For a discrete-time martingale relative to the filtration , each is -measurable, , and for . The tower property then gives the corresponding identity conditional on any earlier time.
A bounded previsible process has measurable at time and . Its martingale transform is adapted and integrable, since each increment has absolute expectation at most . Moreover,
Thus is a martingale, with .
For the Itô integral, begin with a bounded adapted step integrand on . Its integral is . Each summand has zero conditional mean. For , conditioning the cross product on makes its expectation zero. The conditional second moment of a Brownian increment is , hence
Approximate the predictable integrand in by such step processes. The same identity makes their integrals Cauchy in and defines the limit; both its first and second moments converge. This gives the requested mean-zero result and the Itô isometry:
For the Ornstein-Uhlenbeck process, the deterministic integrating factor has no quadratic covariation with . The Itô formula therefore gives . Integrating and dividing by yields
The mean-zero integral and Itô isometry give
At these formulas are read by continuity: , with mean and variance .
For the CIR model, let and use the usual positive parameters. Before a possible visit to zero, the derivatives of are and . Applying the Itô formula gives the Square root of a CIR diffusion:
The PDF's claim under is false: that choice leaves the nonzero drift , which is not affine in . The actual interior cancellation condition is . Even then the principal square root is nonnegative and cannot be an unrestricted Gaussian OU process.
To give the intended corrected construction, let a signed process solve , with . The Itô formula gives
Writing , with either sign chosen at zero, produces a Brownian motion; the diffusion term is . Thus is a CIR model with , while is reflected at zero. The reflection contributes local time of a semimartingale, so the interior SDE alone is not a global unrestricted OU equation. Its second moment is , agreeing with the CIR mean for that corrected parameter choice.
The requested mean under the parameters actually printed is still available without the false OU assertion. Taking expectation in the original CIR equation gives ; the stochastic integral has zero mean after the usual moment/localization justification. Consequently, for the printed choice as well as general admissible parameters,
For minimizing an energy , simulated annealing at temperature uses the Boltzmann distribution . At each temperature run a Metropolis–Hastings algorithm or Gibbs transition preserving that law, then reduce . With a symmetric proposal, accept with probability : downhill moves are accepted and uphill moves remain possible. Keep the best state encountered. As temperature decreases, equilibrium mass favors global minima, provided the target is proper and the relevant concentration hypotheses hold. An arbitrary finite cooling run does not by itself prove global convergence.
Let , and . For the Poisson distribution model the observations must be nonnegative integers. Its log-likelihood is
For , differentiation gives and . If , the likelihood is decreasing and the maximizer is the boundary value zero. For the normal distribution model write , giving
For each , this is maximized at , and differentiating there gives . Hence
The normal variance estimate has denominator , not . A finite positive normal maximum requires ; if all observations coincide, the likelihood is unbounded as . The normal annealing calculation below assumes .
To maximize likelihood, choose the energy , so the annealing law is , not . This sign is important in the Poisson request: literally inserting a positive log-likelihood into a minimizing Boltzmann law would give a nonnormalizable probability density function growing like at infinity. We use likelihood-power annealing with the maximizing sign.
For the normal distribution model, take base measure . The joint annealing probability density function is
Completing the square gives the Gibbs sampling conditional for . Matching the conditional probability density function to the supplied inverse-gamma distribution convention gives
Integrating out contributes a factor , so the marginal law is
Thus this joint Boltzmann law is proper for . Its variance marginal and the normal conditional both concentrate at the stated estimates as . For example, for its marginal mean and variance are
Also .
There is a direct convergence proof for the actual changing-temperature Gibbs algorithm, beyond convergence of its equilibrium probability densities. Start at a finite deterministic , choose deterministic with , draw conditional on , then draw conditional on using the two boxed distributions. Conditional normal moments give
The inverse-gamma distribution first and second moments consequently yield
and
In the first recursion the coefficient of the preceding mean is at most ; in the second the coefficient of the preceding second moment is at most . The means therefore stay bounded, and then so do the second moments. Letting in these formulas gives and . Together with , this proves
This Gaussian likelihood Gibbs annealing convergence uses the explicit conditional structure and does not assume that one Gibbs sampling sweep is an exact equilibrium draw.
For the Poisson distribution model, the likelihood-power law with respect to has the normalized probability density function
so
The mean tends to and the variance to zero, proving convergence to a point mass at the Poisson maximum-likelihood estimate, including the zero-boundary case.
The Akaike information criterion is , counting one parameter for Poisson and two for normal. With and , the two forms are
and
Choose the smaller criterion. Likelihood constants that differ between models, including the factorial and terms, cannot be discarded.
For a fully specified trans-dimensional annealing for penalized likelihood, let or and define
Use the combined target, relative to counting measure on models and Lebesgue measure on their parameters,
where are fixed reference weights. For and both model components are integrable. The factor is a convenient temperature convention keeping the within-model laws exactly those already derived. The extra factor makes the zero-temperature objective AIC rather than unpenalized likelihood.
Use, for example, probability for a model-change attempt in either model and probability for a within-model update. Within the Poisson distribution model draw a fresh gamma value from its boxed law; within the normal distribution model perform the boxed Gibbs sampling sweep. For a Poisson-to-normal proposal, draw with a fixed , and set
The inverse normal-to-Poisson proposal is with reverse auxiliary . This dimension matching uses one auxiliary variable to augment the one-dimensional Poisson parameter space. The absolute Jacobian determinant is
If the forward model-jump selection probability is and the reverse probability is , the reversible-jump Markov chain Monte Carlo acceptance ratio is
Accept with probability . For the reverse move compute and use
For the suggested equal selection probabilities, and those factors cancel. The probability density function must still be included, despite symmetry of its normal distribution: it is an auxiliary draw in only the forward dimension-increasing proposal. Pairing a move with its inverse makes the accepted forward and reverse probability flows equal after the change of variables, proving detailed balance for each fixed-temperature model-switch kernel.
Cool through proper temperatures, allow enough model and parameter exploration at each stage, and retain the smallest observed . The minima of the two energies are exactly the two AIC values above. Low-temperature equilibrium mass concentrates on the better criterion and its fitted parameters; if they tie, the minimizing set contains both models. Unlike the direct normal Gibbs argument, this multimodel exploration claim needs an appropriate cooling/mixing justification and is not guaranteed by an arbitrary short run.
Finally, an exact statistical comparison must put both likelihoods on the same observable and reference-measure convention. A Poisson distribution model is a probability mass function on counts, whereas the displayed normal distribution model is a continuous probability density function. The formulas above are the question's formal Gaussian working-likelihood comparison. A coherent exact normal approximation for count observations can instead use probabilities of rounding intervals; that changes its likelihood and generally loses the simple analytic normal estimates. This distinction matters when interpreting the numerical AIC difference, while leaving the dimension-matching and annealing construction applicable to properly specified likelihoods.

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