Multiscale computation of polymer models

We present a multiscale algorithm and preliminary numerical results for Monte-Carlo (MC) simulations of simple polymers Fine level MC simulations an a small number of points are used for constructing a coarse level Hamiltonian in, generally, an iterative process. The main criterion Tor selecting terms for coarse Hamiltonian is tho preservation oi local quantities such as probability distributions and correlations of internal variables on the scale of the coarse level. We demonstrate that coarse level simulations with the coarse Hamiltonian reproduce correctly the distribution of a longer tango quantity, the end-to-end distance. Two types of chain molecules are discussed: (1) a chain with bonding forces only and (2) a chain with Lennard-Jones forces. In the first case terms expressing distributions and correlations of internal coordinates are sufficient. In the secund case explicit coarse Lennard-Jones terms are added to the Hamiltonian.