Let on . The Radon-Nikodym derivative is nonnegative and . For ,
Uniqueness of the restricted Radon-Nikodym derivative therefore identifies
For , the tower property of conditional expectation gives , and . Thus is a nonnegative true martingale, indeed a uniformly integrable martingale, since it consists of conditional expectations of one integrable random variable. This is the Radon-Nikodym density martingale. Under the usual conditions for a filtration, it has the standard càdlàg version; continuity is the additional assumption in part (c), not a consequence of absolute continuity alone.
Set , the stochastic logarithm of . On each finite horizon, a strictly positive continuous path has its reciprocal bounded; stopping at levels of , and rigorously localizes this stochastic integral. Thus is a zero-start continuous local martingale, with
The Itô formula for the ordinary logarithm gives
Accordingly, with the exponential convention printed in this question,
The initial constant in contributes no quadratic variation. Under the standard normalized Doléans-Dade exponential convention, which subtracts the driver's initial value in the exponent, the same formula is written .
For uniqueness, suppose another continuous driver gives the printed exponential. Evaluating at zero forces . Applying the Itô formula to its exponential gives . Since , associativity of stochastic integration yields . Thus , up to indistinguishability of stochastic processes.
There is an initial-integrability qualification if “local martingale” is used with the usual integrable-initial-value definition. The displayed is a continuous local martingale when , in particular when is a positive deterministic number. An arbitrary positive integrable does not ensure this. For example, take an -measurable positive integer with and put for all . This is a bounded strictly positive martingale, but any representing driver must have , which is not integrable. Thus the unqualified claim is false under that definition. For general , the always-valid local-martingale statement is the zero-start representation ; the printed representation additionally requires the stated integrability or the convention that allows arbitrary finite initial constants.
For , put . The process , where , is a bounded martingale by part (a). Also almost surely: before exit the endpoint lies inside the radius- ball, whose Gaussian probability tends to zero as . Path continuity makes the terminal radius either or . If , the dominated convergence theorem applied to this bounded stopped martingale gives
Solving yields
The denominator is nonzero because is strictly decreasing for .
As , the events increase to . In one direction membership already implies a finite hit of ; in the other direction a continuous path up to its finite hit of has a finite maximum radius, so it lies in the event for every sufficiently large . Continuity from below of probability and now give
Write and use the ordinary Euclidean Laplacian , so the generator of standard Brownian motion is half that operator. For a radial function , the Laplacian in polar coordinates is . With ,
Thus is a harmonic function on the punctured space.
We first justify that the singularity is not visited. For , stop at . The Itô formula makes a local martingale, and it is bounded on this annulus, hence a true martingale. The exit time is finite almost surely: by the Gaussian transition density. Taking the bounded terminal limit yields
If the origin were reached before , every smaller radius would first be reached, so letting shows that event has probability zero. Taking a countable sequence proves almost surely. Localization on increasing annuli now applies the Itô formula globally and gives the Bessel power local martingale
Stopping this continuous local martingale at any , including , preserves its local-martingale property.
For , the stopped path stays at radius at least , so . The bounded local martingale criterion makes it a true martingale. For , Brownian radial oscillations at its starting sphere give almost surely and the stopped process is constant. Thus the same true-martingale conclusion holds for .
For , the process is a strict local martingale. To prove this, write in distribution, with a standard -dimensional normal vector. For every shift ,
Indeed the integrand outside the unit ball is at most one, while inside it the Gaussian density is at most and the radial integral is a constant times . Brownian scaling therefore gives
A true martingale would have expectation at every time, which is impossible.
The printed range also permits , and this case must not be grouped with the bounded case. Here almost surely, by the same Gaussian bound for exit from a bounded ball. Moreover,
since the second expectation is at most . It too is a strict local martingale. The complete stopped inverse radial power martingale classification is therefore
Work first with . More generally, the assertion needs and then applies to . Without this assumption, an integrable but non-square-integrable initial random variable kept constant in time has zero quadratic variation but is not bounded in L2 space.
Choose increasing stopping times
They tend to infinity almost surely, because the continuous path and its quadratic variation are finite on every compact time interval. The stopped process is a bounded martingale, and is bounded and a local martingale. Taking expectations gives
Suppose . These squared expectations are at most . For each fixed , the stopped values converge almost surely to and are uniformly integrable, since they are uniformly bounded in L2 space. They therefore converge in . Passing in the stopped conditional-expectation identity, using contraction of conditional expectation in , gives for . Thus is a true martingale. The Fatou lemma gives , uniformly in , so it is an L2-bounded continuous martingale.
Conversely, suppose is a true martingale with . For fixed , bounded-time optional sampling theorem gives . Conditional Jensen inequality then yields
The brackets increase to . Applying the monotone convergence theorem first in and then in proves .
The exact identities and terminal limits can also be justified. On each fixed horizon , the Doob L2 maximal inequality gives
This is an integrable dominating variable for , so the dominated convergence theorem and increasing convergence of the brackets remove the stopping:
The L2 martingale convergence theorem supplies with almost surely and in L2 space. Consequently the squares converge in , while . We obtain the integrable terminal bracket criterion with the precise terminal identity
For rather than zero, the final identity becomes .
As with quadratic variation, the increasing process must be normalized by . Suppose have this initial value and both and are continuous local martingales. Since and these differences are continuous, both and are continuous. Their difference is a finite-variation process, and
is a continuous local martingale. The result proved in part (a) makes this difference constant. Its initial value is zero, so and are indistinguishable. This is the uniqueness of an increasing square compensator; it uses no existence argument.
If no initial-value normalization is imposed, the literal uniqueness assertion is false: whenever works, works for every deterministic constant , since subtracting a constant preserves the local martingale property. The exact unnormalized conclusion is for all , up to indistinguishability of stochastic processes.
The initial-value convention is essential. A continuous finite-variation local martingale is constant, so the conclusion is up to indistinguishability of stochastic processes. The stated zero conclusion uses the usual zero-start convention ; without it, is a counterexample.
Put . Let be its total-variation process, and stop when either or reaches . The resulting process is a bounded continuous local martingale, hence a true martingale, and its total variation on every time interval is bounded by . Fix and take deterministic partitions whose mesh tends to zero. Orthogonality of martingale increments gives
Pathwise,
by uniform continuity of a continuous path on . The sums are bounded by , so the dominated convergence theorem makes their expectations tend to zero. Hence almost surely. Take the countable intersection over rational and integer , use path continuity, and then let . The stopping times tend to infinity because the original path and its finite variation are bounded on every compact time interval. Thus
Since , invert the image clock before exit and put . Its quadratic variation is and its initial value is zero, because . By the Lévy characterization of Brownian motion, or the localized Dambis-Dubins-Schwarz theorem,
for a standard Brownian motion up to the image exit time. If that lifetime is finite, the stopped Brownian motion can be extended beyond it; only the stopped law is used here.
In this clock the image mapping-out functions obey the ordinary Chordal Loewner equation with driver . Their generated trace therefore has the law of SLE with parameter six, stopped when it leaves . Meanwhile the original image segment is exactly up to exit from . The clock change is increasing, so it disappears after passing to . This proves the local stopped-curve equality required in part (a).
For completeness, the original law is scale invariant: the maps have driver , whose law is again by Brownian scaling. Hence has the original trace law. Together with the stopped conformal-change argument, this establishes
The conclusion concerns local unparameterized segments up to exit; the capacity parameter itself changes according to .
Let map out the original filled hull and let map out the filled hull generated by . Both have hydrodynamic normalization at infinity. Write
Although the original is specified only on , this conjugacy is defined in the surviving neighborhood of the tip. Boundary preservation gives a real analytic extension there by the Schwarz reflection principle, with . The identities below are first applied on compactly localized intervals before exit and then continued up to .
Set , the capacity clock of the image hull. The Chordal Loewner equation for each flow and the chain rule give
Analyticity at the driving point cancels the coefficient of , forcing . The constant term, computed as in Question 3, is then . For a general SLE driver , the Itô formula consequently gives the transformed SLE driving function
At , this becomes
Thus is a continuous local martingale up to the exit time. If the source notation indexes the image driver by original time, it denotes . If it uses the image's own half-plane-capacity time, it denotes , with the inverse clock; that time-changed process is also a local martingale. This distinguishes the two clocks without changing the requested conclusion.
Regard a chord as an oriented curve modulo increasing reparameterization, and write for that class. Scale invariance of means that has the same law for each . The locality of a chord law under conformal changes of neighborhoods is the following stopped-curve property: if are neighborhoods of the starting boundary point and is a boundary-preserving conformal isomorphism with , then
The two stopped curves are sampled from the same original law ; equality is for their unparameterized initial segments. A normalization of can be imposed in an equivalent formulation, since scale invariance removes the positive dilation. In a hull-removal formulation, the Locality property of SLE says that a modification away from the starting point leaves the initial law unchanged until the curve reaches the modified part of the domain.
Put . The neighborhood assumptions at zero and infinity mean that is a bounded hull away from the starting boundary point. For , the conjugacy map is the hydrodynamically normalized mapping-out function of the remaining forbidden hull after removing . Its closure avoids the driver, so the boundary derivative is an excursion avoidance probability yields . Thus , and the bounded local martingale criterion makes the localized martingales uniformly bounded and uniformly integrable.
Choose localization times , also bounded by . Optional stopping gives . The supplied terminal limit and continuity of the power function give
Since these variables are bounded by one, dominated convergence yields
As usual for a chord starting at zero, the time-zero point is understood as the boundary prime end of ; the avoidance assertion concerns the subsequent curve. The boundedness step is what upgrades the local martingale calculation to an equality of probabilities.
Use the PDF's starred image hull and write . For , this map is analytic across a real neighborhood of the driver. Set
Let . The classical conformal change of half-plane capacity gives , and the image driver, indexed by original time, is . Differentiating the conjugacy relation gives
These are precisely the standard classical identities allowed in this question. The singularities cancel at . To calculate the remaining derivative, put and expand at fixed :
Therefore
where each keeps the spatial variable fixed before evaluation. Applying the Itô formula to gives the boundary derivative diffusion under conformal Loewner conjugacy
At , the third-derivative drift cancels. A second application of the Itô formula to gives
Choosing cancels the remaining drift. Hence the SLE eight-thirds restriction martingale is
Localizing away from and from vanishing derivatives makes the displayed stochastic integral a martingale on each localized interval, so it defines a continuous local martingale up to .
For , use the allowed probability-one event that the whole SLE trace lies in at every positive time. At each deterministic rational , part (b) makes the centered mapped future a fresh SLE with the same property. Consequently, almost surely, for every ,
This inverse relation holds through the trace's boundary-extension construction and puts positive-time future points in the open surviving domain. It therefore excludes every contact with the old filled hull, not just contacts with a fixed old point.
Intersect these probability-one events over all positive rational . If for , choose rational with . The first occurrence belongs to , whereas the second lies in , a contradiction. A return to is excluded by the original positive-imaginary-part assertion. This simplicity of SLE from boundary-avoiding restarts proves
After the fixed time , the mapping-out function of the uncentered mapped future is
Its expansion is , so the elapsed time is still the half-plane-capacity parameterization. Differentiating in yields
Therefore the Loewner transform of is , with starting point .
Centering gives , whose driver is . By independent stationary Brownian increments, is a standard Brownian motion independent of the past through time . Hence
This proves the domain Markov property of a chordal Loewner chain from the composition rule for chordal Loewner driving functions. For possible contacts with the old hull, the mapped trace is understood through the continuous boundary extension or prime ends; on the open half-plane the displayed maps are ordinary conformal maps. The PDF asks first for the driver of the uncentered curve ; the centered driver is included as well.
For chordal SLE with parameter , the Loewner trace starts at zero and is parametrized by half-plane capacity . Let be the unbounded component of , and let be its filled compact H-hull. When the trace is not simple, the hull includes the regions it disconnects from infinity. The associated mapping-out functions of compact H-hulls have expansion
and satisfy the Chordal Loewner equation
where is standard real Brownian motion. The equation is solved until is swallowed by the hull. The Loewner transform records the real boundary point at which the mapped hull grows; its relation to the trace is
Equivalently the tip maps to in the appropriate boundary prime-end sense. Thus the flow encodes the surviving domains, the transform drives that flow, and the trace generates its hulls. For , the driver is identically zero, with the branch asymptotic to , and .
At a real point outside the hull's closure, the Schwarz reflection principle makes the mapping-out function analytic across the boundary and . Use the course's Brownian excursion restriction identity: for an excursion from to infinity in the complex upper half-plane,
One way to obtain this identity is to start its Doob h-transform at : its avoidance probability is , which tends to by reflection. Thus the boundary derivative is an excursion avoidance probability, and in particular it is at most one.
Let be the filled unit half-disc. Since , avoiding implies avoiding . The monotonicity of boundary derivatives of mapping-out functions and part (b) give, for ,
The upper bound is attained by the empty hull and the lower by the filled half-disc, so the comparison is sharp on either real tail.
For the Joukowski map ,
It maps the exterior upper half-disc into . It is injective there: gives , and the second factor cannot vanish when both moduli exceed one. It is onto as well. For , the two roots of have product one. Neither lies on the unit circle, since is real there. Exactly one root has modulus greater than one, and the displayed imaginary-part formula makes its imaginary part positive. That root belongs to and maps to .
Since takes values in , its domain must be contained in . The established injectivity and surjectivity of mean that no proper subdomain of can map onto all of using this same function. Therefore
The half-plane capacity is the coefficient of in the normalized expansion. If hulls are regarded as closed subsets of , take the closure of this half-disc, including its diameter; its relative half-plane hull and capacity are unchanged. This is the half-plane capacity of a half-disc.
Write . By definition of a compact H-hull, is a proper simply connected domain, and it agrees with the complex upper half-plane outside a sufficiently large disc. The Riemann mapping theorem supplies a conformal isomorphism . Choose its boundary normalization so that the prime end at infinity maps to infinity. This prime end is unambiguous because the domain has an ordinary straight real boundary near infinity.
The Schwarz reflection principle, applied there in the coordinate , extends across the real boundary near infinity. It has a simple pole there, with Laurent series
The pole is simple because the reflected map is locally conformal at infinity; its leading coefficient is positive because it maps the upper side to the upper side. Postcomposing with the half-plane automorphism gives
This is hydrodynamic normalization at infinity.
If both satisfy it, is a conformal automorphism of the upper half-plane. The asymptotics force it to fix infinity, so it has the form with and . The same asymptotics force , . Hence the normalized mapping-out function exists and is unique. The argument requires reflection only outside a large disc, not regularity of the hull's entire boundary.
Values at the endpoint one do not affect any of the conclusions, since . Set to define each random variable everywhere, and choose a Borel measurable function agreeing with almost everywhere if necessary. These choices leave all integrals and almost-sure assertions unchanged.
Every dyadic interval at level lies in a dyadic interval at level . Equivalently, is a measurable function of , so : these sigma-algebras form a filtration. Since the uniform distribution of gives , the conditional expectations are integrable and adapted. The tower property of conditional expectation gives
Thus is a martingale, and part (a) proves that it is uniformly integrable.
For , the event is an atom of a sigma-algebra , with probability . The value of the conditional expectation there is therefore
Consequently This also shows directly that .
By the uniformly integrable martingale convergence theorem, there is an integrable such that almost surely and in the L1 norm. To identify this limit, put . Since away from the harmless endpoint, , and is -measurable. Conversely every is a measurable function of . Thus , and both and have -measurable versions.
For , the defining property of conditional expectation gives whenever . Passing to the limit in the L1 norm yields
The increasing union is an algebra of sets generating . The class of events for which this integral identity holds is a Dynkin system, so the pi-lambda theorem extends the identity to every . Since both random variables are measurable there, uniqueness of conditional expectation gives almost surely. We have proved that the dyadic conditional averages recover integrable functions:
because a probability-one assertion for is a full-Lebesgue measure assertion for . Likewise, the uniform distribution of turns convergence in L1 into
No pointwise continuity of is required.
A family of integrable random variables is uniformly integrable if
This controls the contributions of large values uniformly over the family, rather than merely bounding its L1 norms.
Put and . The conditional Jensen inequality gives , hence . The Markov inequality therefore gives
Using the measurability of in the defining integral identity for conditional expectation, for every we get
The first term tends to zero as because is integrable, by the dominated convergence theorem. Given , first choose to make it less than , then choose to make the second term less than . These choices work for every . This proves uniform integrability of conditional expectations:
In fact the argument works for an arbitrary family of sigma-algebras; nesting is not needed for this conclusion.

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