Aldous, David (1989) Probability Approximations via the Poisson Clumping Heuristic. Springer Verlag, New York, Berlin, Heidelberg.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....on the up or down probability, be enough for having a success probability of more than 60 . 4. 5 Aldous Poisson Clumping Heuristic If n becomes large compared to l then the local dependence created by the pattern (with memory l 1) becomes negligible and hence Poisson clumping heuristic (Aldous [2]) becomes the prime candidate for approximation. For coumpound Poisson approximation see also the recent user s guide by Barbour and Chryssaphinou [5] Here is what Poisson Clumping Heuristic can do. We follow the notation of Aldous (see subsection B5) For xed H = H 1 H 2 ; H l let SH ....

....SH be the set of random times k such that X k (l 1) X k (l 2) X k 1 X k = H, and let CH be the number of times pattern H appears in a position overlapping this pattern H including H itself. Knowing P (H 1 ) P (H 2 ) P (H l ) we can directly compute E (CH ) and use what Aldous ([2]) calls the fundamental identity to get the H clump rate, that is H = P (H) E (CH ) 49) This yields the approximation P (T 1 l j) e j H ; j = 0; 1; 2; and hence, by the memoryless property, the process T 1 ; T 2 ; is approximately a homogeneous Poisson process with rate ....

D. Aldous. Probability Approximations via the Poisson Clumping Heuristic. Springer-Verlag, 1989.

....time scales, looking for how similarly the character arrivals cluster at each time scale. Our analysis follows two specific branches, depending on the choice of #. Analysis by Multiscale Block Averages. Here we choose # to be a very simple bump actually the boxcar function #(t) 1 [0,1] depicted in Figure 5, panel (b) As indicated above, we choose p =1,and it then turns out that the coe#cients amount to simple averages of the data over blocks of various lengths and locations. Accordingly, we call this choice of # as corresponding to the analysis of multiscale block averages. ....

....2 (t) #max supp(#) In words, the di#erence between N 1 and N 2 is controlled by the volume in N 1 .Wenow use results about extremes of Poisson processes. If N 1 is the set of cumulative arrivals of a Poisson counting process, then E N 1 (t #) N 1 (t) For more details see [2] and [1]. Based on these two analytical tools, we can easily obtain the results in the previous section: Calculation for Multiscale Block Averages. This is based on the following ingredients. First, symbols emerge at Poisson arrival times t 1 , t N , with rate #. Second, the bump has mean 1 ....

David J. Aldous. Probability Approximations Via the Poisson Clumping Heuristic. SpringerVerlag: New York. January 1989 14

....888# 313 where r = m # 1. Note that this expansion can also be obtained from (3) but with more involved computations. 2.3 Poissonization Poissonization is a widely used technique in stochastic process, summability of divergent sequence, analysis of algorithms, etc. see, for example, [1, 6, 18, 35, 19]. The idea is roughly described as follows. Given a discrete probability distribution k#0 (or, in general, a complex sequence) consider the Poisson generating function: b(#) e # a j C) The usual Poisson heuristic reads: If the sequence k#0 is smooth enough, then a n ....

D. Aldous, Probability approximations via the Poisson clumping heuristic, Springer-Verlag, New York, 1989.

....number of fixed points (cycles of size 1) in a random permutation 2, n . In this case, we have ] z (w 1)z 0#j#n , and thus the approximation rate is super exponential: Many combinatorial examples of Poisson limiting law can be found in Aldous [1], Barbour et al. 12] and Soria [81, p. 135] 6 Conclusions We have demonstrated, through many discrete examples, that precise asymptotic relations can be derived by a direct analytic approach for Poisson approximation distances when analytic generating functions are available. It is applicable ....

Aldous, D. (1989) Probability approximations via the Poisson clumping heuristic. SpringerVerlag, New York.

....a k a bn=2c : This heuristic is indeed a Poisson heuristic, which states that if a n is smooth enough, then a n A(n) A(z) e a k k (For, e Gammaz P b k z =k = A(z=2) thus OE n A(bn=2c) a bn=2c . See Flajolet [9] for more information on Bernoulli sums and Aldous [1] on Poisson heuristics. Polynomial asymptotic transfers. Proposition 2. Assume that OE n satisfies (4) If n = cn ) 15) if n = O(n ) or n = o(n ) then OE n = O(n ) or o(n ) respectively. Proof. A useful heuristic is that if we assume that OE n c 0 n , then by ....

D. Aldous, Probability Approximations via the Poisson Clumping Heuristic, Springer-Verlag, New York (1989).

....888 313 where r = m= 1. Note that this expansion can also be obtained from (3) but with more involved computations. 2.3 Poissonization Poissonization is a widely used technique in stochastic process, summability of divergent sequence, analysis of algorithms, etc. see, for example, [1, 6, 18, 35, 19]. The idea is roughly described as follows. Given a discrete probability distribution fa k g k0 (or, in general, a complex sequence) consider the Poisson generating function: b( e a j ( 2 C) The usual Poisson heuristic reads: If the sequence fa k g k0 is smooth enough, then a ....

D. Aldous, Probability approximations via the Poisson clumping heuristic, Springer-Verlag, New York, 1989.

.... attacks in each summer, the number of students in a class with the same birthday, the number of times of winning the jackpot for the lottery, the number of typos per page made by a secretary, the number of phone calls received by a telephone operator, the number of flaws in a bolt of fabric; see [1, 13, 17, 30] for more information. To probablists, the so called misfortunes never come single may also have a natural connection to Poisson law. The wide spread use of the Poisson distribution lies partly in its simple definition: P (X = k) e Gamma k (k = 0; 1; where 0. Among the ....

....exists and is non normal; see [8] The phase change of Y n at t = 58 is just the tip of an iceberg. We can systematically produce phase changes at other values; see [8, 9] For example, consider the random variables Z n defined by Z 0 = 0, Z n = 1 for 1 n m Gamma 1, where m 3, and Z n = Z [1] In (1) Delta Delta Delta Z [m] In (m) 1 (n m) where the Z [i] n s are independent, identical copies of Z n , and P (I n (1) i 1 ; I n (m) i m ) Gamma1 for all nonnegative tuples (i 1 ; i m ) satisfying i 1 Delta Delta Delta i m = n Gamma m ....

D. Aldous, Probability Approximations via the Poisson Clumping Heuristic, Springer-Verlag, New York, 1989

....sequence a n , b n : 2 n . This heuristic is indeed a Poisson heuristic, which states that if a n is smooth enough, then A(n) A(z) e z k . For, e z b k z k = A(z 2) thus # n A(#n 2#) See Flajolet [9] for more information on Bernoulli sums and Aldous [1] on Poisson heuristics. Polynomial asymptotic transfers. Proposition 2. Assume that # n satisfies (4) If # n = cn ) 15) if # n = O(n ) or # n = o(n ) then # n = O(n ) or o(n ) respectively. Proof. A useful heuristic is that if we assume that # n , then by (1) ....

....g n (y) # n (y) g k (y) n with g 0 (y) g 1 (y) 1, where # n (y) E(e ) # n denoting the number of coin flips used to find a leader. Then (see Prodinger [32] # n (y) # k (y) # 0 (y)# n (y) n with # 0 (y) # 1 (y) 1. The Dickman function #(u) 0, #) ## [0, 1] is defined as the continuous solution to the di#erential di#erence equation u# # (u) #(u 1) 0 (u 1) with the initial condition #(u) 1 for 0 u 1. Since (see Tenenbaum [35] #(u) du = e where # denote the Euler constant, the function e # #(u) is a density function. We call ....

D. Aldous, Probability Approximations via the Poisson Clumping Heuristic, Springer-Verlag, New York (1989).

....number of fixed points (cycles of size 1) in a random permutation of f1; 2; ng. In this case, we have ] z (w Gamma1)z ; and thus the approximation rate is super exponential: Many combinatorial examples of Poisson limiting law can be found in Aldous [1], Barbour et al. 12] and Soria [81, p. 135] 6 Conclusions We have demonstrated, through many discrete examples, that precise asymptotic relations can be derived by a direct analytic approach for Poisson approximation distances when analytic generating functions are available. It is applicable ....

Aldous, D. (1989) Probability approximations via the Poisson clumping heuristic. SpringerVerlag, New York.

.... attacks in each summer, the number of students in a class with the same birthday, the number of times of winning the jackpot for the lottery, the number of typos per page made by a secretary, the number of phone calls received by a telephone operator, the number of flaws in a bolt of fabric; see [1, 13, 17, 30] for more information. To probablists, the so called misfortunes never come single may also have a natural connection to Poisson law. The wide spread use of the Poisson distribution lies partly in its simple definition: P (X = k) e # k (k = 0, 1, where # 0. Among the many ....

....on the e#ciency of the tools used since proving central limit theorems is usually more sophisticated than, say the zero one law. Such a notion will appear repeatedly later in this paper. Also if one is interested in more refined approximations, then more tools and e#orts are needed. 1. 1 Maxima in [0, 1] : from Poisson to constant Multidimensional data have no total ordering. A natural partial order is the following dominance relation: given two points A = a 1 , a d ) and B = b 1 , b d ) where d 1, we say that A dominates B if a i b i for all i = 1, d. The ....

[Article contains additional citation context not shown here]

D. Aldous, Probability Approximations via the Poisson Clumping Heuristic, Springer-Verlag, New York, 1989

....II II is the Lebesgue measure of a set and the random set Az = x : Z(x) z is called the excursion set above the threshold z. This approximation involves unknown EllA[I, which is the mean clump size of the excursion set. The distribution of [IA[I has been estimated for the case of Gaussian [3], X 2, t and F fields [16] but for general random fields, no approximation is available. An alternate approximation based on the expected Euler characteristic (EC) of A is also available. For very high threshold z, it can be shown that P(maxZ(x) z) E x(A) xE where x(Az) is the Euler ....

D. Aldous. Probability Approximations via the Poisson Clumping Heuristic. Springer- Verlag, New Yor, 1989.

....of just 3:4 : 4 14 trading days is enough for having a reasonable success probability. 4. 5 Poisson Clumping Heuristic If n becomes large compared to l then the local dependence created by the pattern (with memory l 1) becomes negligible and hence Poisson clumping heuristic (Aldous [2]) becomes the prime candidate for approximation. For coumpound Poisson approximation see also the recent user s guide by Barbour and Chryssaphinou [4] Here is what Poisson Clumping Heuristic can do. We follow the notation of Aldous (see subsection B5) For xed H = H 1 H 2 ; H l let SH ....

....SH be the set of random times k such that X k (l 1) X k (l 2) X k 1 X k = H, and let CH be the number of times pattern H appears in a position overlapping this pattern H including H itself. Knowing Pr(H 1 ) Pr(H 2 ) Pr(H l ) we can directly compute E (CH ) and use what Aldous ([2]) calls the fundamental identity to get the H clump rate, that is H = Pr(H) E (CH ) 47) This yields the approximation Pr(T 1 l j) e j H ; j = 0; 1; 2; and hence, by the memoryless property, the process T 1 ; T 2 ; is approximately a homogeneous Poisson process with rate ....

D. Aldous. Probability Approximations via the Poisson Clumping Heuristic. Springer-Verlag, 1989.

....T of the process x t from V thre x 0 is approximately given by(see Fig.1) #T # = # ## [H # (V thre ) # (H ## (x 0 ) # exp # 2[H(V thre ) H(x 0 ) # 2 # (3.1) and furthermore T # p(t) 1 #T # exp # t #T # # (3.2) where # 0 is a parameter. We refer the reader to [6] for a detailed proof of Eq. 3.1) and Eq. 3.2) Eq. 3.1) together with Eq. 3.2) tell us that T is exponentially distributed, i.e. the e#erent spike trains of the IF model are a Poisson process with a rate 1 #T #. The dependence of 1 #T # on the function H and model parameters V thre , # ....

Aldous, D. (1989) Probability Approximation Via The Poisson Clumping Heuristic Springer-Verlag: New York.

.... u X l k X u (i; l; u)h u = L(i; k) kC(i; k 1) X u (i; u)h u C(i; k 1) X l k X u (i; l; u) C(i; k 1) h u : 20) By (16) we observe that C(i; k 1) O( q ( k 2 ) p k , k 1, uniformly on i, and by standard properties of Markov chains (see Keilson [10] Aldous [1]) we know that the hitting time to k is such that (we write D i for D i (k) h i = D i (i) O( Actually a Laurent series exists for suciently small) Pr i [T k x] e x=h i ; x 1: 21) We will soon check that D is independent of i. First we note that L(i; k) kC(i; k 1) ....

D. Aldous. Probability Approximations via the Poisson Clumping Heuristic. Springer-Verlag, 1989.

....: so 2 = C 2 (2 C 4 ) With (38) we also derive (k) C 4 O( l) C 4 1 O(z l ) l 1; l 6= k: 22) could be similarly computed, but we will not need its explicit form. However, we will use the normalization R = 1, which gives C 2 1 = 0: 23) By Keilson [26] Aldous [1], Aldous and Brown, 2] 3] we know that the hitting time to a distant state is asymptotically exponential. But, here, we need a precise equivalent. For further use, set C 9 : 2 2 1 =2. Let us rst x n. We analyze, with k = log n) and starting with the stationary distribution, Pr(T k ....

Aldous, D. Probability Approximations via the Poisson Clumping Heuristics, Springer{Verlag, Heidelberg, 1989.

No context found.

Aldous, David (1989) Probability Approximations via the Poisson Clumping Heuristic. Springer Verlag, New York, Berlin, Heidelberg.

No context found.

D. Aldous. Probability Approximations via the Poisson Clumping Heuristic, volume 77 of Applied Mathematical Sciences. Springer Verlag, 1989

No context found.

Aldous, D., Probability Approximations via the Poisson Clumping Heuristic, Springer-Verlag, 1987.

No context found.

Aldous, D. Probability Approximations via the Poisson Clumping Heuristic. Applied Mathematical Sciences, 77. Springer-Verlag, New York-Berlin, 1989.

No context found.

Aldous, David 1989 Probability approximations via the Poisson clumping heuristic. Applied Mathematical Sciences, 77. Springer-Verlag, New YorkBerlin.

No context found.

D. Aldous, Probability Approximations via the Poisson Clumping Heuristic. New York: Springler-Verlag, 1989.

No context found.

D. Aldous. Probability Approximations via the Poisson Clumping Heuristic. Springer-Verlag, 1989.

No context found.

D. Aldous. Probability Approximations via the Poisson Clumping Heuristic. Springer-Verlag, 1989.

No context found.

D. Aldous. Probability Approximations via the Poisson Clumping Heuristic. Springer-Verlag, 1989.

No context found.

D. Aldous. Probability Approximations via the Poisson Clumping Heuristic. Springer-Verlag, New York, 1989.

First 50 documents  Next 50

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC