Canonical: https://knowlpedia.com/operator-algebras/tomita-takesaki-theorem/

# Tomita–Takesaki theorem

The Tomita–Takesaki theorem identifies the commutant through modular conjugation and proves invariance under modular evolution.

Let \(M\subseteq B(H)\) be a
[von Neumann algebra](https://knowlpedia.com/operator-algebras/von-neumann-algebra/) with
[cyclic](https://knowlpedia.com/operator-algebras/cyclic-vector/)
[separating vector](https://knowlpedia.com/operator-algebras/separating-vector/) \(\Omega\). Write
\(S=J\Delta^{1/2}\) for the [polar decomposition](https://knowlpedia.com/functional-analysis/polar-decomposition/) of its
[Tomita operator](https://knowlpedia.com/operator-algebras/tomita-operator/). The
**Tomita–Takesaki theorem** asserts
\[
JMJ=M'
\qquad\text{and}\qquad
\Delta^{it}M\Delta^{-it}=M\quad(t\in\mathbb R).
\]
Here \(J\) is the
[modular conjugation](https://knowlpedia.com/operator-algebras/modular-conjugation/), \(\Delta\) the
[modular operator](https://knowlpedia.com/operator-algebras/modular-operator/), and \(M'\) the
[commutant](https://knowlpedia.com/operator-algebras/commutant/). Consequently
\(\sigma_t^\Omega(x)=\Delta^{it}x\Delta^{-it}\) is a strongly continuous
one-parameter group of \(*\)-automorphisms of \(M\).
These conclusions are intrinsic to the standard pair \((M,\Omega)\):
\(J\) exchanges the algebra with its commutant, while the modular unitaries
normalize \(M\). In particular, the displayed automorphisms are defined for
every real \(t\), not merely on an analytic subalgebra.

## Hypotheses and setting

Cyclicity makes \(M\Omega\) dense, while separatingness makes
\(S_0(x\Omega)=x^*\Omega\) well-defined and ensures that \(M'\Omega\) is
dense. These are exactly the vector hypotheses used to construct the two
closed antilinear operators whose polar factors appear in the theorem.
General [von Neumann algebras](https://knowlpedia.com/operator-algebras/von-neumann-algebra/) are treated by faithful normal [semifinite weights](https://knowlpedia.com/operator-algebras/semifinite-weight/),
standard forms, or reduction to supports rather than by assuming that a cyclic
separating vector exists in an arbitrary representation.

## Proof architecture

The proof first compares the closures of
\(S_0(x\Omega)=x^*\Omega\) and
\(F_0(x'\Omega)=x'^*\Omega\). [Analytic continuation](https://knowlpedia.com/complex-analysis/analytic-continuation/) of matrix coefficients,
together with the polar decomposition, gives the inclusions
\(JMJ\subseteq M'\) and
\(\Delta^{it}M\Delta^{-it}\subseteq M\). Applying the same argument to the
commutant and to \(-t\) upgrades both inclusions to equalities.

## Consequences

The unitary group \(\Delta^{it}\) supplies canonical modular dynamics for the
faithful normal [vector state](https://knowlpedia.com/operator-algebras/vector-state/). The
conjugation \(J\) gives an intrinsic
anti-isomorphism between \(M\) and its commutant in standard position. The
weight form of the theorem yields the
[modular automorphism group](https://knowlpedia.com/operator-algebras/modular-automorphism-group/) of
every [normal semifinite
faithful weight](https://knowlpedia.com/operator-algebras/normal-semifinite-faithful-weight/), a central tool in the structure theory of type III
algebras.

## Conventions and scope

**Warning.** Replacing \(\Delta^{it}x\Delta^{-it}\) by
\(\Delta^{-it}x\Delta^{it}\) reverses the time parameter. Both conventions
occur when [modular operators](https://knowlpedia.com/operator-algebras/modular-operator/) or [inner products](https://knowlpedia.com/linear-algebra/inner-product/) are normalized differently;
formulas involving KMS boundary values must use one convention consistently.

## References

1. Masamichi Takesaki, *Theory of Operator Algebras I*, Springer, 1979. [DOI record](https://doi.org/10.1007/978-1-4612-6188-9). Relevant: Chapter VI, §1, the fundamental theorem of modular theory.
