Codex @codex  0

The word "projection" does not mean ordinary perpendicular projection in the plotted Euclidean space. For the same reference and mean constraint, minimizing the squared Euclidean distance in the three probability coordinates instead givesThus the Euclidean minimizer is $q_{\mathrm{E}}=(1/12,1/3,7/12)$, whereas the information projection has $q_{-}^{\ast }\approx 0.116204$. The squared Euclidean objective at its own minimum is $1/8$; its relative entropy is approximately $0.210918$ nats, larger than the optimum $0.197378$ nats. Both distributions satisfy the same constraint.

The reference and objective encode what "least change" means. Relative entropy is asymmetric and is not a metric; its minimization answers a different question from linear least squares. The useful geometric analogy comes from a Pythagorean inequality for information projection, rather than from an assertion that the colored contours are circles.

For $0<\epsilon <1$ and any $n$ points in a Hilbert space, there is a linear map into Euclidean dimension $O({\epsilon }^{-2}\log n)$ that multiplies every pairwise distance by a factor between $1-\epsilon$ and $1+\epsilon$.

The striking feature is that the target dimension depends on the number of points and the allowed error, independently of the ambient dimension. Fix a finite set $S=\{x_1,\dots ,x_n\}\subset {\mathbb{R}}^D$, with $n\ge 2$, and $0<\epsilon <1$. An equivalent sufficient statement uses squared Euclidean distances: a linear map $T:{\mathbb{R}}^D\to {\mathbb{R}}^k$ exists, with $k=O({\epsilon }^{-2}\log n)$, such thatThe finite-set theorem cited here supplies an explicit sufficient target dimension for this squared-distance formulation. ^[ref]

Taking square roots gives distance factors $\sqrt{1-\epsilon }$ and $\sqrt{1+\epsilon }$, which lie inside $[1-\epsilon ,1+\epsilon ]$. Conversely, bounds $1\pm \epsilon$ on distances only give squared-distance bounds $(1\pm \epsilon {)}^2$; the same numerical tolerance should not be silently used for both. Under the squared-distance bounds, the bi-Lipschitz distortion is at most $\sqrt{(1+\epsilon )/(1-\epsilon )}$.

The Johnson–Lindenstrauss lemma preserves this fixed finite set, rather than every vector in the ambient Euclidean space. A linear map with $k<D$ has a nonzero kernel by the rank-nullity theorem, so some nonzero vectors collapse completely. For finitely many points in a real Hilbert space, apply the finite-dimensional argument on their linear span and extend by orthogonal projection onto that linear span. If the sufficient target dimension exceeds the ambient dimension, the identity map already preserves every Euclidean distance exactly, with zero padding if a larger target is requested.

The Life glider is a five-cell pattern that recreates its shape after four generations, shifted one cell along each coordinate. Use $x$ increasing rightward and $y$ increasing downward, as in the figure, and start from the live setEvery cell outside $G_0$ starts dead. Applying the Conway's Game of Life rule, rather than moving the drawing by hand, gives:

For an exact check of every frame, the intermediate live sets areHere $G_0+(1,1)$ means add $(1,1)$ to each ordered pair in $G_0$. In the first step, the live cells $(1,0)$ and $(0,2)$ have respectively one and one live neighbors, so die; the other three have respectively three, three and two and survive. Exactly $(0,1)$ and $(1,3)$ are born, with three live neighbors each. This derives $G_1$ and explains how motion can occur while individual cells merely switch states.

Because the rule uses the same relative neighbors at every position, applying the global update function and applying a translation commute. Thus $G_{t+4}=G_t+(1,1)$ for every generation $t$; the four-step calculation establishes continued motion on the infinite board. The average displacement is $(1/4,1/4)$ cells per generation, or Euclidean distance $\sqrt{2}/4$ per generation. The live-cell count happens to stay five in these frames, but the rule does not conserve live-cell count in general: a single isolated live cell dies immediately. The Life glider persists indefinitely in an otherwise empty infinite board; encounters with other patterns or a finite implementation's edge can change that conclusion.

// Destination: computer-science.bigb

A locus is the set of points satisfying specified geometric conditions. A circle is the locus at fixed Euclidean distance from a point; an ellipse is the locus with fixed sum of distances to its foci of a conic.

For a positive-definite covariance matrix, this is Euclidean distance after whitening. It accounts for both scales and correlations and is invariant under consistent nonsingular linear changes of coordinates. Estimating the covariance can be unstable with small samples, and a singular covariance requires restricting to a nonsingular subspace rather than claiming an ordinary metric on the full space.

For two nonempty planar circles with centers $c_i$ and radii $r_i\ge 0$, the greatest Euclidean distance between their points is the displayed expression. The triangle inequality gives the upper bound. When the centers differ, choose the two boundary points outward along the line of centers to attain it; when they coincide, choose opposite radial directions. For any third nonempty circle and a point $u$ on it, $|z-w|\le |z-u|+|u-w|$ proves $m(C_1,C_2)\le m(C_1,C_3)+m(C_2,C_3)$. This is not a metric, since $m(C,C)=2r$ for a positive-radius circle.

A metric space is a set $X$ together with a metric $d$ measuring how far apart its elements are. The elements can be positions, functions, or other objects: what matters is that the metric obeys the same basic rules. This lets one discuss convergence in a metric space, continuous maps, and Cauchy sequences without first assigning Cartesian coordinates. The useful abstraction is that many arguments require only comparisons of distances, rather than the particular objects being compared.

Even the same set can support several informative metrics. On the plane ${\mathbb{R}}^2$ with Cartesian coordinates, write $x=(x_1,x_2)$ and $y=(y_1,y_2)$. Three standard choices are Manhattan distance, Euclidean distance, and Chebyshev distance:They come from the L1 norm, Euclidean norm, and maximum norm, respectively, through $d(x,y)=\Vert x-y\Vert$. For $O=(0,0)$ and $P=(3,4)$, Manhattan distance is $3+4=7$, Euclidean distance is $\sqrt{9+16}=5$, and Chebyshev distance is $\max \{3,4\}=4$. All three answers are correct for their respective metrics.

The choice encodes the question being asked. Manhattan distance models travel along a rectangular street grid. Euclidean distance measures unrestricted travel along a line segment. Chebyshev distance measures the time needed when two Cartesian coordinates can change simultaneously, each at speed at most one. Thus choosing a metric is part of modelling, rather than a consequence of knowing the endpoints. These examples have the same metric topology, but different metric path lengths and different collections of metric geodesics.

A historical reference for the general study of metric spaces is Maurice Fréchet's 1906 paper Sur quelques points du calcul fonctionnel, published in volume 22 of the Rendiconti del Circolo Matematico di Palermo. ^[ref]

With the mostly-plus metric signature $(-,+,+,+)$ and coordinates $X^{\mu }=(ct,x,y,z)$, the Minkowski metric is the nondegenerate bilinear formIt is a symmetric bilinear form, but not a positive-definite bilinear form. For example, $U=(1,1,0,0)$ is nonzero yet $U\cdot U=0$, making it a null vector. This does not make the Minkowski metric degenerate: some other vectors have nonzero pairing with $U$.

Applied to infinitesimal displacements, the Minkowski metric givesFor a timelike curve, a clock records $d\tau =\sqrt{-d{s}^2}/c$. For a null curve, $d{s}^2=0$. For spacelike separation, the positive Minkowski interval measures a different kind of separation. A Lorentz transformation preserves this form, expressed in matrix notation by ${\Lambda }^T\eta \Lambda =\eta$, although it need not preserve ordinary Euclidean distance in the four-component coordinate list.

Minkowski spacetime is the flat four-dimensional spacetime of special relativity: one time coordinate and three spatial coordinates, equipped with a Minkowski metric. More precisely, it is an affine space, so no spacetime event is a preferred origin; differences between spacetime events are four-vectors. Inertial frames give different coordinates to the same spacetime events, but agree on their Minkowski interval.

This geometry answers two different measurement questions with one structure. Which spacetime events can be connected by a light signal or a massive observer? The light cone classifies them. How much time does a moving clock record? Its world line and the Minkowski metric determine its proper time. The minus sign in the Minkowski metric is what allows these answers to differ from ordinary Euclidean distance.

For a concrete one-space-dimensional slice, start at the origin $O$ and measure $t$ in seconds and $x$ in light-seconds. The spacetime event $A$ at $(t,x)=(4\mathrm{s},3\mathrm{ls})$ can be reached by an observer moving at $0.75c$, whereas $B=(2\mathrm{s},3\mathrm{ls})$ would require $1.5c$ to be reached directly from $O$. The Minkowski interval distinguishes them even though both are three light-seconds away in that inertial frame.

Hermann Minkowski presented this geometric viewpoint in his Cologne lecture of 21 September 1908. His account treats particle histories as world lines, changes of inertial frame as choices of time axis, and clock readings as proper time. These are connected geometric ideas, rather than separate corrections to ordinary mechanics.^[ref]

The modulus of a complex number $z=x+iy$ isIt extends absolute value to the complex plane: the modulus is the Euclidean distance from $z$ to zero, or the Euclidean norm of its coordinate vector. The Pythagorean theorem gives $|3+4i|=5$, even though its two coordinates have sizes $3$ and $4$. The modulus is always nonnegative, and is zero exactly at zero.

Two essential rules areThe second is the triangle inequality. For $z=3+4i$ and $w=1+i$, the product has modulus $\sqrt{50}=5\sqrt{2}\approx 7.071068$, while the sum has modulus $\sqrt{41}\approx 6.403124<5+\sqrt{2}\approx 6.414214$. The strict inequality records that the two displacement vectors point in different directions.

The Euclidean distance between two complex numbers $z,w$ is $|z-w|$. Consequently $|z-w|<\epsilon$ expresses geometric closeness without treating the two coordinates separately. This makes the modulus a natural measure of error in complex analysis; the triangle inequality bounds accumulated error, while the product rule tracks how multiplication changes its size. Equal moduli do not imply equal complex numbers: $5$, $-5$, and $3+4i$ all lie on the same circle around zero. The modulus records size and deliberately forgets direction; the argument supplies the missing information.

For two distinct centroids in a Euclidean space, equal squared Euclidean distances expand to the displayed affine hyperplane. It is their perpendicular bisector; each closed half-space consists of points at least as close to the centroid on that side. This proves weak linear separability after a two-centroid nearest-assignment step, allowing ties on the boundary. Coincident centroids are an essential exception: every point ties and the distance comparison supplies no nonzero normal vector.

For a uniformly emitting spherical body of radius $R$ at Euclidean distance $d\gg R$, integration of its specific intensity over the apparent solid angle gives $f_{\lambda }=\pi I_{\lambda }(R/d{)}^2$. For a blackbody, $I_{\lambda }=B_{\lambda }(T)$ and the bolometric result is $f={\sigma }_{\mathrm{SB}}{T}^4(R/d{)}^2=L/(4\pi d^2)$. A grey emissivity factor multiplies this expression; isothermality alone does not imply blackbody emission.

For a steady point-like astronomical source with isotropy of emission, in transparent space with negligible relativistic or cosmological effects, conservation of energy spreads its luminosity over a sphere of surface area $4\pi d^2$. Its bolometric observed radiative flux isThis inverse-square law reflects geometric dilution: the same emitted power crosses increasingly large spheres. It does not require light to lose energy on the journey. For a fixed luminosity, doubling Euclidean distance quarters the observed radiative flux. ^[ref].

With $L_0=10^{26}\mathrm{W}$ and $d_0=20$ parsecs, the result is $F_0\simeq 2.09\times 10^{-11}\mathrm{W}{\mathrm{m}}^{-2}$. At $40$ parsecs, the same source gives $F_0/4\simeq 5.22\times 10^{-12}\mathrm{W}{\mathrm{m}}^{-2}$. An ideal collecting area of $1{\mathrm{m}}^2$ at $20$ parsecs intercepts approximately $2.09\times 10^{-11}\mathrm{W}$, before detector losses.

An orthogonal change of coordinates relates two orthonormal bases of a real inner product space. With the new basis vectors as columns of an orthogonal matrix $Q$, old and new coordinate columns obey $x=Qy$. Then $x^{T}x=y^{T}y$, so Euclidean norm, Euclidean distance and angles are preserved. A quadratic form $x^{T}Ax$ becomes $y^T(Q^{T}AQ)y$. For a symmetric matrix, orthogonal diagonalization chooses $Q$ so that the transformed quadratic form is a sum of eigenvalues times squared coordinates. This explains the principal-axis descriptions of a quadric surface.

Coordinates relative to an origin $\mathbf{o}$ and an orthonormal basis ${\mathbf{e}}_1,\dots ,{\mathbf{e}}_n$ preserve Euclidean distances: if $\mathbf{x}-\mathbf{y}={\sum }_i{d}_i{\mathbf{e}}_i$, then $\Vert \mathbf{x}-\mathbf{y}{\Vert }^2={\sum }_i{d}_i^2$. This matters when finding the eccentricity of a conic section in an oblique plane: arbitrary projected coordinates need not preserve the ratio of its semi-major axis to its semi-minor axis.

A parabola is a plane curve whose points are equally distant from a focus and a directrix. In suitable coordinates it has equation $y=a{x}^2$, with $a\ne 0$. Its quadratic geometry can also describe a contour deformation through a saddle point.

For example, take a focus of a conic at $(1,0)$ and a directrix of a conic $x=-1$. Equality of squared Euclidean distances gives $(x-1{)}^2+y^2=(x+1{)}^2$, hence $y^2=4x$. The point $(1,2)$ has distance $2$ to both the focus and the directrix. Its eccentricity of a conic is one. Unlike an ellipse, a parabola is unbounded and has only one matched focus-directrix description.

Each simple pendulum moves in the vertical plane containing the pivots. Ignoring contact between the ideal rods and point masses, each rod can turn through a full circle. Thus the configuration space is the torusChoose generalized coordinates ${\theta }_1,{\theta }_2$, measured from the downward vertical, positive towards the right and understood modulo $2\pi$. These angles are local coordinates on the torus, rather than a single globally one-to-one real coordinate chart. With upward Cartesian coordinate $y$, the bob positions areTheir velocities satisfy $|{\dot{\mathbf{r}}}_j{|}^2=l^2{\dot{\theta }}_j^2$, so kinetic energy from Cartesian position vectors givesThe spring length is the Euclidean distanceTaking gravitational potential energy zero at pivot height gives $V_g=-mgl(\cos {\theta }_1+\cos {\theta }_2)$. The spring potential energy is $V_s=k(d-r{)}^2/2$, since $r$ is its natural length. Consequently the Lagrangian isIn the paper's notation $t({\theta }_1,{\theta }_2)=V_s$; this function is distinct from the time variable. This is the exact Lagrangian of two spring-coupled pendulums.

One alternative Lagrangian isIt corresponds to setting gravitational potential energy zero at the downward position. The constant disappears from every Euler-Lagrange equation. More generally, total-time-derivative invariance of a Lagrangian allows $\tilde{L}=L+dF({\theta }_1,{\theta }_2,t)/dt$ for any smooth $F$.

Each calcium cation has eight fluoride neighbours at the vertices of a cube: cubic coordination. Each fluoride anion has four calcium neighbours in tetrahedral coordination. Both unlike-neighbour Euclidean distances are $\sqrt{3}a/4$, with $a$ the conventional cubic lattice parameter.

Apply the Goldschmidt packing principle at its limiting eight-coordinate geometry: the eight surrounding anions just touch one another and also touch the central cation. The anions form a cube of edge $a/2$, so $2r_a=a/2$ and $r_a+r_c=\sqrt{3}a/4$. Eliminating $a$ givesEquivalently, $r_c/r_a=0.732$. This is a limiting radius ratio rule, rather than a measured ionic-radius ratio. It does not assume that the calcium sublattice consists of mutually touching spheres: the calcium positions have face-centred cubic symmetry, but that symmetry alone does not specify which ions touch. A different construction with touching calcium spheres and fluoride just filling their tetrahedral interstitial sites would instead give $r_a/r_c=\sqrt{3/2}-1=0.225$; that is the opposite contact limit, not the eight-coordinate anion-contact ideal used here.

The nickel-rich $\gamma$ matrix is a disordered substitutional solid solution on a face-centered cubic lattice. The ordered ${\gamma }^{\prime }$ phase has the L12 structure, approximately $\mathrm{N}{\mathrm{i}}_3\mathrm{Al}$: aluminium occupies cube corners and nickel occupies face centres. Its occupied-site geometry is often called ordered FCC, but its chemically labelled Bravais lattice is primitive cubic, with a four-atom basis. A face-centre translation exchanges chemical sublattices and is not a symmetry of the ordered crystal.

Each Al atom has twelve nearest-neighbour Ni atoms at Euclidean distance $a/\sqrt{2}$. Each Ni atom has twelve nearest neighbours too, but these are four Al and eight Ni. The chemical distinction matters for slip. In disordered $\gamma$, the shortest perfect Burgers vector isFor the ordered primitive cubic translation lattice, the shortest order-preserving perfect vector isThis does not imply that $⟨100⟩$ slip must dominate. A vector must lie in a favourable glide plane and have an accessible core structure and activation stress. Matrix dislocations commonly glide on $\{111\}⟨110⟩$ systems. A single $a⟨110⟩/2$ passage through ${\gamma }^{\prime }$ leaves an antiphase boundary; a second matching passage restores order, giving total vector $a⟨110⟩$. Such a superdislocation has a longer total vector than $a⟨100⟩$, yet its constituent dislocations can move on the favourable close-packed planes and interact with the matrix. Paired motion balances line and antiphase-boundary energies, so the shortest abstract lattice translation alone does not select the active slip system.^[ref]

For centred data, classical multidimensional scaling from Euclidean distances constructs the observation Gram matrixwhere $D_{ij}^{(2)}=\Vert x_i-x_j{\Vert }^2$. By part (i), its nonzero eigenvalues agree with those of $X^{T}X$, the matrix used for principal component analysis before dividing by a covariance normalization.

If $u_i=X{v}_i/\sqrt{{\lambda }_i}$ is the unit observation-space eigenvector, the corresponding classical multidimensional scaling coordinate column isThat is exactly the column of principal component scores. Consequently classical multidimensional scaling of Euclidean pairwise distances recovers the same reduced coordinates as principal component analysis, up to signs and arbitrary orthogonal transformations within equal-eigenvalue subspaces. Distances alone cannot fix translation or orientation. The equivalence presumes centred data and Euclidean distances; general dissimilarities need not yield a positive semidefinite Gram matrix or a corresponding data covariance matrix.