An example power-law graph that demonstrates ranking of popularity. To the right is the long tail, and to the left are the few that dominate (also known as the 80–20 rule).
In mathematics and science, a power law is a functional relationship between two quantities, where a relative change in one quantity results in a relative change in the other quantity proportional to the change raised to a constant exponent: one quantity varies as a power of another. The change is independent of the initial size of those quantities.
For instance, the area of a square has a power law relationship with the length of its side, since if the length is doubled, the area is multiplied by 22, while if the length is tripled, the area is multiplied by 32, and so on.[1]
Empirical examples
The distributions of a wide variety of physical, biological, and human-made phenomena approximately follow a power law over a wide range of magnitudes: these include the sizes of craters on the moon and of solar flares,[2] cloud sizes,[3] the foraging pattern of various species,[4] the sizes of activity patterns of neuronal populations,[5] the frequencies of words in most languages, frequencies of family names, the species richness in clades of organisms,[6] the sizes of power outages, volcanic eruptions,[7] human judgments of stimulus intensity[8][9] and many other quantities.[10] Empirical distributions can only fit a power law for a limited range of values, because a pure power law would allow for arbitrarily large or small values. Acoustic attenuation follows frequency power-laws within wide frequency bands for many complex media. Allometric scaling laws for relationships between biological variables are among the best known power-law functions in nature.
Properties
Statistical incompleteness
The power-law model does not obey the treasured paradigm of statistical completeness. Especially probability bounds, the suspected cause of typical bending and/or flattening phenomena in the high- and low-frequency graphical segments, are parametrically absent in the standard model.[11]
Scale invariance
One attribute of power laws is their scale invariance. Given a relation , scaling the argument by a constant factor causes only a proportionate scaling of the function itself. That is,
where denotes direct proportionality. That is, scaling by a constant simply multiplies the original power-law relation by the constant . Thus, it follows that all power laws with a particular scaling exponent are equivalent up to constant factors, since each is simply a scaled version of the others. This behavior is what produces the linear relationship when logarithms are taken of both and , and the straight-line on the log–log plot is often called the signature of a power law. With real data, such straightness is a necessary, but not sufficient, condition for the data following a power-law relation. In fact, there are many ways to generate finite amounts of data that mimic this signature behavior, but, in their asymptotic limit, are not true power laws. Thus, accurately fitting and validating power-law models is an active area of research in statistics; see below.
Lack of well-defined average value
A power-law has a well-defined mean over only if , and it has a finite variance only if ; most identified power laws in nature have exponents such that the mean is well-defined but the variance is not, implying they are capable of black swan behavior.[2] This is exemplified by the effect, on the mean income, of the inclusion of the person with the highest income in the world into a sample of individuals of otherwise similar income.[12] Income is distributed according to a power-law known as the Pareto distribution (for example, the net worth of Americans is distributed according to a power law with an exponent of 2).
数学的には、厳密なべき乗則は確率分布にはなり得ないが、切り捨てられたべき乗関数である分布は可能である。のために指数(ギリシャ文字のアルファ。スケーリング係数と混同しないように。)(上記で使用)が 1 より大きい場合(そうでない場合、尾部の面積は無限大になる)、最小値 is needed otherwise the distribution has infinite area as x approaches 0, and the constant C is a scaling factor to ensure that the total area is 1, as required by a probability distribution. More often one uses an asymptotic power law – one that is only true in the limit; see power-law probability distributions below for details. Typically the exponent falls in the range , though not always.[10]
Examples
More than a hundred power-law distributions have been identified in physics (e.g. sandpile avalanches), biology (e.g. species extinction and body mass), and the social sciences (e.g. city sizes and income).[17] Among them are:
Pareto Q–Q plots compare the quantiles of the log-transformed data to the corresponding quantiles of an exponential distribution with mean 1 (or to the quantiles of a standard Pareto distribution) by plotting the former versus the latter. If the resultant scatterplot suggests that the plotted points asymptotically converge to a straight line, then a power-law distribution should be suspected. A limitation of Pareto Q–Q plots is that they behave poorly when the tail index (also called Pareto index) is close to 0, because Pareto Q–Q plots are not designed to identify distributions with slowly varying tails.[58]
Mean residual life plots
On the other hand, in its version for identifying power-law probability distributions, the mean residual life plot consists of first log-transforming the data, and then plotting the average of those log-transformed data that are higher than the i-th order statistic versus the i-th order statistic, for i=1,...,n, where n is the size of the random sample. If the resultant scatterplot suggests that the plotted points tend to stabilize about a horizontal straight line, then a power-law distribution should be suspected. Since the mean residual life plot is very sensitive to outliers (it is not robust), it usually produces plots that are difficult to interpret; for this reason, such plots are usually called Hill horror plots.[59]
Log-log plots
A straight line on a log–log plot is necessary but insufficient evidence for power-laws, the slope of the straight line corresponds to the power law exponent.
Log–log plots are an alternative way of graphically examining the tail of a distribution using a random sample. Taking the logarithm of a power law of the form results in:[60]
which forms a straight line with slope 対数-対数スケールで。ただし、対数-対数プロットは必要ではあるが、べき乗則関係の十分な証拠にはならないため、注意が必要です。多くの非べき乗則分布は対数-対数プロット上で直線として現れるからです。[ 10 ] [ 61 ]この方法は、分布の特定の数値が発生する確率の推定量の対数を、その特定の数値の対数に対してプロットすることから成ります。通常、この推定量は、データセットでその数値が発生する割合です。プロットの点が x 軸の大きな数値に対して直線に収束する傾向がある場合、研究者は分布がべき乗則の裾を持っていると結論付けます。これらのタイプのプロットの適用例が発表されています。[ 62 ]これらのプロットの欠点は、信頼できる結果を得るためには、膨大な量のデータが必要であることです。さらに、離散的(またはグループ化された)データにのみ適しています。
どこ is the incomplete zeta function. The uncertainty in this estimate follows the same formula as for the continuous equation. However, the two equations for are not equivalent, and the continuous version should not be applied to discrete data, nor vice versa.
Further, both of these estimators require the choice of . For functions with a non-trivial function, choosing too small produces a significant bias in , while choosing it too large increases the uncertainty in , and reduces the statistical power of our model. In general, the best choice of depends strongly on the particular form of the lower tail, represented by above.
More about these methods, and the conditions under which they can be used, can be found in .[10] Further, this comprehensive review article provides usable code (Matlab, Python, R and C++) for estimation and testing routines for power-law distributions.
Kolmogorov–Smirnov estimation
Another method for the estimation of the power-law exponent, which does not assume independent and identically distributed (iid) data, uses the minimization of the Kolmogorov–Smirnov statistic, , between the cumulative distribution functions of the data and the power law:
with
where and denote the cdfs of the data and the power law with exponent , respectively. As this method does not assume iid data, it provides an alternative way to determine the power-law exponent for data sets in which the temporal correlation can not be ignored.[5]
Validating power laws
Although power-law relations are attractive for many theoretical reasons, demonstrating that data does indeed follow a power-law relation requires more than simply fitting a particular model to the data.[34] This is important for understanding the mechanism that gives rise to the distribution: superficially similar distributions may arise for significantly different reasons, and different models yield different predictions, such as extrapolation.
For example, log-normal distributions are often mistaken for power-law distributions:[72]. When you take the log of its probability density function, the log-normal distribution has terms that are constant, log, and log-squared. When the mean is small and variance is large, the constant in front of the log-squared term is very small. In that case, for most of the distribution, it will be linear on a log-log plot. It is only for extreme values that the log-squared term asserts itself and shows that it is not a power-law.
例えば、比例成長過程に関するジブラの法則は、対数正規分布を生成しますが、その対数-対数プロットは限られた範囲で直線に見えます。これを説明すると、対数正規密度関数の対数はlog( x ) の二次式であるため、対数-対数プロットでは「弓形」になりますが、二次項が線形項に比べて小さい場合、結果はほぼ直線に見えることがあります。対数正規分布の挙動は、二次項が支配的な場合にのみ確認でき、そのためにはかなり多くのデータが必要になる場合があります。したがって、わずかに下向きに「弓形」になっている対数-対数プロットは、べき乗則ではなく、対数正規分布を反映している可能性があります。
↑Cannavò, Flavio; Nunnari, Giuseppe (2016-03-01). "On a Possible Unified Scaling Law for Volcanic Eruption Durations". Scientific Reports. 6 22289. Bibcode:2016NatSR...622289C. doi:10.1038/srep22289. ISSN2045-2322. PMC4772095. PMID26926425.
↑Stevens, S. S. (1957). "On the psychophysical law". Psychological Review. 64 (3): 153–181. doi:10.1037/h0046162. PMID13441853.
↑Staddon, J. E. R. (1978). "Theory of behavioral power functions". Psychological Review. 85 (4): 305–320. doi:10.1037/0033-295x.85.4.305. hdl:10161/6003.
↑Andriani, P.; McKelvey, B. (2007). "Beyond Gaussian averages: redirecting international business and management research toward extreme events and power laws". Journal of International Business Studies. 38 (7): 1212–1230. doi:10.1057/palgrave.jibs.8400324. S2CID512642.
↑Lacquaniti, Francesco; Terzuolo, Carlo; Viviani, Paolo (1983). "The law relating the kinematic and figural aspects of drawing movements". Acta Psychologica. 54 (1–3): 115–130. doi:10.1016/0001-6918(83)90027-6. PMID6666647. S2CID5144040.
↑Albert, J. S.; Bart, H. J.; Reis, R. E. "Species richness & cladal diversity". In Albert & Reis (2011), pp.89–104.
↑Yu, Frank H.; Willson, Timothy; Frye, Stephen; Edwards, Aled; Bader, Gary D.; Isserlin, Ruth (2011-02-02). "The human genome and drug discovery after a decade. Roads (still) not taken". Nature. 470 (7333): 163–165. arXiv:1102.0448v2. Bibcode:2011Natur.470..163E. doi:10.1038/470163a. PMID21307913. S2CID4429387.
↑Saravia, Leonardo A.; Doyle, Santiago R.; Bond-Lamberty, Ben (2018-12-10). "Power laws and critical fragmentation in global forests". Scientific Reports. 8 (1): 17766. Bibcode:2018NatSR...817766S. doi:10.1038/s41598-018-36120-w. ISSN2045-2322. PMC6288094. PMID30532065.
↑Corral, A, Osso, A, Llebot, JE (2010). "Scaling of tropical cyclone dissipation". Nature Physics. 6 (9): 693–696. arXiv:0910.0054. Bibcode:2010NatPh...6..693C. doi:10.1038/nphys1725. S2CID67754747.
↑Lorenz RD (2009). "Power Law of Dust Devil Diameters on Earth and Mars". Icarus. 203 (2): 683–684. Bibcode:2009Icar..203..683L. doi:10.1016/j.icarus.2009.06.029.
↑Sutton, J. (1997), "Gibrat's Legacy", Journal of Economic Literature XXXV, 40–59.
↑Li, W. (November 1999). "Random texts exhibit Zipf's-law-like word frequency distribution". IEEE Transactions on Information Theory. 38 (6): 1842–1845. doi:10.1109/18.165464. ISSN0018-9448.
↑Curtis, Vickie (2018-04-20). Online Citizen Science and the Widening of Academia: Distributed Engagement with Research and Knowledge Production. Springer. ISBN978-3-319-77664-4.
↑Croteau, David; Hoynes, William (2013-11-06). Media/Society: Industries, Images, and Audiences. SAGE Publications. ISBN978-1-4833-2355-8.
↑Lewis Fry Richardson (1950). The Statistics of Deadly Quarrels.
↑Berreby, David (July 31, 2014). "Cloudy With a Chance of War". Nautilus Magazine. Retrieved October 22, 2020.
↑Martin, Charles H.; Mahoney, Michael W. (2018-10-02). "Implicit Self-Regularization in Deep Neural Networks: Evidence from Random Matrix Theory and Implications for Learning". arXiv:1810.01075 [cs.LG].
↑Moret, M.; Zebende, G. (2007). "Amino acid hydrophobicity and accessible surface area". Physical Review E. 75 (1 Pt 1). 011920. Bibcode:2007PhRvE..75a1920M. doi:10.1103/PhysRevE.75.011920. PMID17358197.
↑Mackay, D. M. (1963). "Psychophysics of perceived intensity:A theoretical basis for Fechner's and Stevens' laws". Science. 139 (3560): 1213–1216. Bibcode:1963Sci...139.1213M. doi:10.1126/science.139.3560.1213-a. S2CID122501807.
↑Staddon, J. E. R. (1978). "Theory of behavioral power functions"(PDF). Psychological Review. 85 (4): 305–320. doi:10.1037/0033-295x.85.4.305. hdl:10161/6003.
↑John T. Wixted; Shana K. Carpenter. "The Wickelgren Power Law and the Ebbinghaus Savings Function"(PDF). Psychological Science. Archived from the original(PDF) on April 8, 2016. Retrieved August 31, 2016.
↑Jóhannesson, Gudlaugur; Björnsson, Gunnlaugur; Gudmundsson, Einar H. (2006). "Afterglow Light Curves and Broken Power Laws: A Statistical Study". The Astrophysical Journal. 640 (1): L5. arXiv:astro-ph/0602219. Bibcode:2006ApJ...640L...5J. doi:10.1086/503294. S2CID16139116.
↑"Curved-power law". Archived from the original on 2016-02-08. Retrieved 2013-07-07.
↑N. H. Bingham, C. M. Goldie, and J. L. Teugels, Regular variation. Cambridge University Press, 1989
↑ Kendal, WS; Jørgensen, B (2011). "Taylorのべき乗則とゆらぎスケーリングは中心極限のような収束によって説明される". Phys. Rev. E . 83 (6) 066115. Bibcode : 2011PhRvE..83f6115K . doi : 10.1103/physreve.83.066115 . PMID 21797449 .
↑ Kendal, WS; Jørgensen, BR (2011). "Tweedie convergence: a mathematical basis for Taylor's power law, 1/ f noise and multifractality" (PDF) . Phys. Rev. E . 84 (6) 066120. Bibcode : 2011PhRvE..84f6120K . doi : 10.1103/physreve.84.066120 . PMID 22304168 .
↑ Petersen, Casper; Simonsen, Jakob Grue; Lioma, Christina (2016). "情報検索におけるべき乗則分布" . ACM Transactions on Information Systems . 34 (2): 1– 37 – Google Scholar 経由。
↑ Beirlant, J., Teugels, JL, Vynckier, P. (1996) Practical Analysis of Extreme Values , Leuven: Leuven University Press
↑ Coles, S. (2001)極値の統計的モデリング入門. Springer-Verlag, London.
1 2 3 4 Diaz, FJ (1999). "残差分位関数による裾の挙動の識別". Journal of Computational and Graphical Statistics . 8 (3): 493– 509. doi : 10.2307/1390871 . JSTOR 1390871 .
↑ Resnick, SI (1997). "Heavy Tail Modeling and Teletraffic Data" . The Annals of Statistics . 25 (5): 1805–1869 . doi : 10.1214/aos/1069362376 .
↑ Arnold, BC; Brockett, PL (1983). "いつβパーセンタイル残存寿命関数が分布を決定するのか?". Operations Research . 31 (2): 391– 396. doi : 10.1287/opre.31.2.391 .
↑ Joe, H.; Proschan, F. (1984). "パーセンタイル残余寿命関数". Operations Research . 32 (3): 668– 678. doi : 10.1287/opre.32.3.668 .
↑ Joe, H. (1985), "パーセンタイル残存寿命からの寿命分布の特性評価", Ann. Inst. Statist. Math. 37, Part A, 165–172.
↑ Csorgo, S.; Viharos, L. (1992). "パーセンタイル残存寿命の信頼区間" (PDF) . Journal of Statistical Planning and Inference . 30 (3): 327– 337. doi : 10.1016/0378-3758(92)90159-p . hdl : 2027.42/30190 .
↑ Schmittlein, DC; Morrison, DG (1981). "中央値残存寿命:特性定理と応用". Operations Research . 29 (2): 392– 399. doi : 10.1287/opre.29.2.392 .
↑ Morrison, DG; Schmittlein, DC (1980). "Jobs, strikes, and wars: Probability models for duration". Organizational Behavior and Human Performance . 25 (2): 224– 251. doi : 10.1016/0030-5073(80)90065-3 .
↑ Gerchak, Y (1984). "社会科学における失敗率の低下と関連する問題". Operations Research . 32 (3): 537– 546. doi : 10.1287/opre.32.3.537 .
↑ Bauke, H. (2007). "最大尤度法によるべき乗則分布のパラメータ推定". European Physical Journal B . 58 (2): 167– 173. arXiv : 0704.1867 . Bibcode : 2007EPJB...58..167B . doi : 10.1140/epjb/e2007-00219-y . S2CID 119602829 .