Toda's theorem is a result in computational complexity theory that was proven by Seinosuke Toda in his paper "PP is as Hard as the Polynomial-Time Hierarchy"^[1] and was given the 1998 Gödel Prize.

The theorem states that the entire polynomial hierarchy PH is contained in P^PP; this implies a closely related statement, that PH is contained in P^#P.

#P is the problem of exactly counting the number of solutions to a polynomially-verifiable question (that is, to a question in NP), while loosely speaking, PP is the problem of giving an answer that is correct more than half the time. The class P^#P consists of all the problems that can be solved in polynomial time if you have access to instantaneous answers to any counting problem in #P (polynomial time relative to a #P oracle). Thus Toda's theorem implies that for any problem in the polynomial hierarchy there is a deterministic polynomial-time Turing reduction to a counting problem.^[2]

An analogous result in the complexity theory over the reals (in the sense of Blum–Shub–Smale real Turing machines) was proved by Saugata Basu and Thierry Zell in 2009^[3] and a complex analogue of Toda's theorem was proved by Saugata Basu in 2011.^[4]

The proof is broken into two parts.

• First, it is established that

${\displaystyle \Sigma ^{P}\cdot {\mathsf {BP}}\cdot \oplus {\mathsf {P}}\subseteq {\mathsf {BP}}\cdot \oplus {\mathsf {P}}}$

The proof uses a variation of Valiant–Vazirani theorem. Because ${\displaystyle {\mathsf {BP}}\cdot \oplus {\mathsf {P}}}$ contains ${\displaystyle {\mathsf {P}}}$ and is closed under complement, it follows by induction that ${\displaystyle {\mathsf {PH}}\subseteq {\mathsf {BP}}\cdot \oplus {\mathsf {P}}}$.

• Second, it is established that

${\displaystyle {\mathsf {BP}}\cdot \oplus {\mathsf {P}}\subseteq {\mathsf {P}}^{\sharp P}}$

Together, the two parts imply

${\displaystyle {\mathsf {PH}}\subseteq {\mathsf {BP}}\cdot \oplus {\mathsf {P}}\subseteq {\mathsf {P}}\cdot \oplus {\mathsf {P}}\subseteq {\mathsf {P}}^{\sharp P}}$