First work with the filtration , which records the Brownian motion and whether default has occurred. On all subsequent defaultable stock prices vanish. On , the memoryless property of the exponential distribution gives survival from to with conditional probability . The residual survival event is independent of the future independent increments of the Brownian motion. Thus, for deterministic ,
Also , so this is a true martingale, with no localization needed. Its natural filtration is contained in ; applying the tower property of conditional expectation to the displayed identity yields the requested natural-filtration martingale property:
The printed convention keeps the stock alive at . It is a left-continuous convention at that one random time. Replacing it by gives the usual càdlàg zero-recovery default model; since at each fixed , the fixed-time martingale calculations above are unchanged.
Write . Conditional on , independent increments of the Brownian motion give
where has the standard normal distribution and is independent of . Factoring out of the positive part identifies the normalized Black-Scholes call function:
The European call option payoff is integrable, because and . The tower property of conditional expectation therefore proves that is a true martingale. At , the boundary value gives . The answer is the conditional payoff martingale
The natural filtration of is the natural Brownian filtration: since , the equation recovers the entire Brownian motion history from the stock history. For , independent increments and the moment-generating function of a normal distribution give
All these expectations are finite. Since , the martingale identity holds for every exactly when the last exponential equals one. The required logarithmic drift is
This is the distinction between the drift of and the drift of : here the latter is zero.
In natural logarithms, Fano's inequality for a uniform index among alternatives states . The mutual information is at most the average Kullback-Leibler divergence to any fixed reference law. With the supplied packing, nearest-neighbour decoding turns partition error below half the separation into correct decoding. Thus the packing alone proves a bound at radius , with the unavoidable term. It does not automatically prove the much larger printed radius.
A list-decoding Fano inequality supplies the intended near-half error threshold. Assume , since otherwise that threshold is nonpositive and its event is automatic. Take a uniform prior on all unlabelled balanced partitions, and let . A Hamming ball of radius less than around any estimated partition contains at most such partitions. This follows by selecting the representative closer to the estimated group; each unlabelled partition contributes at most one such representative because . The entropy bound for a Hamming ball gives , where is the binary entropy function measured in natural logarithms. Its curvature gives . Also , so
Use the all- independent-edge law as a reference. Only the crossing pairs differ; the preceding Bernoulli distribution divergence estimate gives . Hence . Whenever the denominator is positive, the list-decoding Fano inequality proves the finite-sample answer
For , this is at least . If in addition , it implies the intended form with . This proves the claimed information-theoretic scale with its needed finite-sample qualifications.
As a uniform finite-sample assertion for every positive , the printed form is false. At every partition has the same data law, and a uniformly random balanced guess has a positive chance of being correct, giving error strictly below one. Such randomness can also be extracted from the finite graph sample: assign distinct graph outcomes to the finitely many balanced partitions; under the common law every selected outcome has positive probability. The worst error remains strictly below one for sufficiently small positive , by continuity. The printed lower bound tends to one as for any fixed universal , which contradicts this. The entropy correction cannot be dropped without an asymptotic or signal-size condition.
The Kullback-Leibler divergence of a product measure is the sum of the individual divergences, by expanding the logarithm of the likelihood ratio and taking its expected value. Put and define to be the unordered pairs lying within one group, while is the set of unordered pairs crossing between groups. Then
Thus choosing the printed to mean gives its order of the summands. If denotes the usual graph cut , their two orders must be interchanged. Each unordered edge is counted once.
The inequality gives
Since , each changed-edge divergence is at most . Summing over at most pairs gives . This is the information scale needed for Fano's inequality.
For partitions, the appropriate Hamming distance between unlabelled bipartitions is . It counts vertices assigned to the wrong group after the better global exchange of the two group names. The printed signed-indicator formula does not implement this exchange: negating a zero-one indicator is not taking its complement. It must be read as this partition distance, or written with the membership vectors.
Use the corrected leading eigenvector estimator from 4(b), and orient its sign to minimize . At every wrongly signed coordinate, this difference has magnitude at least . Therefore the sign rounding bound for a unit eigenvector gives
The last inequality uses the projector distance bound . Taking the expected value of this Frobenius norm estimate proves . The assertion relies on the corrected estimator and partition-distance definition.
For a symmetric matrix with simple largest eigenvalue and spectral gap , the Davis-Kahan curvature lemma says , where is its leading unit eigenvector. Applying this to a leading eigenvector of , the maximal Rayleigh quotient and duality of the operator norm and trace norm give ; the projector difference has matrix rank at most two. This is a rank-one eigenprojector perturbation bound.
Here has independent centered bounded upper-triangular entries, apart from symmetry. The spectral norm bound for a centered Bernoulli adjacency matrix gives . Taking and therefore gives, for the largest-eigenvalue estimator,
The printed argmin selects the wrong end of the spectrum. For a minimizing unit eigenvector , compare its Rayleigh quotient with any unit vector orthogonal to . This yields , and hence . Its mean projector error is at least , using . Along with sufficiently large even , the printed parameter restriction holds and this lower bound tends to , contradicting an error of order . Replace argmin by argmax, or negate the entire matrix before minimizing.
Write on and on , and put and . Because the two groups have equal size, are orthonormal. Let and . The expected value of the adjacency matrix of a graph is . Consequently
Both displayed eigenvalues are positive under the stated range of , so the first matrix has matrix rank two. The other eigenvalues are zero. The figure shows the within-group and between-group blocks of this balanced stochastic block model; itself has zero diagonal.
Figure 1.
Expected adjacency blocks and the positive rank-one community signal
. Details.
Let . The columns of have Euclidean norm at most one. Under the usual sub-Gaussian random vector convention, 3(d) justifies the tuning choice , which is independent of . Also , so the Chernoff bound gives with probability at least .
On the intersection of these events, of probability at least , set and . The weighted Young inequality gives for . Applying it to 3(c) and rearranging yields the rigorous bound
If the coefficient-one concentration statement in 3(d) is taken as an extra hypothesis, instead choose ; the identical argument gives precisely the printed coefficient in the first term. Thus the deterministic argument and its risk rate are sound; its concentration constants depend on correcting or strengthening 3(d).
With the usual variance-proxy convention for a sub-Gaussian random vector, every column inner product is a sub-Gaussian random variable with proxy at most . The two-sided Chernoff bound and union bound imply
with probability at least . No independence between these inner products is required.
The printed coefficient-one bound is false under this convention. Take , , and with equal probability. This Rademacher random variable has the stipulated variance proxy, but for the printed threshold is less than , whereas surely. Even a general sub-Gaussian assumption does not imply Gaussian concentration inequality for arbitrary Lipschitz functions. Under the stronger convention , the same union bound proves the printed constants. That convention, however, differs from the binomial variance proxy used in 1(b).
Decompose using the incidence pseudoinverse decomposition. The Cauchy-Schwarz inequality controls the constant component, and the duality of the and norms controls the remaining component:
On the stipulated event, the second term contributes at most after multiplication by . The triangle inequality gives , so the negative penalty in the basic inequality for a penalized least-squares estimator cancels. We obtain
on an event of probability at least .
Let . The defining minimum for graph total variation denoising, compared with the feasible value , gives . Expand the squared Euclidean norm and cancel to obtain the basic inequality for a penalized least-squares estimator:
Each row of the oriented incidence matrix records one signed endpoint difference, hence
Reversing an edge changes the sign of its row, and permuting the edges permutes rows; neither operation changes the sum of absolute differences. Thus the graph total variation is independent of the chosen orientation and ordering. If , values agree at the endpoints of every edge. Any two graph vertices of a connected graph are joined by a graph path, so their values agree. Therefore .
Two independent cyclic intervals intersect with probability at most . For , precisely the starting-point offsets can overlap; for larger , the upper bound is automatic. The cyclic interval overlap bound and the preceding chi-squared divergence estimate therefore give
As in 1(e), the square-root hypothesis is meaningful when . The chi-squared testing lower bound bounds the total variation distance by . The worst individual Type II error dominates the error under the uniform mixture model. Thus
Taking small makes this lower bound informative. This argument uses the actual cyclic-interval alternative, without substituting a different family of subsets.
Define . Expanding its chi-squared divergence by the second moment of a mixture likelihood ratio, and using the supplied one-observation cross moment and independence, gives
Here are independent uniform members of the specified family of cyclic intervals. Their overlap does not have the hypergeometric distribution of arbitrary uniform -subsets. The family's geometry must be retained.
Let . Since and , we obtain . Taking the expected value proves the upper bound. The factor in the last exponent is present in the PDF and missing in the TeX aid.
Put and . Use the quadratic scan statistic and reject when . Under the null hypothesis, each has the chi-squared distribution with degrees of freedom. The chi-squared concentration inequality gives , by the union bound. Since , the proposed Type I error is at most .
For the true set , has the same chi-squared distribution. We need a lower-tail bound that remains positive even when is close to one. Set . The chi-squared Chernoff lower-tail bound gives : indeed for . To verify this last inequality, its derivative is ; the expression in parentheses is strictly concave, starts at zero, and changes sign once, so the minimum occurs at an endpoint, where the inequality holds.
The function is convex with , so on this interval. Consequently implies . Failure to reject then requires , giving Type II error at most , uniformly in . An explicit, deliberately conservative answer is
The sharp constant is unnecessary; the detection scale is .
Use the convention for the centered sub-Gaussian random variable. Applying the two-sided Chernoff bound to gives . Since , this already supplies the requested bound:
For averages of independent random variables, the stronger useful statement is that the centered square of a sub-Gaussian random variable is a sub-exponential random variable. The Bernstein bound for independent sub-exponential variables then give for a universal positive . These are the concentration theorems used for quadratic scan statistics.
Because the scalar signal and the noise vector are independent random variables with normal distributions, their sum is a Gaussian random vector. Its expected value is zero, and the covariance matrix is
Thus this is a rank-one covariance spike. Each summand has expected value , so . Here the empirical matrix is an uncentered empirical second-moment matrix; because the population expected value is known to be zero, it is an unbiased estimator of the covariance matrix. The TeX's apparent factorial denominator is a transcription error.
Let . From the supplied second moment of a mixture likelihood ratio, dropping a nonpositive term gives
The last step uses the assumed separation. It has a real square root only when ; otherwise no parameters satisfy it. The chi-squared testing lower bound follows from the Cauchy-Schwarz inequality: . Every statistical hypothesis testing rule has sum of its Type I error and Type II error at least . Its larger error is at least half the sum, hence
For , the chi-squared divergence is . Write for the likelihood ratios of the complete -ball sample. Since and , the second moment of a mixture likelihood ratio gives
If instead denotes a one-ball likelihood ratio, its cross moment must be raised to the th power for the full sample, by independence.

Pinned article: Introduction to the OurBigBook Project

Welcome to the OurBigBook Project! Our goal is to create the perfect publishing platform for STEM subjects, and get university-level students to write the best free STEM tutorials ever.
Everyone is welcome to create an account and play with the site: ourbigbook.com/go/register. We belive that students themselves can write amazing tutorials, but teachers are welcome too. You can write about anything you want, it doesn't have to be STEM or even educational. Silly test content is very welcome and you won't be penalized in any way. Just keep it legal!
We have two killer features:
  1. topics: topics group articles by different users with the same title, e.g. here is the topic for the "Fundamental Theorem of Calculus" ourbigbook.com/go/topic/fundamental-theorem-of-calculus
    Articles of different users are sorted by upvote within each article page. This feature is a bit like:
    • a Wikipedia where each user can have their own version of each article
    • a Q&A website like Stack Overflow, where multiple people can give their views on a given topic, and the best ones are sorted by upvote. Except you don't need to wait for someone to ask first, and any topic goes, no matter how narrow or broad
    This feature makes it possible for readers to find better explanations of any topic created by other writers. And it allows writers to create an explanation in a place that readers might actually find it.
    Figure 1.
    Screenshot of the "Derivative" topic page
    . View it live at: ourbigbook.com/go/topic/derivative
  2. local editing: you can store all your personal knowledge base content locally in a plaintext markup format that can be edited locally and published either:
    This way you can be sure that even if OurBigBook.com were to go down one day (which we have no plans to do as it is quite cheap to host!), your content will still be perfectly readable as a static site.
    Figure 5. . You can also edit articles on the Web editor without installing anything locally.
    Video 3.
    Edit locally and publish demo
    . Source. This shows editing OurBigBook Markup and publishing it using the Visual Studio Code extension.
  3. https://raw.githubusercontent.com/ourbigbook/ourbigbook-media/master/feature/x/hilbert-space-arrow.png
  4. Infinitely deep tables of contents:
    Figure 6.
    Dynamic article tree with infinitely deep table of contents
    .
    Descendant pages can also show up as toplevel e.g.: ourbigbook.com/cirosantilli/chordate-subclade
All our software is open source and hosted at: github.com/ourbigbook/ourbigbook
Further documentation can be found at: docs.ourbigbook.com
Feel free to reach our to us for any help or suggestions: docs.ourbigbook.com/#contact