parcas ****** more DOUBLECHECKs with PARCAS !!!! * E_kin * x/y/z(t) of both * video of hex -> 100db implement ********* - improve ins_m_atoms (merge to general ins_atoms function maybe) * modified first: try rules to better distribute Si and C *** * general cleanup * type 1 square, type 2 radius! - clean up mdrun - config sanity checks - anneal -> from current to T with rate R - check virial calc, where does the - come from? - angular distribution - structurfactoranalyze (see nordlund paper) - filled atoms need atom attrib?!?!? or aattr pre prerun - bond_analyze to dump xyz file of atoms - improve diff calc - pthreads: * threads nur einmal oeffnen * verteilung auf laufende threads - potentials: * tersoff fast * meam - optimize code! * 6.56 * -> 6.16 (albe_potential_force_calc) unoptimized * -> 7.05 first variable optimization * -> 5.53 2nd variable mod (moving out of blocks) * -> 5.12 source c files (orig albe) /wo float-store (otherwise 9.x) * -> 5.00 source c, fpu_set() (orig albe) * -> 5.04 source c, fpu_set() (albe fast) (same w -finline-functions) * -> 4.55 source c, fpu(), albe orig, fast-math ;) * -> 4.12 source c, fpu(), albe fast * -> 4.06 source c, fpu(), albe fast, arch opts * -> 3.54 ^ + c,d,h,gamma opts * -> 3.50 ^ + static lists * -> 3.37 source c, albe f, arch opts, c,d,h,gamma + lowmem lists (test for bigger + atoms!) * -> 3.31(36) ^ + inline v_calc * -> 4.44 orig albe, lowmem, source c, arch opts todo: - listen ! estimate time - more in fast (also think about -> 'type') - inline simulation runs *************** - start with small nuclei and amorphous surrounding - tctrl only in outer regions - only 1 atom per timestep - melting exps (both, anneal + interface method) - interstitials: - sic ints (compare!) - more interstitial combinations - characterize interstitials by PCF, then inc temperature learn ***** - ab initio / dft itself (find good docs) - checkout abinit in the same time - checkout thight binding - of course md theory! write ***** - results >> thesis!