12.07.2015 Views

1 Studies in the History of Statistics and Probability ... - Sheynin, Oscar

1 Studies in the History of Statistics and Probability ... - Sheynin, Oscar

1 Studies in the History of Statistics and Probability ... - Sheynin, Oscar

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

dem<strong>and</strong> to apply transf<strong>in</strong>ite numbers but was ra<strong>the</strong>r reduced to usualma<strong>the</strong>matical <strong>in</strong>duction.In <strong>the</strong> applied <strong>the</strong>ory <strong>of</strong> probability <strong>the</strong> harmless imp turns out asharp horned devil who favours to corrupt meanly statisticalhomogeneity. So far as we keep to <strong>the</strong> concept <strong>of</strong> ensemble <strong>and</strong> checkthat homogeneity by available methods, we are able at least to reveal<strong>in</strong> time <strong>the</strong> devilish dirty trick whereas, ab<strong>and</strong>on<strong>in</strong>g it, we whollysurrender ourselves to <strong>the</strong> devil’s rule <strong>and</strong> ought to be prepared forsurprises. Thus, from <strong>the</strong> po<strong>in</strong>t <strong>of</strong> view <strong>of</strong> modern probability <strong>the</strong>ory,<strong>the</strong> boundary between science <strong>and</strong> magic is def<strong>in</strong>ed by <strong>the</strong> notion <strong>of</strong>statistical ensemble. It follows that <strong>in</strong>ferences, derived by apply<strong>in</strong>gthat <strong>the</strong>ory when a statistical ensemble <strong>of</strong> experiments is lack<strong>in</strong>g, hasno scientific certa<strong>in</strong>ty.Unlike <strong>the</strong> arsenal <strong>of</strong> magic, <strong>the</strong> tools <strong>of</strong> science must be entirelyjustified. However, when conclud<strong>in</strong>g that, for example, <strong>the</strong> error <strong>of</strong> aresult obta<strong>in</strong>ed from a s<strong>in</strong>gle realization <strong>of</strong> a stochastic process issituated <strong>in</strong> <strong>the</strong> given <strong>in</strong>terval with probability 0.95, we do not know towhat does that probability correspond, − to an ensemble <strong>of</strong> realizationswhich we ought to conjuncture by issu<strong>in</strong>g from <strong>the</strong> s<strong>in</strong>gle observedrealization so as to apply <strong>the</strong> notion <strong>of</strong> stochastic process?But all those o<strong>the</strong>r realizations are irrelevant <strong>and</strong> it is very easy toprovide examples <strong>of</strong> faulty <strong>in</strong>ferences made when apply<strong>in</strong>g <strong>the</strong> <strong>the</strong>ory<strong>of</strong> probability <strong>in</strong> manufactur<strong>in</strong>g, geology, etc where it is senseless todiscuss statistical ensembles. Historically, science emerged frommagic but treats it disda<strong>in</strong>fully <strong>and</strong> would wish to ignore it. However,we should not wholly yield to that temptation ei<strong>the</strong>r.A representative <strong>of</strong> <strong>the</strong> constructive direction <strong>in</strong> ma<strong>the</strong>maticsconsiders <strong>the</strong> usual ma<strong>the</strong>matical analysis a magic. We should ra<strong>the</strong>rdist<strong>in</strong>guish between white <strong>and</strong> black magic <strong>the</strong> latter connected withbe<strong>in</strong>g subjectively unconscionable. At present, we can not ignorehonest attempts to apply probability <strong>the</strong>ory when statistical ensemblesare lack<strong>in</strong>g. I venture to forecast that someth<strong>in</strong>g be<strong>in</strong>g magic todaywill become science tomorrow. It would have been unreasonable tokeep too strongly to <strong>the</strong> established concept <strong>of</strong> statistical homogeneity.However, here I will entirely hold on to that concept s<strong>in</strong>ce nowadaysany o<strong>the</strong>r method <strong>of</strong> obta<strong>in</strong><strong>in</strong>g really plausible results is lack<strong>in</strong>g.1.4. Summary. Thus, while perfection <strong>of</strong> experiment<strong>in</strong>g is go<strong>in</strong>g on<strong>in</strong> one or ano<strong>the</strong>r branch <strong>of</strong> science or technology, a special situation<strong>of</strong>ten arises when statistical stability is present but complete stability<strong>of</strong> <strong>the</strong> results is impossible to achieve. The former is characterized bystability <strong>of</strong> <strong>the</strong> frequencies <strong>of</strong> <strong>the</strong> occurrences <strong>of</strong> <strong>the</strong> various eventsconnected with <strong>the</strong> experiment’s outcome.An exhaust<strong>in</strong>g check <strong>of</strong> such stability (statistical homogeneity,statistical ensemble) is impossible, but <strong>in</strong> many cases <strong>the</strong> presence <strong>of</strong> astatistical ensemble is sufficiently certa<strong>in</strong>. Accord<strong>in</strong>g to modern ideas,<strong>the</strong>se cases <strong>in</strong>deed comprise <strong>the</strong> field <strong>of</strong> scientific applications <strong>of</strong> <strong>the</strong>probability <strong>the</strong>ory.And still <strong>the</strong>re exists a readily understood wish to apply it also <strong>in</strong>o<strong>the</strong>r cases <strong>in</strong> which <strong>the</strong> results <strong>of</strong> <strong>the</strong> experiments are not def<strong>in</strong>ite, but<strong>the</strong> existence <strong>of</strong> a statistically homogeneous ensemble is impossible.For <strong>the</strong> time be<strong>in</strong>g, such applications belong to magic ra<strong>the</strong>r than11

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!