n n (cid:18) (cid:19)(cid:18) (cid:19) n (cid:18) (cid:19)(cid:18) (cid:19) n (cid:18) (cid:19)(cid:18) (cid:19) X n i X n i X n i = c (−1)i ti+c (−1)i−1 ti+···+c (−1)i−n ti 0 1 n i 0 i 1 i n i=0 i=1 i=n " 1 (cid:18) (cid:19)(cid:18) (cid:19)# " n (cid:18) (cid:19)(cid:18) (cid:19)# X n 1 X n n = c + c t1+···+ c tn 0 i i 1 1 n n i=0 i=0 Sincethepowerbasisisalinearlyindependentset,wemusthavethat c = 0 0 1 (cid:18) (cid:19)(cid:18) (cid:19) X n 1 c = 0 i 1 1 i=0 . . . n (cid:18) (cid:19)(cid:18) (cid:19) X n n c = 0 i n n i=0 whichimpliesthatc = c = ··· = c = 0(c isclearlyzero,substitutingthisinthesecondequation 0 1 n 0 10

See more

The list of books you might like

Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.