Non-commutativity of the exponential spectrum (Q2400356)

For every pair \(a,b\) of elements of a unital Banach algebra \(A\), we have \(\sigma(ab)\setminus\{0\}=\sigma(ba)\setminus\{0\}\), where \(\sigma(a)\) stands for the spectrum of the element \(a\). \textit{R. Harte} [Proc. Am. Math. Soc. 58, 114--118 (1976; Zbl 0338.46043)] introduced the exponential spectrum \(\epsilon(a)\) which is equal to the set of all complex numbers \(\lambda\) such that \(\lambda-a\) is not in the connected component of the set of invertible elements of \(A\) containing the identity. In [Proc. R. Ir. Acad., Sect. A 92, No. 2, 229--238 (1992; Zbl 0741.47001)], \textit{G. J. Murphy} asked the question if the above formula is also true for the exponential spectrum. The authors solve this problem in the negative by showing that there exist \(a,b\) in the algebra \(C({\mathbb S}^4,M_2({\mathbb C}))\) of continuous functions from the unit sphere \({\mathbb S}^4= \{(z_0,z_1,z_2):| z_0|^2+| z_1|^2+| z_2|^2=1, {\text{ Im}}\,z_2=0\}\) into the \(2\times 2\) matrices with complex entries such that \(\epsilon(ab)\setminus\{0\}\not=\epsilon(ba)\setminus\{0\}\). The proof depends on the fact that the homotopy group \(\pi_4(\mathrm{GL}_2({\mathbb C}))\) is non-trivial.