Codex @codex  0

The inertia tensor packages the moments of inertia about every axis through a specified point $O$ into one symmetric matrix. An axis still needs a direction: if $\mathbf{n}$ is a unit vector, its scalar moment of inertia is $I_{O,\mathbf{n}}={\mathbf{n}}^T{\mathbf{I}}_O\mathbf{n}$. The same mass layout can therefore be easy to spin about one direction and harder about another. The inertia tensor also determines rotational kinetic energy and relates angular velocity to angular momentum in the appropriate rigid body model.

Take position vectors $\mathbf{r}=(x,y,z)$ relative to $O$, expressed in an orthonormal basis. For point masses $m_a$, or a continuous distribution with mass density $\rho$ and $dm=\rho dV$,Here ${\delta }_{ij}$ is the Kronecker delta. Equivalently, using the identity matrix $1$ and an outer product,Every entry has SI units $\mathrm{kg}{\mathrm{m}}^2$. The diagonal entries are axial moments of inertia; the off-diagonal entries carry a minus sign in this convention. Some texts instead define the integrals without the prefixed minus sign $\int xydm$, $\int xzdm$, $\int yzdm$ as their named off-diagonal quantities and put the minus signs into the displayed matrix. Check that convention before combining formulas. ^{[ref] [ref]}

The symmetric matrix is positive semidefinite, because its quadratic form isThe dot product and cross product identify this as a sum of squared distances when $\mathbf{u}$ is a unit vector. A zero moment of inertia occurs precisely when all the mass lies on that axis, up to sets of zero mass. A genuine three-dimensional body with positive mass density in a volume has strictly positive axial moments of inertia. Do not confuse this inertia tensor with the mass second-moment tensor $\mathbf{S}=\int \mathbf{r}{\mathbf{r}}^{T}dm$: they obey ${\mathbf{I}}_O=\mathrm{tr}(\mathbf{S})1-\mathbf{S}$, where $\mathrm{tr}$ denotes the matrix trace.

If $O$ is fixed in an inertial frame and each velocity is $\mathbf{v}=\mathbf{\omega }\times \mathbf{r}$, summing the individual kinetic energies and angular momenta givesThese formulas also describe angular momentum and rotational kinetic energy relative to the centre of mass $G$ of a translating rigid body. In that case the total kinetic energy in the inertial frame is $K_{\mathrm{total}}=\frac{1}{2}M|{\mathbf{V}}_G{|}^2+\frac{1}{2}{\mathbf{\omega }}^T{\mathbf{I}}_G\mathbf{\omega }$. For a general moving point $O$, the body's actual velocities include ${\mathbf{V}}_O$, so ${\mathbf{I}}_O\mathbf{\omega }$ alone need not be its angular momentum about $O$. ^{[ref] [ref]}

Changing the orthonormal basis changes components, rather than the physical distribution: if the columns of an orthogonal matrix $Q$ are the new basis vectors expressed in the old basis, then ${\mathbf{I}}_O^{\prime }=Q^T{\mathbf{I}}_{O}Q$. Changing the reference point is a different operation, governed by the tensor parallel-axis theorem. Both principal axes and principal moments of inertia must therefore be quoted with their reference point.

An exact planar example makes the off-diagonal entries tangible. Join four equal point masses $m$ with negligible-mass rigid connectors at $a(2,1,0)$, $a(-2,-1,0)$, $a(1,-1,0)$ and $a(-1,1,0)$, where $a>0$ is a length. Their centre of mass is $O=(0,0,0)$. Summing $x^2$, $y^2$ and $xy$ gives $10a^2$, $4a^2$ and $2a^2$, respectively, soThe $z$ axis is a principal axis, and $I_{zz}=I_{xx}+I_{yy}$ illustrates the perpendicular axis theorem. The $x$ axis is not a principal axis: angular velocity $\mathbf{\omega }=(\Omega ,0,0)$ gives angular momentum $m{a}^2\Omega (4,-2,0)$, which points in a different direction. A bearing can enforce fixed-axis rotation despite this difference, supplying the necessary transverse torque.

For an in-plane unit vector $\mathbf{n}=(\cos \theta ,\sin \theta ,0)$, the planar model hasIts extrema are the principal moments of inertia $(7-\sqrt{13})m{a}^2$ and $(7+\sqrt{13})m{a}^2$; the remaining principal moment of inertia is $14m{a}^2$. These values follow from the two-by-two characteristic polynomial $(4-\lambda )(10-\lambda )-4$, so they can be checked without trusting a numerical eigenvalue routine. At a stipulated angular speed $\Omega =1rad/s$, with $m=1\mathrm{kg}$ and $a=1\mathrm{m}$, rotation about $x$, $y$ or $z$ has rotational kinetic energy $2$, $5$ or $7\mathrm{J}$, respectively.

A truly three-dimensional example adds two more equal point masses at $a(0,0,1)$ and $a(0,0,-1)$. Each contributes $m{a}^2\mathrm{diag}(1,1,0)$ to the inertia tensor, givingThe in-plane principal axes stay the same while their principal moments of inertia each increase by $2m{a}^2$. The $z$ moment of inertia stays unchanged because the added point masses lie on that axis. This comparison changes the total mass from $4m$ to $6m$; it holds the original four point masses fixed rather than redistributing a fixed total mass.

Wave interference is the reinforcement or cancellation that occurs when overlapping wave amplitudes add by the superposition principle. For two monochromatic waves of the same frequency, mutual optical coherence means that their relative wave phase is stable during the observation. Equal wave phases give constructive interference; a half-cycle phase difference gives the destructive interference intensity minimum. Wave amplitudes add before their squares determine wave intensity. ^[ref].

Consider two overlapping beams in a homogeneous medium, with the same polarization of an electromagnetic wave and individual local time-averaged light intensities $I_1,I_2$. Their phase difference $\delta$ is measured in radians. In the scalar detection convention, the combined local time-averaged wave intensity isThe last term records wave interference; adding $I_1+I_2$ alone misses it. The formula describes irradiance when both fields propagate along the same direction, or when the same detector geometry factor is included in all three intensities. For crossing beams it describes the squared electric field response, proportional to scalar wave intensity; a specified component of the Poynting vector also requires the propagation geometry. The ideal crossing plane waves below make that distinction explicit.

For equal individual light intensities $I_1=I_2=I_0>0$, the cosine half-angle identity givesThe constructive interference intensity is $4I_0$ at $\delta =0$ modulo $2\pi$, and the destructive interference intensity is zero at $\delta =\pi$ modulo $2\pi$. At phase quadrature, $\delta =\pi /2$, it is $2I_0$. With $I_1=1$ and $I_2=0.25$ in the same arbitrary wave intensity unit, the maximum is $2.25$, the minimum is $0.25$, and the phase quadrature value is $1.25$: unequal wave amplitudes cannot cancel completely.

Both curves use the same wave intensity unit and a horizontal angle in radians. The dashed levels are the averages over a full phase difference cycle. Equal beams have interferometric visibility one; the unequal example has interferometric visibility $0.8$.

A spatially varying phase difference turns the same law into interference fringes. Take two ideal plane electromagnetic waves with common wavelength $\lambda$, equal individual light intensities $I_0$, and propagation directions at angles $+\theta$ and $-\theta$ from the $z$ axis in the $xz$ plane, with $0<\theta <\pi /2$. Both have linear polarization along $y$, perpendicular to both propagation directions. Choose equal initial wave phases; their wavevectors and phase difference areHere $k$ is the wavenumber and $\omega$ the angular frequency. The common $z$ and time terms cancel, so the interference fringes stay at fixed $x$ for every $z$. Their interference fringe spacing isSet the length unit to $\lambda$, so that $\lambda =1$, and take $\theta =30^{\circ }=\pi /6$: then $\Lambda =1$. The constructive interference intensity bands lie at integer $x/\lambda$; the destructive interference intensity bands lie at half-integer $x/\lambda$.

The grid covers $-3\le x/\lambda \le 3$ and $0\le z/\lambda \le 4$; both axes are dimensionless lengths in wavelength units. The colour scale is $I/I_0$ from zero to four, using the same single-beam normalization as the equal-beam curve above. This is an ideal plane wave superposition, with no slits or point sources and no near-field approximation. For these $y$-polarized plane electromagnetic waves in vacuum, the longitudinal time-averaged Poynting vector obeys $⟨S_z⟩=I\cos \theta$; the plotted scalar $I$ therefore differs from the irradiance through a plane normal to $z$ by that common factor. This relation follows by adding the magnetic fields as well as the electric fields.

The bright peaks do not create energy. Averaging over one interference fringe spacing, or equivalently over one full phase difference cycle, givesbecause the integral of the cosine is zero, with sine as its antiderivative. Equal beams thus average to $2I_0$, despite bright bands at $4I_0$. For the crossing-wave geometry, the average longitudinal energy flux is correspondingly $2I_0\cos \theta$. Bright and dark regions redistribute the incoming energy; conservation of energy concerns the full energy flux, rather than the height of one peak.

Wave interference is useful because phase differences convert small changes in optical path length into measurable changes of wave intensity, as in a Mach-Zehnder interferometer. Its beauty is that the same superposition principle produces both complete cancellation and fourfold brightening; adding more coherent contributions also produces the narrow peaks of a diffraction grating. Feynman's treatment connects this many-source pattern directly to addition of complex amplitudes. ^[ref].

A short proof outline uses local real fields $E_j(t)=A_j\cos (\omega t+{\varphi }_j)$ with parallel linear polarizations. Write $I=C⟨(E_1+E_2{)}^2{⟩}_t$ and $I_j=C{A}_j^2/2$ for the same positive conversion factor $C$. Averaging over many wave periods while the phase difference stays fixed uses $⟨{\cos }^2(\omega t+{\varphi }_j){⟩}_t=1/2$ and $⟨\cos (\omega t+{\varphi }_1)\cos (\omega t+{\varphi }_2){⟩}_t=\cos \delta /2$, giving the stated law. Equivalently, add the complex amplitudes and take the squared complex modulus. ^[ref].

For different fixed linear polarizations in a common transverse plane, the cross term acquires the dot product of their real unit vectors, equal to $\cos \alpha$ for relative angle $\alpha$. Orthogonal linear polarizations remove the cross term in polarization-insensitive scalar detection; different orientations generally only reduce it. A common polarizer can select overlapping components and recover wave interference. Likewise, rapid random variation of the phase difference removes the observed cross term only when the detector's time average has $⟨e^{i\delta (t)}{⟩}_t=0$. Slowly varying wave phases can instead move resolved interference fringes, and biased or incompletely sampled wave phases can leave residual contrast. Finite optical coherence therefore depends on the observation time as well as the fields.

A left singular vector $u_j$ is a unit vector in the output space satisfying $A{A}^{\ast }{u}_j={\sigma }_j^2{u}_j$. For positive singular values, the pairing also satisfies $A^{\ast }{u}_j={\sigma }_j{v}_j$. Thus $u_1=(1,0{)}^T$ and $u_2=(0,1{)}^T$ describe the output axes of the preceding ellipse, while the right singular vectors describe its input directions. At zero singular values, division by ${\sigma }_j$ is impossible: complete the orthonormal bases independently in the null spaces of $A$ and $A^{\ast }$. Repeated singular values allow different matched orthonormal bases within the corresponding eigenspaces. The same positive pairing extends to a compact operator through its singular value system.

Even distinct singular values do not fix arrow signs: replacing both $u_j$ and $v_j$ by their negatives preserves $A{v}_j={\sigma }_j{u}_j$ and the outer product $u_j{v}_j^{\ast }$. Reversing only one arrow would change that term's sign. The paired directions, rather than independently chosen arrow orientations, encode the matrix.

The lever arm of a nonzero force about an origin is the shortest distance from that origin to the line through the application point parallel to the force. Its length is $\ell =|\mathbf{r}\times \hat{\mathbf{F}}|$, with unit vector $\hat{\mathbf{F}}=\mathbf{F}/|\mathbf{F}|$. Thus torque magnitude is $F\ell$, maximal at a right angle for fixed $r$ and $F$, and zero if the line passes through the origin. Sliding the application point along that line leaves torque unchanged, since $(\mathbf{r}+s\hat{\mathbf{F}})\times \mathbf{F}=\mathbf{r}\times \mathbf{F}$. The same perpendicular-distance construction gives angular momentum magnitude $p\ell$ for a line parallel to nonzero momentum through the particle.

A low-rank approximation replaces an $m\times n$ matrix $A$ by a matrix $B$ of matrix rank at most a chosen integer $r$, with $0\le r\le \min (m,n)$, while making an explicitly chosen error norm small. The point is to describe many entries using a few shared patterns: an image compression may retain broad brightness patterns while discarding finer ones. The approximation is useful when the saved storage or computation justifies the lost detail. A small matrix rank alone says nothing about the error.

The compact representation is a sum of outer products, equivalently a product of two thin matrices:Here $x_j,y_j$ are the columns of $X,Y$, and the star denotes the conjugate transpose; over the real numbers use the transpose. The rank bound for a matrix product gives $\mathrm{rank}B\le r$. Conversely, choose a basis of the column space of $B$ as the columns of $X$, and put the coordinates of each column of $B$ into $Y^{\ast }$; pad with zero columns if the matrix rank is smaller than $r$. Each nonzero outer product is a rank-one matrix: it repeats a single column pattern at different strengths across the columns.

Keeping the two factors stores $r(m+n)$ scalars instead of $mn$. This saves space only when $r(m+n)<mn$. The usual singular value decomposition representation stores $r(m+n+1)$ scalars because it keeps the $r$ singular values separately; absorbing them into one factor removes that extra array. These are representation sizes at equal scalar precision, before file overhead. See Uschmajew and Vandereycken, §9.2: thin factors and their storage, and Illinois CS 357, Low-rank approximation: storing singular vectors and values.

The factors also give $Bz=X(Y^{\ast }z)$ without forming the full matrix $B$. Direct evaluation uses $nr+mr$ scalar multiplications, compared with $mn$ for a dense matrix applied to a vector. For $m=1000,n=800,r=10$, the two factors contain $18,000$ scalars rather than $800,000$, and the same counts describe these multiplication costs. Constructing the factors is a separate cost. Moreover, a low-rank approximation need not be a sparse matrix: the constant $8\times 8$ matrix whose entries are all $1$ has matrix rank $1$ despite having no zero entries.

For the standard unweighted problem, write a singular value decomposition with $p=\min (m,n)$ and singular values ${\sigma }_1\ge \cdots \ge {\sigma }_p\ge 0$. Retain the first $r$ paired left singular vectors and right singular vectors:The Eckart-Young-Mirsky theorem says that this same $A_r$ minimizes both of the following error norms over all matrices with matrix rank at most $r$:where ${\sigma }_{p+1}=0$ and an empty sum is zero. Here the Frobenius norm is $\Vert E{\Vert }_F=({\sum }_{i,j}|E_{ij}{|}^2{)}^{1/2}$, whereas the operator norm is specifically induced by the Euclidean norm, $\Vert E{\Vert }_2={\max }_{\Vert z{\Vert }_2=1}\Vert Ez{\Vert }_2$. The first aggregates errors in all entries; the second bounds the worst output error for a unit vector input. These formulas apply to real and complex matrices. ^[ref]

For $A\ne 0$, a relative Frobenius norm tolerance $\epsilon$ is met precisely when the discarded squared singular values sum to at most ${\epsilon }^2{\sum }_j{\sigma }_j^2$. Thus one can choose the smallest suitable $r$ before reconstructing the matrix. Many individually small discarded singular values can still have a large combined error. An operator norm tolerance instead tests only the largest discarded singular value, ${\sigma }_{r+1}$. The two questions can therefore select different ranks.

The following stipulated $8\times 8$ matrix makes the tradeoff visible. Define four real vectors of length $8$:The vectors $h_j/\sqrt{8}$ are orthonormal, so this is already a singular value decomposition written as outer products, with positive singular values $8,3,1,1/2$. All entries lie between $7/16$ and $25/16$ and can be displayed as brightnesses. The first component is uniform brightness, the second distinguishes broad blocks, and the last two supply finer alternating patterns. This is a designed arithmetic example, not measured image data.

Here $A_1=h_0{h}_0^T$ has every entry equal to $1$, while $A_2=h_0{h}_0^T+3h_1{h}_1^T/8$ has entries $11/8$ within the two diagonal blocks and $5/8$ in the other two blocks. For instance, $A_{11}=25/16$ becomes $(A_2{)}_{11}=11/8$: its error is $3/16$. The full residual is $h_2{h}_2^T/8+h_3{h}_3^T/16$, whose Frobenius norm is $\sqrt{1+(1/2{)}^2}=\sqrt{5}/2$ and whose operator norm is $1$. Since $\Vert A{\Vert }_F=\sqrt{297}/2$, the relative Frobenius norm error is $\sqrt{5/297}\approx 12.97\%$. Two thin factors use $32$ scalars, or $34$ if the normalized left singular vectors, right singular vectors, and singular values are stored separately, instead of $64$ dense entries.

Retaining three components reduces the Frobenius norm error to $1/2$ while using $48$ scalars in two factors. Retaining all four reconstructs $A$ exactly but uses $64$ scalars, giving no saving in this generic representation; keeping the singular values separately uses $68$. Our particular sign patterns could be encoded more economically, but that would exploit an additional structure beyond the general low-rank approximation. The striking feature is that numerical error and storage can both be read from the decomposition, rather than judged only by how smooth the picture looks.

The constraint is deliberately matrix rank at most $r$. The truncation has $\mathrm{rank}{A}_r=\min (r,\mathrm{rank}A)$. If $\mathrm{rank}A<r$, the best approximation is $A$ itself, with zero error. Requiring matrix rank exactly $r$ can destroy attainment: for $A=\mathrm{diag}(1,0)$ and $r=2$, the matrices $B_{\delta }=\mathrm{diag}(1,\delta )$ have matrix rank $2$ for $\delta >0$, and both error norms equal $\delta$. Their error has infimum zero, but no rank-two matrix attains it. If $r\le \mathrm{rank}A$, the displayed $A_r$ does have matrix rank exactly $r$ and attains the same minimum under either constraint.

Optimality also does not always mean uniqueness. For $0<r<\mathrm{rank}A$, the best Frobenius norm approximation is unique when ${\sigma }_r>{\sigma }_{r+1}$. ^[ref] A positive tie across the cutoff permits different choices: for $A=\mathrm{diag}(3,3,1)$ and $r=1$, both $\mathrm{diag}(3,0,0)$ and $\mathrm{diag}(0,3,0)$ have Frobenius norm error $\sqrt{10}$ and operator norm error $3$. More generally, $3w{w}^T$ is optimal for every real unit vector $w$ in the first two coordinate directions. Rotating which tied direction is kept changes the approximation. Repeated singular values entirely inside the retained group do not change its sum of outer products. At $r=0$ the only feasible matrix is zero; at $r\ge \mathrm{rank}A$ the unique zero-error approximation is $A$, even if some singular values are tied at zero.

A strict gap does not guarantee uniqueness for the operator norm. For $A=\mathrm{diag}(3,1)$ and $r=1$, every $B_t=\mathrm{diag}(t,0)$ with $2\le t\le 4$ has matrix rank $1$ and error $\Vert A-B_t{\Vert }_2=\max (|3-t|,1)=1$, which is optimal. Only $t=3$ minimizes the Frobenius norm, since $\Vert A-B_t{\Vert }_F=\sqrt{(3-t{)}^2+1}$. An unchanged worst direction can hide extra error in another direction under the operator norm.

Finally, these guarantees concern the stated norms and the sole matrix rank constraint. They do not establish that a discarded pattern is noise or irrelevant to a particular task: the finer patterns in our constructed example were put there intentionally. Adding entrywise weights or restrictions on the factors changes the problem, so the Eckart-Young-Mirsky theorem above does not automatically solve it. For historical context, the original 1936 matrix problem is documented in the existing Eckart-Young-Mirsky theorem section.

For incident light with linear polarization, an ideal linear polarizer transmitswhere $I$ is light intensity and $\theta$ is the angle between the incident electric field's line of oscillation and the polarizer's pass axis. The input and output are compared in the same medium. The classical explanation is an orthogonal projection: for a unit vector $\mathbf{u}$ along the pass axis,The dot product reduces the transmitted wave amplitude by $\cos \theta$. For a plane electromagnetic wave in the same medium, light intensity is proportional to squared electric field wave amplitude, giving Malus's law. For example, $\theta =60^{\circ }$ transmits one quarter of the incident light intensity, although the transmitted wave amplitude is one half. ^[ref].

Start with light already in linear polarization along the horizontal $0^{\circ }$ axis and let its light intensity be $I_0$. Two ideal absorbing linear polarizers with pass axes $0^{\circ }$ and $90^{\circ }$ give $I_0{\cos }^2{0}^{\circ }{\cos }^{2}90^{\circ }=0$. Insert a $45^{\circ }$ polarizer between them: the first leaves $I_0$ unchanged, the middle transmits $I_0/2$ with linear polarization along $45^{\circ }$, and the last transmits $I_0/4$ with linear polarization along $90^{\circ }$. Each angle in Malus's law is relative to the preceding output, so the final step uses $90^{\circ }-45^{\circ }=45^{\circ }$:

For these ideal absorbing polarizers, the rejected electric field component transfers energy to the filter. In the three-filter arrangement the middle absorbs the fraction $I_0/2$, and the last absorbs $I_0/4$, leaving $I_0/4$ transmitted. Conservation of energy holds at each step. Inserting the middle polarizer changes the linear polarization arriving at the last one; it does not create light. The effect illustrates why successive orthogonal projections cannot generally be replaced by a single comparison of the first and last axes.

The initially horizontal assumption matters. For unpolarized light of light intensity $I_0$, the first ideal linear polarizer transmits $I_0/2$: averaging ${\cos }^2\theta$ over uniformly distributed incident orientations gives $1/2$. The same $0^{\circ },45^{\circ },90^{\circ }$ sequence then transmits $I_0/8$. This extra first-stage factor does not apply to the already horizontally polarized input in the figure.

For three unit vectors, the squared Euclidean norm of their sum is $3+2S$, proving the displayed bound; equality holds exactly when $u+v+w=0$. For fixed unit vectors $u,v$ with $u+v\ne 0$, the Cauchy-Schwarz inequality gives the unique minimizing third vector $w=-(u+v)/\Vert u+v\Vert$, and the minimum is $u\cdot v-\sqrt{2+2u\cdot v}$. Three coplanar directions separated by $120$ degrees attain the global bound.

For a positive-definite symmetric matrix $A$, the ellipsoid $x^{T}Ax=1$ has nearest points to the origin exactly at $x={\lambda }_{\max }^{-1/2}u$, where $u$ ranges over unit vectors in the largest-eigenvalue eigenspace. Indeed, the spectral theorem for real symmetric matrices gives $x^{T}Ax\le {\lambda }_{\max }|x{|}^2$, with equality precisely in that eigenspace. A simple largest eigenvalue gives two opposite nearest points; multiplicity $m>1$ gives a scaled $(m-1)$-sphere of nearest points.

For bounded self-adjoint $T$, put $B=\beta I-T\ge 0$ with $\beta =\sup W(T)$. Unit vectors with $⟨B{u}_n,u_n⟩\to 0$ satisfy $\Vert B{u}_n{\Vert }^2\le \Vert B\Vert ⟨B{u}_n,u_n⟩$ by positive-form Cauchy-Schwarz inequality. Thus $\beta$ is an approximate eigenvalue. Applying the same argument to $T-\alpha I$ proves the result at $\alpha =\inf W(T)$.

In a right-handed reference frame, the rotation matrix from an orbital plane frame whose first axis points along the ascending node is $R_z(\Omega )R_x(I)$. Its columns are the orbital frame's orthonormal basis expressed in reference coordinates. The unit vector normal is $(\sin \Omega \sin I,-\cos \Omega \sin I,\cos I)$; its dot product with another such normal gives the cosine of their mutual inclination.

On the smooth boundary of an orientable surface, let $n$ be the chosen unit normal vector and $t$ the positively oriented unit tangent vector used by Stokes theorem. The outward conormal is $u=t\times n$. It is a unit vector, tangent to the surface and orthogonal to its boundary. This convention turns $(n\times v)\cdot d\mathbf{r}$ into $u\cdot vds$ by cyclic invariance of the scalar triple product.

Set $w=(x-y)/2$. The sum condition then gives $x=a+w$ and $y=a-w$, and their dot product isTherefore $|w{|}^2=|a{|}^2-c$. Since $c<|a{|}^2$, this is a strictly positive squared radius. The complete solution isEvery choice of unit vector $n$ satisfies both conditions, and every solution has this form. The endpoints lie on the sphere with center $a$ and radius $R$, and their midpoint is $a$. They are therefore opposite ends of one diameter of that sphere.

For the stated angle, $q=\cos (2\pi /3)=-1/2$, so $\Vert u+v\Vert =\sqrt{2+2q}=1$. The minimizing unit vector is therefore $w=-(u+v)$, andIndeed all three pairwise dot products equal $-1/2$, so the vectors lie in one plane, separated by $120$ degrees. This is also the global minimum pairwise dot-product sum of three unit vectors: for any three unit vectors,so $S\ge -3/2$, with equality exactly when $u+v+w=0$.

The triangle inequality for the two unit vectors gives $0<\Vert u+v\Vert \le 2$. Taking reciprocals and using the negative sign in $\lambda =-1/\Vert u+v\Vert$ yieldsIn fact equality would require $u=v$, by the equality condition in the Cauchy-Schwarz inequality. Since the two vectors here are linearly independent, the inequality is strict for every admissible pair.

Put $q=u\cdot v$ and $d=u+v$. The linear independence assumption implies $d\ne 0$. Since $w$ is a unit vector, the Cauchy-Schwarz inequality givesEquality holds exactly when $w$ points opposite to $d$. Its required unit length then fixes the scale:Here $\Vert u+v{\Vert }^2=2+2q$ because both vectors have unit norm. The minimizing vector is unique, and the minimum pairwise dot-product sum of three unit vectors for these fixed $u,v$ is

Two nonzero vectors are linearly independent when $ax+by=0$ implies $a=b=0$. Their span then has dimension two. If they are linearly dependent, one is a scalar multiple of the other, so their span is a nonzero line and has dimension one. Thus the requested dimensions are two and one, respectively.

The dot product and corresponding Euclidean norm areThe Cauchy-Schwarz inequality states $\boxed{|x\cdot y|\le \Vert x\Vert \Vert y\Vert }$. If $y=0$ this is immediate. Otherwise the nonnegativity of a squared norm, with $t=(x\cdot y)/\Vert y{\Vert }^2$, givesMultiplying by $\Vert y{\Vert }^2$ and taking square roots proves the inequality. Equality holds precisely when the vectors are linearly dependent. Expanding another squared norm and using the inequality givesand hence the triangle inequality $\boxed{\Vert x+y\Vert \le \Vert x\Vert +\Vert y\Vert }$.

In ${\mathbb{R}}^3$, the two vectors lie in a common plane. If their directions make angle $\alpha$, the law of cosines gives $\Vert x-y{\Vert }^2=\Vert x{\Vert }^2+\Vert y{\Vert }^2-2\Vert x\Vert \Vert y\Vert \cos \alpha$. Comparing with the dot-product expansion yieldsThe sketch shows the same relation as a projection: multiply the length of $x$ by the signed component of $y$ in its direction. That component is negative for obtuse $\alpha$.

A unit vector is a vector of Euclidean norm one, equivalently $u\cdot u=1$. In particular the dot product of two unit vectors is the cosine of their mutual angle.

The unit vectors of plane polar coordinates obey ${\dot{\mathbf{e}}}_r=\dot{\theta }{\mathbf{e}}_{\theta }$ and ${\dot{\mathbf{e}}}_{\theta }=-\dot{\theta }{\mathbf{e}}_r$, as follows by differentiating their Cartesian sine/cosine components. Since $\mathbf{x}=r{\mathbf{e}}_r$,At a regular point of nonzero speed, rotating this velocity vector through a right angle produces a normal vector. Thus, up to its orientation,Its scalar product with $\mathbf{x}$ gives the distance from the origin to the tangent line:Where $r$ is a differentiable function of $\theta$ and $\dot{\theta }\ne 0$, substitute $\dot{r}=(dr/d\theta )\dot{\theta }$ to obtainA choice of principal normal is undefined on a locally straight segment, but the perpendicular normal line and the tangent-distance formula remain meaningful.

For a central force, the torque vanishes. Use physical signed angular momentum $h=m{r}^2\dot{\theta }$, with $h\ne 0$. The preceding geometric formula implies $|h|=mpv$, where $v=|\dot{\mathbf{x}}|$. Conservation of energy gives $E=m{v}^2/2+V(r)$. Eliminate $v$ to obtain the pedal equation for a central-force orbit:Equivalently the radial energy equation isOn an outward radial branch, divide $\dot{\theta }=h/(m{r}^2)$ by $\dot{r}=\sqrt{2[E-V_{\mathrm{eff}}]/m}$ and integrate:On an inward radial branch the integrand acquires a minus sign. At a turning point the two branches must be joined; the quoted positive-root integral is a local branch representation, not a single global time-parametrized orbit.

For $V=c/r^2$ with $c>0$, defineA finite moving orbit has $E>0$. In the radial integral, set $s=r_{\min }/r$. Because $dr/r^2=-ds/r_{\min }$, the outward integral, measured from the radial minimum, isJoining the inward and outward branches givesThe orientation in time is determined by the sign of $h$. The repulsive inverse-square potential generates force $-V^{\prime }(r)=2c/r^3$, an outward inverse-cube central force, not an inverse-square force. The orbit approaches infinity along two asymptotic directions, turns once at $r_{\min }$ and escapes again. The total polar-angle sweep is $\pi /\beta <\pi$, and the repulsive scattering deflection isThe effective potential $C/r^2$ decreases strictly to zero, so it admits no finite circular or bounded orbit. The sketch shows this scattering in a repulsive inverse-square potential. In the radial case $h=0$, the formulas dividing by $h$ are inapplicable; the direction is fixed, the turning radius is $\sqrt{c/E}$ and the particle reverses on the same ray.

Expanding the cross product component by component gives its cross-product matrix:It is a real skew-symmetric matrix and annihilates $\mathbf{n}$. By the vector triple product,so $A^2=\mathbf{n}{\mathbf{n}}^T-I$ for a unit vector $\mathbf{n}$. Choose a unit vector $\mathbf{u}\perp \mathbf{n}$ and put $\mathbf{v}=\mathbf{n}\times \mathbf{u}$. Then $(\mathbf{u},\mathbf{v},\mathbf{n})$ is an orthonormal basis and $A\mathbf{u}=\mathbf{v}$, $A\mathbf{v}=-\mathbf{u}$, $A\mathbf{n}=0$. Thus the characteristic polynomial is $\lambda ({\lambda }^2+1)$ and the cross-product matrix spectrum isOver the complex numbers, corresponding eigenvectors are $\mathbf{n}$, $\mathbf{u}-i\mathbf{v}$, and $\mathbf{u}+i\mathbf{v}$. Only the zero eigenvalue has real eigenvectors.

For the specified third coordinate axis, direct multiplication givesIts zero eigenspace is $\mathrm{span}\{(0,0,1)\}$, and its $-1$ eigenspace is the whole plane $\mathrm{span}\{(1,0,0),(0,1,0)\}$. Hence the eigenvalues of $A^2$ are $-1,-1,0$, with every nonzero vector in the appropriate eigenspace an eigenvector.