# If P = NP, why wouldn't $\emptyset$ and $\Sigma^*$ be NP-complete?

Apparently, if ${\sf P}={\sf NP}$, all languages in ${\sf P}$ except for $\emptyset$ and $\Sigma^*$ would be ${\sf NP}$-complete.

Why these two languages in particular? Can't we reduce any other language in ${\sf P}$ to them by outputting them when accepting or not accepting?

As there are no strings in $\emptyset$, any machine that computes it always rejects, so we can't map Yes-instance of other problems to anything. Similarly for $\Sigma^{\ast}$ there's nothing to map No-instances to.

You need a polynomial reduction from the problem $A$ to the problem $B$ if you want to prove that $B$ is "harder" than $A$. We build a polynomial reduction by transforming any instance $x$ of $A$ into an instance $f(x)$ of $B$ such that $x∈A$ iff $f(x)∈B$.

The function $f$ must and can be polynomial. If $\mathsf{P}=\mathsf{NP}$ and $A$ is a NP problem, then $f$ can itself solve the problem $A$ of the problem and embed any $x∈A$ into some element $y$ of $B$ and any $x\notin A$ into some element $z$ that is not in $B$.

If $B$ is either $∅$ or $Σ^*$ then $y$ or $z$ cannot exist, otherwise the reasoning above shows that $B$ is harder than $A$.

Just a note: the previous answers are ok, however you're not too far from the correct trivial reduction:

if $\sf{P} = \sf{NP}$ then any $L \in \sf{NP}$ is Karp reducible to the language $\{1\}$ (just map in polynomial time every $x \in L$ to 1, every $x \notin L$ to 0), which is trivially a sparse language

The converse direction: "if an $\sf{NP}$ complete language is Karp reducible to a sparse set then $\sf{P}=\sf{NP}$" is certainly more interesting and it is known as the Mahaney's theorem:

Let $c$ be a constant and $A$ be set such that for all $n$, $A$ has at most $n^c$ strings of length $n$. If $A$ is $\sf{NP}$-complete then $\sf{P}=\sf{NP}$.