Linear response
From the generalized susceptibility
The static linear response can be obtained from the frequency dependent generalized susceptibility \(\chi^{PH}_{\bar{a}b\bar{c}d}(\omega_0, \nu_n, \nu_m)\) or the one-time dependent bosonic propagator \(\langle (\bar{a}b)(\tau) (\bar{c}d) \rangle\) by summing/integrating over frequency/time
From an applied external field
The linear response of a system is related to its response to the application of an external field. Hence, a brute force approach to compute the static linear response is to apply a general quadratic field \(F_{\bar{a}b} ( \bar{a} b + \bar{b} a)\) to the system and compute the derivative of the single particle density matrix \(\rho_{\bar{a}b} \equiv \langle \bar{a}b \rangle\) with respect to the field \(F_{\bar{a}b}\) in the limit \(F_{\bar{a}b} \rightarrow 0\). In other words we can compute the system response \(R_{\bar{a}b\bar{c}d}\) as
However, the central difference between the system response \(R_{\bar{a}b\bar{c}d}\) and the generalized susceptibility \(\chi^{PH}_{\bar{a}b\bar{c}d}\) is that the applied field has to be Hermitian. Whence, they are related by
and the response \(R_{\bar{a}b\bar{c}d}\) is symmetric in \(\bar{a}b\) transposes \(R_{\bar{a}b\bar{c}d} = R_{\bar{b}a\bar{c}d}\).
Can we reconstruct \(\chi\) from \(R\)? We use the (anti-)periodicity and anticommutation relations yielding
We can compute four relations involving pair-wise permutations of \(\chi^{PH}_{\bar{a}b\bar{c}d}\) from \(R_{\bar{a}b\bar{c}d}\):
however, these relations are not linearly independent and can not be solved for the \(\chi^{PH}\) components.