From c6b0e1aba4a68c8c00323eded0d50a599be4167a Mon Sep 17 00:00:00 2001 From: linjiez Date: Tue, 15 Jun 2021 20:46:58 +0200 Subject: [PATCH] fromJ --- thesis/2_Introduction/introduction.tex | 10 +++------- thesis/3/.DS_Store | Bin 10244 -> 10244 bytes thesis/6_backmatter/references.bib | 2 +- thesis/thesis.log | 2 +- thesis/thesis.pdf | Bin 31308682 -> 31308682 bytes 5 files changed, 5 insertions(+), 9 deletions(-) diff --git a/thesis/2_Introduction/introduction.tex b/thesis/2_Introduction/introduction.tex index 0f55a2b..c2b8445 100644 --- a/thesis/2_Introduction/introduction.tex +++ b/thesis/2_Introduction/introduction.tex @@ -1363,12 +1363,8 @@ $N$ replicas ($C_i, i = 1, 2, ..., N$) of the same system are simulated in paral The time evolution of each replica is independent with each other but exchanges of configurations between adjacent replicas $C_i$ and $C_j$, where $T_i < T_j$ and $i = j - 1$ are permitted at regular time intervals. The choice of the extreme temperatures $T_1$ and $T_N$ is very important for the algorithm to be optimal. -The lowest temperature ($T_1$) should be the one at which usual simulations are blocked and the highest -temperature ($T_N$) should be chosen so that all significant energy barriers can be overcome during the -simulation. Moreover, the temperatures between $T_1$ and $T_N$ must be chosen to lead to sufficient overlap -between the density of states of the adjacent replicas. Indeed, if this overlap is too small, the probability of -exchange is very low, which makes the PTMD simulations inefficient and leads to a bad exploration of the PES. -In contrast, if the overlap is too large, a large amount of redundant information will be produced, which will +The lowest temperature ($T_1$) should be the one at which usual simulations are blocked in basins and the highest temperature ($T_N$) should be chosen so that all significant energy barriers can be overcome during the simulation. Moreover, the temperatures between $T_1$ and $T_N$ must be chosen to lead to sufficient overlap between the density of states of the adjacent replicas. Indeed, if this overlap is too small, the probability of exchange is very low, which makes the PTMD simulations inefficient and leads to a bad exploration of the PES. +In contrast, if the overlap is too large, a large significant of redundant information will be produced, which will cost unnecessary computational resources. Configurations between two neighbouring replicas at different $T$ are exchanged based on the Metropolis–Hastings criterion with probability: \begin{align} @@ -1400,7 +1396,7 @@ all particles can be renormalized as follows: %======= %>>>>>>> 92023a10c3aa8b7dc4ace43987c1d571fb99a738 -\textbf{Global Optimization} refers to the determination of the lowest energy point on a PES, \textit{i.e.} the global minimum. As this latter usually +\textbf{Global optimization} refers to the determination of the lowest energy point on a PES, \textit{i.e.} the global minimum. As this latter usually includes a large number of stationary points, it is not straightforward to find the global minimum. Local optimization methods do not make it possible to cross the energy barriers between local minima. Therefore, a global optimization scheme such as MD or Monte Carlo simulations is needed to perform a more exhaustive exploration of the PES to get to the lowest energy minimum. diff --git a/thesis/3/.DS_Store b/thesis/3/.DS_Store index a4c8622ebf094a982a7feca4af325fa51bb70cf4..6620adeafd9d4d546c4dbfbfbd5795f1b4eb59a3 100644 GIT binary patch delta 74 zcmZn(XbIR5F2Hzla)dyGrbKnMp@FH6g0Ycdt&T#qxw(Okf{C$NZ7nBM8XF_ge eRdr2m-ONdX@{F@KD+oU0+sv-;i)E82Gcy2Bu@<5L delta 75 zcmZn(XbIR5F2HzVa)dyGhIn-ny2~dMLYEp~Z)S)i* zs80hL(ul@1p()L1P77Mniq^EDE$wJe2RhP;&UB$G-RMpadeV#D^r0{P=+6KKGKj$p zVJO2G&Im>_iqVW=EaMo@1ST?x$xLA?FpcTVU?#JOXEtUIbD76{7O;>-EM^HyS;lf! zu##1*W({ju$9gufkxgu73tQR7c6P9nUF>ELd)dc+4seh|9Oei|ImU5LaFSD;<_u>^ z;2h_S|UJKW_S_j$lW9`TqbJmneBdBICw@tXH|pAYzukNB8R z_>|B1oGJSd~$v{NG>23 zlncp)GICkDoE$8N$f0t1IZUn~hszOiq#Pwzlq<=V?*+lZVSA z*WpdMtPIGS>7UVmAA>;P<|vomY>K^|K9XjDiX69^8D35%#2R5|*8N<#LzuQjyHI{>Wqi~s-t delta 1260 zcmajVV|x$^0EXe~&2}x@9GddBY}>YNJZZgKwr#ss%eHxJvyM;aADj>Ow>$Rjn+&o4 znTkk4Qj(FJ6r>~-sYydx(vhAFWF!-r$wF4Lk)0d_kds{GCJ%YZM}7)WkU|uu2t_GI zaY|5DN@94S|lE6Y{ns&X~C zx?Dq!lB4Aqxu#r8t}WM*>&o@y`f>xgq1;GrEH{yx%FX2Fatpbo+)8dOw~^b*?d0}y z2f3r%N$xCnk-N&>~{_+5Mpgc$(EDw=~%ERR0@(6jPJW3ud zkCDg9|>>@;upcoE$IDmlwzjSI8^nRq|?ijl5P~C$E<`$Q$KN@@9F9yj9*NZ@zNeqa8tr6wk0p5u9wg3PC