Skip to main content
added 347 characters in body
Source Link
Narutaka OZAWA
  • 10.1k
  • 1
  • 42
  • 50

For a commuting pair of C*-algebras $A,B\subset B(H)$, the multiplication map is injective if and only if there are no nonzero elements $a\in A$ and $b\in B$ with $ab=0$. In particular, if $A$ is simple (and non-degenerate on $H$), then the multiplication map is injective.

This is because the pure states on C*-algebras are "excised"; for any pure state $\phi$ on $A$, there is a net $e_i$ in $A$ with $0\le e_i\le 1$ and $\|e_i\|=1$ that satisfies $\|e_iae_i - \phi(a)e_i^2\|\to0$ for all $a\in A$. It follows that any pure states on $\phi$ on $A$ and $\psi$ on $B$ give rise to a state $\phi\times\psi$ on $C^*(A,B)$ as long as $ab\neq0$ for any nonzero $a\in A_+$ and $b\in B_+$ (which immediately implies $\|ab\|=\|a\| \|b\|$ for every pair $(a,b)\in A_+\times B_+$ by functional calculus). See "Another proof of Proposition 3.4.7" in my book with Brown (p. 82).

Also, at the algebraic level, it is known that if $A$ is a central simple algebra (any simple C*-algebra is central simple), then every ideal in the algebraic tensor product $A \otimes B$ (over the relevant field) is of the form $A \otimes I$ for some ideal $I$ in $B$. See e.g., Drozd & Kirichenko "Finite Dimensional Algebras" Theorem 4.3.2.

For a commuting pair of C*-algebras $A,B\subset B(H)$, the multiplication map is injective if and only if there are no nonzero elements $a\in A$ and $b\in B$ with $ab=0$. In particular, if $A$ is simple (and non-degenerate on $H$), then the multiplication map is injective.

This is because the pure states on C*-algebras are "excised"; for any pure state $\phi$ on $A$, there is a net $e_i$ in $A$ with $0\le e_i\le 1$ and $\|e_i\|=1$ that satisfies $\|e_iae_i - \phi(a)e_i^2\|\to0$ for all $a\in A$. It follows that any pure states on $\phi$ on $A$ and $\psi$ on $B$ give rise to a state $\phi\times\psi$ on $C^*(A,B)$ as long as $ab\neq0$ for any nonzero $a\in A_+$ and $b\in B_+$ (which immediately implies $\|ab\|=\|a\| \|b\|$ for every pair $(a,b)\in A_+\times B_+$ by functional calculus). See "Another proof of Proposition 3.4.7" in my book with Brown (p. 82).

For a commuting pair of C*-algebras $A,B\subset B(H)$, the multiplication map is injective if and only if there are no nonzero elements $a\in A$ and $b\in B$ with $ab=0$. In particular, if $A$ is simple (and non-degenerate on $H$), then the multiplication map is injective.

This is because the pure states on C*-algebras are "excised"; for any pure state $\phi$ on $A$, there is a net $e_i$ in $A$ with $0\le e_i\le 1$ and $\|e_i\|=1$ that satisfies $\|e_iae_i - \phi(a)e_i^2\|\to0$ for all $a\in A$. It follows that any pure states on $\phi$ on $A$ and $\psi$ on $B$ give rise to a state $\phi\times\psi$ on $C^*(A,B)$ as long as $ab\neq0$ for any nonzero $a\in A_+$ and $b\in B_+$ (which immediately implies $\|ab\|=\|a\| \|b\|$ for every pair $(a,b)\in A_+\times B_+$ by functional calculus). See "Another proof of Proposition 3.4.7" in my book with Brown (p. 82).

Also, at the algebraic level, it is known that if $A$ is a central simple algebra (any simple C*-algebra is central simple), then every ideal in the algebraic tensor product $A \otimes B$ (over the relevant field) is of the form $A \otimes I$ for some ideal $I$ in $B$. See e.g., Drozd & Kirichenko "Finite Dimensional Algebras" Theorem 4.3.2.

Source Link
Narutaka OZAWA
  • 10.1k
  • 1
  • 42
  • 50

For a commuting pair of C*-algebras $A,B\subset B(H)$, the multiplication map is injective if and only if there are no nonzero elements $a\in A$ and $b\in B$ with $ab=0$. In particular, if $A$ is simple (and non-degenerate on $H$), then the multiplication map is injective.

This is because the pure states on C*-algebras are "excised"; for any pure state $\phi$ on $A$, there is a net $e_i$ in $A$ with $0\le e_i\le 1$ and $\|e_i\|=1$ that satisfies $\|e_iae_i - \phi(a)e_i^2\|\to0$ for all $a\in A$. It follows that any pure states on $\phi$ on $A$ and $\psi$ on $B$ give rise to a state $\phi\times\psi$ on $C^*(A,B)$ as long as $ab\neq0$ for any nonzero $a\in A_+$ and $b\in B_+$ (which immediately implies $\|ab\|=\|a\| \|b\|$ for every pair $(a,b)\in A_+\times B_+$ by functional calculus). See "Another proof of Proposition 3.4.7" in my book with Brown (p. 82).