For $y^{\prime }=y-A{y}^p$, $A>0$ and $0<p<1$, put $u=y^{1-p}$ while $y>0$. Then $u^{\prime }=(1-p)(u-A)$. If $0<y_0<A^{1/(1-p)}$, $u$ decreases to zero at the displayed finite time. The positive branch joins the zero solution with zero first derivative, giving a nonnegative forward solution. The equilibrium at zero is one-sided stable even though the vector field is not locally Lipschitz continuous there; the positive equilibrium is unstable.

Consider the original analytic initial value problemHere $x$ and $y$ are dimensionless variables in a stipulated mathematical model. The trajectories below are exact solutions of a differential equation, with no measured data or empirical parameter fit. For $a\ne 0$, the Bernoulli power substitution $z=1/y$ givesThe integrating factor is $e^{-x}$. Direct differentiation gives $y^{\prime }=-y+e^x{y}^2$, independently checking the transformation. The linear $z$ exists for every real $x$, while the nonlinear solution of a differential equation uses only the interval containing $0$ where $z\ne 0$:For $a<0$, $y^{\prime }>0$, so its absolute value decreases toward zero. The lost zero solution of a differential equation is recovered by substitution into the original equation. Since $f(x,y)=-y+e^x{y}^2$ is locally Lipschitz in $y$ everywhere, the Picard-Lindelöf theorem ensures uniqueness through every finite point, including $y=0$. Thus a nonzero trajectory never meets or crosses zero at a finite $x$.

For $a=1/2$, the exact solution of a differential equation is $y=e^{-x}/(2-x)$ on $(-\infty ,2)$. It first decreases, reaches a local minimum $y(1)=e^{-1}$, then increases without bound. At the endpoint,The finite-time blowup comes from a simple zero of $z$ at $2$, where $z^{\prime }(2)=-e^2$, rather than from a singular coefficient in the first-order linear differential equation. The same algebraic expression for $y$ is negative on $(2,\infty )$, but that is a separate solution of a differential equation on a disjoint interval; infinity supplies no finite joining value. For $a=-1$, $y=-e^{-x}/(1+x)$ on $(-1,\infty )$, giving a distinct decaying branch whose finite blowup is instead in the past.

The direction field also explains the positive turning point. Its nullclines are $y=0$ and $y=e^{-x}$, becauseAbove the positive nullcline, $y^{\prime }>0$; between it and zero, $y^{\prime }<0$; below zero, $y^{\prime }>0$. The moving nullcline $y=e^{-x}$ is not a solution of a differential equation: its derivative is $-e^{-x}$, whereas the prescribed slope there is zero. The $a=1/2$ trajectory crosses it at $x=1$, exactly its local minimum.

For arbitrary real $n\ne 0,1$, work first on an interval where $y>0$. Define exponentiation by $y^r=\exp (r\log y)$, using the real logarithm. The change of variable is differentiable and invertible there, and the chain rule givesMultiplying the Bernoulli differential equation by $(1-n)y^{-n}$ yields the first-order linear differential equationThis is an exact transformation of nonzero solutions of a differential equation on the chosen branch; it is valid far from any reference state and does not discard higher powers as a local linearization would.

Given the initial condition $y(x_0)=y_0>0$, the integrating factor and transformed solution of a differential equation areThe product rule verifies $(\mu z{)}^{\prime }=(1-n)\mu Q$. Recover $y=z^{1/(1-n)}$ only on the connected component of $\{x:z(x)>0\}$ containing $x_0$, inside the original coefficient interval. The first-order linear differential equation can have a longer existence interval than this inverse transformation allows.^[ref]

Negative values need a separately justified real branch. For integer $n$, the required powers are real and differentiable for $y<0$ too, but the inverse must retain the sign selected by the initial condition. For $n=3$, $z=y^{-2}>0$ represents either $y=+1/\sqrt{z}$ or $y=-1/\sqrt{z}$; both square roots are taken of positive $z$. For $n=2$, $z=1/y$ retains the sign directly. For a general noninteger exponent, use the positive branch unless a specific differentiable real negative branch has been established. Never evaluate a real fractional power of a negative value merely because the linear formula has become negative.

Division by $y^n$ excludes $y=0$. When $n>0$ and the original real power is defined at zero, $y\equiv 0$ is an additional solution of a differential equation and must be checked in the original equation. When $n<0$, the equation as written is undefined at zero. On a positive branch the right-hand side is locally Lipschitz in $y$, so the Picard-Lindelöf theorem gives local uniqueness. At zero, the same hypothesis needs a fresh check: it holds for the polynomial $n=2$ example, whereas for $0<n<1$ a nonzero power coefficient prevents local Lipschitz continuity there. If $z\downarrow 0$, the inverse sends $y\to \infty$ for $n>1$, but sends $y\downarrow 0$ for $0<n<1$. The latter endpoint can lead to finite-time extinction in a sublinear Bernoulli equation, where the original equation determines how to join to zero.

Suppose a connection polynomial $C$ of degree at most $L$, with recurrence span $L$ gives zero recurrence discrepancies through index $r-1$ but a nonzero discrepancy at $r$. If a polynomial $D$ of span $L^{\prime }$ successfully explains the prefix through $r$ and $L+L^{\prime }\le r$, evaluate the coefficient at $r$ in the convolution $(CD)x$. Applying $D$ first makes it zero. Applying $C$ first leaves its nonzero discrepancy at $r$, because the earlier discrepancies vanish and $D$ has constant coefficient one. This contradiction gives the displayed lower bound. It is the minimality step used by the Berlekamp-Massey algorithm; the algorithm cancels the new discrepancy with a shifted earlier failed polynomial.

In the dilute-vortex description of the two-dimensional XY model, $K={\rho }_s/(k_{B}T)$ is dimensionless stiffness and $y$ is phase vortex fugacity. A phase vortex has logarithmic energy $\pi {\rho }_s\log (L/a)$, while its placement entropy is $2k_B\log (L/a)$. Small phase vortex pairs screen the logarithmic interaction, lowering the stiffness. Near $K=2/\pi$, the rescaled variables $X=2-\pi K$, $Y=4\pi y$ obey $X^{\prime }=Y^2$, $Y^{\prime }=XY$ to leading order. The invariant $X^2-Y^2$ leads to an essential correlation-length singularity $\xi \sim a\exp (C/\sqrt{T-T_c})$, not a finite power-law exponent.

Here $\ell =\log (r/a)$ increases toward larger observation distances, and the vortex fugacity $y$ is a dimensionless weight per unit core, conventionally initialized as $y_0\simeq e^{-E_c/(k_{B}T)}$ with the microscopic measure fixed. Numerical coefficients depend on this normalization. In the displayed convention, differentiating $X=2-\pi K$ using the original flow givesOnly replacing the prefactor $(2-X{)}^2/4$ by one near $X=0$ gives the scaled leading model. Its first integral follows directly:Thus this first integral belongs to the stated scaled model, not exactly to the original $K,y$ equations. For $Y\ge 0$, the critical separatrix is $X=-Y$. The bound side has $X<-Y$ and flows to $Y=0$, $X<0$; the unbinding side ultimately runs to large vortex fugacity, where the dilute approximation ceases to be quantitative.

To derive the unusual correlation length scaling without fitting an exponent, write the invariant on a nearby unbinding trajectory as $X^2-Y^2=-c^2$. Then $X^{\prime }=X^2+c^2$ and the time to cross fixed small limits $-b$ to $b$ is $2\arctan (b/c)/c\sim \pi /c$. If a physical one-parameter path crosses the separatrix transversely, its $c^2$ is proportional to the positive reduced temperature. Since distance grows as $e^{\ell }$, this gives $\xi \sim a\exp (b_0/\sqrt{t})$, with a nonuniversal $b_0>0$. This is the dimensionless-temperature form of the singularity above.

For odd $n$, multiplication by two permutes the residues modulo $n$. Its orbit through $i$ is the binary cyclotomic coset modulo an odd integer. If $\alpha$ has order $n$ in a finite field whose field characteristic is two, the Frobenius automorphism sends ${\alpha }^i$ to ${\alpha }^{2i}$. Hence the minimal polynomial of an algebraic element ${\alpha }^i$ over ${\mathbb{F}}_2$ is ${\prod }_{j\in C_i}(X-{\alpha }^j)$. Its coefficients are fixed by Frobenius and lie in ${\mathbb{F}}_2$, and its degree is the orbit size. Thus a binary cyclic code containing one defining zero of a cyclic code must contain the full orbit as zeros.

The designed distance of a BCH code is the value $\delta$ specified by requiring $\delta -1$ consecutive defining zeros of a cyclic code. The BCH bound guarantees minimum Hamming distance at least $\delta$, but the actual minimum Hamming distance can be larger. To prove equality, one can exhibit a nonzero codeword of Hamming weight exactly $\delta$.

For a proper prior distribution $\pi (\theta )$ independent of coalescent times $T$, let $q(T)={\prod }_{j=2}^n{\lambda }_j{e}^{-{\lambda }_j{T}_j}$ with ${\lambda }_j=\left(\genfrac{}{}{0ex}{}{j}{2}\right)$ and $L=\sum j{T}_j$. Observing $S=k$ under an infinite sites mutation model gives posterior density $\pi (\theta )q(T)e^{-\theta L/2}(\theta L/2{)}^k/(k!m_k)$, where $m_k$ is the prior predictive probability of $S=k$. Multiplication by the conditional Poisson mass and normalization is an application of Bayes' theorem.

// Target: statistical-inference.bigb

For continuously distributed uncertain quantities, define rank one to be the largest. A joint posterior distribution induces a discrete distribution on each quantity's rank. Estimate it by ranking the whole vector in each joint draw and averaging rank indicators; the probability of rank one is the probability of being largest. Ranking posterior means does not recover this uncertainty. The joint draws must preserve any posterior dependence, and tie conventions must be specified for models with atoms.

For one uniformly sampled integer $Y\in \{1,\dots ,N\}$, the likelihood function for the unknown endpoint is $N^{-1}{1}_{N\ge y}$. A flat improper prior on positive integers gives an improper posterior distribution, since the harmonic series diverges. A reciprocal prior instead gives a proper posterior proportional to $N^{-2}$ for $N\ge y$, with approximate survival probability $y/n$ and finite quantiles, but infinite posterior mean. A uniform prior cut off at $M$ gives posterior mean $(M-y+1)/(H_M-H_{y-1})\sim M/\log M$, demonstrating sensitivity to that arbitrary cutoff.

For independent observations uniform on $(\theta -d,\theta +d)$ and a uniform prior on $(A,B)$, the likelihood is constant wherever all data-compatible intervals overlap. The posterior is therefore uniform on the displayed intersection, provided it has positive length. Both squared-error and absolute-error Bayes estimates are its midpoint. An empty or zero-length interval has zero marginal likelihood and cannot be normalized into this ordinary posterior density.

For one observation from a continuous uniform distribution on $[0,\theta ]$ and prior density $\theta e^{-\theta }$, the likelihood cancels the factor $\theta$ and leaves $e^{-\theta }{1}_{\{\theta \ge x\}}$. Normalization gives posterior density $e^{-(\theta -x)}{1}_{\{\theta \ge x\}}$, for $x\ge 0$. Its mean is $x+1$ and variance one. The Bayes estimator under squared error loss is therefore $x+1$, unchanged by multiplying the loss by a positive constant.

A prior predictive check simulates parameters from the prior distribution and observations from their sampling model, then examines whether those observations are plausible before conditioning on data. This reveals unintended information from apparently broad priors, especially after nonlinear transformations or when scales and correlations are coupled. It differs from a posterior predictive check, which uses the posterior distribution.

For a common success chance with beta prior and conditionally independent binomial observations of sizes $m_i$, put $M={\sum }_i{m}_i$ and $s={\sum }_i{x}_i$. Beta-binomial conjugacy gives posterior mean $(\alpha +s)/(\alpha +\beta +M)$, equal to $Z(s/M)+(1-Z)\alpha /(\alpha +\beta )$. Multiplying by the next exposure predicts its expected count. The factor tends to one as exposure increases and to zero as prior concentration increases at fixed prior mean.

For a common gamma distribution (shape $\alpha$, rate $\beta$) frequency prior distribution and conditionally independent Poisson distribution yearly counts of means $w_j\Theta$, observed average amounts $y_j$ determine counts $k_j=w_j{y}_j/c_j$. The posterior distribution shape is $\alpha +{\sum }_j{k}_j$ and rate is $\beta +{\sum }_j{w}_j$. The future per-policy mean is the display, equal to credibility factor $W/(W+\beta )$ times ${\sum }_j(w_j/W)(c_{n+1}/c_j)y_j$ plus the complementary weight times $c_{n+1}\alpha /\beta$. The data must lie on their count lattice; impossible noninteger recovered counts have zero likelihood.

// Destination: probability-theory.bigb

conditionally independent normal distribution observations of mean $\theta$ and known variance ${\sigma }_1^2$, with normal prior mean $\mu$ and variance ${\sigma }_2^2$, give normal-normal conjugacy with known observation variance. Their posterior mean has the displayed exact credibility estimate form. It minimizes posterior squared-error loss and predicts the conditional expectation of the next observation.

A Bayes act minimizes $\int L(\theta ,a)\Pi (d\theta )$ for a specified parameter distribution $\Pi$. Applying the same minimization to the posterior after observing data gives a Bayes decision rule. The optimal data-based Bayes risk is the expectation of the optimal posterior no-data loss.

For loss $L(\theta ,a)=w(\theta )|\theta -a|$, assume $w\ge 0$, $0<\mathbb{E}[w(\Theta )\mid x]<\infty$ and $\mathbb{E}[w(\Theta )|\Theta |\mid x]<\infty$. The posterior expected loss is a positive constant times absolute-error loss under the posterior distribution reweighted by $w$. Thus its minimizers are the medians of that weighted posterior distribution. More precisely, the weighted cumulative distribution function $F_w$ satisfies $F_w(a^{-})\le 1/2\le F_w(a)$; the one-sided slopes of the convex loss give this condition.

For relative absolute loss on positive parameters, $w(\theta )=1/\theta$. With a gamma distribution posterior of shape two and rate $b>0$, reweighting produces an exponential distribution of rate $b$, so the estimate is $(\log 2)/b$. The weighted first-moment condition ensures finite loss; without it every finite action can have infinite loss, making a median characterization uninformative.

For integrable real $g$, including points where $g<0$ decreases the integral and including points where $g>0$ increases it. Thus the negative set minimizes the integral, up to null sets and arbitrary choices where $g=0$. This elementary signed-integral fact derives the Bayes classifier and cost-sensitive Bayes classifier.

With zero cost for correct classifications, assign class one when the displayed inequality holds and class two otherwise. It compares conditional expected losses; ties may be resolved arbitrarily. For positive costs and densities, the log likelihood ratio threshold is $\log (c(1\mid 2){\pi }_2/(c(2\mid 1){\pi }_1))$.