<br />
<script language="javascript" src="/js/readmail.js"></script>&nbsp;<style>
body{font-size: 14px;}</style>
<p>Dear Eduardo&nbsp;and Stefano,</p>
<p>Thank you for clear and careful explaination of my phonon problem. I&nbsp;am now doing some tests and getting more experiences, because I am a freshman in phonon calculation. </p>
<p>It seems to me that ASR can "force" the acoustic phonon at Gamma point to zero. A naive thought is that can we just ignore it? In this case, are there any problem that the slope of the phonon around Gamma is incorrect, because ASR cure the problem of k=0, but what about other points near it?&nbsp; </p>
<p>Regards</p>
<p>Wenxu Zhang</p>
<p>State Key Lab. of thin films and integrated&nbsp;devices&nbsp;,</p>
<p>Univ. of Elec. Sci. &amp; Tech. of China.</p>
<p>Chengdu, 610054, China</p>
<p>&gt;Eduardo&nbsp;Ariel&nbsp;Menendez&nbsp;Proupin&nbsp;wrote:<br />&gt;&gt;&nbsp;This&nbsp;happens&nbsp;because<br />&gt;&gt;&nbsp;a)&nbsp;0&nbsp;frecuencies&nbsp;at&nbsp;Gamma&nbsp;occurs&nbsp;because&nbsp;the&nbsp;energy&nbsp;is&nbsp;is&nbsp;invariant&nbsp;<br />&gt;&gt;&nbsp;with&nbsp;respect&nbsp;to&nbsp;arbitrary&nbsp;translations&nbsp;of&nbsp;the&nbsp;whole&nbsp;crystal&nbsp;(the&nbsp;same&nbsp;<br />&gt;&gt;&nbsp;translation&nbsp;for&nbsp;all&nbsp;atoms).&nbsp;From&nbsp;this&nbsp;assumption&nbsp;the&nbsp;accoustic&nbsp;sum&nbsp;<br />&gt;&gt;&nbsp;rule&nbsp;(ASR)&nbsp;is&nbsp;obtained&nbsp;theoretically.<br />&gt;&gt;&nbsp;b)&nbsp;The&nbsp;plane&nbsp;wave&nbsp;DFT&nbsp;calculations&nbsp;do&nbsp;not&nbsp;have&nbsp;this&nbsp;invariance&nbsp;<br />&gt;&gt;&nbsp;exactly,&nbsp;because&nbsp;of&nbsp;the&nbsp;plane&nbsp;wave&nbsp;cutoff.&nbsp;The&nbsp;energy&nbsp;is&nbsp;invariant&nbsp;not&nbsp;<br />&gt;&gt;&nbsp;for&nbsp;arbitrary&nbsp;translations,&nbsp;but&nbsp;for&nbsp;translations&nbsp;in&nbsp;the&nbsp;FFT&nbsp;grid.<br />&gt;&gt;&nbsp;c)&nbsp;I&nbsp;guess&nbsp;that&nbsp;the&nbsp;absolute&nbsp;translation&nbsp;invariance&nbsp;is&nbsp;lost&nbsp;in&nbsp;the&nbsp;<br />&gt;&gt;&nbsp;back&nbsp;and&nbsp;forth&nbsp;FFTs&nbsp;of&nbsp;the&nbsp;exchange-correlation&nbsp;potential,&nbsp;which&nbsp;is&nbsp;<br />&gt;&gt;&nbsp;non&nbsp;linear,&nbsp;and&nbsp;probably&nbsp;in&nbsp;some&nbsp;part&nbsp;related&nbsp;with&nbsp;the&nbsp;ultrasoft&nbsp;<br />&gt;&gt;&nbsp;pseudopotentials.&nbsp;Someone&nbsp;correct&nbsp;me&nbsp;if&nbsp;I&nbsp;am&nbsp;wrong,&nbsp;please.<br />&gt;everything&nbsp;correct,&nbsp;except&nbsp;for&nbsp;the&nbsp;clarification&nbsp;of&nbsp;a&nbsp;detail.<br />&gt;The&nbsp;problem&nbsp;is&nbsp;actually&nbsp;not&nbsp;in&nbsp;the&nbsp;finite&nbsp;kinetic&nbsp;energy&nbsp;cutoff&nbsp;but&nbsp;in&nbsp;<br />&gt;the&nbsp;fact&nbsp;that&nbsp;the&nbsp;exchange&nbsp;and&nbsp;correlation&nbsp;energy&nbsp;is&nbsp;computed&nbsp;in&nbsp;real&nbsp;<br />&gt;space&nbsp;on&nbsp;a&nbsp;discrete&nbsp;grid&nbsp;and&nbsp;hence&nbsp;the&nbsp;total&nbsp;energy&nbsp;is&nbsp;invariant&nbsp;(as&nbsp;you&nbsp;<br />&gt;said)&nbsp;only&nbsp;for&nbsp;translation&nbsp;in&nbsp;the&nbsp;FFT&nbsp;grid.<br />&gt;Increasing&nbsp;the&nbsp;charge&nbsp;density&nbsp;cutoff&nbsp;increases&nbsp;the&nbsp;grid&nbsp;density&nbsp;thus&nbsp;<br />&gt;making&nbsp;the&nbsp;integral&nbsp;more&nbsp;exact&nbsp;thus&nbsp;reducing&nbsp;the&nbsp;problem....&nbsp;<br />&gt;unfortunately&nbsp;rather&nbsp;slowly...<br />&gt;This&nbsp;problem&nbsp;is&nbsp;usually&nbsp;more&nbsp;severe&nbsp;for&nbsp;GGA&nbsp;&nbsp;than&nbsp;with&nbsp;LDA&nbsp;because&nbsp;the&nbsp;<br />&gt;GGA&nbsp;functionals&nbsp;have&nbsp;functional&nbsp;forms&nbsp;that&nbsp;vary&nbsp;more&nbsp;strongly&nbsp;with&nbsp;&nbsp;the&nbsp;<br />&gt;position;&nbsp;particularly&nbsp;so&nbsp;for&nbsp;isolated&nbsp;molecules&nbsp;or&nbsp;system&nbsp;with&nbsp;<br />&gt;significant&nbsp;portions&nbsp;of&nbsp;"vaccum"&nbsp;because&nbsp;in&nbsp;the&nbsp;exponential&nbsp;tail&nbsp;of&nbsp;the<br />&gt;charge&nbsp;density&nbsp;a)&nbsp;the&nbsp;finite&nbsp;cutoff&nbsp;&nbsp;(hence&nbsp;there&nbsp;is&nbsp;an&nbsp;effect&nbsp;due&nbsp;to&nbsp;<br />&gt;cutoff)&nbsp;induces&nbsp;oscillations&nbsp;in&nbsp;rho&nbsp;and&nbsp;b)&nbsp;the&nbsp;reduced&nbsp;gradient&nbsp;is&nbsp;diverging<br />&gt;<br />&gt;stefano<br />&gt;&nbsp;<br />&gt;&gt;&nbsp;You&nbsp;may&nbsp;increase&nbsp;the&nbsp;density&nbsp;cutoff&nbsp;(ecutrho)&nbsp;and&nbsp;see&nbsp;if&nbsp;the&nbsp;first&nbsp;<br />&gt;&gt;&nbsp;three&nbsp;frequencies&nbsp;get&nbsp;closer&nbsp;to&nbsp;zero.&nbsp;First&nbsp;do&nbsp;it&nbsp;with&nbsp;one&nbsp;of&nbsp;the&nbsp;<br />&gt;&gt;&nbsp;ESPRESSO&nbsp;examples,&nbsp;that&nbsp;run&nbsp;in&nbsp;minutes.&nbsp;I&nbsp;guess&nbsp;that&nbsp;I&nbsp;made&nbsp;that&nbsp;test&nbsp;<br />&gt;&gt;&nbsp;some&nbsp;years&nbsp;ago.<br />&gt;&gt;&nbsp;Using&nbsp;larger&nbsp;cutoffs&nbsp;increase&nbsp;a&nbsp;lot&nbsp;the&nbsp;calculation&nbsp;time.&nbsp;That&nbsp;is&nbsp;why&nbsp;<br />&gt;&gt;&nbsp;you&nbsp;have&nbsp;the&nbsp;tricky&nbsp;option&nbsp;to&nbsp;impose&nbsp;the&nbsp;ASR&nbsp;in&nbsp;postprocessing&nbsp;(matdyn.x).<br />&gt;&gt;<br />&gt;&gt;&nbsp;Practical&nbsp;way.&nbsp;If&nbsp;your&nbsp;"wrong"&nbsp;frecuencies&nbsp;are&nbsp;smaller&nbsp;than&nbsp;50&nbsp;cm^-1,&nbsp;<br />&gt;&gt;&nbsp;this&nbsp;is&nbsp;normal.&nbsp;Apply&nbsp;the&nbsp;ASR&nbsp;in&nbsp;the&nbsp;input&nbsp;for&nbsp;matdyn.&nbsp;Check&nbsp;that&nbsp;the&nbsp;<br />&gt;&gt;&nbsp;other&nbsp;frequencies&nbsp;do&nbsp;not&nbsp;change&nbsp;more&nbsp;than&nbsp;1&nbsp;or&nbsp;2&nbsp;cm^-1.&nbsp;Then,&nbsp;if&nbsp;&nbsp;this&nbsp;<br />&gt;&gt;&nbsp;makes&nbsp;you&nbsp;happy,&nbsp;go&nbsp;to&nbsp;write&nbsp;your&nbsp;report<br />&gt;&gt;<br />&gt;&gt;&nbsp;&gt;I&nbsp;have&nbsp;another&nbsp;question&nbsp;about&nbsp;the&nbsp;output:&nbsp;What&nbsp;is&nbsp;the&nbsp;&gt;difference&nbsp;<br />&gt;&gt;&nbsp;between&nbsp;the&nbsp;total&nbsp;magnetic&nbsp;moment&nbsp;and&nbsp;the&nbsp;&gt;absolute&nbsp;magnetic&nbsp;moment.<br />&gt;&gt;&nbsp;This&nbsp;has&nbsp;been&nbsp;asked&nbsp;in&nbsp;the&nbsp;forum&nbsp;before.&nbsp;Please,&nbsp;google&nbsp;for&nbsp;this.&nbsp;If&nbsp;I&nbsp;<br />&gt;&gt;&nbsp;remember&nbsp;well,&nbsp;the&nbsp;absolute&nbsp;magnetic&nbsp;moment&nbsp;is&nbsp;the&nbsp;integral&nbsp;of&nbsp;the&nbsp;<br />&gt;&gt;&nbsp;absolute&nbsp;value&nbsp;of&nbsp;the&nbsp;magnetization.<br />&gt;&gt;<br />&gt;&gt;&nbsp;Regards<br />&gt;&gt;&nbsp;Eduardo&nbsp;Menendez<br />&gt;&gt;<br />&gt;&gt;&nbsp;On&nbsp;Sat,&nbsp;Jun&nbsp;14,&nbsp;2008&nbsp;at&nbsp;2:57&nbsp;AM,&nbsp;<xwzhang /></xwzhang />&gt;&gt;&nbsp;<xwzhang /></xwzhang />&gt;&nbsp;wrote:<br />&gt;&gt;<br />&gt;&gt;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;Dear&nbsp;my&nbsp;pwscf&nbsp;friends,<br />&gt;&gt;<br />&gt;&gt;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;Recently,&nbsp;I&nbsp;am&nbsp;trying&nbsp;to&nbsp;do&nbsp;phonon&nbsp;calculations&nbsp;in&nbsp;Heusler.&nbsp;As&nbsp;a<br />&gt;&gt;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;first&nbsp;step,&nbsp;I&nbsp;tried&nbsp;to&nbsp;reproduce&nbsp;the&nbsp;results&nbsp;of&nbsp;a&nbsp;paper&nbsp;by<br />&gt;&gt;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;Zayak(PRB,&nbsp;72,&nbsp;054113,&nbsp;2005).&nbsp;I&nbsp;choose&nbsp;Ni2MnGe&nbsp;as&nbsp;an&nbsp;example.&nbsp;To<br />&gt;&gt;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;my&nbsp;surprise,&nbsp;the&nbsp;acoustic&nbsp;branches&nbsp;at&nbsp;Gamma&nbsp;point&nbsp;gives&nbsp;a<br />&gt;&gt;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;frequency&nbsp;of&nbsp;about&nbsp;1&nbsp;THz,&nbsp;which&nbsp;should&nbsp;be&nbsp;zero.&nbsp;Other&nbsp;branches<br />&gt;&gt;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;agree&nbsp;with&nbsp;the&nbsp;published&nbsp;data.<br />&gt;&gt;<br />&gt;&gt;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;I&nbsp;wonder&nbsp;what&nbsp;might&nbsp;be&nbsp;the&nbsp;reason?&nbsp;I&nbsp;attached&nbsp;my&nbsp;scf,&nbsp;and&nbsp;phonon<br />&gt;&gt;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;input&nbsp;files,&nbsp;and&nbsp;the&nbsp;phonon&nbsp;out&nbsp;put&nbsp;files.<br />&gt;&gt;<br />&gt;&gt;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;I&nbsp;have&nbsp;another&nbsp;question&nbsp;about&nbsp;the&nbsp;output:&nbsp;What&nbsp;is&nbsp;the&nbsp;difference<br />&gt;&gt;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;between&nbsp;the&nbsp;total&nbsp;magnetic&nbsp;moment&nbsp;and&nbsp;the&nbsp;absolute&nbsp;magnetic&nbsp;moment.<br />&gt;&gt;<br />&gt;&gt;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;Thank&nbsp;you&nbsp;and&nbsp;wish&nbsp;you&nbsp;a&nbsp;happy&nbsp;weekend!<br />&gt;&gt;<br />&gt;&gt;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;Wenxu&nbsp;Zhang<br />&gt;&gt;<br />&gt;&gt;<br />&gt;&gt;&nbsp;------------------------------------------------------------------------</p>