Jordan and Einstein frames are studied under the light of Hamiltonian formalism. Dirac's constraint theory for Hamiltonian systems is applied to Brans-Dicke theory. In both Jordan and Einstein frames, Brans-Dicke theory has four secondary first class constraints, whose constraint algebra is different. The Weyl (conformal) transformation, between the two frames, is not a canonical transformation, addressing quantum mechanical inequivalence as well.

A canonical analysis of RG improved action of the Einstein-Hilbert functional is performed. The gravitational and cosmological constants as function of the space-time coordinates are treated as external non-geometrical fields. Dirac's constraint analysis is performed, in the general case, up to secondary constraints. The constraints are second class and, in general, the problem appears to be technically complicated. This fact suggests studying the Dirac's constraint analysis of the related Brans-Dicke theory. It exhibits a Dirac's constraint algebra similar to Einstein's geometrodynamics except that the Poisson Brackets between Hamiltonian-Hamiltonian constraints is not only linear combination of the momentum constraints but also of a term note reducible to linear combination of the constraint and proportional to the extrinsic curvature. This shows that Branse-Dicke geometrodynamics is inequivalent to Einstein General Relativity geometrodynamics.

Light dark matter($\simeq 1-30~\rm{MeV}$) particles pair produced in electron-positron annihilation $ e^-e^+ \stackrel{\gamma}{\longrightarrow} \chi \bar{\chi}$ inside the supernova core can take away the energy released in the supernova SN1987A explosion. Working within the formalism of $q$-deformed statistics (with the average value of the supernovae core temperature(fluctuating) being $T_{SN} = 30~\rm{MeV}$) and using the Raffelt's criterion on the emissivity for any new channel $\dot{\varepsilon}(e^+ e^- \to \chi \overline{\chi}) \le 10^{19}~{erg~g^{-1}s^{-1}}$, we find that as the deformation parameter $q$ changes from $1.0$ (undeformed scenario) to $1.1$(deformed scenario), the lower bound on the scale $\Lambda$ of the dark matter effective theory varies from $3.3\times 10^6$ TeV to $3.2 \times 10^7$ TeV for a dark matter fermion of mass $m_\chi = 30~\rm{MeV}$. Using the optical depth criteria on the free streaming of the dark matter fermion, we find the lower bound on $\Lambda \sim 10^{8}~\rm{TeV}$ for $m_\chi = 30~\rm{MeV}$. In a scenerio,where the dark matter fermions are pair produced in the outermost sector of the supernova core (with radius $0.9 R_c \le r \le R_c$, $R_c (=10~\rm{km})$ being the supernova core radius or the radius of proto-neutron star), we find that the bound on $\Lambda$ ($\sim 3 \times 10^7$ TeV)obtained from SN cooling criteria (Raffelt's criteria) is comparable with the bound obtained from free streaming (optical depth criterion) for light fermion dark matter of mass $m_{\chi}=10 - 30$ MeV.