A note on generalization of bent Boolean functions (Q2025371)

In this paper the authors investigate the space of Boolean functions \(f: \mathrm{GF}(2)^n\to \mathrm{GF}(2)\), denoted \(B_n\). Let \(wt(x)\) denote the Hamming weight of a vector \(x\in \mathrm{GF}(2)^n\). For these functions, the usual Walsh-Hadamard transform \(W_f(a)\) (\(a\in \mathrm{GF}(2)^n\)) is modified so that some weights are preferred over others: for each \(p\) with \(0<p<1\), define \[ W_f^{(p)}(a) = \sum_{x\in \mathrm{GF}(2)^n} \rho^{wt(x)} (-1)^{f(x) + a\cdot x}, \] where \(\rho = \frac{p}{1-p}\). The function \[ \mu_p(x) = p^{wt(x)}(1-p)^{n-wt(x)}, \] defines a probability distribution on \(\mathrm{GF}(2)^n\). Note that \(p=1/2\) is the usual uniform probability distribution. Note the above definition of \(W_f^{(p)}\) makes sense even for complex \(p\) and \(\rho\), though one loses the probabilistic interpretation in that case. A function \(f\in B_n\) is called \textit{\(\mu_p\)-bent} if \(|W_f^{(p)}(a)|=(1+|\rho|^2)^{n/2}\) for all \(x\in \mathrm{GF}(2)^n\). The authors prove a number of technical identities for \(W_f^{(p)}\) and then use these to investigate various classes of Boolean functions. For example, the authors classify those affine functions that are \(\mu_p\)-bent. Another example is that of the Maiorana-McFarland functions. Let \(x,y\in \mathrm{GF}(2)^k\), \(\pi\) be a permutation of \(\mathrm{GF}(2)^k\), and let \(g\in B_k\) be any Boolean function. The Maiorana-McFarland function \(f=f_{\pi,g}\) defined by \(f(x,y) = x\cdot \pi(y)+g(y)\) on \(\mathrm{GF}(2)^{2k}\) is known to be bent if \(p=1/2\). In one of their main results, the authors almost, but don't quite, completely classify those (complex) \(p\) for which \(f_{\pi,g}\) is \(\mu_p\)-bent. In more detail, the authors give very explicit conditions on \(p\not= 1/2\) for which \(f_{\pi,g}\) is \textit{not} \(\mu_p\)-bent. However, they stop short of showing the existence of a \(p\not= 1/2\) for which \(f_{\pi,g}\) is \(\mu_p\)-bent, nor do they prove such a value cannot exist. For precise details, please see the paper itself.