Computing norms of free operators with matrix coefficients
The paper provides an interesting generalization of the \textit{Ch. A. Akemann} and \textit{Ph. A. Ostrand} [Am. J. Math. 98, 1015-1047 (1976; Zbl 0342.22008)] formula for norms of convolution operators \(\sum^n_{i=1}\alpha_i\lambda(g_i)\), \(\alpha_i \in \mathbb C\), to matrix valued cases. The author continues here the work started in his paper [Colloq. Math. 74, No. 2, 321-328 (1997; Zbl 0911.22001)] where he obtained upper bounds for norms of free operators with operator coefficients. Let \(g_1, \dots, g_n\) be the generators of the free product group \(\Gamma_n = \mathbb Z_2\ast \mathbb Z_2 \ast\dots \ast \mathbb Z_2\), let \(a_0,a_1,\dots,a_n\) be hermitian \(m\times m\) matrices and assume moreover that \(a_0\) is positive semidefinite. The author proves in Theorem 1.1 that \[ \left\| a_0 \otimes I+\sum^n_{i=1} a_i \otimes\lambda(g_i)\right\| _{\min} = \underset{z>0}{\inf}\left\| 2z+a_0+\sum^n_{i=1}z^{\frac{1}{2}}((I_m+(z^{-\frac{1}{2}}a_iz^{-\frac{1}{2}})^2)^{\frac{1}{2}}-I_m)z^{\frac{1}{2}}\right\| , \] where the infimum is taken over all positive definite invertible \(m\times m\) matrices \(z\) and \(n\geq 2\). In order to see that the norm of the operator is dominated by the expression on the right-hand side, the author uses a matrix valued Gabber's Lemma as he did in his paper cited above. Here he extends the infimum over the whole range of positive definite invertible matrices to reach the norm of the operator. The proof of the lower bound involves asymptotic eigenvectors. The complete proof is transferred to operators on the Fock space \(\mathcal{T}(\mathbb C^n)\) obtaining in Theorem 1.3 \[ \left\| a_0 \otimes I+\sum^n_{i=1} a_i \otimes\l_i+ a_i^\ast\otimes\l_i^\ast\right\| _{\min} = \underset{z>0}{\inf}\left\| z+a_0+\sum^n_{i=1}a_i^\ast z^{-1}a_i\right\| , \] where \(l_1,l_2,\dots,l_n\) and \(l_1^\ast,l_2^\ast,\dots,l_n^\ast\) are the canonical set of creation and annihilation operators. It is also explained how the above formulae can be used with matrices replaced by operators taken from \(\mathcal{B}(H)\) for an arbitrary Hilbert space \(H\). The author shows that his matrix valued formula can be used to compute the norm of finitely supported convolution operators on free groups and on Fock spaces. The paper includes in the last section some theoretical background which can be used to perform numerical computations.
- The norm of polynomials in large random and deterministic matrices
- Scattering in quantum dots via noncommutative rational functions
- Local laws for polynomials of Wigner matrices
- Eigenvalues of random lifts and polynomials of random permutation matrices
- A random matrix approach to the lack of projections in \(C_{\mathrm {red}}^{\ast}(\mathbb F_2)\)
- Can you compute the operator norm?
- A Short Computation of the Norms of Free Convolution Operators
- Norms of Free Operators
- Free operators with operator coefficients
- Dilations of unitary tuples
- On the computation of spectra in free probability
- Matrix concentration inequalities and free probability
- Spectra of infinite graphs via freeness with amalgamation
- Universality and sharp matrix concentration inequalities
- A new approach to strong convergence
- Matrix concentration inequalities and free probability. II: Two-sided bounds and applications
- Sharp concentration for sums of matrices with Markovian dependence through universality
- Random matrices. Abstracts from the workshop held June 7--12, 2026
This page was built for publication: Computing norms of free operators with matrix coefficients
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4254164)
