Mixed f-divergence for multiple pairs of measures

Size: px
Start display at page:

Download "Mixed f-divergence for multiple pairs of measures"

Transcription

1 Mxed f-dvergece for multple pars of measures Elsabeth M. Werer ad Depg Ye Abstract I ths paper, the cocept of the classcal f-dvergece for a par of measures s exteded to the mxed f-dvergece for multple pars of measures. The mxed f-dvergece provdes a way to measure the dfferece betwee multple pars of probablty measures. Propertes for the mxed f-dvergece are establshed, such as permutato varace ad symmetry dstrbutos. A Alexadrov-Fechel type equalty ad a sopermetrc equalty for the mxed f-dvergece are proved. 1 Itroducto I applcatos such as patter matchg, mage aalyss, statstcal learg, ad formato theory, oe ofte eeds to compare two probablty measures ad eeds to kow whether they are smlar to each other. Hece, fdg the rght quatty to measure the dfferece betwee two probablty measures P ad Q s cetral. Tradtoally, people use the classcal L p dstaces betwee P ad Q, such as the varatoal dstace ad the L 2 dstace. However, the famly of f-dvergeces s ofte more sutable to fulfll the goal tha the classcal L p dstace of measures. The f-dvergece D f P, Q of two probablty measures P ad Q was frst troduced 8] ad depedetly 2, 30] ad was defed by p D f P, Q f q dµ. 1.1 q Here, p ad q are desty fuctos of P ad Q wth respect to a measure µ o. The dea behd the f-dvergece s to replace, for stace, the fucto ft t 1 the varatoal dstace by a geeral covex fucto f. Hece the f-dvergece cludes varous wdely used dvergeces as specal cases, such as, the varatoal dstace, the Kullback-Lebler dvergece 16], the Bhattacharyya dstace 5] ad may more. Cosequetly, the f-dvergece receves cosderable atteto ot oly the formato theory e.g., 3, 7, 14, 17, 31] but also may other areas. We oly meto covex geometry. Wth the last few years, amazg coectos have bee dscovered betwee otos ad cocepts from covex geometry ad formato theory, e.g., 9,10,15,24,25,32], leadg to a totally ew Keywords: Alexadrov-Fechel equalty, f-dssmlarty, f-dvergece, sopermetrc equalty. Mathematcs Subject Classfcato Number: 28, 52, 60. Partally supported by a NSF grat Supported by a NSERC grat ad a start-up grat from Memoral Uversty of Newfoudlad 1

2 pot of vew ad troducg a whole ew set of tools the area of covex geometry. I partcular, t was observed 38] that oe of the most mportat affe varat otos, the L p -affe surface area for covex bodes, e.g., 18 20, 22, 34], s Réy etropy from formato theory ad statstcs. Réy etropes are specal cases of f-dvergeces ad cosequetly those were the troduced for covex bodes ad ther correspodg etropy equaltes have bee establshed 39]. We also refer to, for stace 4], for more refereces related to the f-dvergece. Exteso of the f-dvergece from two probablty measures to multple probablty measures s fudametal may applcatos, such as statstcal hypothess test ad classfcato, ad much research has bee devoted to that, for stace 28,29,42]. Such extesos clude, e.g., the Matusta s affty 26,27], the Toussat s affty 37], the formato radus 36] ad the average dvergece 35]. The f-dssmlarty D f P 1,, P l for probablty measures P 1,, P l, troduced 11,12] for a covex fucto f : R l R, s a atural geeralzato of the f-dvergece. It s defed as D f P 1,, P l fp 1,, p l dµ, where the p s are desty fuctos of the P s that are absolutely cotuous wth respect to µ. For a covex fucto f, the fucto fx, y yf x y s also covex o x, y > 0, ad D f P, Q s equal to the classcal f-dvergece defed formula 1.1. Note that the Matusta s affty s related to fx 1,, x l l x 1/l, ad the Toussat s affty s related to fx 1,, x l l l a 1. xa, where a 0 ad such that Here, we troduce specal f-dssmlartes, amely the mxed f-dvergece ad the -th mxed f- dvergece, whch ca be vewed as vector forms of the usual f-dvergece. We establsh some basc propertes of these quattes, such as permutato varace ad symmetry dstrbutos. We prove a sopermetrc type equalty ad a Alexadrov-Fechel type equalty for the mxed f-dvergece. Alexadrov-Fechel equalty s a fudametal equalty covex geometry ad may mportat equaltes such as the Bru-Mkowsk equalty ad Mkowsk s frst equalty follow from t see, e.g., 9, 33]. The paper s orgazed as follows. I Secto 2 we establsh some basc propertes of the mxed f-dvergece, such as permutato varace ad symmetry dstrbutos. I Secto 3 we prove the geeral Alexadrov-Fechel equalty ad sopermetrc equalty for the mxed f-dvergece. Secto 4 s dedcated to the -th mxed f-dvergece ad ts related sopermetrc type equaltes. 2 The Mxed f-dvergece Throughout ths paper, let, µ be a fte measure space. For 1, let P p µ ad Q q µ be probablty measures o that are absolutely cotuous wth respect to the measure µ. Moreover, we assume that for all 1,,, p ad q are ozero µ-a.e. We use P ad Q to deote the vectors of probablty measures, or, short, probablty vectors, P P 1, P 2,, P, Q Q1, Q 2,, Q. 2

3 We use p ad q to deote the vectors of desty fuctos, or desty vectors, for P ad Q respectvely, d P dµ p p 1, p 2,, p, d Q dµ q,,, q. We make the coveto that 0 0. Deote by R + {x R : x 0}. Let f : 0, R + be a o-egatve covex or cocave fucto. The -adjot fucto f : 0, R + of f s defed by f t tf1/t. It s obvous that f f ad that f s aga covex, respectvely cocave, f f s covex, respectvely cocave. Let f : 0, R +, 1, be ether covex or cocave fuctos. Deote by f f 1, f 2,, f the vector of fuctos. We wrte to be the -adjot vector for f. f f 1, f 2,, f Now we troduce the mxed f-dvergece for f, P, Q as follows. Defto 2.1. Let, µ be a measure space. Let P ad Q be two probablty vectors o wth desty vectors p ad q respectvely. The mxed f-dvergece D f P, Q for f, P, Q s defed by D f P, Q Smlarly, we defe the mxed f-dvergece for f, Q, P by D f Q, P p f q dµ. 2.2 q q f p dµ. 2.3 p A specal case s whe all dstrbutos P ad Q are detcal ad equal to a probablty dstrbuto P. I ths case, D f P, Q D f1,f 2,,f P, P,, P, P, P,, P f 1] 1. Let π S deote a permutato o {1, 2,, } ad deote π p p π1, p π2,, p π. Oe mmedate result from Defto 2.1 s the followg permutato varace for D f P, Q. 3

4 Proposto 2.1 Permutato varace. Let the vectors f, P, Q be as above, ad let π S be a permutato o {1, 2,, }. The D f P, Q D π f π P, π Q. Whe all f, P, Q are equal to f, P, Q, the mxed f-dvergece s equal to the classcal f- dvergece, deoted by D f P, Q, whch takes the form p D f P, Q D f,f,,f P, P,, P, Q, Q,, Q f qdµ. q As f t tf1/t, oe easly obtas a fudametal property for the classcal f-dvergece D f P, Q, amely, D f P, Q D f Q, P, for all f, P, Q. Smlar results hold true for the mxed f-dvergece. We show ths ow. Let 0 k. We wrte D f,k P, Q for D f,k P, Q k p f q q k+1 Clearly, D f, P, Q D f P, Q ad D f,0 P, Q D f Q, P, where f f 1, f 2,, f. f q The we have the followg result for chagg order of dstrbutos. p p ] 1 dµ. Proposto 2.2 Prcple for chagg order of dstrbutos. Let f, P, Q be as above. The, for ay 0 k, oe has D f P, Q D f,k P, Q. I partcular, Proof. Let 0 k. The, D f P, Q D f P, Q D f Q, P. k k D f,k P, Q, p f q q p f q q where the secod equalty follows from f p q q f k+1 k+1 q p p. p f q dµ q f q p p ] 1 dµ A drect cosequece of Proposto 2.2 s the followg symmetry prcple for the mxed f- dvergece. 4

5 Proposto 2.3 Symmetry dstrbutos. Let f, P, Q be as above. The, D f P, Q+D f P, Q s symmetrc P ad Q, amely, D f P, Q + D f P, Q D f Q, P + D f Q, P. Remark. Proposto 2.2 says that D f P, Q remas the same f oe replaces ay trple f, P, Q by f, Q, P. It s also easy to see that, for all 0 k, l, oe has Hece, for all 0 k, l, s symmetrc P ad Q. D f P, Q D f,k P, Q D f,l Q, P D f Q, P. D f,k P, Q + D f,l P, Q D f P, Q + D f P, Q Hereafter, we oly cosder the mxed f-dvergece D f P, Q defed formula 2.2. Propertes for the mxed f-dvergece D f Q, P defed 2.3 follow alog the same les. Now we lst some mportat mxed f-dvergeces. Examples. The total varato s a wdely used f-dvergece to measure the dfferece betwee two probablty measures P ad Q o, µ. It s related to fucto ft t 1. Smlarly, the mxed total varato s defed by D T V P, Q p q 1 dµ. It measures the dfferece betwee two probablty vectors P ad Q. For a R, we deote by a + max{a, 0}. The mxed relatve etropy or mxed Kullback Lebler dvergece of P ad Q s defed by D KL P, Q Df+,,f + P, Q p l where ft t l t. Whe P P pµ ad Q Q qµ for all 1, 2,,, we get the followg modfed relatve etropy or Kullback Lebler dvergece ] q D KL P Q p l dµ. p + For the covex ad/or cocave fuctos f α t t α, α R for 1, the mxed Hellger tegrals s defed by D fα1,f α2,,f α P, Q 5 p α q 1 α q p ] dµ. ] 1 + dµ,

6 I partcular, D t α,t α,,t α P, Q p α α dµ. Those tegrals are related to the Toussat s affty 37], ad ca be used to defe the mxed α-réy dvergece D α {P Q } 1 α 1 l p α α dµ 1 α 1 l ] D t α,t α,,t α P, Q. The case α 1 2, for all 1, 2,,, gves the mxed Bhattacharyya coeffcet or mxed Bhattacharyya dstace of P, Q, D t, t,, t P, Q p Ths tegral s related to the Matusta s affty 26, 27]. For more formato o the correspodg f-dvergeces we refer to e.g. 17]. v I vew of exstg coectos betwee formato theory ad covex geometry e.g., 32, 38, 39], we defe the mxed f-dvergeces for covex bodes covex ad compact subsets R wth oempty terors K wth postve curvature fuctos f K, 1, s va the measures dµ. dp K 1 h K dσ ad dq K f K h K dσ, 1. Here, σ s the sphercal measure of the ut sphere S 1, h K u max x K x, u s the support fucto of K, ad f K u s the curvature fucto of K at u S 1, the recprocal of the Gauss curvature at x o the boudary of K wth ut outer ormal u. If f : 0, R +, 1, are covex ad/or cocave fuctos, the D f PK1,..., P K, Q K1,..., Q K S 1 f 1 f K h +1 K f K h K ] 1 dσ, are the geeral mxed affe surface areas troduced 41]. We refer to 33] for more detals o covex bodes. 3 Iequaltes The classcal Alexadrov-Fechel equalty for mxed volumes of covex bodes s a fudametal result covex geometry. A geeral verso of ths equalty for mxed volumes of covex bodes ca be foud 1, 6, 33]. Alexadrov-Fechel type equaltes for mxed affe surface areas ca be foud 21, 22, 40, 41]. Now we prove a equalty for the mxed f-dvergece for measures, whch we call 6

7 a Alexadrov-Fechel type equalty because of ts formal resemblace to be a Alexadrov-Fechel type equalty for covex bodes. Followg 13], we say that two fuctos f ad g are effectvely proportoal f there are costats a ad b, ot both zero, such that af bg. Fuctos f 1,..., f m are effectvely proportoal f every par f, f j, 1, j m s effectvely proportoal. A ull fucto s effectvely proportoal to ay fucto. These otos wll be used the ext theorems. For a measure space, µ ad probablty destes p ad q, 1, we put ad for j 0,, m 1, g 0 u g j+1 u m For a vector p, we deote by p,k the followg vector p f q, 3.4 q p j f j q j. 3.5 q j p,k p 1,, p m, p k,, p }{{ k, k > m. } m Theorem 3.1. Let, µ be a measure space. For 1, let P ad Q be probablty measures o, µ wth desty fuctos p ad q respectvely µ-a.e. Let f : 0, R +, 1, be covex fuctos. The, for 1 m, D f P, Q ] m k m+1 D f,k P,k, Q,k. Equalty holds f ad oly f oe of the fuctos g 1 m 0 g, 1 m, s ull or all are effectvely proportoal µ-a.e. If m, D f P, Q] D f P, Q, wth equalty f ad oly f oe of the fuctos f j pj q j q j, 0 j, s ull or all are effectvely proportoal µ-a.e. Remarks. I partcular, equalty holds Theorem 3.1 f all P, Q cocde, ad f λ f for some covex postve fucto f ad λ 0, 1, 2,,. Theorem 3.1 stll holds true f the fuctos f are cocave. 7

8 Proof. We let g 0 ad g j+1, j 0,, m 1 as 3.4 ad 3.5. By Hölder s equalty see 13] D f P, Q] m m 1 j0 m g 0 ug 1 u g m u dµ m 1 j0 k m+1 g 0 ug j+1 u m ] 1 m dµ m g 0 ugj+1u m dµ D f,k P,k, Q,k. Equalty holds Hölder s equalty, f ad oly f oe of the fuctos g 1 m 0 g, 1 m, s ull or all are effectvely proportoal µ-a.e. I partcular, ths s the case, f for all 1,,, P, Q P, Q ad f λ f for some covex fucto f ad λ 0. We requre some propertes of f-dvergeces for our ext result. Let f : 0, R + be a covex fucto. By Jese s equalty, p D f P, Q f q dµ f p dµ f1, 3.6 q for all pars of probablty measures P, Q o, µ wth ozero desty fuctos p ad q respectvely µ-a.e. Whe f s lear, equalty holds trvally 3.6. Whe f s strctly covex, equalty holds true f ad oly f p q µ-a.e. If f s a cocave fucto, Jese s equalty mples D f P, Q p f q dµ f q p dµ f1, 3.7 for all pars of probablty measures P, Q. Aga, whe f s lear, equalty holds trvally. Whe f s strctly cocave, equalty holds true f ad oly f p q µ-a.e. For the mxed f-dvergece wth cocave fuctos, oe has the followg result. Theorem 3.2. Let, µ be a measure space. For all 1, let P ad Q be probablty measures o whose desty fuctos p ad q are ozero µ-a.e. Let f : 0, R +, 1, be cocave fuctos. The D f P, Q] D f P, Q f If addto, all f are strctly cocave, equalty holds f ad oly f there s a probablty desty p such that for all 1, 2,, p q p, µ a.e. 8

9 Proof. Theorem 3.1 ad the remark after mply that for all cocave fuctos f, D f P, Q] D f P, Q f 1, where the secod equalty follows from equalty 3.7 ad f 0. Suppose ow that for all, p q p, µ-a.e., where p s a fxed probablty desty. The equalty holds trvally 3.8. Coversely, suppose that equalty holds 3.8. The, partcular, equalty holds Jese s equalty whch, as oted above, happes f ad oly f p q for all. Thus, D f P, Q 1] f 1/ / 1... q 1/ dµ. Note also that f all f : 0, R + are strctly cocave, f 1 0 for all 1. Equalty characterzato Hölder s equalty mples that all q are effectvely proportoal µ-a.e. As all q are probablty measures, they are all equal µ-a.e. to a probablty measure wth desty fucto say p. Remark. If f t a t + b are all lear ad postve, the equalty holds f ad oly f all p, q are equal µ-a.e. as covex combatos,.e., f ad oly f for all, j a a + b p + b a + b q a j a j + b j p j + b j a j + b j q j, µ a.e. 4 The -th mxed f-dvergece Let, µ be a measure space. Throughout ths secto, we assume that the fuctos f 1, f 2 : 0, {x R : x > 0}, are covex or cocave, ad that P 1, P 2, Q 1, Q 2 are probablty measures o wth desty fuctos p 1, p 2,, whch are ozero µ-a.e. We also wrte f f 1, f 2, P P1, P 2, Q Q1, Q 2. Defto 4.1. Let R. The -th mxed f-dvergece for f, P, Q, deoted by D f P, Q;, s defed as ] ] D f P, Q; p1 p2 f 1 f 2 dµ. 4.9 Remarks. Note that the -th mxed f-dvergece s defed for ay combato of covexty ad cocavty of f 1 ad f 2, amely, both f 1 ad f 2 cocave, or both f 1 ad f 2 covex, or oe s covex the other s cocave. It s easly checked that D f P, Q; D f2,f 1 P2, Q 2, P 1, Q 1 ;. 9

10 If 0 s a teger, the the trple f 1, P 1, Q 1 appears -tmes whle the trple f 2, P 2, Q 2 appears tmes D f P, Q;. Note that f 0, the D f P, Q; D f2 P 2, Q 2, ad f the D f P, Q; D f1 P 1, Q 1. Aother specal case s whe P 2 Q 2 µ almost everywhere ad µ s also a probablty measure. The such a -th mxed f-dvergece, deoted by D f 1, P 1, Q 1, ; f 2, has the form D f 1, P 1, Q 1, ; f 2 f2 1] 1 / ] p1 f 1 dµ. Examples ad Applcatos. For ft t 1, we get the -th mxed total varato D T V P, Q; p 1 p2 dµ. For f 1 t f 2 t t l t] +, we get the modfed -th mxed relatve etropy or -th mxed Kullback Lebler dvergece D KL P, Q; p 1 l p1 ] + p 2 l p2 ] For the covex or cocave fuctos f αj t t α j, j 1, 2, we get the -th mxed Hellger tegrals I partcular, for α j α, for j 1, 2, + dµ. D fα1,f α2 P, Q; p α 1 1 q1 α 1 1 p α 2 2 q1 α 2 2 dµ. D fα,f P α, Q; p α 1 q1 1 α p α 2 α 2 Ths tegral ca be used to defe the -th mxed α-réy dvergece D α P, Q; 1 α 1 l D fα,f P α, Q; ]. The case α 1 2 for all gves D t, t P, Q; p 1 2 p2 2 dµ, the -th mxed Bhattacharyya coeffcet or -th mxed Bhattacharyya dstace of p ad q. v Importat applcatos are aga the theory of covex bodes. As secto 2, let K 1 ad K 2 be covex bodes wth postve curvature fucto. For l 1, 2, let dp Kl 1 h K l dσ ad dq Kl f Kl h Kl dσ. 10 dµ.

11 Let f l : 0, R, l 1, 2, be postve covex fuctos. The, we defe the -th mxed f-dvergece for covex bodes K 1 ad K 2 by D f PK1, P K2, Q K1, Q K2 ; S 1 f 1 1 f K1 h +1 K 1 f K1 h K1 ] f 2 These are the geeral -th mxed affe surface areas troduced 41]. 1 f K2 h +1 K 2 f K2 h K2 ] The followg result holds for all possble combatos of covexty ad cocavty of f 1 ad f 2. Proposto 4.1. Let f, P, Q be as above. If j k or k j, the D f P, Q; D f P, Q; j ] k k j D f P, Q; k ] j k j. Equalty holds trvally f k or j. Otherwse, equalty holds f ad oly f oe of the fuctos f p q q, 1, 2, s ull, or f p1 1 ad f p2 2 are effectvely proportoal µ-a.e. I partcular, ths holds f P 1, Q 1 P 2, Q 2 ad f 1 λf 2 for some λ > 0. dσ. Proof. By formula 4.9, oe has D f P, Q; ] p1 f 1 f 2 p2 ] dµ { ] j ] j } k k j p1 p2 f 1 f 2 { ] k ] k } j k j p1 p2 f 1 f 2 dµ D f P, Q; j ] k k j D f P, Q; k ] j k j, where the last equalty follows from Hölder s equalty ad formula 4.9. The equalty characterzato follows from the oe Hölder equalty. I partcular, f P 1, Q 1 P 2, Q 2, ad f 1 λf 2 for some λ > 0, equalty holds. Corollary 4.1. Let f 1 ad f 2 be postve, cocave fuctos o 0,. The for all P, Q ad for all 0, D f P, Q; ] f1 1] f 2 1]. If addto, f 1 ad f 2 are strctly cocave, equalty holds ff p 1 p 2 µ-a.e. Proof. Let j 0 ad k Proposto 4.1. The for all 0, D f P, Q; ] Df1 P 1, Q 1 ] D f2 P 2, Q 2 ] f 1 1] f 2 1], 11

12 where the last equalty follows from equalty 3.7. To have equalty, the above equaltes should be equaltes. Proposto 4.1 mples that f 1 p1 ad f 2 p2 are effectvely proportoal µ-a.e. As both f 1 ad f 2 are strctly cocave, Jese s equalty requres that p 1 ad p 2 µ-a.e. Therefore, equalty holds f ad oly f f 1 1 ad f 2 1 are effectvely proportoal µ-a.e. As both f 1 1 ad f 2 1 are ot zero, equalty holds ff p 1 p 2 µ-a.e. Remark. If f 1 t a 1 t + b 1 ad f 2 t a 2 t + b 2 are both lear, equalty holds Corollary 4.1 f ad oly f p, q, 1, 2, are equal as covex combatos,.e., a 1 a 1 + b 1 p 1 + b 1 a 1 + b 1 a 2 a 2 + b 2 p 2 + b 2 a 2 + b 2, µ a.e. Ths proof ca be used to establsh the followg result for D f 1, P 1, Q 1, ; f 2. Corollary 4.2. Let, µ be a probablty space. Let f 1 be a postve cocave fucto o 0,. The for all P 1, Q 1, for all cocave or covex postve fuctos f 2, ad for all 0, D f1, P 1, Q 1, ; f 2 ] f1 1] f 2 1]. If f 1 s strctly cocave, equalty holds f ad oly f P 1 Q 1 µ. Whe f 1 t at + b s lear, equalty holds f ad oly f ap 1 + b a + b µ-a.e. Corollary 4.3. Let f 1 be a postve covex fucto ad f 2 be a postve cocave fucto o 0,. The, for all P, Q, ad for all k, D f P, Q; k ] f1 1] k f 2 1] k. If addto, f 1 s strctly covex ad f 2 s strctly cocave, equalty holds f ad oly f p 1 p 2 µ-a.e. Proof. O the rght had sde of Proposto 4.1, let ad j 0. Let k. The D f P, Q; k ] Df1 P 1, Q 1 ] k D f2 P 2, Q 2 ] k f 1 1] k f 2 1] k. Here, the last equalty follows from equaltes 3.6, 3.7 ad k. To have equalty, the above equaltes should be equaltes. Proposto 4.1 mples that f p1 1 ad f p2 2 are effectvely proportoal µ-a.e. As f 1 s strctly covex ad f 2 s strctly cocave, Jese s equalty mples that p 1 ad p 2 µ-a.e. Therefore, as both f 1 1 ad f 2 1 are ot zero, equalty holds f ad oly f p 1 p 2 µ-a.e. Remark. If f 1 t a 1 t + b 1 ad f 2 t a 2 t + b 2 are both lear, equalty holds Corollary 4.3 f ad oly f p, q, 1, 2, are equal µ-a.e. as covex combatos,.e., a 1 a 1 + b 1 p 1 + b 1 a 1 + b 1 a 2 a 2 + b 2 p 2 + b 2 a 2 + b 2, µ a.e. Ths proof ca be used to establsh the followg result for D f 1, P 1, Q 1, k; f 2. 12

13 Corollary 4.4. Let, µ be a probablty space. Let f 1 be a postve covex fucto o 0,. The for all P 1, Q 1, for all postve cocave or covex fuctos f 2, ad for all k, D f1, P 1, Q 1, k; f 2 ] f1 1] k f 2 1] k. If f 1 s strctly covex, equalty holds f ad oly f P 1 Q 1 µ. Whe f 1 t at + b s lear, equalty holds f ad oly f ap 1 + b a + b µ-a.e. Corollary 4.5. Let f 1 be a postve cocave fucto ad f 2 be a postve covex fucto o 0,. The for all P, Q, ad for all k 0, D f P, Q; k ] f1 1] k f 2 1] k. If addto, f 1 s strctly cocave ad f 2 s strctly covex, equalty holds ff p 1 p 2 µ-a.e. Proof. Let 0 ad j Proposto 4.1. The D f P, Q; k ] D f1 P 1, Q 1 ] k D f2 P 2, Q 2 ] k f 1 1] k f 2 1] k. Here, the last equalty follows from equaltes 3.6, 3.7, ad k 0. To have equalty, the above equaltes should be equaltes. Proposto 4.1 mples that f 1 p1 ad f 2 p2 are effectvely proportoal µ-a.e. As f 1 s strctly cocave ad f 2 s strctly covex, Jese s equalty requres that p 1 ad p 2. Therefore, equalty holds f ad oly f f 1 1 ad f 2 1 are effectvely proportoal µ-a.e. As both f 1 1 ad f 2 1 are ot zero, equalty holds f ad oly f p 1 p 2 µ-a.e. Ths proof ca be used to establsh the followg result for D f 1, P 1, Q 1, k; f 2. Corollary 4.6. Let f 1 be a cocave fucto o 0,. The for all P 1, Q 1, for all cocave or covex fuctos f 2, ad for all k 0, D f1, P 1, Q 1, k; f 2 ] f1 1] k f 2 1] k. If f 1 s strctly cocave, equalty holds f ad oly f P 1 Q 1 µ. Whe f 1 t at + b s lear, equalty holds f ad oly f ap 1 + b a + b µ-a.e. Refereces 1] A.D. Aleksadrov, O the theory of mxed volumes of covex bodes. II. New equaltes betwee mxed volumes ad ther applcatos, Mat. Sb. N. S Russa] 2] M.S. Al ad D. Slvey, A geeral class of coeffcets of dvergece of oe dstrbuto from aother, J. R. Stat. Soc. B ] A.R. Barro, L. Györf ad E.C. va der Meule, Dstrbuto estmates cosstet total varato ad two types of formato dvergece, IEEE Tras. Iform. Theory

14 4] M. Bassevlle, Dvergece measures for statstcal data processg, Techcal Report PI 1961, IRISA, November URL 5] A. Bhattacharyya, O some aalogues to the amout of formato ad ther uses statstcal estmato, Sakhya ] H. Busema, Covex surface, Iterscece Tracts Pure ad Appl. Math., No. 6, Iterscece, New York, MR 21 # ] T. Cover ad J. Thomas, Elemets of formato theory, secod ed., Wley-Iterscece, Joh Wley ad Sos, Hoboke, NJ, ] I. Csszár, Ee formatostheoretsche Uglechug ud hre Awedug auf de Bewes der Ergodztät vo Markoffsche Kette, Publ. Math. Ist. Hugar. Acad. Sc. ser. A, ] R. J. Garder, The Bru-Mkowsk Iequalty, Bull. Amer. Math. Soc. 39, 2002, ] O.G. Guleryuz, E. Lutwak, D. Yag ad G. Zhag, Iformato theoretc equaltes for cotoured probablty dstrbutos, IEEE Tras. Iform. Theory ] L. Györf ad T. Nemetz, f-dssmlarty: A geeral class of separato measures of several probablty measures, I I. Csszár ad P. Elas, edtors, Topcs Iformato Theory, volume 16 of Colloqua Mathematca Socetats Jáos Bolya, pages North-Hollad, ] L. Györf ad T. Nemetz, f-dssmlarty: A geeralzato of the affty of several dstrbutos, A. Ist. Statst. Math ] G.H. Hardy, J.E. Lttlewood ad G. Pólya, Iequaltes, 2d ed., Cambrdge Uv. Press, ] P. Harremoes ad F. Topsoe, Iequaltes betwee etropy ad the dex of cocdece derved from formato dagrams, IEEE Tras. Iform. Theory ] J. Jekso ad E. Werer, Relatve etropes for covex bodes, Tras. Amer. Math. Soc ] S. Kullback ad R. Lebler, O formato ad suffcecy, A. Math. Statst ] F. Lese ad I. Vajda, O Dvergeces ad Iformato Statstcs ad Iformato Theory, IEEE Tras. Iform. Theory ] M. Ludwg, Geeral affe surface areas, Adv. Math ] M. Ludwg ad M. Retzer, A characterzato of affe surface area, Adv. Math ] M. Ludwg ad M. Retzer, A classfcato of SL varat valuatos, Aals of Math ] E. Lutwak, Mxed affe surface area, J. Math. Aal. Appl ] E. Lutwak, The Bru-Mkowsk-Frey theory. II. affe ad geommal surface areas, Adv. Math ] E. Lutwak, D. Yag ad G. Zhag, The Cramer-Rao equalty for star bodes, Duke Math. J ] E. Lutwak, D. Yag ad G. Zhag, Momet-etropy equaltes, A. Probab

15 25] E. Lutwak, D. Yag ad G. Zhag, Cramer-Rao ad momet-etropy equaltes for Rey etropy ad geeralzed Fsher formato, IEEE Tras. Iform. Theory ] K. Matusta, O the oto of affty of several dstrbutos ad some of ts applcatos, A. Ist. Statst. Math ] K. Matusta, Some propertes of affty ad applcatos, A. Ist. Statst. Math ] M.L. Meédez, J.A. Pardo, L. Pardo ad K. Zografos, A prelmary test classfcato ad probabltes of msclassfcato, Statstcs ] D. Morales, L. Pardo ad K. Zografos, Iformatoal dstaces ad related statstcs mxed cotuous ad categorcal varables, J. Statst. Pla. Iferece ] T. Mormoto, Markov processes ad the H-theorem, J. Phys. Soc. Jap ] F. Österrecher ad I. Vajda, A ew class of metrc dvergeces o probablty spaces ad ts applcablty statstcs, A. Ist. Statst. Math ] G. Paours ad E. Werer, Relatve etropy of coe measures ad L p cetrod bodes, Proc. Lodo Math. Soc ] R. Scheder, Covex Bodes: The Bru-Mkowsk theory, Cambrdge Uv. Press, ] C. Schütt ad E. Werer, Surface bodes ad p-affe surface area, Adv. Math ] A. Sgarro, Iformatoal dvergece ad the dssmlarty of probablty dstrbutos, Calcolo ] R. Sbso, Iformato radus, Probab. Theory Related Felds ] G.T. Toussat, Some propertes of Matusta s measure of affty of several dstrbutos, A. Ist. Statst. Math ] E. Werer, Réy Dvergece ad L p -affe surface area for covex bodes, Adv. Math ] E. Werer, f-dvergece for covex bodes, Proceedgs of the Asymptotc Geometrc Aalyss workshop, Felds Isttute, Toroto ] E. Werer ad D. Ye, Iequaltes for mxed p-affe surface area, Math. A ] D. Ye, Iequaltes for geeral mxed affe surface areas, J. Lodo Math. Soc ] K. Zografos, f-dssmlarty of several dstrbutos testg statstcal hypotheses, A. Ist. Statst. Math Elsabeth Werer, elsabeth.werer@case.edu Departmet of Mathematcs Uversté de Llle 1 Case Wester Reserve Uversty UFR de Mathématque Clevelad, Oho 44106, U. S. A Vlleeuve d Ascq, Frace Depg Ye, depg.ye@mu.ca Departmet of Mathematcs ad Statstcs Memoral Uversty of Newfoudlad St. Joh s, Newfoudlad, Caada A1C 5S7 15

PROJECTION PROBLEM FOR REGULAR POLYGONS

PROJECTION PROBLEM FOR REGULAR POLYGONS Joural of Mathematcal Sceces: Advaces ad Applcatos Volume, Number, 008, Pages 95-50 PROJECTION PROBLEM FOR REGULAR POLYGONS College of Scece Bejg Forestry Uversty Bejg 0008 P. R. Cha e-mal: sl@bjfu.edu.c

More information

MAX-MIN AND MIN-MAX VALUES OF VARIOUS MEASURES OF FUZZY DIVERGENCE

MAX-MIN AND MIN-MAX VALUES OF VARIOUS MEASURES OF FUZZY DIVERGENCE merca Jr of Mathematcs ad Sceces Vol, No,(Jauary 0) Copyrght Md Reader Publcatos wwwjouralshubcom MX-MIN ND MIN-MX VLUES OF VRIOUS MESURES OF FUZZY DIVERGENCE RKTul Departmet of Mathematcs SSM College

More information

Q-analogue of a Linear Transformation Preserving Log-concavity

Q-analogue of a Linear Transformation Preserving Log-concavity Iteratoal Joural of Algebra, Vol. 1, 2007, o. 2, 87-94 Q-aalogue of a Lear Trasformato Preservg Log-cocavty Daozhog Luo Departmet of Mathematcs, Huaqao Uversty Quazhou, Fua 362021, P. R. Cha ldzblue@163.com

More information

3. Basic Concepts: Consequences and Properties

3. Basic Concepts: Consequences and Properties : 3. Basc Cocepts: Cosequeces ad Propertes Markku Jutt Overvew More advaced cosequeces ad propertes of the basc cocepts troduced the prevous lecture are derved. Source The materal s maly based o Sectos.6.8

More information

Summary of the lecture in Biostatistics

Summary of the lecture in Biostatistics Summary of the lecture Bostatstcs Probablty Desty Fucto For a cotuos radom varable, a probablty desty fucto s a fucto such that: 0 dx a b) b a dx A probablty desty fucto provdes a smple descrpto of the

More information

A New Measure of Probabilistic Entropy. and its Properties

A New Measure of Probabilistic Entropy. and its Properties Appled Mathematcal Sceces, Vol. 4, 200, o. 28, 387-394 A New Measure of Probablstc Etropy ad ts Propertes Rajeesh Kumar Departmet of Mathematcs Kurukshetra Uversty Kurukshetra, Ida rajeesh_kuk@redffmal.com

More information

The Mathematical Appendix

The Mathematical Appendix The Mathematcal Appedx Defto A: If ( Λ, Ω, where ( λ λ λ whch the probablty dstrbutos,,..., Defto A. uppose that ( Λ,,..., s a expermet type, the σ-algebra o λ λ λ are defed s deoted by ( (,,...,, σ Ω.

More information

The Brunn Minkowski Inequality, Minkowski s First Inequality, and Their Duals 1

The Brunn Minkowski Inequality, Minkowski s First Inequality, and Their Duals 1 Joural of Mathematcal Aalyss ad Applcatos 245, 502 52 2000 do:0.006 jmaa.2000.6774, avalable ole at http: www.dealbrary.com o The Bru Mkowsk Iequalty, Mkowsk s Frst Iequalty, ad Ther Duals R. J. Garder

More information

STK4011 and STK9011 Autumn 2016

STK4011 and STK9011 Autumn 2016 STK4 ad STK9 Autum 6 Pot estmato Covers (most of the followg materal from chapter 7: Secto 7.: pages 3-3 Secto 7..: pages 3-33 Secto 7..: pages 35-3 Secto 7..3: pages 34-35 Secto 7.3.: pages 33-33 Secto

More information

Generalization of the Dissimilarity Measure of Fuzzy Sets

Generalization of the Dissimilarity Measure of Fuzzy Sets Iteratoal Mathematcal Forum 2 2007 o. 68 3395-3400 Geeralzato of the Dssmlarty Measure of Fuzzy Sets Faramarz Faghh Boformatcs Laboratory Naobotechology Research Ceter vesa Research Isttute CECR Tehra

More information

The Arithmetic-Geometric mean inequality in an external formula. Yuki Seo. October 23, 2012

The Arithmetic-Geometric mean inequality in an external formula. Yuki Seo. October 23, 2012 Sc. Math. Japocae Vol. 00, No. 0 0000, 000 000 1 The Arthmetc-Geometrc mea equalty a exteral formula Yuk Seo October 23, 2012 Abstract. The classcal Jese equalty ad ts reverse are dscussed by meas of terally

More information

Lecture 3 Probability review (cont d)

Lecture 3 Probability review (cont d) STATS 00: Itroducto to Statstcal Iferece Autum 06 Lecture 3 Probablty revew (cot d) 3. Jot dstrbutos If radom varables X,..., X k are depedet, the ther dstrbuto may be specfed by specfyg the dvdual dstrbuto

More information

9 U-STATISTICS. Eh =(m!) 1 Eh(X (1),..., X (m ) ) i.i.d

9 U-STATISTICS. Eh =(m!) 1 Eh(X (1),..., X (m ) ) i.i.d 9 U-STATISTICS Suppose,,..., are P P..d. wth CDF F. Our goal s to estmate the expectato t (P)=Eh(,,..., m ). Note that ths expectato requres more tha oe cotrast to E, E, or Eh( ). Oe example s E or P((,

More information

Complete Convergence for Weighted Sums of Arrays of Rowwise Asymptotically Almost Negative Associated Random Variables

Complete Convergence for Weighted Sums of Arrays of Rowwise Asymptotically Almost Negative Associated Random Variables A^VÇÚO 1 32 ò 1 5 Ï 2016 c 10 Chese Joural of Appled Probablty ad Statstcs Oct., 2016, Vol. 32, No. 5, pp. 489-498 do: 10.3969/j.ss.1001-4268.2016.05.005 Complete Covergece for Weghted Sums of Arrays of

More information

ρ < 1 be five real numbers. The

ρ < 1 be five real numbers. The Lecture o BST 63: Statstcal Theory I Ku Zhag, /0/006 Revew for the prevous lecture Deftos: covarace, correlato Examples: How to calculate covarace ad correlato Theorems: propertes of correlato ad covarace

More information

Lecture Notes Types of economic variables

Lecture Notes Types of economic variables Lecture Notes 3 1. Types of ecoomc varables () Cotuous varable takes o a cotuum the sample space, such as all pots o a le or all real umbers Example: GDP, Polluto cocetrato, etc. () Dscrete varables fte

More information

Point Estimation: definition of estimators

Point Estimation: definition of estimators Pot Estmato: defto of estmators Pot estmator: ay fucto W (X,..., X ) of a data sample. The exercse of pot estmato s to use partcular fuctos of the data order to estmate certa ukow populato parameters.

More information

Complete Convergence and Some Maximal Inequalities for Weighted Sums of Random Variables

Complete Convergence and Some Maximal Inequalities for Weighted Sums of Random Variables Joural of Sceces, Islamc Republc of Ira 8(4): -6 (007) Uversty of Tehra, ISSN 06-04 http://sceces.ut.ac.r Complete Covergece ad Some Maxmal Iequaltes for Weghted Sums of Radom Varables M. Am,,* H.R. Nl

More information

Non-uniform Turán-type problems

Non-uniform Turán-type problems Joural of Combatoral Theory, Seres A 111 2005 106 110 wwwelsevercomlocatecta No-uform Turá-type problems DhruvMubay 1, Y Zhao 2 Departmet of Mathematcs, Statstcs, ad Computer Scece, Uversty of Illos at

More information

On Submanifolds of an Almost r-paracontact Riemannian Manifold Endowed with a Quarter Symmetric Metric Connection

On Submanifolds of an Almost r-paracontact Riemannian Manifold Endowed with a Quarter Symmetric Metric Connection Theoretcal Mathematcs & Applcatos vol. 4 o. 4 04-7 ISS: 79-9687 prt 79-9709 ole Scepress Ltd 04 O Submafolds of a Almost r-paracotact emaa Mafold Edowed wth a Quarter Symmetrc Metrc Coecto Mob Ahmad Abdullah.

More information

arxiv:math/ v1 [math.gm] 8 Dec 2005

arxiv:math/ v1 [math.gm] 8 Dec 2005 arxv:math/05272v [math.gm] 8 Dec 2005 A GENERALIZATION OF AN INEQUALITY FROM IMO 2005 NIKOLAI NIKOLOV The preset paper was spred by the thrd problem from the IMO 2005. A specal award was gve to Yure Boreko

More information

Lecture 7. Confidence Intervals and Hypothesis Tests in the Simple CLR Model

Lecture 7. Confidence Intervals and Hypothesis Tests in the Simple CLR Model Lecture 7. Cofdece Itervals ad Hypothess Tests the Smple CLR Model I lecture 6 we troduced the Classcal Lear Regresso (CLR) model that s the radom expermet of whch the data Y,,, K, are the outcomes. The

More information

ON THE LOGARITHMIC INTEGRAL

ON THE LOGARITHMIC INTEGRAL Hacettepe Joural of Mathematcs ad Statstcs Volume 39(3) (21), 393 41 ON THE LOGARITHMIC INTEGRAL Bra Fsher ad Bljaa Jolevska-Tueska Receved 29:9 :29 : Accepted 2 :3 :21 Abstract The logarthmc tegral l(x)

More information

Generalized Measure for Two Utility Distributions

Generalized Measure for Two Utility Distributions Proceedgs of the World Cogress o Egeerg 00 Vol III WCE 00, Jue 30 - July, 00, Lodo, U.K. Geeralzed Meure for Two Utlty Dstrbutos J. S. BHULLAR MEMBER IAENG, O. P. VINOCHA, MANISH GUPTA Abstract The meure

More information

18.413: Error Correcting Codes Lab March 2, Lecture 8

18.413: Error Correcting Codes Lab March 2, Lecture 8 18.413: Error Correctg Codes Lab March 2, 2004 Lecturer: Dael A. Spelma Lecture 8 8.1 Vector Spaces A set C {0, 1} s a vector space f for x all C ad y C, x + y C, where we take addto to be compoet wse

More information

Journal of Mathematical Analysis and Applications

Journal of Mathematical Analysis and Applications J. Math. Aal. Appl. 365 200) 358 362 Cotets lsts avalable at SceceDrect Joural of Mathematcal Aalyss ad Applcatos www.elsever.com/locate/maa Asymptotc behavor of termedate pots the dfferetal mea value

More information

Lecture 9: Tolerant Testing

Lecture 9: Tolerant Testing Lecture 9: Tolerat Testg Dael Kae Scrbe: Sakeerth Rao Aprl 4, 07 Abstract I ths lecture we prove a quas lear lower boud o the umber of samples eeded to do tolerat testg for L dstace. Tolerat Testg We have

More information

Multiple Linear Regression Analysis

Multiple Linear Regression Analysis LINEA EGESSION ANALYSIS MODULE III Lecture - 4 Multple Lear egresso Aalyss Dr. Shalabh Departmet of Mathematcs ad Statstcs Ida Isttute of Techology Kapur Cofdece terval estmato The cofdece tervals multple

More information

Third handout: On the Gini Index

Third handout: On the Gini Index Thrd hadout: O the dex Corrado, a tala statstca, proposed (, 9, 96) to measure absolute equalt va the mea dfferece whch s defed as ( / ) where refers to the total umber of dvduals socet. Assume that. The

More information

Research Article Some Strong Limit Theorems for Weighted Product Sums of ρ-mixing Sequences of Random Variables

Research Article Some Strong Limit Theorems for Weighted Product Sums of ρ-mixing Sequences of Random Variables Hdaw Publshg Corporato Joural of Iequaltes ad Applcatos Volume 2009, Artcle ID 174768, 10 pages do:10.1155/2009/174768 Research Artcle Some Strog Lmt Theorems for Weghted Product Sums of ρ-mxg Sequeces

More information

Entropy ISSN by MDPI

Entropy ISSN by MDPI Etropy 2003, 5, 233-238 Etropy ISSN 1099-4300 2003 by MDPI www.mdp.org/etropy O the Measure Etropy of Addtve Cellular Automata Hasa Aı Arts ad Sceces Faculty, Departmet of Mathematcs, Harra Uversty; 63100,

More information

Extreme Value Theory: An Introduction

Extreme Value Theory: An Introduction (correcto d Extreme Value Theory: A Itroducto by Laures de Haa ad Aa Ferrera Wth ths webpage the authors ted to form the readers of errors or mstakes foud the book after publcato. We also gve extesos for

More information

{ }{ ( )} (, ) = ( ) ( ) ( ) Chapter 14 Exercises in Sampling Theory. Exercise 1 (Simple random sampling): Solution:

{ }{ ( )} (, ) = ( ) ( ) ( ) Chapter 14 Exercises in Sampling Theory. Exercise 1 (Simple random sampling): Solution: Chapter 4 Exercses Samplg Theory Exercse (Smple radom samplg: Let there be two correlated radom varables X ad A sample of sze s draw from a populato by smple radom samplg wthout replacemet The observed

More information

Research Article A New Iterative Method for Common Fixed Points of a Finite Family of Nonexpansive Mappings

Research Article A New Iterative Method for Common Fixed Points of a Finite Family of Nonexpansive Mappings Hdaw Publshg Corporato Iteratoal Joural of Mathematcs ad Mathematcal Sceces Volume 009, Artcle ID 391839, 9 pages do:10.1155/009/391839 Research Artcle A New Iteratve Method for Commo Fxed Pots of a Fte

More information

Chapter 5 Properties of a Random Sample

Chapter 5 Properties of a Random Sample Lecture 6 o BST 63: Statstcal Theory I Ku Zhag, /0/008 Revew for the prevous lecture Cocepts: t-dstrbuto, F-dstrbuto Theorems: Dstrbutos of sample mea ad sample varace, relatoshp betwee sample mea ad sample

More information

MATH 247/Winter Notes on the adjoint and on normal operators.

MATH 247/Winter Notes on the adjoint and on normal operators. MATH 47/Wter 00 Notes o the adjot ad o ormal operators I these otes, V s a fte dmesoal er product space over, wth gve er * product uv, T, S, T, are lear operators o V U, W are subspaces of V Whe we say

More information

ONE GENERALIZED INEQUALITY FOR CONVEX FUNCTIONS ON THE TRIANGLE

ONE GENERALIZED INEQUALITY FOR CONVEX FUNCTIONS ON THE TRIANGLE Joural of Pure ad Appled Mathematcs: Advaces ad Applcatos Volume 4 Number 205 Pages 77-87 Avalable at http://scetfcadvaces.co. DOI: http://.do.org/0.8642/jpamaa_7002534 ONE GENERALIZED INEQUALITY FOR CONVEX

More information

UNIVERSITY OF OSLO DEPARTMENT OF ECONOMICS

UNIVERSITY OF OSLO DEPARTMENT OF ECONOMICS UNIVERSITY OF OSLO DEPARTMENT OF ECONOMICS Postpoed exam: ECON430 Statstcs Date of exam: Jauary 0, 0 Tme for exam: 09:00 a.m. :00 oo The problem set covers 5 pages Resources allowed: All wrtte ad prted

More information

Part 4b Asymptotic Results for MRR2 using PRESS. Recall that the PRESS statistic is a special type of cross validation procedure (see Allen (1971))

Part 4b Asymptotic Results for MRR2 using PRESS. Recall that the PRESS statistic is a special type of cross validation procedure (see Allen (1971)) art 4b Asymptotc Results for MRR usg RESS Recall that the RESS statstc s a specal type of cross valdato procedure (see Alle (97)) partcular to the regresso problem ad volves fdg Y $,, the estmate at the

More information

hp calculators HP 30S Statistics Averages and Standard Deviations Average and Standard Deviation Practice Finding Averages and Standard Deviations

hp calculators HP 30S Statistics Averages and Standard Deviations Average and Standard Deviation Practice Finding Averages and Standard Deviations HP 30S Statstcs Averages ad Stadard Devatos Average ad Stadard Devato Practce Fdg Averages ad Stadard Devatos HP 30S Statstcs Averages ad Stadard Devatos Average ad stadard devato The HP 30S provdes several

More information

D KL (P Q) := p i ln p i q i

D KL (P Q) := p i ln p i q i Cheroff-Bouds 1 The Geeral Boud Let P 1,, m ) ad Q q 1,, q m ) be two dstrbutos o m elemets, e,, q 0, for 1,, m, ad m 1 m 1 q 1 The Kullback-Lebler dvergece or relatve etroy of P ad Q s defed as m D KL

More information

ENGI 4421 Joint Probability Distributions Page Joint Probability Distributions [Navidi sections 2.5 and 2.6; Devore sections

ENGI 4421 Joint Probability Distributions Page Joint Probability Distributions [Navidi sections 2.5 and 2.6; Devore sections ENGI 441 Jot Probablty Dstrbutos Page 7-01 Jot Probablty Dstrbutos [Navd sectos.5 ad.6; Devore sectos 5.1-5.] The jot probablty mass fucto of two dscrete radom quattes, s, P ad p x y x y The margal probablty

More information

TESTS BASED ON MAXIMUM LIKELIHOOD

TESTS BASED ON MAXIMUM LIKELIHOOD ESE 5 Toy E. Smth. The Basc Example. TESTS BASED ON MAXIMUM LIKELIHOOD To llustrate the propertes of maxmum lkelhood estmates ad tests, we cosder the smplest possble case of estmatg the mea of the ormal

More information

NP!= P. By Liu Ran. Table of Contents. The P versus NP problem is a major unsolved problem in computer

NP!= P. By Liu Ran. Table of Contents. The P versus NP problem is a major unsolved problem in computer NP!= P By Lu Ra Table of Cotets. Itroduce 2. Prelmary theorem 3. Proof 4. Expla 5. Cocluso. Itroduce The P versus NP problem s a major usolved problem computer scece. Iformally, t asks whether a computer

More information

Estimation of Stress- Strength Reliability model using finite mixture of exponential distributions

Estimation of Stress- Strength Reliability model using finite mixture of exponential distributions Iteratoal Joural of Computatoal Egeerg Research Vol, 0 Issue, Estmato of Stress- Stregth Relablty model usg fte mxture of expoetal dstrbutos K.Sadhya, T.S.Umamaheswar Departmet of Mathematcs, Lal Bhadur

More information

THE COMPLETE ENUMERATION OF FINITE GROUPS OF THE FORM R 2 i ={R i R j ) k -i=i

THE COMPLETE ENUMERATION OF FINITE GROUPS OF THE FORM R 2 i ={R i R j ) k -i=i ENUMERATON OF FNTE GROUPS OF THE FORM R ( 2 = (RfR^'u =1. 21 THE COMPLETE ENUMERATON OF FNTE GROUPS OF THE FORM R 2 ={R R j ) k -= H. S. M. COXETER*. ths paper, we vestgate the abstract group defed by

More information

Chapter 4 Multiple Random Variables

Chapter 4 Multiple Random Variables Revew for the prevous lecture: Theorems ad Examples: How to obta the pmf (pdf) of U = g (, Y) ad V = g (, Y) Chapter 4 Multple Radom Varables Chapter 44 Herarchcal Models ad Mxture Dstrbutos Examples:

More information

X X X E[ ] E X E X. is the ()m n where the ( i,)th. j element is the mean of the ( i,)th., then

X X X E[ ] E X E X. is the ()m n where the ( i,)th. j element is the mean of the ( i,)th., then Secto 5 Vectors of Radom Varables Whe workg wth several radom varables,,..., to arrage them vector form x, t s ofte coveet We ca the make use of matrx algebra to help us orgaze ad mapulate large umbers

More information

Ordinary Least Squares Regression. Simple Regression. Algebra and Assumptions.

Ordinary Least Squares Regression. Simple Regression. Algebra and Assumptions. Ordary Least Squares egresso. Smple egresso. Algebra ad Assumptos. I ths part of the course we are gog to study a techque for aalysg the lear relatoshp betwee two varables Y ad X. We have pars of observatos

More information

Functions of Random Variables

Functions of Random Variables Fuctos of Radom Varables Chapter Fve Fuctos of Radom Varables 5. Itroducto A geeral egeerg aalyss model s show Fg. 5.. The model output (respose) cotas the performaces of a system or product, such as weght,

More information

NP!= P. By Liu Ran. Table of Contents. The P vs. NP problem is a major unsolved problem in computer

NP!= P. By Liu Ran. Table of Contents. The P vs. NP problem is a major unsolved problem in computer NP!= P By Lu Ra Table of Cotets. Itroduce 2. Strategy 3. Prelmary theorem 4. Proof 5. Expla 6. Cocluso. Itroduce The P vs. NP problem s a major usolved problem computer scece. Iformally, t asks whether

More information

arxiv: v1 [math.st] 24 Oct 2016

arxiv: v1 [math.st] 24 Oct 2016 arxv:60.07554v [math.st] 24 Oct 206 Some Relatoshps ad Propertes of the Hypergeometrc Dstrbuto Peter H. Pesku, Departmet of Mathematcs ad Statstcs York Uversty, Toroto, Otaro M3J P3, Caada E-mal: pesku@pascal.math.yorku.ca

More information

Rademacher Complexity. Examples

Rademacher Complexity. Examples Algorthmc Foudatos of Learg Lecture 3 Rademacher Complexty. Examples Lecturer: Patrck Rebesch Verso: October 16th 018 3.1 Itroducto I the last lecture we troduced the oto of Rademacher complexty ad showed

More information

Strong Convergence of Weighted Averaged Approximants of Asymptotically Nonexpansive Mappings in Banach Spaces without Uniform Convexity

Strong Convergence of Weighted Averaged Approximants of Asymptotically Nonexpansive Mappings in Banach Spaces without Uniform Convexity BULLETIN of the MALAYSIAN MATHEMATICAL SCIENCES SOCIETY Bull. Malays. Math. Sc. Soc. () 7 (004), 5 35 Strog Covergece of Weghted Averaged Appromats of Asymptotcally Noepasve Mappgs Baach Spaces wthout

More information

Research Article Multidimensional Hilbert-Type Inequalities with a Homogeneous Kernel

Research Article Multidimensional Hilbert-Type Inequalities with a Homogeneous Kernel Hdaw Publshg Corporato Joural of Iequaltes ad Applcatos Volume 29, Artcle ID 3958, 2 pages do:.55/29/3958 Research Artcle Multdmesoal Hlbert-Type Iequaltes wth a Homogeeous Kerel Predrag Vuovć Faculty

More information

Some identities involving the partial sum of q-binomial coefficients

Some identities involving the partial sum of q-binomial coefficients Some dettes volvg the partal sum of -bomal coeffcets Bg He Departmet of Mathematcs, Shagha Key Laboratory of PMMP East Cha Normal Uversty 500 Dogchua Road, Shagha 20024, People s Republc of Cha yuhe00@foxmal.com

More information

Lecture 4 Sep 9, 2015

Lecture 4 Sep 9, 2015 CS 388R: Radomzed Algorthms Fall 205 Prof. Erc Prce Lecture 4 Sep 9, 205 Scrbe: Xagru Huag & Chad Voegele Overvew I prevous lectures, we troduced some basc probablty, the Cheroff boud, the coupo collector

More information

CIS 800/002 The Algorithmic Foundations of Data Privacy October 13, Lecture 9. Database Update Algorithms: Multiplicative Weights

CIS 800/002 The Algorithmic Foundations of Data Privacy October 13, Lecture 9. Database Update Algorithms: Multiplicative Weights CIS 800/002 The Algorthmc Foudatos of Data Prvacy October 13, 2011 Lecturer: Aaro Roth Lecture 9 Scrbe: Aaro Roth Database Update Algorthms: Multplcatve Weghts We ll recall aga) some deftos from last tme:

More information

Bounds on the expected entropy and KL-divergence of sampled multinomial distributions. Brandon C. Roy

Bounds on the expected entropy and KL-divergence of sampled multinomial distributions. Brandon C. Roy Bouds o the expected etropy ad KL-dvergece of sampled multomal dstrbutos Brado C. Roy bcroy@meda.mt.edu Orgal: May 18, 2011 Revsed: Jue 6, 2011 Abstract Iformato theoretc quattes calculated from a sampled

More information

A Study on Generalized Generalized Quasi hyperbolic Kac Moody algebra QHGGH of rank 10

A Study on Generalized Generalized Quasi hyperbolic Kac Moody algebra QHGGH of rank 10 Global Joural of Mathematcal Sceces: Theory ad Practcal. ISSN 974-3 Volume 9, Number 3 (7), pp. 43-4 Iteratoal Research Publcato House http://www.rphouse.com A Study o Geeralzed Geeralzed Quas (9) hyperbolc

More information

Mu Sequences/Series Solutions National Convention 2014

Mu Sequences/Series Solutions National Convention 2014 Mu Sequeces/Seres Solutos Natoal Coveto 04 C 6 E A 6C A 6 B B 7 A D 7 D C 7 A B 8 A B 8 A C 8 E 4 B 9 B 4 E 9 B 4 C 9 E C 0 A A 0 D B 0 C C Usg basc propertes of arthmetc sequeces, we fd a ad bm m We eed

More information

Some Notes on the Probability Space of Statistical Surveys

Some Notes on the Probability Space of Statistical Surveys Metodološk zvezk, Vol. 7, No., 200, 7-2 ome Notes o the Probablty pace of tatstcal urveys George Petrakos Abstract Ths paper troduces a formal presetato of samplg process usg prcples ad cocepts from Probablty

More information

MEASURES OF DISPERSION

MEASURES OF DISPERSION MEASURES OF DISPERSION Measure of Cetral Tedecy: Measures of Cetral Tedecy ad Dsperso ) Mathematcal Average: a) Arthmetc mea (A.M.) b) Geometrc mea (G.M.) c) Harmoc mea (H.M.) ) Averages of Posto: a) Meda

More information

Entropies & Information Theory

Entropies & Information Theory Etropes & Iformato Theory LECTURE II Nlajaa Datta Uversty of Cambrdge,U.K. See lecture otes o: http://www.q.damtp.cam.ac.uk/ode/223 quatum system States (of a physcal system): Hlbert space (fte-dmesoal)

More information

Probabilistic Meanings of Numerical Characteristics for Single Birth Processes

Probabilistic Meanings of Numerical Characteristics for Single Birth Processes A^VÇÚO 32 ò 5 Ï 206 c 0 Chese Joural of Appled Probablty ad Statstcs Oct 206 Vol 32 No 5 pp 452-462 do: 03969/jss00-426820605002 Probablstc Meags of Numercal Characterstcs for Sgle Brth Processes LIAO

More information

Solving Constrained Flow-Shop Scheduling. Problems with Three Machines

Solving Constrained Flow-Shop Scheduling. Problems with Three Machines It J Cotemp Math Sceces, Vol 5, 2010, o 19, 921-929 Solvg Costraed Flow-Shop Schedulg Problems wth Three Maches P Pada ad P Rajedra Departmet of Mathematcs, School of Advaced Sceces, VIT Uversty, Vellore-632

More information

v 1 -periodic 2-exponents of SU(2 e ) and SU(2 e + 1)

v 1 -periodic 2-exponents of SU(2 e ) and SU(2 e + 1) Joural of Pure ad Appled Algebra 216 (2012) 1268 1272 Cotets lsts avalable at ScVerse SceceDrect Joural of Pure ad Appled Algebra joural homepage: www.elsever.com/locate/jpaa v 1 -perodc 2-expoets of SU(2

More information

Arithmetic Mean and Geometric Mean

Arithmetic Mean and Geometric Mean Acta Mathematca Ntresa Vol, No, p 43 48 ISSN 453-6083 Arthmetc Mea ad Geometrc Mea Mare Varga a * Peter Mchalča b a Departmet of Mathematcs, Faculty of Natural Sceces, Costate the Phlosopher Uversty Ntra,

More information

Median as a Weighted Arithmetic Mean of All Sample Observations

Median as a Weighted Arithmetic Mean of All Sample Observations Meda as a Weghted Arthmetc Mea of All Sample Observatos SK Mshra Dept. of Ecoomcs NEHU, Shllog (Ida). Itroducto: Iumerably may textbooks Statstcs explctly meto that oe of the weakesses (or propertes) of

More information

A New Family of Transformations for Lifetime Data

A New Family of Transformations for Lifetime Data Proceedgs of the World Cogress o Egeerg 4 Vol I, WCE 4, July - 4, 4, Lodo, U.K. A New Famly of Trasformatos for Lfetme Data Lakhaa Watthaacheewakul Abstract A famly of trasformatos s the oe of several

More information

Generalized Convex Functions on Fractal Sets and Two Related Inequalities

Generalized Convex Functions on Fractal Sets and Two Related Inequalities Geeralzed Covex Fuctos o Fractal Sets ad Two Related Iequaltes Huxa Mo, X Su ad Dogya Yu 3,,3School of Scece, Bejg Uversty of Posts ad Telecommucatos, Bejg,00876, Cha, Correspodece should be addressed

More information

Can we take the Mysticism Out of the Pearson Coefficient of Linear Correlation?

Can we take the Mysticism Out of the Pearson Coefficient of Linear Correlation? Ca we tae the Mstcsm Out of the Pearso Coeffcet of Lear Correlato? Itroducto As the ttle of ths tutoral dcates, our purpose s to egeder a clear uderstadg of the Pearso coeffcet of lear correlato studets

More information

Assignment 5/MATH 247/Winter Due: Friday, February 19 in class (!) (answers will be posted right after class)

Assignment 5/MATH 247/Winter Due: Friday, February 19 in class (!) (answers will be posted right after class) Assgmet 5/MATH 7/Wter 00 Due: Frday, February 9 class (!) (aswers wll be posted rght after class) As usual, there are peces of text, before the questos [], [], themselves. Recall: For the quadratc form

More information

Exercises for Square-Congruence Modulo n ver 11

Exercises for Square-Congruence Modulo n ver 11 Exercses for Square-Cogruece Modulo ver Let ad ab,.. Mark True or False. a. 3S 30 b. 3S 90 c. 3S 3 d. 3S 4 e. 4S f. 5S g. 0S 55 h. 8S 57. 9S 58 j. S 76 k. 6S 304 l. 47S 5347. Fd the equvalece classes duced

More information

Simple Linear Regression

Simple Linear Regression Statstcal Methods I (EST 75) Page 139 Smple Lear Regresso Smple regresso applcatos are used to ft a model descrbg a lear relatoshp betwee two varables. The aspects of least squares regresso ad correlato

More information

Econometric Methods. Review of Estimation

Econometric Methods. Review of Estimation Ecoometrc Methods Revew of Estmato Estmatg the populato mea Radom samplg Pot ad terval estmators Lear estmators Ubased estmators Lear Ubased Estmators (LUEs) Effcecy (mmum varace) ad Best Lear Ubased Estmators

More information

Marcinkiewicz strong laws for linear statistics of ρ -mixing sequences of random variables

Marcinkiewicz strong laws for linear statistics of ρ -mixing sequences of random variables Aas da Academa Braslera de Cêcas 2006 784: 65-62 Aals of the Brazla Academy of Sceces ISSN 000-3765 www.scelo.br/aabc Marckewcz strog laws for lear statstcs of ρ -mxg sequeces of radom varables GUANG-HUI

More information

Chain Rules for Entropy

Chain Rules for Entropy Cha Rules for Etroy The etroy of a collecto of radom varables s the sum of codtoal etroes. Theorem: Let be radom varables havg the mass robablty x x.x. The...... The roof s obtaed by reeatg the alcato

More information

Lecture 8: Linear Regression

Lecture 8: Linear Regression Lecture 8: Lear egresso May 4, GENOME 56, Sprg Goals Develop basc cocepts of lear regresso from a probablstc framework Estmatg parameters ad hypothess testg wth lear models Lear regresso Su I Lee, CSE

More information

CHAPTER VI Statistical Analysis of Experimental Data

CHAPTER VI Statistical Analysis of Experimental Data Chapter VI Statstcal Aalyss of Expermetal Data CHAPTER VI Statstcal Aalyss of Expermetal Data Measuremets do ot lead to a uque value. Ths s a result of the multtude of errors (maly radom errors) that ca

More information

UNIVERSITY OF OSLO DEPARTMENT OF ECONOMICS

UNIVERSITY OF OSLO DEPARTMENT OF ECONOMICS UNIVERSITY OF OSLO DEPARTMENT OF ECONOMICS Exam: ECON430 Statstcs Date of exam: Frday, December 8, 07 Grades are gve: Jauary 4, 08 Tme for exam: 0900 am 00 oo The problem set covers 5 pages Resources allowed:

More information

n -dimensional vectors follow naturally from the one

n -dimensional vectors follow naturally from the one B. Vectors ad sets B. Vectors Ecoomsts study ecoomc pheomea by buldg hghly stylzed models. Uderstadg ad makg use of almost all such models requres a hgh comfort level wth some key mathematcal sklls. I

More information

Chapter 14 Logistic Regression Models

Chapter 14 Logistic Regression Models Chapter 4 Logstc Regresso Models I the lear regresso model X β + ε, there are two types of varables explaatory varables X, X,, X k ad study varable y These varables ca be measured o a cotuous scale as

More information

Beam Warming Second-Order Upwind Method

Beam Warming Second-Order Upwind Method Beam Warmg Secod-Order Upwd Method Petr Valeta Jauary 6, 015 Ths documet s a part of the assessmet work for the subject 1DRP Dfferetal Equatos o Computer lectured o FNSPE CTU Prague. Abstract Ths documet

More information

Analysis of Variance with Weibull Data

Analysis of Variance with Weibull Data Aalyss of Varace wth Webull Data Lahaa Watthaacheewaul Abstract I statstcal data aalyss by aalyss of varace, the usual basc assumptos are that the model s addtve ad the errors are radomly, depedetly, ad

More information

arxiv:math/ v2 [math.gr] 26 Feb 2001

arxiv:math/ v2 [math.gr] 26 Feb 2001 arxv:math/0101070v2 [math.gr] 26 Feb 2001 O drft ad etropy growth for radom walks o groups Aa Erschler (Dyuba) e-mal: aad@math.tau.ac.l, erschler@pdm.ras.ru 1 Itroducto prelmary verso We cosder symmetrc

More information

ECONOMETRIC THEORY. MODULE VIII Lecture - 26 Heteroskedasticity

ECONOMETRIC THEORY. MODULE VIII Lecture - 26 Heteroskedasticity ECONOMETRIC THEORY MODULE VIII Lecture - 6 Heteroskedastcty Dr. Shalabh Departmet of Mathematcs ad Statstcs Ida Isttute of Techology Kapur . Breusch Paga test Ths test ca be appled whe the replcated data

More information

Extend the Borel-Cantelli Lemma to Sequences of. Non-Independent Random Variables

Extend the Borel-Cantelli Lemma to Sequences of. Non-Independent Random Variables ppled Mathematcal Sceces, Vol 4, 00, o 3, 637-64 xted the Borel-Catell Lemma to Sequeces of No-Idepedet Radom Varables olah Der Departmet of Statstc, Scece ad Research Campus zad Uversty of Tehra-Ira der53@gmalcom

More information

Large and Moderate Deviation Principles for Kernel Distribution Estimator

Large and Moderate Deviation Principles for Kernel Distribution Estimator Iteratoal Mathematcal Forum, Vol. 9, 2014, o. 18, 871-890 HIKARI Ltd, www.m-hkar.com http://dx.do.org/10.12988/mf.2014.4488 Large ad Moderate Devato Prcples for Kerel Dstrbuto Estmator Yousr Slaou Uversté

More information

1 Onto functions and bijections Applications to Counting

1 Onto functions and bijections Applications to Counting 1 Oto fuctos ad bectos Applcatos to Coutg Now we move o to a ew topc. Defto 1.1 (Surecto. A fucto f : A B s sad to be surectve or oto f for each b B there s some a A so that f(a B. What are examples of

More information

Lecture 3. Sampling, sampling distributions, and parameter estimation

Lecture 3. Sampling, sampling distributions, and parameter estimation Lecture 3 Samplg, samplg dstrbutos, ad parameter estmato Samplg Defto Populato s defed as the collecto of all the possble observatos of terest. The collecto of observatos we take from the populato s called

More information

Bounds for the Connective Eccentric Index

Bounds for the Connective Eccentric Index It. J. Cotemp. Math. Sceces, Vol. 7, 0, o. 44, 6-66 Bouds for the Coectve Eccetrc Idex Nlaja De Departmet of Basc Scece, Humates ad Socal Scece (Mathematcs Calcutta Isttute of Egeerg ad Maagemet Kolkata,

More information

Multivariate Transformation of Variables and Maximum Likelihood Estimation

Multivariate Transformation of Variables and Maximum Likelihood Estimation Marquette Uversty Multvarate Trasformato of Varables ad Maxmum Lkelhood Estmato Dael B. Rowe, Ph.D. Assocate Professor Departmet of Mathematcs, Statstcs, ad Computer Scece Copyrght 03 by Marquette Uversty

More information

About k-perfect numbers

About k-perfect numbers DOI: 0.47/auom-04-0005 A. Şt. Uv. Ovdus Costaţa Vol.,04, 45 50 About k-perfect umbers Mhály Becze Abstract ABSTRACT. I ths paper we preset some results about k-perfect umbers, ad geeralze two equaltes

More information

A Remark on the Uniform Convergence of Some Sequences of Functions

A Remark on the Uniform Convergence of Some Sequences of Functions Advaces Pure Mathematcs 05 5 57-533 Publshed Ole July 05 ScRes. http://www.scrp.org/joural/apm http://dx.do.org/0.436/apm.05.59048 A Remark o the Uform Covergece of Some Sequeces of Fuctos Guy Degla Isttut

More information

Lecture Note to Rice Chapter 8

Lecture Note to Rice Chapter 8 ECON 430 HG revsed Nov 06 Lecture Note to Rce Chapter 8 Radom matrces Let Y, =,,, m, =,,, be radom varables (r.v. s). The matrx Y Y Y Y Y Y Y Y Y Y = m m m s called a radom matrx ( wth a ot m-dmesoal dstrbuto,

More information

Bayes (Naïve or not) Classifiers: Generative Approach

Bayes (Naïve or not) Classifiers: Generative Approach Logstc regresso Bayes (Naïve or ot) Classfers: Geeratve Approach What do we mea by Geeratve approach: Lear p(y), p(x y) ad the apply bayes rule to compute p(y x) for makg predctos Ths s essetally makg

More information

Lebesgue Measure of Generalized Cantor Set

Lebesgue Measure of Generalized Cantor Set Aals of Pure ad Appled Mathematcs Vol., No.,, -8 ISSN: -8X P), -888ole) Publshed o 8 May www.researchmathsc.org Aals of Lebesgue Measure of Geeralzed ator Set Md. Jahurul Islam ad Md. Shahdul Islam Departmet

More information

Discrete Mathematics and Probability Theory Fall 2016 Seshia and Walrand DIS 10b

Discrete Mathematics and Probability Theory Fall 2016 Seshia and Walrand DIS 10b CS 70 Dscrete Mathematcs ad Probablty Theory Fall 206 Sesha ad Walrad DIS 0b. Wll I Get My Package? Seaky delvery guy of some compay s out delverg packages to customers. Not oly does he had a radom package

More information

Chapter 8. Inferences about More Than Two Population Central Values

Chapter 8. Inferences about More Than Two Population Central Values Chapter 8. Ifereces about More Tha Two Populato Cetral Values Case tudy: Effect of Tmg of the Treatmet of Port-We tas wth Lasers ) To vestgate whether treatmet at a youg age would yeld better results tha

More information