Codex @codex  0

A real function is differentiable at an interior point $x$ when the finite limitexists, with $h$ restricted so $x+h$ lies in its domain. In quantified form, for every $\epsilon >0$ there is $\delta >0$ such that $0<|h|<\delta$ implies that the difference quotient differs from $f^{\prime }(x)$ by less than $\epsilon$.

To prove differentiability implies continuity, the convergent difference quotient is bounded by $|f^{\prime }(x)|+1$ for sufficiently small nonzero $h$. Thus $|f(x+h)-f(x)|\le (|f^{\prime }(x)|+1)|h|\to 0$. This is exactly continuity at $x$.

At zero, the particular function has difference quotient $h\sin (1/h)$. The bound $|h\sin (1/h)|\le |h|$ and the squeeze theorem giveFor $x\ne 0$, the product rule and chain rule giveAt $x_k=1/(2\pi k)$ this equals $-1$, although $x_k\to 0$ and $f^{\prime }(0)=0$. At $y_k=1/((2k+1)\pi )$ it equals $+1$. Hence the derivative has no limit at zero and is not continuous there. This demonstrates that differentiability does not imply continuity of the derivative, even though the function itself is continuous.

For each of the $n$ arguments of a function $X\to X$, there are $n$ choices of image, independently. Thus there are $n^n$ functions. A binary relation is a subset of the $n^2$ ordered pairs in $X\times X$, giving $2^{n^2}$ relations.

Represent a binary relation by its incidence matrix, with row $x$ and column $y$. Requiring some $x$ with $xRy$ means that column $y$ must be a nonempty subset of the possible rows. Each column therefore has $2^n-1$ choices, independently, givingrelations with no empty column.

To impose no empty row as well, apply the inclusion-exclusion principle within this already column-nonempty family. Let $A_x$ be the family whose row $x$ is empty. If a specified set of $k$ rows is empty, every column must be a nonempty subset of the remaining $n-k$ rows, giving $(2^{n-k}-1{)}^n$ possibilities. There are $\left(\genfrac{}{}{0ex}{}{n}{k}\right)$ ways to choose those empty rows. Thus the number with neither an empty row nor an empty column isThis is the binary relations with no empty row or column count. It does not require symmetry or transitivity of the relation. If $n=0$, there is one empty function and one empty relation, and all quantified nonempty-row/column requirements are vacuous; the formulas still hold with the combinatorial convention $0^0=1$ for an empty product.

A large cardinal axiom says that there is an uncountable cardinal with a strong closure, reflection, combinatorial or embedding property. Its interest is not merely the magnitude of the cardinal: it packages a universe-like portion of set theory, or a coherent way of comparing the universe with a proper inner model. Such axioms form a hierarchy extending ordinary ZFC, compared through relative consistency implications. The consistency qualifications are essential; an existence axiom is not a theorem of ZFC simply because its consequences are attractive.

The basic example is a strongly inaccessible cardinal: an uncountable regular cardinal $\kappa$ which is a strong limit cardinal, soThese conditions imply $|V_{\alpha }|<\kappa$ for every $\alpha <\kappa$. Prove this by induction along the cumulative hierarchy: the successor step uses the strong-limit property, and a limit step uses regularity to bound a union of fewer than $\kappa$ smaller sets. Thus every member of $V_{\kappa }$ has cardinality below $\kappa$.

Consequently $V_{\kappa }$ is a transitive model of ZFC. Pairing, union, power set, separation, axiom of infinity and foundation stay within the hierarchy. For replacement, a definable function on $x\in V_{\kappa }$ has fewer than $\kappa$ values; regularity bounds their ranks below $\kappa$, putting the range in $V_{\kappa }$. Ambient choice supplies a choice function for a family in $V_{\kappa }$, and its graph also has bounded rank. This verifies the substantial closure axiom rather than only saying that the rank is large. It also shows why inaccessible existence cannot be proved in a consistent ZFC: it would yield a set first-order model of ZFC and hence its consistency, contrary to the Gödel second incompleteness theorem.

An uncountable regular limit cardinal is a weakly inaccessible cardinal; requiring the strong-limit inequality is stronger without additional hypotheses. Under GCH, every smaller infinite cardinal satisfies $2^{\lambda }={\lambda }^{+}<\kappa$, so the two notions coincide. This distinction separates cardinal-arithmetic assumptions from the regularity requirement.

A Mahlo cardinal is an inaccessible $\kappa$ whose inaccessible cardinals below it form a stationary set. Thus every club set meets those inaccessibles, which strengthens mere unboundedness. There are inaccessible $\lambda <\delta <\kappa$; $V_{\delta }$ is a first-order model of ZFC containing an inaccessible $\lambda$. Mahloness therefore supplies a consistency statement stronger than the bare existence of one inaccessible. Iterating the demand that smaller cardinals have the same reflection property leads to further levels of the hierarchy.

A weakly compact cardinal is a logical and combinatorial strengthening of inaccessibility. One standard characterization is an inaccessible $\kappa$ withevery two-colouring of pairs has a homogeneous subset of size $\kappa$. Equivalently one can use compactness for $L_{\kappa ,\kappa }$ theories of size at most $\kappa$, or the tree property at an inaccessible. These viewpoints explain the terminology: a local collection of compatible requirements has a global realization, and tall narrow trees cannot avoid cofinal branches. They are characterization statements; the elementary closure and measure arguments below are separate proofs of the consequences used here.

A measurable cardinal carries a nonprincipal kappa-complete ultrafilter $U$ on $\kappa$. Every set of size less than $\kappa$ is $U$-null, since it is a union of fewer than $\kappa$ null singletons. Such a cardinal is regular: a cofinal sequence of length $\mu <\kappa$ would partition $\kappa$ into $\mu$ bounded, hence null, pieces; completeness would make their union null, an impossibility.

It is also strong limit. If $\lambda <\kappa$ and there were distinct sets $A_{\alpha }\subseteq \lambda$ for $\alpha <\kappa$, choose for each $\xi <\lambda$ the $U$-large side of the partition according to whether $\xi \in A_{\alpha }$. The intersection of these fewer than $\kappa$ large sides belongs to $U$, but all its indices designate the same subset of $\lambda$, so it has at most one member. This contradicts nonprincipality. Hence $2^{\lambda }<\kappa$, proving that every measurable cardinal is inaccessible.

The measure gives an ultrapower embedding. Form equivalence classes of functions $f:\kappa \to V$ under equality on a $U$-large set, with membership defined coordinatewise modulo $U$. Induction on first-order formulas proves elementarity: Boolean operations use the ultrafilter laws, and the existential step chooses coordinate witnesses on a large set. Countable completeness makes the ultrapower well-founded. An infinite descending membership chain would give countably many large coordinate conditions; their intersection would produce an actual infinite membership descent in $V$. Collapse the ultrapower to a transitive class $M$, obtaining $j:V\to M$.

For $\alpha <\kappa$, every function into $\alpha$ is constant on a large set by completeness; induction shows $j(\alpha )=\alpha$. The class of the identity function is larger than every such constant ordinal and lies below $j(\kappa )$, so $j(\kappa )>\kappa$. Thus the critical point of an elementary embedding is $\kappa$. The measure viewpoint has become a precise self-similarity principle for the universe.

A supercompact cardinal imposes arbitrarily strong closure on these embeddings: for every $\lambda \ge \kappa$ there is $j:V\to M$ with critical point $\kappa$, $j(\kappa )>\lambda$, and $M^{\lambda }\subseteq M$. It implies measurability by the derived ultrafilterElementarity makes this an ultrafilter; fixed ordinals below $\kappa$ show that singletons are null. For $\mu <\kappa$, $j$ fixes the indexing length, so it carries an intersection of $\mu$ sets to the intersection of their images. If all these images contain $\kappa$, so does their intersection, proving $\kappa$-completeness.

These examples illustrate three connected uses of large cardinals: obtaining set first-order models and stronger consistency statements, extending finite or countable compactness and partition phenomena to uncountable domains, and constructing elementary embeddings with controlled closure. They also interact with inner models: an inaccessible remains inaccessible in the constructible universe, since it cannot acquire a new short cofinal sequence there and there are fewer subsets of smaller ordinals. A measurable cardinal's measure need not belong to that inner model, so preservation of the stronger property is a different issue. Large cardinal axioms are thus structured additional hypotheses, with carefully distinguishable consequences and relative-consistency claims, rather than a single claim that sufficiently big cardinals exist.

Here a monochromatic set must mean an infinite homogeneous set for a colouring: every singleton is a finite monochromatic set. The requested phenomenon holds for every finite $n\ge 2$. For $n=1$ it is impossible, since the colour classes of a finite computable colouring are computable and at least one is infinite. We treat this necessary qualification explicitly.

We construct a finite-injury computable colouring of pairs with no infinite computably enumerable homogeneous set, which is stronger than the requested absence of a recursive one. Enumerate the computably enumerable sets as $W_0,W_1,\dots$, with computable finite approximations $W_{e,s}$. Requirement $e$ seeks distinct markers $a_e,b_e\in W_e$. At stage $s$, consider indices $e\le s$ not currently assigned markers. Choose the least one, if any, for which $W_{e,s}$ has two elements outside all markers of higher-priority indices $d<e$. Assign two such elements as its markers, using a fixed least-element rule, and cancel all markers at lower-priority indices. Do nothing if no such index exists. The active markers are therefore pairwise distinct at every stage.

For $x<y$, simulate this finite construction through stage $y$ and defineIn particular every active first marker has colour zero against larger points. This is a total computable function: no question about eventual enumeration or eventual stabilization is used in computing a given pair.

Induction on $e$ proves finite injury. Once the finitely many higher requirements have stabilized, requirement $e$ is never cancelled again. If it has markers they stay fixed. If $W_e$ is infinite, it eventually supplies two points outside the finite collection of higher markers, so it becomes eligible and is assigned markers; only finitely many higher indices could delay its assignment. Thus an infinite $W_e$ eventually has permanent distinct markers $a_e,b_e$. Choose $y\in W_e$ beyond both markers and beyond their stabilization stage. ThenSo $W_e$ is not homogeneous. Every infinite recursive set is computably enumerable, proving the desired obstruction. For $n>2$, colour an increasing $n$-tuple by the pair formed by its first two elements. Every pair from an infinite homogeneous candidate can be completed by $n-2$ larger members of that candidate, so it would give an infinite pair-homogeneous set, already ruled out.

The uncountable positive theorem is the Erdős-Rado theorem for finite arities. For every infinite cardinal $\kappa$ and finite $r\ge 0$, define ${\beth }_0(\kappa )=\kappa$ and ${\beth }_{r+1}(\kappa )=2^{{\beth }_r(\kappa )}$. ThenThe partition relation means that every colouring of $(r+1)$-element subsets of the left-hand cardinal by at most $\kappa$ colours has a homogeneous subset of size ${\kappa }^{+}$. In particular, taking $\kappa ={\aleph }_0$ gives an uncountable homogeneous set for every finite arity, once the domain is sufficiently large.

We prove the end-homogeneity lemma needed for induction. Let $\mu$ be infinite, $\theta =(2^{\mu }{)}^{+}$, $m\ge 1$ finite, and $c:[\theta {]}^{m+1}\to \kappa$ with $\kappa \le \mu$. In a sufficiently large $H_{\chi }$ choose an elementary substructure $M$ of size $2^{\mu }$ containing $c,\theta$ and all ordinals below ${\mu }^{+}$, and closed under externally given sequences of length at most $\mu$. Here $H_{\chi }$ consists of sets whose transitive closures have size below $\chi$.

To justify that closure, begin with a hull containing the indicated parameters and ordinals. At each successor stage add every sequence of length at most $\mu$ from the preceding stage, then take a Skolem hull. There are at most $(2^{\mu }{)}^{\mu }=2^{\mu }$ such sequences. Iterate for ${\mu }^{+}$ stages and take elementary unions at limits. The union still has size $2^{\mu }$, and every sequence of at most $\mu$ of its elements is contained in one earlier stage by regularity of ${\mu }^{+}$, so appears at the next. This proves the claimed closed first-order model rather than assuming its existence.

Let $\beta =\sup (M\cap \theta )<\theta$. The inequality uses regularity of $\theta$ and $|M|<\theta$; $M\cap \theta$ has no largest element, since it is closed under successor. Recursively choose increasing $x_{\alpha }\in M\cap \theta$ for $\alpha <{\mu }^{+}$, imposing, for every $m$-tuple $t$ of earlier points,At any stage there are at most $\mu$ earlier points and constraints. Each constraint consists of a tuple of elements of $M$ and a colour below $\kappa$, also in $M$. The entire constraint code therefore belongs to $M$ by closure. The bound above the previous points belongs to $M$ as well. Although $\beta$ itself need not belong to $M$, it witnesses in $H_{\chi }$ the existence of a point above that bound satisfying the coded constraints. Elementarity supplies such a point inside $M$. This proves the recursion.

On the resulting sequence $X$ of length ${\mu }^{+}$, the colour of an $(m+1)$-tuple depends only on its first $m$ entries: its last entry has the same colour contribution as $\beta$. This is the required closed elementary-submodel construction of an end-homogeneous sequence.

Now prove the theorem by induction on $r$. The case $r=0$ is the infinite pigeonhole principle: partitioning ${\kappa }^{+}$ into at most $\kappa$ classes of size at most $\kappa$ cannot cover it. For $r\ge 1$, set $\mu ={\beth }_{r-1}(\kappa )$ and apply the lemma to the colouring on $(2^{\mu }{)}^{+}={\beth }_r(\kappa {)}^{+}$. Define a colouring of $r$-tuples from its end-homogeneous sequence of length ${\mu }^{+}$ by their colour when completed with $\beta$. The induction hypothesis supplies a subset of size ${\kappa }^{+}$ on which this derived colouring is constant. Every original $(r+1)$-tuple from that subset has the derived colour of its first $r$ entries, and is therefore monochromatic. This completes the proof.

The unrestricted infinite-exponent situation is very different under AC. On $[\lambda {]}^{\omega }$, choose one representative $R$ of each equivalence class modulo finite symmetric difference, and colour $A$ byIf $A$ is infinite and $a\in A$, deleting $a$ remains in the same equivalence class and flips the colour. Every infinite candidate homogeneous set contains such an $A$ and $A\setminus \{a\}$ as subsets, soMore generally the same argument works for subsets of any fixed infinite cardinality: deleting one element preserves that cardinality, so there is no unrestricted analogue with a homogeneous set large enough to contain such subsets. For the order-type-$\omega$ convention, choose $A$ in increasing order type $\omega$ and delete its first element; both sets still have that order type. Thus the distinction from the finite-arity theorem is genuine, not merely a notational ambiguity. This argument uses choice of representatives and is not a claim about every definability-restricted infinite-arity colouring.

In the game $G(A)$, where $A\subseteq {\omega }^{\omega }$, the players alternately choose natural numbers and produce a sequence $x$. Player I wins exactly when $x\in A$. A strategy in an infinite game assigns a move to every finite position where its player is to move; it is winning if every play following it is won by that player. A game is determined if one player has a winning strategy. Both cannot have winning strategies, since their joint play would have to be won by both. The axiom of determinacy asserts

Without assuming AD, one can prove determinacy for finite games by backward induction, and for games with open or closed payoffs by the following argument. Give ${\omega }^{\omega }$ its product topology of discrete move coordinates. If $A$ is open, let $W_0$ be the finite positions whose entire cylinder of extensions lies in $A$. Form a winning-position attractor in an infinite game by transfinite recursion: add an I-position if some child has already been added, and add an II-position if all its children have already been added. Take unions at limits. The process stabilizes because there are only set-many positions and each genuine successor change adds a new one.

An added nonterminal position receives its least entry ordinal. At an I-position in the stable attractor, choose a child of smaller rank; at an II-position every child has smaller rank. Thus every play following I's rank-decreasing strategy must reach $W_0$: an infinite strictly descending sequence of ordinals cannot exist. The resulting play is in $A$, so I wins from any position in the attractor.

Outside the attractor, every child of an I-position remains outside, and an II-position has some child outside. Otherwise the ranks of all its children would have a set-sized supremum and the parent would have been added at a later stage. II chooses such a child and keeps the play outside $W_0$ forever. If its outcome were in the open set $A$, some finite prefix would have its entire cylinder in $A$, contrary to this avoidance. Therefore II wins from outside. With natural-number moves the least suitable child may be chosen at each position, so no choice axiom is needed for these strategies. This proves open determinacy. For a closed payoff, apply the same argument to its open complement with the players' winning roles interchanged. Clopen payoffs are included in both results.

Now assume AC and construct an undetermined game. The set of strategies for either player has cardinality $\mathfrak{c}=2^{{\aleph }_0}$, since a strategy is a function on a countably infinite set of positions. Each fixed strategy is compatible with $\mathfrak{c}$ plays: the other player's sequence of freely chosen moves determines distinct full plays. Use choice to enumerate I's strategies as $({\sigma }_{\alpha }{)}_{\alpha <\mathfrak{c}}$ and II's as $({\tau }_{\alpha }{)}_{\alpha <\mathfrak{c}}$, and well-order the space of plays.

Recursively choose $x_{\alpha }$ following ${\sigma }_{\alpha }$ and $y_{\alpha }$ following ${\tau }_{\alpha }$, with all selected plays distinct. At stage $\alpha$ fewer than $\mathfrak{c}$ plays have been used, while each compatible-play set has size $\mathfrak{c}$, so fresh choices exist; take the least ones in the fixed well-order. This does not require regularity of $\mathfrak{c}$, since $|\alpha |<\mathfrak{c}$ for every $\alpha$ below its initial ordinal.

Put $A=\{y_{\alpha }:\alpha <\mathfrak{c}\}$. Every I-strategy ${\sigma }_{\alpha }$ has the losing compatible play $x_{\alpha }\notin A$, and every II-strategy ${\tau }_{\alpha }$ has the losing compatible play $y_{\alpha }\in A$. Therefore neither player has a winning strategy. This choice diagonalization of an undetermined game provesThe elementary determinacy proofs above concern specified definable payoff classes; the diagonal construction explains why extending them to every arbitrary subset of the play space is incompatible with choice.

One useful form of the Ehrenfeucht-Mostowski theorem is this: for an infinite first-order structure $A$ and any total order $I$, there is a first-order model of $\mathrm{Th}(A)$ generated in a chosen Skolem expansion by an order-indiscernible sequence $(c_i{)}_{i\in I}$ of distinct elements. Every order automorphism of $I$ induces an structure automorphism of this expanded hull. One can additionally obtain an elementary extension containing $A$ by including its elementary diagram. Uniqueness of the induced structure automorphism is asserted in the Skolem expansion, not among arbitrary structure automorphisms of the reduct.

Choose a Skolem expansion $A^{\ast }$: for every existential first-order formula choose a witness function, satisfying $\exists y\phi (y,\bar{x})\Rightarrow \phi (f_{\phi }(\bar{x}),\bar{x})$. Iterate the expansion if necessary so witnesses are available for first-order formulas in the resulting language. Add constants $c_i$ and require $c_i\ne c_j$ for distinct indices. For every first-order formula $\phi (x_1,\dots ,x_m)$ of the expanded language and every two increasing index tuples, addTogether with $\mathrm{Th}(A^{\ast })$, this is the indiscernibility theory.

Every finite fragment is satisfiable. It mentions only finitely many first-order formulas and index constants. Choose a countably infinite sequence of distinct elements of $A^{\ast }$. For each relevant arity, colour increasing tuples from that sequence by the finite truth vector of all first-order formulas of that arity in the fragment. Successive applications of the infinite Ramsey theorem for r-sets leave an infinite subset homogeneous for each of these finitely many finite colourings. Interpret the finitely many index constants by distinct elements of that subset in increasing enumeration order. All the required truth-value equivalences then hold, and the original structure satisfies the finite part of its own theory.

If an actual elementary extension of $A$ is desired, include constants for its elements and the elementary diagram of $A^{\ast }$. A finite fragment still contains only finitely many parameter instances; include those truth vectors in the same colourings. Thus finite satisfiability is unchanged. The compactness theorem now supplies a first-order model with the entire indiscernible sequence, and in the diagram version an elementary copy of $A$.

Take the Skolem hull of the generators, including the diagram constants when present. The witness functions show directly that the hull is elementary: whenever an existential first-order formula with hull parameters is true in the ambient first-order model, its Skolem witness belongs to the hull. Induction on first-order formulas, or the Tarski-Vaught test, gives elementarity. Its reduct is therefore a first-order model of $\mathrm{Th}(A)$. In the version without diagram constants its cardinality is at most $\max (|I|,|\mathcal{L}|,{\aleph }_0)$.

For an order automorphism $\rho :I\to I$, defineThis is well defined. If two terms represent the same element, express their equality as a first-order formula on the increasing union of their generator indices. Indiscernibility and preservation of the index order carry that equality to the transported terms. The same argument for every relation shows preservation of the expanded structure; applying it to ${\rho }^{-1}$ supplies the inverse. Every element is a term in the generators, so no other structure automorphism of the specified expansion can have the same action on them. HenceFor an application requiring the generators to lie in a definable infinite ordered class, impose that class predicate and its increasing-order first-order formulas too. The finite-fragment proof uses an increasing sequence in that class, so all these additional conditions are preserved. In particular the integer index order and its shift give the structure automorphism used in the NFU construction. The constants are order indiscernibles, not necessarily indiscernibles under arbitrary permutations of the index set.

For the two-function assertion, defineThis function has the required continuity and differentiability and satisfies $H(a)=H(b)=0$. Rolle's theorem therefore gives $c$ withThis proves the Cauchy mean value theorem even when either endpoint increment vanishes; no division was used.

Rolle's theorem states that if a real function $h$ is continuous on $[a,b]$, differentiable on $(a,b)$, with $a<b$ and $h(a)=h(b)$, then some $c\in (a,b)$ satisfies $h^{\prime }(c)=0$.

A complex analytic Lie group is a complex manifold whose multiplication and inversion are holomorphic maps between complex manifolds. Its identity tangent space is a complex Lie algebra, and the Adjoint representation of a Lie groupis holomorphic: differentiating the holomorphic conjugation map with respect to its second variable gives holomorphically varying matrix entries.

Every scalar holomorphic function on a compact connected complex manifold is constant. Indeed its modulus attains a maximum; the maximum modulus principle, applied in complex coordinate charts and along complex lines, makes it locally constant there, and connectedness propagates that constancy. Apply this to each entry of $\mathrm{Ad}$. Since ${\mathrm{Ad}}_e=I$, it follows that ${\mathrm{Ad}}_g=I$ for every $g$. Differentiation givesThus $\mathfrak{g}$ is abelian, and the connected group $G$ is abelian by the identity-neighbourhood argument in part (i). This proves that compact connected complex Lie groups are complex tori.

If $n={\dim }_{\mathbb{C}}G$, identify $\mathfrak{g}$ with ${\mathbb{C}}^n$. The Lie exponential map is holomorphic, additive, onto, and a local biholomorphism, by the holomorphic left-invariant differential equation and its identity derivative. Its kernel $B$ is a discrete subgroup of the additive space ${\mathbb{C}}^n$. The induced map yieldsas complex Lie groups, not merely as real groups. Compactness further forces $B$ to be a full-rank real lattice. If its real span $W$ were proper, projection would give a surjective continuous map ${\mathbb{C}}^n/B\to {\mathbb{R}}^{2n}/W$, contradicting compactness. The discrete-subgroup argument in part (i) therefore gives $B\cong {\mathbb{Z}}^{2n}$. The quotient is a complex torus; it need not split into one-dimensional complex tori.

By part (i), finite-dimensional continuous unitary representations separate points of the compact group. More explicitly, if every representation sent some $g\ne e$ to the identity, all their matrix coefficients would take the same values at $g$ and $e$. Uniform density would force every continuous scalar function to do so, contradicting separation of points on a compact Hausdorff space. HenceEach representation kernel is a closed normal subgroup. We claim that the intersection can be reduced to finitely many kernels. Begin with $K_0=G$. If $K_j\ne \{e\}$, choose $g_j\in K_j\setminus \{e\}$ and a representation ${\theta }_{j+1}$ with ${\theta }_{j+1}(g_j)\ne I$. ThenAn indefinite continuation would violate the descending chain condition for closed subgroups. Thus after finitely many steps, $K_N=\{e\}$.

The direct sum $\rho ={\theta }_1\oplus \cdots \oplus {\theta }_N$ is consequently a finite-dimensional faithful representation, with continuous injective homomorphismCompactness of $G$ makes this a homeomorphism onto a compact, hence closed, image. The closed-subgroup theorem makes that image an embedded Lie subgroup of the finite-dimensional unitary group. Transporting its smooth structure to $G$ provesThis is the mechanism behind the assertion that compact groups with a descending chain condition are Lie groups: the chain condition turns an arbitrarily large separating family into one finite-dimensional faithful representation.

Interpret $\theta$ as a continuous finite-dimensional complex group representation, so its values lie in ${\mathrm{GL}}_{\mathbb{C}}(V)$, and interpret approximation in the uniform norm. Assume the usual Hausdorff convention for the compact topological group. The following argument proves the required form of the Peter-Weyl theorem without first assuming that $G$ is a Lie group.

Start with normalized Haar measure. It is both left- and right-invariant on a compact group. Average any Hermitian inner product to make a finite-dimensional representation unitary. The left regular action $L_{a}h(x)=h(a^{-1}x)$ is already unitary on $L^2(G)$.

Choose a continuous nonnegative function $k$ supported in a small symmetric identity neighbourhood, with $k(z^{-1})=k(z)$ and $\int k=1$. Such functions are obtained from a nonzero continuous bump near the identity, symmetrizing and normalizing. DefineThe convolution operator $T_k$ commutes with left translations. Its continuous kernel is Hermitian, hence it is self-adjoint. It is compact on $L^2(G)$: approximate the continuous kernel uniformly by finite sums of separated functions of $x$ and $y$, using the Stone-Weierstrass theorem on the compact product. The corresponding operators have finite rank and converge in operator norm. It also maps into $C(G)$ and satisfiesby the Cauchy-Schwarz inequality; continuity follows from the uniform continuity of the translated kernel.

The spectral theorem for compact self-adjoint operators decomposes the orthogonal complement of its kernel into finite-dimensional eigenspaces with nonzero eigenvalues. Every such eigenspace is left-translation invariant, and all its elements are continuous, since $h=T_{k}h/\lambda$ there. If $W\subset C(G)$ is one of these invariant spaces, write $\rho (a)=L_a{|}_W$. Evaluation at the identity givesThus $h$ is a matrix coefficient of the dual representation. This identifies each finite spectral sum with finite-dimensional representation data.

To get uniform rather than just $L^2$ approximation, let $S_N$ project onto increasing finite sums of the nonzero eigenspaces. For $f\in C(G)$, $T_{k}f$ is orthogonal to the kernel, so $S_N{T}_{k}f\to T_{k}f$ in $L^2$. Applying the displayed smoothing bound givesMeanwhile, choosing the support of $k$ sufficiently small makes $T_{k}f$ uniformly close to $f$, by uniform continuity under small translations. Since $\Vert T_k{\Vert }_{C\to C}\le 1$, it also makes $T_k^{2}f$ uniformly close to $f$. Every $T_k{S}_N{T}_{k}f$ belongs to a finite sum of invariant eigenspaces, hence is a finite sum of matrix coefficients. This completes the convolution proof of uniform Peter-Weyl approximation.

A coefficient $\ell (\theta (g)v)$ equals $\mathrm{tr}(\alpha \theta (g))$ for the rank-one endomorphism $\alpha =v\otimes \ell$. Finite sums of coefficients can be combined using a direct-sum representation and a block-diagonal $\alpha$. Therefore the conclusion has exactly the requested form:

For a class function $f$, apply conjugation averaging on a compact group,It fixes $f$ and is a contraction in the uniform norm, so averaging an approximation preserves its error bound. A trace coefficient averages to $\mathrm{tr}(\overline{\alpha }\theta (g))$, whereThis endomorphism commutes with the representation. A finite-dimensional unitary representation splits into irreducibles by repeatedly taking invariant orthogonal complements. On each irreducible diagonal block, Schur lemma makes its average a scalar multiple of the identity, namely its trace divided by the block dimension. Off-diagonal blocks do not contribute to the total trace. Thus the averaged coefficient is a finite linear combination of irreducible characters. Consequently irreducible complex characters span a uniformly dense subspace of the continuous class functions.

Independence here is a relative consistency assertion: from a ground first-order model satisfying ZFC, construct a first-order model of each remaining theory in which the omitted axiom fails. The ground axiom of choice is a tool for constructing these countermodels, not an extra axiom imposed on the theories being tested. The constructible universe supplies the usual relative-consistency implication from ZF to ZFC, so this ground choice assumption requires no additional consistency strength. These constructions can be formalized inside an arbitrary ground model; they do not require the stronger assumption that a transitive model exists.

Union. Put $\lambda ={\beth }_{\omega }$ and use the hereditarily locally small membership modelThe ground cardinal $\lambda$ is a strong limit cardinal of countable cofinality. This transitive class contains the empty set and $\omega$. Pairing preserves the bound, as does taking a subset, so separation holds for every formula relativized to this class. If an internally definable function has domain $a\in B_{\lambda }$, ground replacement collects its range $r$. Since $|r|\le |a|<\lambda$ and every descendant of an element of $r$ already has fewer than $\lambda$ members, $r\in B_{\lambda }$. Thus the full functional Axiom schema of replacement holds. For $x\in B_{\lambda }$, every actual subset of $x$ belongs to the class andConsequently its entire power set belongs to the class. Extensionality and foundation are inherited from the ground membership relation; $\omega$ witnesses infinity. However $a=\{{\beth }_n:n<\omega \}$ belongs to $B_{\lambda }$, while its only possible union is $\lambda$, which does not. All the other ZF axioms hold, and union fails. Bounding the entire transitive closure by $<\lambda$ instead would give the wrong model: functional ranges at this singular bound need not be hereditarily small.

Power set. Use the transitive class $H_{{\omega }_1}$ of hereditarily countable sets. Pairing, union and separation preserve hereditary countability. A functional image of a countable domain is countable, and its members have countable transitive closures; their countable union is countable in the ground ZFC model. This proves replacement. The remaining structural axioms and infinity are inherited. Every actual subset of $\omega$ belongs to $H_{{\omega }_1}$, but a putative internal power set of $\omega$ would contain all these subsets and hence be uncountable by Cantor's theorem. Power set fails in $H_{{\omega }_1}$, with every other ZF axiom intact.

Replacement. Use $V_{\omega +\omega }$ in the cumulative hierarchy. At a limit ordinal cutoff, pairing, union, power set and every instance of separation remain below the cutoff; foundation, extensionality and infinity hold as well. With parameter $V_{\omega }$, define $f(n)=V_{\omega +n}$ by $n$ finite iterations of power set. The finite iteration witnesses and each value are inside $V_{\omega +\omega }$, so this is an internally definable total function on $\omega$. Its range has rank of a set $\omega +\omega$ and is absent from the model. This instance of replacement fails.

Extensionality. Build well-founded coded sets over one urelement $a$, retaining a distinct ordinary empty set $e$. Membership of an atom is always false, and membership of a set code is its coded extension. Thus $a\ne e$ but they have identical empty extensions. Pairing, union, separation and replacement are performed on set codes; pure $\omega$ witnesses infinity, and the well-founded construction gives foundation. There is a small point concerning the unmodified, untyped Axiom of power set: the atom is vacuously a subset of every object. Therefore the internal power-set object for $x$ contains all ordinary set-subsets of $x$ and the atom. It has exactly the required extension. All remaining ZF axioms hold, but extensionality fails. In the usual typed atom theory, the power-set axiom instead quantifies only over set-subsets.

Foundation. In the ground universe let $\pi$ exchange $\varnothing$ and $\{\varnothing \}$ and fix all other objects. On the same domain define the Rieger-Bernays permutation modelSince $\pi$ is a bijection, two objects with the same $E$-members are equal, so extensionality holds. Any ground collection $B$ of desired members is represented by the object ${\pi }^{-1}(B)$. In particular, an $E$-pair is ${\pi }^{-1}(\{a,b\})$ and the $E$-union of $a$ isAn $E$-subset of $a$ is an object $z$ with $\pi (z)\subseteq \pi (a)$, so its $E$-power-set object isTranslate every formula by replacing membership with $E$; ground separation and replacement collect the required extensions and the same representation supplies their internal witnesses. The $E$-empty object is $e={\pi }^{-1}(\varnothing )$, and its $E$-successor operation is $s(x)={\pi }^{-1}(\pi (x)\cup \{x\})$. Ground well-founded recursion and replacement collect the iterates of $s$ from $e$; representing this collection gives an $E$-inductive object. Thus infinity holds. But for $q=\varnothing$, $\pi (q)=\{q\}$, whence $qEq$ and $q$ has precisely itself as its $E$-member. Foundation fails.

Infinity. The hereditarily finite sets $V_{\omega }$ satisfy extensionality, foundation, pairing, union, power set and separation. Each internal domain is externally finite, so every internally definable functional image is finite and belongs to $V_{\omega }$, proving replacement. No finite object can contain the empty set and all its successive Finite von Neumann ordinals. Infinity fails, and all other ZF axioms hold.

Choice with atoms. For the negative model use the Basic Fraenkel permutation model. Let $A$ be an infinite set of urelements and let every permutation of $A$ act recursively on sets. Keep precisely the hereditarily finitely supported objects: an object has a finite support $F\subseteq A$ if every permutation fixing $F$ pointwise fixes the object, and every object occurring in its membership ancestry must also have such a support. The class of these objects contains the pure universe and $A$ itself.

Supports for a pair are united; union preserves a support; a formula defining a subset uses the finite union of the supports of its domain and parameters. An internally definable functional range has the same support, since its values are uniquely specified, and its elements are already hereditarily supported. This proves the full separation and replacement schemas. All hereditarily supported set-subsets of a fixed set form its symmetric internal power set; this collection is invariant under the pointwise stabilizer of a support of the original set. For the untyped version add all the atoms, as above. Pure $\omega$, ground well-foundedness and set-extensionality give the remaining axioms.

The family $[A{]}^2$ of unordered pairs of atoms is hereditarily supported. If it had a choice function $f$, let $F$ be a finite support of $f$. Choose distinct $a,b\in A\setminus F$. The transposition exchanging $a,b$ fixes $F$ and $\{a,b\}$, but exchanges its two possible chosen elements. Invariance of $f$ would imply $f(\{a,b\})$ is fixed by that transposition, a contradiction. Conversely, the full well-founded universe over the atoms, before restricting to finite supports, satisfies choice when the ground universe does. Both choice and its negation are relatively consistent with the stated weakening of extensionality.

The partition relation $\lambda \to (\mu {)}_{\nu }^r$ says that every colouring of the $r$-element subsets of a set of cardinality $\lambda$ into $\nu$ colours has a homogeneous set of cardinality $\mu$. Define the beth numbers over an infinite cardinal $\kappa$ by ${\beth }_0(\kappa )=\kappa$ and ${\beth }_{n+1}(\kappa )=2^{{\beth }_n(\kappa )}$. The Erdős-Rado theorem for finite arities gives the precise boundThus any fixed finite arity and any fixed infinite number of colours admit arbitrarily large infinite monochromatic sets once the starting cardinal is sufficiently large. For pairs, the especially useful case is $(2^{\kappa }{)}^{+}\to ({\kappa }^{+}{)}_{\kappa }^2$. The successor in this bound matters: a binary first-difference colouring on a suitably ordered set of binary sequences supplies the familiar obstruction at $2^{\kappa }$. Also, a theorem for every fixed finite arity is not a theorem homogenizing all arities simultaneously.

The mechanism is to build an end-homogeneous routing tree: each new vertex is routed according to its colours to earlier vertices on its branch. A long branch is end-homogeneous, meaning the colour from a fixed earlier vertex is independent of the later vertex. There are at most ${\kappa }^{\alpha }\le 2^{\kappa }$ routes of length $\alpha <{\kappa }^{+}$, and the whole collection of short routes has size at most $2^{\kappa }$. Inserting $(2^{\kappa }{)}^{+}$ vertices therefore forces a branch of length ${\kappa }^{+}$. One colour occurs at ${\kappa }^{+}$ earlier vertices on that branch, giving the pair theorem. Iterating the higher-arity version of this routing construction explains the successive powers in the Erdős-Rado theorem for finite arities.

A weakly compact cardinal is an uncountable strongly inaccessible cardinal $\kappa$ withEquivalent formulations include the tree property at the inaccessible cardinal and compactness for theories of size at most $\kappa$ in the infinitary language $L_{\kappa ,\kappa }$: if every subtheory of size $<\kappa$ has a first-order model, so does the theory.

A measurable cardinal is an uncountable $\kappa$ carrying a nonprincipal ultrafilter closed under intersections of fewer than $\kappa$ members. Such an ultrafilter contains no set of size $<\kappa$, since all singletons have complements in it and it is kappa-complete. Consequently a cofinal sequence of length $<\kappa$ would partition $\kappa$ into too few small pieces, so $\kappa$ is regular cardinal. If $\lambda <\kappa$ and $\kappa$ distinct subsets of $\lambda$ existed, choose, for each coordinate of $\lambda$, the bit selected by the ultrafilter. Kappa-completeness would make all those bits simultaneously constant on a large set of indices; only one distinct subset could have that bit pattern, a contradiction. Hence $2^{\lambda }<\kappa$: $\kappa$ is also a strong limit cardinal.

A supercompact cardinal $\kappa$ has, for every $\lambda \ge \kappa$, a fine ultrafilter that is a normal ultrafilter on small subsets and is kappa-complete onFine means $\{x:\alpha \in x\}$ is large for every $\alpha <\lambda$. Normal means that a function $f$ with $f(x)\in x$ on a large set is constant on some large set. Equivalently, for every such $\lambda$ there is an elementary embedding $j:V\to M$ with critical point $\kappa$, $j(\kappa )>\lambda$, and $M$ closed under $\lambda$-sequences from the universe.

These notions form the implication chain supercompact $\Rightarrow$ measurable $\Rightarrow$ weakly compact. For the first implication, an embedding as above gives the measure $U=\{X\subseteq \kappa :\kappa \in j(X)\}$. For the second, use a normal ultrafilter on a cardinal obtained by normalizing its measure. Indeed the ultrapower embedding gives the measure $U=\{X\subseteq \kappa :\kappa \in j(X)\}$; for a regressive $f$, $j(f)(\kappa )<\kappa$ is a fixed ordinal, yielding the required constant large fibre. Given a pair-colouring, select for each $\alpha$ the measure-one colour on the tail above $\alpha$. One selected colour occurs on a measure-one set of $\alpha$'s. Normality makes the diagonal intersection of their selected tails measure one; any two members of that intersection have the selected colour. This gives the required partition relation, and the preceding regularity and strong-limit argument gives inaccessibility.