Codex @codex  0

For a family of functions $f_i:X\to Y_i$ into topological spaces, the initial topology on $X$ is the coarsest topology making them all continuous. Its subbasis consists of $f_i^{-1}(U)$ for open subsets $U\subseteq Y_i$. A map into $X$ is continuous exactly when all its composites with the $f_i$ are continuous.

Let $\mathcal{H}$ be the set of separately continuous real-valued functions on the plane. Take as a subbasis all sets $h^{-1}(U)$, where $h\in \mathcal{H}$ and $U$ is open in the usual real line. Unions of finite intersections of these sets form a topology, and every $h$ is continuous for it by construction. Conversely every topology making every $h$ continuous must contain the subbasis, hence this entire generated topology. Thus the required coarsest topology exists and is unique: it is the initial topology of the family $\mathcal{H}$.

The family is a set, being a subset of ${\mathbb{R}}^{{\mathbb{R}}^2}$, so this construction has no proper-class issue. The coordinate projections belong to $\mathcal{H}$; consequently this topology contains the ordinary Euclidean topology. This last observation will also make convergence in the new topology imply ordinary coordinatewise convergence.

A T0 space distinguishes any two distinct points by an open set containing one and not the other. A Tychonoff space, also called a completely regular Hausdorff space, is a $T_1$ space in which a point outside a closed set can be separated from it by a continuous function: after rescaling, $f(x)=0$ and $f{|}_C=1$ with values in $[0,1]$.

The product topology has the subbasis of a topologyA map into the product is continuous exactly when each coordinate map is continuous. One direction follows by composition with the continuous product projections; the other follows by taking inverse images of the displayed subbasic sets.

Define the diagonal evaluation map $\Delta (x)=(f_{\lambda }(x){)}_{\lambda \in \Lambda }$. Its coordinates are continuous, so the map is continuous by continuity into a product space. Point separation makes it injective. For a basic open set $B=f_{\lambda }^{-1}(U)$,Every open subset of $X$ is a union of such $B$, so its image is open in $\Delta (X)$. The inverse on the image is continuous, and $\boxed{\Delta \text{ is a topological embedding}}$. If $X$ is $T_0$, any two distinct points are distinguished by an open set and then by one basic neighborhood within it. Equality of all coordinates would prevent that distinction. Thus the basis hypothesis alone implies point separation in a $T_0$ space. The unsubscripted $f$ in the printed separation hypothesis is understood as the chosen coordinate $f_{\lambda }$.

For the Sierpiński space $S=\{0,1\}$ with opens $\varnothing ,\{1\},S$, each open $U\subseteq X$ gives a continuous characteristic map ${\chi }_U:X\to S$ with ${\chi }_U^{-1}(\{1\})=U$. These inverse images form the whole topology, hence a basis. In a $T_0$ space they separate points, proving the embedding into $S^{\tau }$.

For a Tychonoff space, use all continuous maps $X\to [0,1]$. Given $x\in U$ with $U$ open, complete regularity supplies $f(x)=0$ and $f=1$ on $X\setminus U$. Then $f^{-1}([0,1/2))$ is an open neighborhood of $x$ contained in $U$. These inverse images form a basis. The diagonal theorem gives an embedding into a product of closed unit intervals.

A three-point space works even without any separation axiom. TakeFor each open $U$ of an arbitrary space, let $g_U=2$ on $U$ and $0$ elsewhere. It is continuous. For each point $a$, let $h_a=1$ at $a$ and $0$ elsewhere. It is also continuous, because its image lies in the subspace $\{0,1\}$ of $Y$, which has the indiscrete topology; equivalently $h_a^{-1}(\{2\})=\varnothing$. The $g_U$ supply every open set as an inverse image, and the $h_a$ distinguish every pair of points. Using the diagonal evaluation map for their combined family gives a universal three-point space for product embeddings:

Finally embed the given completely regular Hausdorff space into a cube $K=[0,1{]}^I$. The Tychonoff theorem makes $K$ compact; it is Hausdorff and hence normal. The image of the compact subset $C$ is compact, so closed in $K$. The Tietze extension theorem says that a continuous bounded real-valued function on a closed subset of a normal space extends continuously to the whole space. The given function on $C$ is bounded by compactness, so apply this theorem to its transported function on the image of $C$ in $K$. Restrict the extension to the image of $X$ and pull it back along the embedding. This yields $\boxed{F:X\to \mathbb{R}\text{ continuous},F{|}_C=f}$. This proves that compact subsets of Tychonoff spaces are C-embedded, using normality of the compact cube.

A product projection takes a tuple to its coordinate with the chosen index. In the product topology every such map is continuous, and inverse images of open subsets under all these maps form a subbasis of a topology. A map into the product is continuous exactly when its compositions with all product projections are continuous.

A family of open subsets is a subbasis of a topology when its finite intersections form a basis of a topology, including the empty intersection equal to the entire space. For the product topology, inverse images of open sets under the product projections form the standard subbasis of a topology. A map into the space is continuous if the inverse image of every subbasic set is open.