In crystal plasticity there still is a major gap between what we know about single dislocations and how the macroscopic plastic characteristics emerge from the collective behaviour of many dislocations. Continuum dislocation dynamics has been an important step towards closing this gap in providing a kinematic framework for evolving tensorial dislocation density variables. However, progress on the constitutive, i.e., kinetic modelling has remained scarce and rather conceptually [1]. In the current work we show how the correlation tensors employed for the local density approximation of the average dislocation energy in [1] may be obtained from ensemble averaged discrete dislocation data. We achieve this by means of the maximum information entropy principle which is used to construct a higher dimensional correlation function from spatially projected dislocation alignment and pair density tensors, which may be directly extracted from discrete dislocation dynamics simulations [2]. Examples of this closure are provided, and the advantages over a conventional reconstruction of the correlation function in terms of truncated series expansions are discussed. Furthermore, we discuss the ensuing local density approximation of the dislocation energy. From such an approximation, evolution equations for the dislocation density variables may be derived by standard methods from continuum thermodynamics. REFERENCES [1] M. Zaiser. “Local density approximation for the energy functional of three-dimensional dislocation systems.” In: Phys. Rev. B 92 (17 Nov. 2015), p. 174120. doi: 10.1103/PhysRevB.92.174120 [2] T. Hochrainer, B. Weger, and S. Gupta. “Making sense of dislocation correlations.” In: Materials Theory 6.1 (Feb. 2022). doi: 10.1186/s41313-021-00040-6