Content-Type: multipart/mixed; boundary="-------------0309260838251" This is a multi-part message in MIME format. ---------------0309260838251 Content-Type: text/plain; name="03-441.comments" Content-Transfer-Encoding: 7bit Content-Disposition: attachment; filename="03-441.comments" AMS-Code:35Q55,37K40,37K05,70K40 ---------------0309260838251 Content-Type: text/plain; name="03-441.keywords" Content-Transfer-Encoding: 7bit Content-Disposition: attachment; filename="03-441.keywords" nonlinear Schr\"odinger equations, dynamics, solitons, Hamiltonian systems, stability, modulation, variational principle, energy estimates ---------------0309260838251 Content-Type: application/x-tex; name="tdyn.tex" Content-Transfer-Encoding: 7bit Content-Disposition: inline; filename="tdyn.tex" % First comes an example EPS file -- just ignore it and % proceed on the \documentclass line \begin{filecontents*}{ngamma2.eps} %!PS-Adobe-2.0 EPSF-1.2 %%BoundingBox: 91 3 322 146 %%HiResBoundingBox: 91.5625 3.1875 321.938 145.5 %%Creator: (Mathematica 4.2 for X) %%CreationDate: (Monday, September 15, 2003) (18:18:05) %%Title: Clipboard %%DocumentNeededResources: font Courier %%DocumentSuppliedResources: %%DocumentNeededFonts: Courier %%DocumentSuppliedFonts: %%DocumentFonts: Courier %%EndComments 0 148.688 translate 1 -1 scale 150 dict begin /Mfixwid true def /Mrot 0 def /Mpstart { MathPictureStart } bind def /Mpend { MathPictureEnd } bind def /Mscale { 0 1 0 1 5 -1 roll MathScale } bind def /Plain /Courier findfont def /Bold /Courier-Bold findfont def /Italic /Courier-Oblique findfont def /MathPictureStart { /Mimatrix matrix currentmatrix def gsave newpath Mleft Mbottom translate /Mtmatrix matrix currentmatrix def Plain Mfontsize scalefont setfont 0 setgray 0 setlinewidth } bind def /MathPictureEnd { grestore } bind def /MathSubStart { Momatrix Mgmatrix Mtmatrix Mleft Mbottom Mwidth Mheight 9 -2 roll moveto Mtmatrix setmatrix currentpoint Mgmatrix setmatrix 11 -2 roll moveto Mtmatrix setmatrix currentpoint 2 copy translate /Mtmatrix matrix currentmatrix def /Mleft 0 def /Mbottom 0 def 3 -1 roll exch sub /Mheight exch def sub /Mwidth exch def } bind def /MathSubEnd { /Mheight exch def /Mwidth exch def /Mbottom exch def /Mleft exch def /Mtmatrix exch def dup setmatrix /Mgmatrix exch def /Momatrix exch def } bind def /Mdot { moveto 0 0 rlineto stroke } bind def /Mtetra { moveto lineto lineto lineto fill } bind def /Metetra { moveto lineto lineto lineto closepath gsave fill grestore 0 setgray stroke } bind def /Mistroke { flattenpath 0 0 0 { 4 2 roll pop pop } { 4 -1 roll 2 index sub dup mul 4 -1 roll 2 index sub dup mul add sqrt 4 -1 roll add 3 1 roll } { stop } { stop } pathforall pop pop currentpoint stroke moveto currentdash 3 -1 roll add setdash } bind def /Mfstroke { stroke currentdash pop 0 setdash } bind def /Mrotsboxa { gsave dup /Mrot exch def Mrotcheck Mtmatrix dup setmatrix 7 1 roll 4 index 4 index translate rotate 3 index -1 mul 3 index -1 mul translate /Mtmatrix matrix currentmatrix def grestore Msboxa 3 -1 roll /Mtmatrix exch def /Mrot 0 def } bind def /Msboxa { newpath 5 -1 roll Mvboxa pop Mboxout 6 -1 roll 5 -1 roll 4 -1 roll Msboxa1 5 -3 roll Msboxa1 Mboxrot [ 7 -2 roll 2 copy [ 3 1 roll 10 -1 roll 9 -1 roll ] 6 1 roll 5 -2 roll ] } bind def /Msboxa1 { sub 2 div dup 2 index 1 add mul 3 -1 roll -1 add 3 -1 roll mul } bind def /Mvboxa { Mfixwid { Mvboxa1 } { dup Mwidthcal 0 exch { add } forall exch Mvboxa1 4 index 7 -1 roll add 4 -1 roll pop 3 1 roll } ifelse } bind def /Mvboxa1 { gsave newpath [ true 3 -1 roll { Mbbox 5 -1 roll { 0 5 1 roll } { 7 -1 roll exch sub (m) stringwidth pop .3 mul sub 7 1 roll 6 -1 roll 4 -1 roll Mmin 3 -1 roll 5 index add 5 -1 roll 4 -1 roll Mmax 4 -1 roll } ifelse false } forall { stop } if counttomark 1 add 4 roll ] grestore } bind def /Mbbox { 0 0 moveto false charpath flattenpath pathbbox newpath } bind def /Mmin { 2 copy gt { exch } if pop } bind def /Mmax { 2 copy lt { exch } if pop } bind def /Mrotshowa { dup /Mrot exch def Mrotcheck Mtmatrix dup setmatrix 7 1 roll 4 index 4 index translate rotate 3 index -1 mul 3 index -1 mul translate /Mtmatrix matrix currentmatrix def Mgmatrix setmatrix Mshowa /Mtmatrix exch def /Mrot 0 def } bind def /Mshowa { 4 -2 roll moveto 2 index Mtmatrix setmatrix Mvboxa 7 1 roll Mboxout 6 -1 roll 5 -1 roll 4 -1 roll Mshowa1 4 1 roll Mshowa1 rmoveto currentpoint Mfixwid { Mshowax } { Mshoway } ifelse pop pop pop pop Mgmatrix setmatrix } bind def /Mshowax { 0 1 4 index length -1 add { 2 index 4 index 2 index get 3 index add moveto 4 index exch get show } for } bind def /Mshoway { 3 index Mwidthcal 5 1 roll 0 1 4 index length -1 add { 2 index 4 index 2 index get 3 index add moveto 4 index exch get [ 6 index aload length 2 add -1 roll { pop Strform stringwidth pop neg exch add 0 rmoveto } exch kshow cleartomark } for pop } bind def /Mwidthcal { [ exch { Mwidthcal1 } forall ] [ exch dup Maxlen -1 add 0 1 3 -1 roll { [ exch 2 index { 1 index Mget exch } forall pop Maxget exch } for pop ] Mreva } bind def /Mreva { [ exch aload length -1 1 {1 roll} for ] } bind def /Mget { 1 index length -1 add 1 index ge { get } { pop pop 0 } ifelse } bind def /Maxlen { [ exch { length } forall Maxget } bind def /Maxget { counttomark -1 add 1 1 3 -1 roll { pop Mmax } for exch pop } bind def /Mwidthcal1 { [ exch { Strform stringwidth pop } forall ] } bind def /Strform { /tem (x) def tem 0 3 -1 roll put tem } bind def /Mshowa1 { 2 copy add 4 1 roll sub mul sub -2 div } bind def /MathScale { Mwidth Mheight Mlp translate scale /yscale exch def /ybias exch def /xscale exch def /xbias exch def /Momatrix xscale yscale matrix scale xbias ybias matrix translate matrix concatmatrix def /Mgmatrix matrix currentmatrix def } bind def /Mlp { 3 copy Mlpfirst { Mnodistort { Mmin dup } if 4 index 2 index 2 index Mlprun 11 index 11 -1 roll 10 -4 roll Mlp1 8 index 9 -5 roll Mlp1 4 -1 roll and { exit } if 3 -1 roll pop pop } loop exch 3 1 roll 7 -3 roll pop pop pop } bind def /Mlpfirst { 3 -1 roll dup length 2 copy -2 add get aload pop pop pop 4 -2 roll -1 add get aload pop pop pop 6 -1 roll 3 -1 roll 5 -1 roll sub dup /MsaveAx exch def div 4 1 roll exch sub dup /MsaveAy exch def div } bind def /Mlprun { 2 copy 4 index 0 get dup 4 1 roll Mlprun1 3 copy 8 -2 roll 9 -1 roll { 3 copy Mlprun1 3 copy 11 -3 roll /gt Mlpminmax 8 3 roll 11 -3 roll /lt Mlpminmax 8 3 roll } forall pop pop pop pop 3 1 roll pop pop aload pop 5 -1 roll aload pop exch 6 -1 roll Mlprun2 8 2 roll 4 -1 roll Mlprun2 6 2 roll 3 -1 roll Mlprun2 4 2 roll exch Mlprun2 6 2 roll } bind def /Mlprun1 { aload pop exch 6 -1 roll 5 -1 roll mul add 4 -2 roll mul 3 -1 roll add } bind def /Mlprun2 { 2 copy add 2 div 3 1 roll exch sub } bind def /Mlpminmax { cvx 2 index 6 index 2 index exec { 7 -3 roll 4 -1 roll } if 1 index 5 index 3 -1 roll exec { 4 1 roll pop 5 -1 roll aload pop pop 4 -1 roll aload pop [ 8 -2 roll pop 5 -2 roll pop 6 -2 roll pop 5 -1 roll ] 4 1 roll pop } { pop pop pop } ifelse } bind def /Mlp1 { 5 index 3 index sub 5 index 2 index mul 1 index le 1 index 0 le or dup not { 1 index 3 index div .99999 mul 8 -1 roll pop 7 1 roll } if 8 -1 roll 2 div 7 -2 roll pop sub 5 index 6 -3 roll pop pop mul sub exch } bind def /intop 0 def /inrht 0 def /inflag 0 def /outflag 0 def /xadrht 0 def /xadlft 0 def /yadtop 0 def /yadbot 0 def /Minner { outflag 1 eq { /outflag 0 def /intop 0 def /inrht 0 def } if 5 index gsave Mtmatrix setmatrix Mvboxa pop grestore 3 -1 roll pop dup intop gt { /intop exch def } { pop } ifelse dup inrht gt { /inrht exch def } { pop } ifelse pop /inflag 1 def } bind def /Mouter { /xadrht 0 def /xadlft 0 def /yadtop 0 def /yadbot 0 def inflag 1 eq { dup 0 lt { dup intop mul neg /yadtop exch def } if dup 0 gt { dup intop mul /yadbot exch def } if pop dup 0 lt { dup inrht mul neg /xadrht exch def } if dup 0 gt { dup inrht mul /xadlft exch def } if pop /outflag 1 def } { pop pop} ifelse /inflag 0 def /inrht 0 def /intop 0 def } bind def /Mboxout { outflag 1 eq { 4 -1 roll xadlft leadjust add sub 4 1 roll 3 -1 roll yadbot leadjust add sub 3 1 roll exch xadrht leadjust add add exch yadtop leadjust add add /outflag 0 def /xadlft 0 def /yadbot 0 def /xadrht 0 def /yadtop 0 def } if } bind def /leadjust { (m) stringwidth pop .5 mul } bind def /Mrotcheck { dup 90 eq { yadbot /yadbot xadrht def /xadrht yadtop def /yadtop xadlft def /xadlft exch def } if dup cos 1 index sin Checkaux dup cos 1 index sin neg exch Checkaux 3 1 roll pop pop } bind def /Checkaux { 4 index exch 4 index mul 3 1 roll mul add 4 1 roll } bind def /Mboxrot { Mrot 90 eq { brotaux 4 2 roll } if Mrot 180 eq { 4 2 roll brotaux 4 2 roll brotaux } if Mrot 270 eq { 4 2 roll brotaux } if } bind def /brotaux { neg exch neg } bind def /Mabsproc { 0 matrix defaultmatrix dtransform idtransform dup mul exch dup mul add sqrt } bind def /Mabswid { Mabsproc setlinewidth } bind def /Mabsdash { exch [ exch { Mabsproc } forall ] exch setdash } bind def /MBeginOrig { Momatrix concat} bind def /MEndOrig { Mgmatrix setmatrix} bind def /sampledsound where { pop} { /sampledsound { exch pop exch 5 1 roll mul 4 idiv mul 2 idiv exch pop exch /Mtempproc exch def { Mtempproc pop } repeat } bind def } ifelse /MFill { 0 0 moveto Mwidth 0 lineto Mwidth Mheight lineto 0 Mheight lineto fill } bind def /MPlotRegion { 3 index Mwidth mul 2 index Mheight mul translate exch sub Mheight mul /Mheight exch def exch sub Mwidth mul /Mwidth exch def } bind def /Mcharproc { currentfile (x) readhexstring pop 0 get exch div } bind def /Mshadeproc { dup 3 1 roll { dup Mcharproc 3 1 roll } repeat 1 eq { setgray } { 3 eq { setrgbcolor } { setcmykcolor } ifelse } ifelse } bind def /Mrectproc { 3 index 2 index moveto 2 index 3 -1 roll lineto dup 3 1 roll lineto lineto fill } bind def /_Mcolorimage { 7 1 roll pop pop matrix invertmatrix concat 2 exch exp 1 sub 3 1 roll 1 1 2 index { 1 1 4 index { dup 1 sub exch 2 index dup 1 sub exch 7 index 9 index Mshadeproc Mrectproc } for pop } for pop pop pop pop } bind def /_Mimage { pop matrix invertmatrix concat 2 exch exp 1 sub 3 1 roll 1 1 2 index { 1 1 4 index { dup 1 sub exch 2 index dup 1 sub exch 7 index Mcharproc setgray Mrectproc } for pop } for pop pop pop } bind def /Mimage { 4 index 4 index mul 1600 gt { image } { _Mimage } ifelse } def /Mcolorimage { 6 index 6 index mul 1600 gt { colorimage } { _Mcolorimage } ifelse } def /g { setgray} bind def /k { setcmykcolor} bind def /p { gsave} bind def /r { setrgbcolor} bind def /w { setlinewidth} bind def /C { curveto} bind def /F { fill} bind def /L { lineto} bind def /rL { rlineto} bind def /P { grestore} bind def /s { stroke} bind def /S { show} bind def /N {currentpoint 3 -1 roll show moveto} bind def /Msf { findfont exch scalefont [1 0 0 -1 0 0 ] makefont setfont} bind def /m { moveto} bind def /Mr { rmoveto} bind def /Mx {currentpoint exch pop moveto} bind def /My {currentpoint pop exch moveto} bind def /X {0 rmoveto} bind def /Y {0 exch rmoveto} bind def /pp_save save def /showpage {} def gsave /MacintoshRomanEncoding [ /.notdef/.notdef/.notdef/.notdef/.notdef/.notdef/.notdef/.notdef /.notdef/.notdef/.notdef/.notdef/.notdef/.notdef/.notdef/.notdef /.notdef/.notdef/.notdef/.notdef/.notdef/.notdef/.notdef/.notdef /.notdef/.notdef/.notdef/.notdef/.notdef/.notdef/.notdef/.notdef /space/exclam/quotedbl/numbersign/dollar/percent/ampersand/quotesingle /parenleft/parenright/asterisk/plus/comma/hyphen/period/slash /zero/one/two/three/four/five/six/seven /eight/nine/colon/semicolon/less/equal/greater/question /at/A/B/C/D/E/F/G /H/I/J/K/L/M/N/O /P/Q/R/S/T/U/V/W /X/Y/Z/bracketleft/backslash/bracketright/asciicircum/underscore /grave/a/b/c/d/e/f/g /h/i/j/k/l/m/n/o /p/q/r/s/t/u/v/w /x/y/z/braceleft/bar/braceright/asciitilde/.notdef /Adieresis/Aring/Ccedilla/Eacute/Ntilde/Odieresis/Udieresis/aacute /agrave/acircumflex/adieresis/atilde/aring/ccedilla/eacute/egrave /ecircumflex/edieresis/iacute/igrave/icircumflex/idieresis/ntilde/oacute /ograve/ocircumflex/odieresis/otilde/uacute/ugrave/ucircumflex/udieresis /dagger/degree/cent/sterling/section/bullet/paragraph/germandbls /registered/copyright/trademark/acute/dieresis/.notdef/AE/Oslash /.notdef/plusminus/.notdef/.notdef/yen/.notdef/.notdef/.notdef /.notdef/.notdef/.notdef/ordfeminine/ring/.notdef/ae/oslash /questiondown/exclamdown/logicalnot/.notdef/florin/.notdef/.notdef/guillemotleft /guillemotright/ellipsis/.notdef/Agrave/Atilde/Otilde/OE/oe /endash/emdash/quotedblleft/quotedblright/quoteleft/quoteright/divide/.notdef /ydieresis/Ydieresis/.notdef/currency/guilsinglleft/guilsinglright/.notdef/.notdef /daggerdbl/dotaccent/quotesinglbase/quotedblbase/perthousand/Acircumflex/Ecircumflex/Aacute /Edieresis/Egrave/Iacute/Icircumflex/Idieresis/Igrave/Oacute/Ocircumflex /.notdef/Ograve/Uacute/Ucircumflex/Ugrave/dotlessi/circumflex/tilde /macron/.notdef/.notdef/.notdef/ogonek/.notdef/.notdef/.notdef ] def /WindowsANSIEncoding [ /.notdef/.notdef/.notdef/.notdef/.notdef/.notdef/.notdef/.notdef /.notdef/.notdef/.notdef/.notdef/.notdef/.notdef/.notdef/.notdef /.notdef/.notdef/.notdef/.notdef/.notdef/.notdef/.notdef/.notdef /.notdef/.notdef/.notdef/.notdef/.notdef/.notdef/.notdef/.notdef /space/exclam/quotedbl/numbersign/dollar/percent/ampersand/quotesingle /parenleft/parenright/asterisk/plus/comma/hyphen/period/slash /zero/one/two/three/four/five/six/seven /eight/nine/colon/semicolon/less/equal/greater/question /at/A/B/C/D/E/F/G /H/I/J/K/L/M/N/O /P/Q/R/S/T/U/V/W /X/Y/Z/bracketleft/backslash/bracketright/asciicircum/underscore /grave/a/b/c/d/e/f/g /h/i/j/k/l/m/n/o /p/q/r/s/t/u/v/w /x/y/z/braceleft/bar/braceright/asciitilde/.notdef /.notdef/.notdef/quotesinglbase/florin/quotedblbase/ellipsis/dagger/daggerdbl /circumflex/perthousand/Scaron/guilsinglleft/OE/.notdef/Zcaron/.notdef /.notdef/quoteleft/quoteright/quotedblleft/quotedblright/bullet/endash/emdash /tilde/trademark/scaron/guilsinglright/oe/.notdef/zcaron/Ydieresis /.notdef/exclamdown/cent/sterling/currency/yen/brokenbar/section /dieresis/copyright/ordfeminine/guillemotleft/logicalnot/hyphen/registered/macron /degree/plusminus/twosuperior/threesuperior/acute/.notdef/paragraph/dotaccent /ogonek/onesuperior/ring/guillemotright/onequarter/onehalf/threequarters/questiondown /Agrave/Aacute/Acircumflex/Atilde/Adieresis/Aring/AE/Ccedilla /Egrave/Eacute/Ecircumflex/Edieresis/Igrave/Iacute/Icircumflex/Idieresis /Eth/Ntilde/Ograve/Oacute/Ocircumflex/Otilde/Odieresis/multiply /Oslash/Ugrave/Uacute/Ucircumflex/Udieresis/Yacute/Thorn/germandbls /agrave/aacute/acircumflex/atilde/adieresis/aring/ae/ccedilla /egrave/eacute/ecircumflex/edieresis/igrave/iacute/icircumflex/idieresis /eth/ntilde/ograve/oacute/ocircumflex/otilde/odieresis/divide /oslash/ugrave/uacute/ucircumflex/udieresis/yacute/thorn/ydieresis ] def /Mnodistort true def 1.000000 1.000000 scale 91.562500 145.500000 translate 1.000000 -1.000000 scale -0.000000 -0.000000 translate /Mleft 0.000000 def /Mbottom 0.000000 def /Mwidth 230.375000 def /Mheight 142.312500 def 0 setgray 0 setlinewidth /Courier findfont 12 scalefont setfont /Mfontsize 12 def /Plain /Courier findfont def %! %%Creator: Mathematica %%AspectRatio: .61803 MathPictureStart /Mabs { Mgmatrix idtransform Mtmatrix dtransform } bind def /Mabsadd { Mabs 3 -1 roll add 3 1 roll add exch } bind def %% Graphics %%IncludeResource: font Courier %%IncludeFont: Courier /Courier findfont 10 scalefont setfont % Scaling calculations 0.0604396 0.457875 0.103747 0.296774 [ [ 0 0 0 0 ] [ 1 .61803 0 0 ] ] MathScale % Start of Graphics 1 setlinecap 1 setlinejoin newpath 0 g .25 Mabswid [ ] 0 setdash 0 .10375 m 1 .10375 L s .06044 0 m .06044 .61803 L s 0 0 m 1 0 L 1 .61803 L 0 .61803 L closepath clip newpath .5 Mabswid .06044 .1631 m .06157 .1631 L .0626 .16309 L .06379 .16308 L .06491 .16306 L .0669 .16301 L .06907 .16295 L .07146 .16285 L .07399 .16272 L .07852 .16243 L .0876 .1616 L .09759 .16033 L .10767 .15869 L .11851 .15655 L .1381 .15174 L .17774 .13887 L .21592 .12339 L .25265 .10661 L .29173 .08776 L .32935 .06967 L .36934 .05162 L .40786 .03645 L .44493 .02493 L .4641 .02046 L .47388 .01864 L .48436 .01704 L .48943 .01641 L .49423 .0159 L .49891 .01549 L .50324 .01519 L .50733 .01497 L .5094 .01488 L .51168 .01481 L .51279 .01478 L .51396 .01475 L .51458 .01474 L .51525 .01473 L .51642 .01472 L .51753 .01472 L .51872 .01472 L .51985 .01472 L .52087 .01473 L .52202 .01474 L .52309 .01476 L .52553 .01483 L .52791 .01491 L .53047 .01503 L .53509 .01533 L .53932 .01568 L .54957 .01688 L Mistroke .55931 .0185 L .56855 .02047 L .57839 .02305 L .59609 .02901 L .61671 .03822 L .63612 .04924 L .67324 .07717 L .71271 .11767 L .75073 .16823 L .78729 .2285 L .82621 .30626 L .86368 .39533 L .9035 .50642 L Mfstroke .9035 .50642 m .938 .61803 L s [ .04 .05 ] 0 setdash .06044 .25213 m .06157 .25213 L .0626 .25212 L .06379 .25211 L .06491 .25209 L .0669 .25205 L .06907 .25198 L .07146 .25188 L .07399 .25175 L .07852 .25146 L .0876 .25063 L .09759 .24936 L .10767 .24772 L .11851 .24558 L .1381 .24077 L .17774 .22791 L .21592 .21242 L .25265 .19564 L .29173 .1768 L .32935 .1587 L .36934 .14065 L .40786 .12549 L .44493 .11396 L .4641 .10949 L .47388 .10767 L .48436 .10607 L .48943 .10544 L .49423 .10494 L .49891 .10452 L .50324 .10422 L .50733 .104 L .5094 .10391 L .51168 .10384 L .51279 .10381 L .51396 .10379 L .51458 .10378 L .51525 .10377 L .51642 .10375 L .51753 .10375 L .51872 .10375 L .51985 .10375 L .52087 .10376 L .52202 .10378 L .52309 .1038 L .52553 .10386 L .52791 .10395 L .53047 .10407 L .53509 .10436 L .53932 .10471 L .54957 .10592 L Mistroke .55931 .10753 L .56855 .1095 L .57839 .11208 L .59609 .11804 L .61671 .12725 L .63612 .13827 L .67324 .16621 L .71271 .2067 L .75073 .25726 L .78729 .31754 L .82621 .39529 L .86368 .48436 L .9035 .59546 L Mfstroke .9035 .59546 m .91048 .61803 L s [ ] 0 setdash .83883 .61803 m .83883 .10375 L s gsave .97619 .07407 -66 -8.90625 Mabsadd m 1 1 Mabs scale currentpoint translate 0 17.8125 translate 1 -1 scale /g { setgray} bind def /k { setcmykcolor} bind def /p { gsave} bind def /r { setrgbcolor} bind def /w { setlinewidth} bind def /C { curveto} bind def /F { fill} bind def /L { lineto} bind def /rL { rlineto} bind def /P { grestore} bind def /s { stroke} bind def /S { show} bind def /N {currentpoint 3 -1 roll show moveto} bind def /Msf { findfont exch scalefont [1 0 0 -1 0 0 ] makefont setfont} bind def /m { moveto} bind def /Mr { rmoveto} bind def /Mx {currentpoint exch pop moveto} bind def /My {currentpoint pop exch moveto} bind def /X {0 rmoveto} bind def /Y {0 exch rmoveto} bind def 63.000 11.250 moveto %%IncludeResource: font Courier %%IncludeFont: Courier /Courier findfont 10.000 scalefont [1 0 0 -1 0 0 ] makefont setfont 0.000 0.000 0.000 setrgbcolor (X) show 1.000 setlinewidth grestore [(1)] .83883 .07407 0 0 Mshowa gsave .25275 .07407 -66 -8.90625 Mabsadd m 1 1 Mabs scale currentpoint translate 0 17.8125 translate 1 -1 scale /g { setgray} bind def /k { setcmykcolor} bind def /p { gsave} bind def /r { setrgbcolor} bind def /w { setlinewidth} bind def /C { curveto} bind def /F { fill} bind def /L { lineto} bind def /rL { rlineto} bind def /P { grestore} bind def /s { stroke} bind def /S { show} bind def /N {currentpoint 3 -1 roll show moveto} bind def /Msf { findfont exch scalefont [1 0 0 -1 0 0 ] makefont setfont} bind def /m { moveto} bind def /Mr { rmoveto} bind def /Mx {currentpoint exch pop moveto} bind def /My {currentpoint pop exch moveto} bind def /X {0 rmoveto} bind def /Y {0 exch rmoveto} bind def 63.000 11.250 moveto %%IncludeResource: font Courier %%IncludeFont: Courier /Courier findfont 10.000 scalefont [1 0 0 -1 0 0 ] makefont setfont 0.000 0.000 0.000 setrgbcolor (p) show 1.000 setlinewidth grestore gsave .02381 .25213 -66 -8.90625 Mabsadd m 1 1 Mabs scale currentpoint translate 0 17.8125 translate 1 -1 scale /g { setgray} bind def /k { setcmykcolor} bind def /p { gsave} bind def /r { setrgbcolor} bind def /w { setlinewidth} bind def /C { curveto} bind def /F { fill} bind def /L { lineto} bind def /rL { rlineto} bind def /P { grestore} bind def /s { stroke} bind def /S { show} bind def /N {currentpoint 3 -1 roll show moveto} bind def /Msf { findfont exch scalefont [1 0 0 -1 0 0 ] makefont setfont} bind def /m { moveto} bind def /Mr { rmoveto} bind def /Mx {currentpoint exch pop moveto} bind def /My {currentpoint pop exch moveto} bind def /X {0 rmoveto} bind def /Y {0 exch rmoveto} bind def 63.000 11.250 moveto %%IncludeResource: font Courier %%IncludeFont: Courier /Courier findfont 10.000 scalefont [1 0 0 -1 0 0 ] makefont setfont 0.000 0.000 0.000 setrgbcolor (Q) show 1.000 setlinewidth grestore gsave .02381 .1631 -66 -8.90625 Mabsadd m 1 1 Mabs scale currentpoint translate 0 17.8125 translate 1 -1 scale /g { setgray} bind def /k { setcmykcolor} bind def /p { gsave} bind def /r { setrgbcolor} bind def /w { setlinewidth} bind def /C { curveto} bind def /F { fill} bind def /L { lineto} bind def /rL { rlineto} bind def /P { grestore} bind def /s { stroke} bind def /S { show} bind def /N {currentpoint 3 -1 roll show moveto} bind def /Msf { findfont exch scalefont [1 0 0 -1 0 0 ] makefont setfont} bind def /m { moveto} bind def /Mr { rmoveto} bind def /Mx {currentpoint exch pop moveto} bind def /My {currentpoint pop exch moveto} bind def /X {0 rmoveto} bind def /Y {0 exch rmoveto} bind def 63.000 11.250 moveto %%IncludeResource: font Courier %%IncludeFont: Courier /Courier findfont 10.000 scalefont [1 0 0 -1 0 0 ] makefont setfont 0.000 0.000 0.000 setrgbcolor (q) show 1.000 setlinewidth grestore % End of Graphics MathPictureEnd %%PSTrailer grestore %%EPS Trailer pp_save restore end %%Trailer %%EOF \end{filecontents*} \documentclass[12pt]{article} \usepackage{amsmath} \usepackage{amsthm} \usepackage{amssymb} \usepackage{graphicx} \usepackage{psfrag} %\include{macro_dyn} \addtolength{\textwidth}{6mm} %%%%%%%% macros \makeatletter \def\fixNumberingInArticle{ \@addtoreset{equation}{section} \renewcommand{\theequation}{\thesection.\arabic{equation}} } \def\fixNumberingInAppendix{ \setcounter{section}{0} \renewcommand{\theequation}{\Alph{section}.\arabic{equation}} \renewcommand{\thesection}{\Alph{section}} \setcounter{equation}{0} } \makeatother \newcommand{\Am}{A_{\freq,0}} \newcommand{\RII}[1]{R_{\sol}^{(2)}(#1)} \newcommand{\RIII}[1]{R_{\sol}^{(3)}(#1)} \newcommand{\NII}[1]{N_{\sol}(#1)} \newcommand{\pRIII}[1]{{R_{\sol}^{(3)}}'(#1)} \newcommand{\RotSpace}{\mathrm{SO}} \newcommand{\dotp}[2]{\langle #1,#2\rangle} \newcommand{\dotC}[2]{\dotp{#1}{#2}_{\mathbb{C}}} \newcommand{\dotRE}[2]{\dotp{#1}{#2}_{\mathrm{Re}}} \newcommand{\dotR}[2]{\dotp{#1}{#2}_{\mathbb{R}}} \newcommand{\Mm}{n} \newcommand{\Mt}{\tilde{n}} \newcommand{\conv}{*} \newcommand{\freq}{\mu} \newcommand{\om}{\Omega_{\sol}} \newcommand{\iom}{\Omega_{\sol}^{-1}} \newcommand{\piom}{\Omega_{\sol_{\Par}}^{-1}} \newcommand{\Miom}{\left. \iom\right|_{\{z_{k}\}}} \newcommand{\PS}{\Gamma} \newcommand{\qq}{\varphi} \newcommand{\pp}{\varpi} \newcommand{\xx}{\psi} \newcommand{\TG}[1][\null]{\mathrm{T}_{#1}\PS} \newcommand{\TCG}[1][\null]{\mathrm{T}_{#1}^{\mathbb{C}}\PS} \newcommand{\TGx}{\mathrm{T}_{\xx}\PS} \newcommand{\QQ}{q} \newcommand{\PP}{p} \newcommand{\QQx}{\QQ_{\xx}} \newcommand{\PPx}{\PP_{\xx}} \newcommand{\XX}{z} \newcommand{\XXx}{\XX_{\xx}} \newcommand{\YY}{y} \newcommand{\TopD}{\mathcal{D}} \newcommand{\stf}{\omega} \newcommand{\J}{J} \newcommand{\Jx}{J_{\xx}} \newcommand{\bone}{\mathchoice {\rm 1\mskip-4mu l} {\rm 1\mskip-4mu l} {\rm 1\mskip-4.5mu l} {\rm 1\mskip-5mu l}} \newcommand{\Ham}{\mathcal{H}} \newcommand{\GG}{\mathcal{G}} \newcommand{\Pt}[1]{\mathcal{P}_{\theta:#1}} \newcommand{\CF}{\set{F}_\sol} \newcommand{\xvp}{z_\xx} \newcommand{\sol}{\eta} \newcommand{\solp}{\sol_\sigma} \newcommand{\solw}{\sol_\freq} \newcommand{\pstf}{\omega_{\PS}} \newcommand{\pGG}{\mathcal{G}_{\psi}} \newcommand{\pJ}{J_{\PS}} \newcommand{\stfrm}{\omega} \newcommand{\srfrm}{\stfrm_{\sol_{\Par}}} \newcommand{\tfrm}[2]{\stfrm(#1,#2)} \newcommand{\tfrmp}[2]{\stfrm_\psi(#1,#2)} \newcommand{\rfrm}[2]{\srfrm(#1,#2)} \newcommand{\cf}{{\it cf.\/}\ } \newcommand{\vip}{v} \newcommand{\symb}[1]{\boldsymbol{#1}} \newcommand{\set}[1]{\mathrm{#1}} \newcommand{\fat}[1]{\mbox{\boldmath $#1$}} \newcommand{\sfat}[1]{{\mbox{\scriptsize \boldmath $#1$}}} \newcommand{\Hess}{\mathop{\mathrm{Hess}}} \renewcommand{\exp}[1]{\mathrm{e}^{#1}} \newcommand{\RE}{\mathop{\set{Re}}} \newcommand{\IM}{\mathop{\set{Im}}} \renewcommand{\i}{{\rm i}} \newcommand{\supp}{\mathop{\mathrm{supp}}} \newcommand{\diag}{\mathop{\mathrm{diag}}} \newcommand{\domain}[1]{\set{D}( #1 )} \newcommand{\Range}[1]{\set{Ran}( #1 )} \newcommand{\Null}[1]{\set{N}( #1 )} \newcommand{\closure}[1]{\left\{ #1 \right\}_{\mathrm{cl}} } \newcommand{\adj}{*} \newcommand{\adjoint}[1]{#1^{\adj}} \newcommand{\conj}[1]{\bar{#1}} \newcommand{\commute}[2]{[#1,#2]} \newcommand{\Res}[2]{\mathrm{Res}\left\{ #1; #2 \right\}} \newcommand{\Eq}{Eq.\/~} \newcommand{\eq}{eq.\/~} \newcommand{\Eref}[1]{\Eq\eqref{#1}} \newcommand{\eref}[1]{\eq\eqref{#1}} \newcommand{\etal}{{\it et.\/ al.\/}, } \newcommand{\ie}{{\it i.e.\/}, } \newcommand{\etc}{{\it etc.\/}} \newcommand{\eg}{{\it e.g.\/}, } \newcommand{\diff}{\mathop{\mathrm{\mathstrut{d}}}\!}%Diff op. \newcommand{\ess}{\sigma_{\mathrm{ess}}} \newcommand{\Sch}{Schr\"{o}dinger } \newcommand{\Hone}{\set{H}_1} \newcommand{\Htwo}{\set{H}_2} \newcommand{\Sob}[1]{\set{H}_{#1}} \newcommand{\C}[1]{\set{C}^{#1}} \newcommand{\Ltwo}{\set{L}^2} \newcommand{\Lp}[1]{\set{L}^{#1}} \newcommand{\DF}{\mathcal{D}} \newcommand{\sind}[1]{{\text{{\tiny $#1$}}}} \newcommand{\eps}{{\epsilon_{\sind{V}}}} \newcommand{\epsE}{{\epsilon_{\sind{0}}}} \newtheorem{proposition}{Proposition} \newtheorem{step}{Step} \newtheorem{lemma}{Lemma} \newtheorem{remark}{Remark} \newtheorem{que}{Question} \newtheorem{theorem}{Theorem} \newtheorem{corollary}{Corollary}[theorem] \newtheorem{ass}{Assumption}[theorem] \newtheorem{work}{Exercise} \newtheorem{hint}{Hint}[work] \newtheorem{Def}{Definition} \newcommand{\Hae}{\mathcal{H}_{V}} \newcommand{\EV}{\En_{V}} \newcommand{\Sn}{\mathcal{S}} \newcommand{\SV}{\Sn_{V}} \newcommand{\En}{\mathcal{E}} \newcommand{\Nn}{\mathcal{N}} \newcommand{\Ew}{\En_{\freq}} \newcommand{\EwN}{\En_{\freq_0}} \newcommand{\Ee}{\En_{\freq,{\eps}}} \newcommand{\DEe}{\En'_{\freq,{\eps}}} \newcommand{\vn}{\sol_{\freq_0}} \newcommand{\nEw}{\En_{\freq_0}} \newcommand{\DEw}{\En_{\freq}'} \newcommand{\dEw}{d\En_{\freq}} \newcommand{\DE}{\En'} \newcommand{\LL}{\mathcal{L}_\sol} \newcommand{\LLm}{\mathcal{L}_{\sol_{\freq}}} \newcommand{\Le}{\mathcal{L}_{\sol,\sigma}} \newcommand{\Mf}{\set{M}_{\mathrm{s}}} \newcommand{\TM}{\set{T}_{\sol}\Mf} \newcommand{\Span}{\mathop{\mathrm{span}}} \newcommand{\Oh}{\mathcal{O}} \newcommand{\nrmHo}[1]{\|#1\|_{\Hone}} \newcommand{\nrmHt}[1]{\|#1\|_{\Htwo}} \newcommand{\nrmL}[1]{\|#1\|_{\Ltwo}} \newcommand{\nrmD}[1]{\|#1\|_{\Sob{-1}}} \newcommand{\tnrm}[1]{|\|#1\||} \newcommand{\tnrmHo}[1]{|\|#1\||_{\Hone}} \newcommand{\Laplace}{\Delta} \newcommand{\vw}{\sol_{\freq}} \newcommand{\Lth}{\theta} \newcommand{\Pth}{\theta} \newcommand{\sPth}{\theta} \newcommand{\hzvec}{\xi} \newcommand{\zvec}{z} \newcommand{\za}{\zvec_{\mathrm{t}}} \newcommand{\zg}{\zvec_{\mathrm{g}}} \newcommand{\zt}{\zvec_{\mathrm{b}}} \newcommand{\zn}{\zvec_{\mathrm{s}}} \newcommand{\htr}{\hzvec_{\mathrm{t}}} \newcommand{\hgu}{\hzvec_{\mathrm{g}}} \newcommand{\hga}{\hzvec_{\mathrm{b}}} \newcommand{\hsa}{\hzvec_{\mathrm{s}}} \newcommand{\tr}{\za} \newcommand{\gu}{\zg} \newcommand{\ga}{\zt} \newcommand{\sa}{\zn} \newcommand{\Pro}{\mathcal{P}_{\sol_{\Par}}} \newcommand{\bP}{\bar{\Pro}} \newcommand{\nP}{\Pro_{0}} \newcommand{\bnP}{\bar{\Pro}_{0}} \newcommand{\Q}{\mathcal{Q}} \newcommand{\bQ}{\bar{\mathcal{Q}}} \newcommand{\psol}{\psi_\mathrm{sol}} \newcommand{\psolo}{\psi_{\mathrm{sol},0}} \newcommand{\nrm}[1]{\|#1\|} \newcommand{\rdot}[2]{\langle #1,#2 \rangle} \newcommand{\gN}{\set{N}_{\mathrm{gen}}} \newcommand{\TB}{\mathcal{S}_{a \vip \gamma}} \newcommand{\TOB}{\hat{\mathcal{S}}_{a \vip \gamma \freq}} \newcommand{\Tb}{\hat{\mathcal{S}}_{a \vip \gamma \freq}} \newcommand{\iTback}{\Sc^{-1}\circ\TB^{-1}} \newcommand{\Qr}{m(\freq)} \newcommand{\pQQ}{m'(\freq)} \newcommand{\mm}{m} \newcommand{\mw}{\mm(\freq)} \newcommand{\mmp}{\mm'} \newcommand{\mpw}{\mm'(\freq)} \newcommand{\mtw}{\mm^2(\freq)} \newcommand{\VR}{\mathcal{R}_{V}} \newcommand{\tVR}{\mathcal{R}_{V}} \newcommand{\W}{|\|w\||} \newcommand{\WW}{|\|w\||_{\Hone}} \newcommand{\WWT}{|\|w\||_{\cHone}} \newcommand{\setPar}{\Sigma} \newcommand{\Par}{\sigma} \newcommand{\spow}{s} \newcommand{\smT}{\alpha_{2d+1}} \newcommand{\smF}{\alpha_{2d+2}} \newcommand{\Anrm}[1]{|#1|_{\infty}} \newcommand{\anrm}[1]{|#1|} \newcommand{\Lpp}{L_+} \newcommand{\enu}{e_{-\nu}} \newcommand{\lk}{\mathcal{K}} \newcommand{\lss}{\underline{\lk}} \newcommand{\Ass}{\underline{\alpha}} \newcommand{\Lal}{\alpha\cdot \lk} \newcommand{\Sal}{\Ass\cdot \lss} \newcommand{\Rot}{\mathcal{T}^{\textrm{g}}_{\gamma}} \newcommand{\Tr}{\mathcal{T}_a^{\textrm{tr}}} \newcommand{\Bo}{\mathcal{T}_{\vip}^{\textrm{gal}}} \newcommand{\Sc}{\mathcal{T}_{\freq}^{\textrm{s}}} \newcommand{\Ph}{\mathcal{T}_{\vip}^{\textrm{b}}} \newcommand{\SRot}{\mathcal{T}^{\textrm{r}}_{R}} \newcommand{\CC}{\mathcal{T}^{\textrm{c}}} \newcommand{\Man}{\set{M}} \newcommand{\TMan}[1]{\set{T}_{#1}\set{M}} \newcommand{\Ip}[1]{I_{#1}} \newcommand{\dIp}[1]{\dot{I}_{#1}} %%%%%%%%%%%%%% document \fixNumberingInArticle \title{Solitary wave dynamics in an external potential} \author{J. Fr\"{o}hlich$^1$, S. Gustafson$^{2}$ \thanks{supported by NSERC grant 22R80976}, B. L. G. Jonsson$^{3,4,}$\thanks{the support of Wenner-Gren Foundation is gratefully acknowledged}, I. M. Sigal$^{3,5,}$\thanks{Supported by NSERC under NA7601}\\ {\small $^1$Institute f\"{u}r Theoretische Physik, ETH H\"{o}nggerberg, Z\"{u}rich, Schwitzerland.} \\ {\small $^2$Department of Mathematics, University of British Columbia, Vancouver, Canada.} \\ {\small %BC V6T 1Z2 $^3$Department of Mathematics, University of Toronto, Toronto, Canada.}\\ {\small $^4$The Alfv\'enlaboratory, Royal Institute of Technology, Stockholm, Sweden.}\\ % SE-100 44 {\small $^5$Department of Mathematics, University of Notre Dame, Notre Dame, IN, USA.}} %IN 46556 \date{\today} \begin{document} \maketitle \begin{abstract} We study the behavior of solitary-wave solutions of some generalized nonlinear Schr\"odinger equations with an external potential. The equations have the feature that in the absence of the external potential, they have solutions describing inertial motions of stable solitary waves. We construct solutions of the equations with a non-vanishing external potential corresponding to initial conditions close to one of these solitary wave solutions and show that, over a large interval of time, they describe a solitary wave whose center of mass motion is a solution of Newton's equations of motion for a point particle in the given external potential, up to small corrections corresponding to radiation damping. \end{abstract} %\keywords{nonlinear Schr\"odinger equations, dynamics, solitons, %Hamiltonian systems, stability, modulation, variational principle, %energy estimates} \section{Introduction} \label{sec:intro} In this paper we study the effective dynamics of solitary wave solutions of a class of generalized nonlinear Schr\"{o}dinger equations with an external potential. These equations have the form \begin{equation} \i \partial_t \psi = (-\Laplace + V)\psi - f(\psi) \; , \label{eq:sp1} \end{equation} where $\psi:\mathbb{R}^{d}\times \mathbb{R}\mapsto \mathbb{C}$, $x$ denotes a point in space $\mathbb{R}^d$, $t\in \mathbb{R}$ is time, $\partial_{t} = \frac{\partial}{\partial t}$, $\Laplace=\sum_{j=1}^d \frac{\partial^2}{\partial x_j^2}$ is the spatial Laplacian, $V(x)$ is the external potential and is a real-valued, bounded slowly varying function on $\mathbb{R}^d$, and $f$ is a map from the complex Sobolev space $\Hone(\mathbb{R}^{d},\mathbb{C})$ to $\Sob{-1}(\mathbb{R}^{d},\mathbb{C})$ such that $f(0)=0$, and $f(\bar{\psi})=\overline{f(\psi)}$, describing a nonlinear ``self-interaction''. Precise assumptions on $V$ and $f$ will be given in Section~\ref{sec:res}. Examples of nonlinearities of interest include local nonlinearities such as \begin{equation} f(\psi)=\lambda |\psi|^{2s}\psi, \ \ 00, \label{eq:lnl} \end{equation} and Hartree-type nonlinearities \begin{equation} f(\psi) = \lambda (W*|\psi|^2)\psi,\ \ \lambda>0, \label{eq:nlnl} \end{equation} where $W$ is of positive type, continuous, spherically symmetric potential function which tends to 0, as $|x|\rightarrow \infty$ and $W*g(x):=\int W(x-y)g(y)\diff^d y$, denotes (spatial) convolution. Of course, $\lambda$ can be scaled out by rescaling $\psi$. We assume that the nonlinearity in \eqref{eq:sp1} is such that the Cauchy problem has a unique, global solution, $\psi(x,t)$, in the space $\set{C}(\mathbb{R}^+;\Hone(\mathbb{R}^d,\mathbb{C}))\cap \C{1}(\mathbb{R}^+;\Sob{-1}(\mathbb{R}^d,\mathbb{C}))$, given an initial condition $\psi(x,0)=\psi_0(x)\in\Hone(\mathbb{R}^d,\mathbb{C})$. Results on the Cauchy problem associated with \eqref{eq:sp1} can be found in \cite{Ginibre+Velo1979,Kato1987,Cazenave1996} (see Section~\ref{sec:res} for a discussion). For $V\equiv 0$, eq.~\eqref{eq:sp1} is the usual generalized nonlinear Schr\"odinger (or Hartree) equation. For self-focusing nonlinearities (as in example \eqref{eq:lnl} or \eqref{eq:nlnl} with $W$ of positive type and $\lambda>0$ large enough), it can have stable solitary wave solutions of the form \begin{equation} \sol_{\Par}(x,t):= \exp{\i(\frac{1}{2}v\cdot(x-a)+\gamma)}\solw(x-a), \label{eq:sp2a} \end{equation} where $\Par:=\{a,\vip,\gamma,\freq\}$, and $a=vt+a_0$, $\gamma=\freq t + \frac{v^2}{4}t+ \gamma_{0}$, with $\gamma_0 \in [0,2\pi)$, $a_0,\vip\in\mathbb{R}^{d}$ and $\freq\in\mathbb{R}^{+}$, constant, and where $\solw$ is a positive solution of the nonlinear eigenvalue problem \begin{equation} (-\Laplace + \freq) \solw - f(\solw) = 0 \label{eq:sp2} \end{equation} (see Section~\ref{sec:res}). Solutions of eq.~\eqref{eq:sp1} of the form $\eqref{eq:sp2a}$ describe solitary waves traveling through space with a constant velocity $\vip$, and with an oscillatory phase given by $\freq t-\frac{1}{4}v^2 t$. Existence of such solutions, for a large class of nonlinearities has been established in \cite{Strauss1977,Berestycki+Lions+Peletier1981,Berestycki+LionsI1983,Berestycki+LionsII1983,Berestycki+Gallouet+Kavian1983,Jones+Kupper1986,Lions1986b,Adachi2002}. See Section~\ref{sec:res} for an outline of results relevant for this paper. In analyzing soliton-like solutions of eq.~\eqref{eq:sp1} we encounter two length scales: the size $\propto \freq^{-1/2}$ of the support of the function $\solw$, which is determined by our choice of initial condition $\psi_0$, and the length scale $\propto (\sup |\nabla V|)^{-1}$ over which the external potential $V$ varies appreciably. We will assume that the ratio \begin{equation}\label{eq:scales} \eps = \frac{\sup |\nabla V(x)|}{\sqrt{\freq}}, \end{equation} is {\em small}; \ie that the potential $V$ varies little over the support of a solitary wave solution. When $V$ does not vanish, the wave $\sol_{\Par}(x)$ in \eqref{eq:sp2a} does {\em not} solve eq.~\eqref{eq:sp1}. However, we expect that if the initial condition \begin{equation} \psi_0(x) := \psi(x,0) \end{equation} for \eqref{eq:sp1} is close to $\sol_{\Par_0}(x)$ for some $\Par_{0}$, in the sense that \begin{equation} \nrmHo{\exp{-\i \frac{1}{2}\vip_0\cdot x}(\psi_0-\sol_{\Par_{0}}) } \leq \epsE, \end{equation} then for {\em all times} $0\leq t \leq \frac{T}{\eps+\epsE^2}$, where $T$ is some positive constant, the solution $\psi(x,t)$ remains {\em close} to a solitary wave of the form \eqref{eq:sp2a} for some {\em time-dependent} parameters $\freq,\vip,a$ and $\gamma$ determined by $V$ and $\psi_0$. We will show, more specifically, that these parameters can be chosen to be solutions of the following system of ordinary differential equations \begin{equation}\label{eq:new} \dot{a} = \vip \; , \ \ \frac{\dot{\vip}}{2} = - \nabla V(a) \; , \ \ \dot{\freq} = 0 \; , \end{equation} and \begin{equation}\label{eq:new2} \dot{\gamma} = \freq +\frac{\vip^2}{4} - V(a) \; . \end{equation} (up to error terms of size $\Oh(\eps^2+\epsE^2)$), with initial conditions given by $a(0)=a_0$, $\vip(0)=\vip_0$, $\freq(0)=\freq_0$, and $\gamma(0)=\gamma_0$, with $a_0,\vip_0,\freq_0,\gamma_0$ as in \eqref{eq:sp2a}. We observe that the first two equations in \eqref{eq:new} are {\em Newton's equations of motion} for the trajectory ($a(t),\vip(t)=\dot a(t)$) of a point particle of mass $\frac{1}{2}$ moving in the external potential $V(a)$. The center of a solitary wave solution of \eqref{eq:sp1} follows this trajectory, up to deviations $\Oh(\eps+\epsE)$ due to ``radiation damping''. We state here rigorously the result discussed above for a special class of nonlinearities. The general class of nonlinearities is introduced and discussed in the next section. We assume the external potential $V(x)$ satisfies the conditions \begin{equation} V\in \C{2} \ \text{and}\ |\partial_x^\alpha V(x)| \leq C_\alpha \eps^{|\alpha|}, \ \ \text{for}\ |\alpha|\leq 2. \label{con:V} \end{equation} In \eqref{con:V}, $\eps>0$ is the small parameter introduced in \eqref{eq:scales}. In other words, we find it convenient to fix the size of the support (the `width') of the solitary wave solution at $\Oh(1)$ and assume that the external potential $V(x)$ varies slowly. Let $\epsilon:=\eps +\epsE$. \begin{theorem} Assume that the nonlinearity $f$ is given by \eqref{eq:lnl}, and assume that the potential $V$ satisfies \eqref{con:V} with $\eps\ll 1$. Let $I_0$ be any closed, bounded interval in $(0,\infty)$. Let $\epsE\ll 1$ and the initial condition $\psi_{0}$ satisfy \begin{equation} \nrmHo{\exp{-\i\frac{1}{2}\vip_{0}\cdot x} (\psi_{0}-\sol_{\Par_{0}})}<\epsE, \end{equation} for some $\Par_{0}\in \mathbb{R}^{d}\times\mathbb{R}^{d}\times [0,2\pi)\times I_0$. Then there is a constant $T>0$, independent of $\eps$, $\epsE$ but possibly dependent on $I_0$, such that for times $0\leq t\leq T(\eps+\epsE^{2})^{-1}$, the solution to \eqref{eq:sp1} with this initial condition is of the form \begin{equation} \psi(x,t) =\exp{\i(\frac{1}{2}\vip\cdot (x-a) + \gamma)} (\solw(x-a) + w(x-a,t)), \end{equation} where \begin{equation} \nrmHo{w}=\Oh(\epsilon), \end{equation} and where the parameters $\vip, a, \gamma$ and $\freq$ satisfy the differential equations \begin{align} \frac{1}{2}\dot{\vip} & = - (\nabla V)(a) + \Oh(\epsilon^2), \\ \dot{a} & = \vip + \Oh(\epsilon^2), \\ \dot{\gamma} & = \freq - V(a) + \frac{1}{4}\vip^2 + \Oh(\epsilon^2), \\ \dot{\freq} &= \Oh(\epsilon^2). \end{align} \end{theorem} The same conclusions hold for nonlinearities of the form $f(\psi)=g(|\psi|^{2})\psi+(W*|\psi|^{2})\psi$ where $W$ and $g$ satisfy explicit conditions (see the discussion of the conditions in Section~\ref{sec:res}), provided an additional spectral condition is satisfied (see Condition~\ref{con:g.spec} in Section~\ref{sec:res}). The first result of this type was proved by Fr\"{o}hlich, Tsai and Yau \cite{Frohlich+Tsai+Yau2000,Frohlich+Tsai+Yau2002} for the Hartree equation (\eqref{eq:sp1} with \eqref{eq:nlnl}) under a spectral condition (see Condition~\ref{con:g.spec} of Section 2). The choice of the Hartree type nonlinearity plays an important role in \cite{Frohlich+Tsai+Yau2000,Frohlich+Tsai+Yau2002}. For local, pure power nonlinearities and a small parameter, $\eps$, Bronski and Jerrard \cite{Bronski+Jerrard2000} have shown that if an initial condition is of form \eqref{eq:sp2a}, with $t=0$, then the solution $\psi(x,t)$ of \eref{eq:sp1} satisfies \begin{equation} \eps^{-d}|\psi(\frac{x}{\eps},\frac{t}{\eps})|^{2} \diff^{d} x \rightarrow \nrmL{\solw}^{2}\delta_{a(t)} \end{equation} in the $\set{C}^{1*}$ (dual to $\C{1}$) topology, provided $a(t)$ satisfies the equation $\frac{1}{2}\ddot{a}=\nabla U(a)$, where $V(x)=U(\eps x)$ (see \eqref{eq:new}). Our approach is built on important developments in the nonlinear Schr\"odinger equation (NLS) in the last 20 years (see \cite{Cazenave1996,Sulem1999} for reviews). We outline the landmark developments briefly here. Orbital stability of NLS solitary waves for $V=0$ was proved by Cazenave and Lions \cite{Cazenave+Lions82} and M.~Weinstein \cite{Weinstein1985,Weinstein1986}, whose result was significantly extended by Grillakis, Shatah and Strauss \cite{Grillakis+Shatah+Strauss1987,Grillakis+Shatah+Strauss1990}. The next significant step was made by A.~Soffer and M.~Weinstein \cite{Soffer+Weinstein1990} who proved, under some restrictive conditions, asymptotic stability of nonlinear ground states for \eref{eq:sp1} and by V.~Buslaev and G.~Perel'man \cite{BP92} who, motivated by \cite{Soffer+Weinstein1990}, proved asymptotic stability of NLS solitary waves (V=0) in one dimension, again under certain restrictive conditions. These results were significantly extended by Tsai and Yau \cite{Tsai+Yau2002,Tsai+Yau2002b,Tsai+Yau2002c}, Soffer and Weinstein \cite{Soffer+Weinstein1992,Soffer+Weinstein2003}, Cuccagna \cite{Cuccagna2001,Cuccagna2002}, Buslaev and Perel'man \cite{Buslaev+Perelman1995}, and Buslaev and C.~Sulem \cite{Buslaev+Sulem2002}. Furthermore Perel'man \cite{Perelman2001}, and Rodnianski, Schlag and Soffer \cite{RSS}, have obtained the first results on soliton scattering. Many of the issues touched upon in the present paper were studied also in work of Gustafson and Sigal \cite{GS} on dynamics of magnetic vortices. We also mention interesting non-rigorous results by Pelinovsky, Afanasjev and Kiv\-shar \cite{Pelinovsky1996} and Pelinovsky and Grimshaw \cite{Pelinovsky+Grimshaw1997} on dynamics of NLS solitons near the borderline for stability. Nonrigorous results on dynamics of `center of mass' of solitons were obtained by E.~van Groesen and F.~Mainardi and G.~Derks and E.~van Groesen \cite{Groesen+Mainardi,Derks+Groesen}. Our paper is organized as follows. In Section~\ref{sec:res} we present the general hypotheses on the class of nonlinearities and formulate our main result under these hypothesis. In Section~\ref{sec:sp}, we explain the Hamiltonian and variational aspects of the dynamics given by eq.~\eqref{eq:sp1} which play a role in our proof. In Section~\ref{sec:man}, we describe the symmetries of \eref{eq:sp1} for $V\equiv 0$ and find the ``zero modes'' associated with these symmetries. Furthermore, we introduce and analyze a finite-dimensional manifold, $\Mf$, of stable solitary wave solutions to \eref{eq:sp1}. In Section~\ref{sec:on}, we introduce a convenient parameterization of functions in a small neighborhood of $\Mf$ in phase space. In Section~\ref{sec:mf}, we transform the equation \eqref{eq:sp1} to a moving frame, and then rewrite the resulting equation in terms of the parameters introduced in Section~\ref{sec:on} In Sections~\ref{sec:Lf} and~\ref{sec:lL}, we control solutions of our equations of motion in a moving frame by constructing an approximately conserved Lyapunov functional. The proof of our main result is completed in Section~\ref{sec:MT}. Some material of technical or review nature is collected in four appendices. Remark about the notation: in this paper we consider equations, maps and functionals on complex spaces such as $\Hone(\mathbb{R}^{d},\mathbb{C})$, which sometimes are identified with real spaces; \eg $\Hone(\mathbb{R}^{d},\mathbb{R})= \Hone(\mathbb{R}^{d},\mathbb{R})\oplus\Hone(\mathbb{R}^{d},\mathbb{R})$, under the identification $\psi \leftrightarrow (\RE \psi,\IM \psi)$. In this case the operator of multiplication by $\i^{-1}$ is identified with the operator \begin{equation} \J = \begin{pmatrix} 0 & 1 \\ -1 & 0 \end{pmatrix} \end{equation} (which defines a complex structure on the corresponding real spaces). Thus a real function $\solw$ is sometimes written as $(\solw,0)$ and similarly an imaginary function as $\i \solw$ as $(0,\solw)$ (or even as $\J\solw$). Fr\'echet derivatives are always understood to be defined on real spaces. They are denoted by primes. $c$ will denote various constants which may depend only on the interval $I_0$ for the parameter $\freq_0$. \section*{Acknowledgments} B.L.G.J. and I.M.S. are grateful to J. Colliander for useful discussions and remarks and to ETH-Z{\"u}rich for hospitality during their work on this paper. J.F. thanks T.-P. Tsai and H.-T. Yau for very useful discussions and correspondence which led to the results in \cite{Frohlich+Tsai+Yau2000,Frohlich+Tsai+Yau2002}. \section{Main Result} \label{sec:res} In this section we formulate general assumptions on the nonlinearity $f$ in \eqref{eq:sp1} and state our main result under general assumptions. Our assumptions are rather abstract, each responsible for certain aspect of the problem and of our approach. Then we discuss specific nonlinearities for which these assumptions are satisfied \renewcommand{\theenumi}{(\Alph{enumi})} \begin{enumerate} \item \label{con:g.ham} (Energy) There exists a $\C{3}$-functional $F:\Hone\mapsto \mathbb{R}$ such that $F'(\psi)=f(\psi)$, $\sup_{\nrmHo{u}\leq M} \nrm{F''(u)}_{\mathrm{B}(\Hone,\Sob{-1})}<\infty$ and $\sup_{\nrmHo{u}\leq M} \nrm{F'''(u)}_{\Hone\mapsto \mathrm{B}(\Hone,\Sob{-1})}\leq \infty$ (here $\mathrm{B}$ is the space of bounded linear operators); \item \label{con:g.sym} (Symmetries) $F(\mathcal{T}\psi)=F(\psi)$, where $\mathcal{T}$ is a translation $\Tr:u(x)\mapsto u(x-a)$ $\forall a\in \mathbb{R}^{d}$, a rotation $\SRot:u(x)\mapsto u(R^{-1}x)$ $\forall R\in \RotSpace(d)$, a gauge transform $\Rot:u(x)\mapsto \exp{\i \gamma}u(x)$ $\forall \gamma\in[0,2\pi)$, a boost transform $\Ph:u(x)\mapsto \exp{\frac{\i}{2}v\cdot x}u(x)$ $\forall v\in\mathbb{R}^{d}$, or a complex conjugation $\CC: u(x)\mapsto \bar{u}(x)$; \item \label{con:g.Omega} (Existence of solitons) There is an interval $I\subset \mathbb{R}$ such that $\forall \freq\in I$ \eref{eq:sp2} has a positive, spherically symmetric, $\Ltwo\cap \C{2}$ solution $\solw$, such that $\nrm{|x|^{3}\solw}+\nrm{|x|^{2}|\nabla\solw|}+ \nrm{|x|^{2}\partial_{\freq}\solw}<\infty$ $\forall \freq\in I$ (here $\nrm{\cdot}$ denotes the $L^2$ norm); \item \label{con:g.stab} (Orbital stability) The solutions $\solw$, $\freq\in I$, of \eqref{eq:sp2} described in~\ref{con:g.Omega}, satisfy \begin{equation} \partial_\freq \int \solw^{2} \diff^d x>0; \label{eq:Qwbbl} \end{equation} \item \label{con:g.sol} (GWP) \Eref{eq:sp1} is globally well-posed in $\Sob{1}$ and in $\Sob{2}$; \item \label{con:g.spec} (Null space condition) $\forall \mu\in I$, \begin{equation} \Null{\LLm} = \Span\{ \big(0, \solw\big), \big(\partial_{x_j}\solw, 0\big), j=1,...,d \}\; \label{eq:NullL} \end{equation} where $\LLm:= -\Laplace +\freq -f'(\solw)$, the Fr\'echet derivative of the map $\psi\mapsto (-\Laplace + \freq)\psi-f(\psi)$ at $\solw$. \end{enumerate} \renewcommand{\theenumi}{\roman{enumi}} We now discuss conditions~\ref{con:g.ham}--\ref{con:g.spec}, beginning with general remarks. Condition~\ref{con:g.ham} allows us to define the conserved energy or Hamiltonian \begin{equation}\label{eq:HV} \Ham_{V}(\psi) = \frac{1}{2} \int (|\nabla \psi|^{2} + V|\psi|^{2}) \diff^{d} x - F(\psi), \end{equation} and gives us the following estimates on the nonlinearities \begin{equation}\label{R} |\RII{w}|\leq c(M)\nrmHo{w}^{2}, \ \ |\RIII{w}|\leq c(M)\nrmHo{w}^{3}, \end{equation} and \begin{equation}\label{eq:N} \nrmD{\NII{w}}\leq C(M)\nrmHo{w}^{2} \end{equation} for any $\sol\in\Htwo(\mathbb{R}^{d})$ and $w\in \Hone(\mathbb{R}^{d})$. with $\nrmHo{\sol}+\nrmHo{w} \leq M$. Here \begin{eqnarray} \RII{w}:=F(\sol+w)-F(\sol)-\dotp{F'(\sol)}{w}, \\ \RIII{w}:=F(\sol+w)-F(\sol)-\dotp{F'(\sol)}{w}- \frac{1}{2}\dotp{F''(\sol)w}{w} \label{eq:RIII} \end{eqnarray} and \begin{equation} \NII{w}=F'(\sol+w)-F'(\sol)-F''(\sol)w. \end{equation} Note that \begin{equation} \NII{w}=\pRIII{w}. \end{equation} We can assume without loss of generality that $f'(0)=0$. Then one can show that $I\subset \mathbb{R}_+$. As was mentioned in the introduction, the nonlinearities of interest to us are local nonlinearities, \begin{equation}\label{eq:l2} f(\psi)(x)=h(|\psi(x)|^2)\psi(x), \end{equation} for some real function $h$ on $\mathbb{R}_{+}$, and Hartree-type nonlinearities, \begin{equation}\label{eq:nl2} f(\psi) = (W * |\psi|^{2})\psi, \end{equation} where $W$ is a fixed, real valued, spherically symmetric function, tending to 0 at $\infty$. More generally we consider nonlinearities the form \begin{equation} f(\psi)(x) = h(|\psi(x)|^2)\psi(x) + (W*|\psi|^2)(x)\psi(x). \label{eq:non} \end{equation} We discuss now under which conditions on $h$ and $W$ in \eqref{eq:non} Conditions~\ref{con:g.ham}--\ref{con:g.spec} are satisfied. Condition~\ref{con:g.ham}. For nonlinearities of type \eqref{eq:non} the functional $F$ is given by \begin{equation} F(\psi) = \frac{1}{2} \int H(|\psi|^{2}) +\frac{1}{2}(W*|\psi|^{2}) \diff^d x, \end{equation} where $H(s)=\int_{0}^{s}h(p)\diff p$. The functional $F$ is $\C{3}$ with the inequalities specified in Condition~\ref{con:g.ham} satisfied, provided $h(s)$ is $\C{2}$ with $h^{(k)}(s) \leq c(1+s^{q-k})$ ($k=0,1,2$), for some $q < 2/(d-2)$, and $d<4$, and $W \in \Lp{p} + \Lp{\infty}$ for some $p > d/2$. Condition~\ref{con:g.sym}. This condition is trivially satisfied for \eqref{eq:non} with $W(x)=W(|x|)$. Condition~\ref{con:g.Omega}. Existence of a positive, spherically symmetric solitary wave solution to Eq \eqref{eq:sp2}, was proved in \cite{Berestycki+LionsI1983,Berestycki+LionsII1983,Berestycki+Lions+Peletier1981} for local nonlinearities \eqref{eq:l2} satisfying \begin{align} -\infty < \lim_{r\rightarrow 0} h(r) < \freq, \nonumber \\ -\infty \leq\lim_{r\rightarrow\infty}r^{-\alpha}h(r) \leq C , \label{eq:Uclass1} \end{align} for $0<\alpha < 2/(d-2)$, if $d>2$ and $\alpha \in(0,\infty)$, if $d=1,2$, \begin{equation*} \exists \zeta>0, \ \text{such that}\ \int^\zeta_0 h(r) \diff r>\freq\zeta. \nonumber \end{equation*} Additional results, for a large class of nonlinearities can be found in \cite{Strauss1977,Berestycki+Gallouet+Kavian1983,Jones+Kupper1986,Lions1986b,Adachi2002}. For nonlocal nonlinearities \eqref{eq:nl2} (Hartree equation) with \begin{equation} W\in \Lp{p}_{\mathrm{loc}},\ p\geq\frac{d}{2},\ W\rightarrow 0\ \text{as}\ |x|\rightarrow \infty, \label{eq:Uclass2} \end{equation} existence of solutions was proved in \cite{Cazenave1996,Ginibre+Velo1980,Frohlich+Tsai+Yau2002,Li1990,Li+Li1993,Adachi2002}. Moreover, for nonlinearities \eqref{eq:l2} or \eqref{eq:nl2} satisfying conditions \eqref{eq:Uclass1} or \eqref{eq:Uclass2}, solutions to \eqref{eq:sp2} (solitary waves) are exponentially decaying at $\infty$ as $\Oh(\exp{-\sqrt{\freq}|x|})$; see \cite{Berestycki+LionsI1983,Peletier+Serrin1983}. Condition~\ref{con:g.stab}. Condition \eqref{eq:Qwbbl} is a sufficient condition for orbital stability of solitary waves for (generalized) nonlinear Schr\"{o}dinger equations (see \cite{Grillakis+Shatah+Strauss1987,Grillakis+Shatah+Strauss1990}). This condition is to be checked for each nonlinearity. In the special case of a pure power nonlinearity, $f(\psi)=\lambda|\psi|^{2s}\psi$, we have $\solw(x) = \freq^{\frac{1}{2s}}\sol_{\freq=1}(x\sqrt{\freq})$. Thus condition~\ref{con:g.stab} reduces to the condition that $s<2/d$. Condition~\ref{con:g.sol}. Assume that $f$ is of the form \eqref{eq:non} with $h: [0,\infty)\mapsto \mathbb{R}$, smooth and satisfying $h(0)=0$, and \begin{equation} |h'(r)|\leq C (1+r^{\alpha-1}) \; , \label{eq:hp} \end{equation} for some $\alpha\in [0, \frac{2}{d-2})$, for $d\geq 3$ and $\alpha\in [0,\infty)$ for $d=1,2$ and \begin{equation} h(r)\leq C(1+r^{\beta}) \; , \label{eq:hpp} \end{equation} for some $\beta\in [0,\frac{2}{d})$. Furthermore assume that $W:\mathbb{R}^d\mapsto \mathbb{R}$ is an even potential such that \begin{equation} W\in \Lp{q}(\mathbb{R}^d) + \Lp{\infty}(\mathbb{R}^d), \end{equation} for some $q\geq 1$, $q>d/4$ and let $W_+=\max(0,W)$ s.t. \begin{equation} W_+ \in \Lp{r}(\mathbb{R}^d) + \Lp{\infty}(\mathbb{R}^d), \end{equation} for some $r\geq 1$, $r\geq d/2$ for $d>2$ and $r>1$ for $d=2$. Then \eqref{eq:sp1} is globally well-posed in $\Hone$ and $\Htwo$ (see \cite[Chapter 6]{Cazenave1996}). Condition~\ref{con:g.spec}. This condition is more delicate. First we observe that \begin{equation} (\partial_{x_j} \solw,0), (0,\solw) \in \Null{\LLm}\; , \ \ \forall j=1,...d \; , \label{eq:nulls} \end{equation} due to the fact that $\solw$ breaks the translation and gauge symmetry of \eqref{eq:sp2} (see Section~\ref{sec:man}). Now we list some facts which are discussed in Appendix~\ref{app:nullL}. If $\solw$ is spherically symmetric and $f$ is a local nonlinearity, $f(\psi)(x)=f(\psi(x))$, then $\LL$ can have at most one zero eigenfunction in addition to \eqref{eq:nulls}. Hence \begin{equation} d+1\leq \dim \Null{\LLm} \leq d+2. \end{equation} This extra zero eigenfunction is spherically symmetric and is also a zero eigenfunction of the ordinary differential operator \begin{equation} A_{\freq,0} = - \Laplace_r +\freq - f'(\solw) \end{equation} on $\Ltwo(\mathbb{R}_+, r^{d-1}\diff r)$, where $\Laplace_r$ is the radial Laplacian, \begin{equation} \Laplace_r = \partial_r^2 + \frac{d-1}{r}\partial_r \; . \end{equation} Thus \begin{equation} \Null{\LLm} = \Span\{(0, \solw), (\partial_{x_j}\solw, 0),(\Null{A_{\freq,0}},0) \}. \label{eq:nullL} \end{equation} For local nonlinearities \eqref{eq:l2}, $\Null{A_{\freq,0}}=\{0\}$, if either $d=1$ or \begin{equation}\label{Uclass} h'(r)+h''(r)r^2>0. \end{equation} Alternative conditions on $h$ for $d>1$ are given in Appendix~\ref{app:nullL}. Thus, for local nonlinearities, if either $d=1$ or \eqref{Uclass} holds, then condition \ref{con:g.spec} is satisfied (see also \cite{McLeod+Serrin1987,Weinstein1985,McLeod1993,Stuart2001}). In any case this extra degeneracy, if it happens, is non-generic. If \eqref{eq:NullL} holds for some $f$, then it also holds for small perturbations of $f$. On the other hand we expect that if \eqref{eq:NullL} fails for a given $f$ then there are arbitrarily small perturbations of $f$ for which \eqref{eq:NullL} is satisfied. It is easy to check that Condition~\ref{con:g.ham}--\ref{con:g.spec} are satisfied for nonlinearity \eqref{eq:lnl}. Let $\setPar:=\mathbb{R}^d\times \mathbb{R}^d\times [0,2\pi)\times I$ and $\setPar_0:=\mathbb{R}^d\times \mathbb{R}^d\times [0,2\pi)\times I_0$, where $I$ is a closed, bounded interval on the positive real axis, specified in Condition~\ref{con:g.Omega}, and $\bar{I_0}$ is a closed interval contained in $I\backslash\partial I$. Recall $\epsilon:=\eps+\epsE$. Our main result is \begin{theorem}\label{thm:main} Assume \eqref{con:V} and \ref{con:g.ham}--\ref{con:g.spec} are satisfied. Given $\epsE>0$ and $\eps>0$ defined in \eqref{eq:scales} such that $\epsilon\ll 1$ and an initial condition $\psi_0$ satisfying \begin{equation} \nrmHo{\exp{-\frac{\i}{2}\vip_0 \cdot x}(\psi_0 - \sol_{\Par_{0}})} \leq \epsE \; , \end{equation} for some $\Par_0:=\{a_0, \vip_0, \gamma_0, \freq_0\} \in \setPar_0$. Then there is a constant $T>0$ (independent of $\eps, \epsE$ but possibly dependent on $I_0$) such that for times $t(\eps+\epsE^2) \leq T$ the solution to \eqref{eq:sp1} with this initial condition can be written in the form \begin{equation} \psi(x,t) =\exp{\i(\frac{1}{2}\vip\cdot (x-a) + \gamma)} (\solw(x-a) + w(x-a,t)) \end{equation} where \begin{equation} \nrmHo{w} = \Oh(\epsilon) \end{equation} and the parameters $\vip, a, \gamma$ and $\freq$ satisfy the differential equations \begin{align} \frac{1}{2}\dot{\vip} & = - (\nabla V)(a) + \Oh(\epsilon^2), \\ \dot{a} & = \vip + \Oh(\epsilon^2), \\ \dot{\gamma} & = \freq - V(a) + \frac{1}{4}\vip^2 + \Oh(\epsilon^2), \\ \dot{\freq} &= \Oh(\epsilon^2). \end{align} \end{theorem} The proof of this theorem is given in Section~\ref{sec:MT}. \section{The Hamiltonian and variational structure of generalized nonlinear Schr\"odinger equations} \label{sec:sp} It is well known that the generalized nonlinear Schr\"odinger equation is a Hamiltonian equation of motion on an infinite-dimensional phase space. In this paper we make extensive use of this fact and of the symplectic geometry of phase space, in particular of certain ``submanifolds'' in this space. We outline briefly some facts which are relevant for us. In this section we consider the generalized nonlinear Schr\"odinger equation \eref{eq:sp1}, \begin{equation*} \i \partial_t \psi = (-\Laplace + V)\psi - f(\psi) \; , \end{equation*} under Conditions~\ref{con:g.ham} and~\ref{con:g.sym} only. We study \eref{eq:sp1} on the space $\Hone(\mathbb{R}^{d},\mathbb{C})$. This space, considered as a real space ($\Hone(\mathbb{R}^{d},\mathbb{R}^{2})=\Hone(\mathbb{R}^{d},\mathbb{R}) \oplus\Hone(\mathbb{R}^{d},\mathbb{R})$, $\psi \leftrightarrow (\RE \psi,\IM \psi)$), and equipped with the symplectic form \begin{equation} \tfrm{u}{v} := \IM \int u\bar{v} \diff^{d}x \end{equation} (defined for $u$, $v$ in the tangent space at a given point in $\Hone(\mathbb{R}^d;\mathbb{R}^2)$, which we can identify with $\Hone(\mathbb{R}^{d},\mathbb{R}^{2})$ in our case), is a symplectic space. Define the Hamiltonian functional on $\Hone(\mathbb{R}^{d},\mathbb{C})$ as \begin{equation}\label{eq:haM} \Ham_{V}(\psi):=\frac{1}{2} \int |\nabla \psi|^{2} + V|\psi|^{2} \diff^{d} x -F(\psi), \end{equation} where $F(\psi)$ is as in Condition~\ref{con:g.ham}; \ie $F'(\psi)=f(\psi)$. With these definitions \eref{eq:sp1} can be written as \begin{equation} \partial_{t}\psi = J \Ham_{V}'(\psi), \end{equation} where $J$ is the operator on $\Hone(\mathbb{R}^{d},\mathbb{R}^{2})$ (strictly speaking $J:(\set{T}_{\psi}\Hone)^{*}\mapsto\set{T}_{\psi}\Hone$) given by \begin{equation}\label{eq:Jreal} \J:=\begin{pmatrix} 0 & 1 \\ -1 & 0 \end{pmatrix} \end{equation} in block-diagonal notation. Thus $\J$ is a complex structure on $\Hone(\mathbb{R}^{d},\mathbb{R}^{2})$ corresponding to the operator $\i^{-1}$ on $\Hone(\mathbb{R}^{d},\mathbb{C})$. Observe that the space $\Hone(\mathbb{R}^{d},\mathbb{C})$ also has a real inner product (Riemannian metric) \begin{equation} \dotp{u}{v}:=\RE \int u\bar{v}\diff^{d}x, \end{equation} so that $\tfrm{u}{v}=\dotp{u}{\J^{-1}v}$. Since the Hamiltonian $\Ham_{V}(\psi)$ is autonomous (it does not explicitly depend on time $t$) and invariant under the gauge transformation $\Ham_{V}(\exp{\i \gamma}\psi)=\Ham_{V}(\psi)$ for all $\gamma\in [0,2\pi)$, we have conservation of energy, $\Ham_{V}(\psi)=\text{const}$, and ``mass'' or ``number of particles'', $\Nn(\psi)=\text{const}$, where \begin{equation} \Nn(\psi) := \frac{1}{2}\int |\psi|^{2} \diff^{d} x \end{equation} under the evolution of \eref{eq:sp1}. Due to Condition~\ref{con:g.stab} the solitary wave profiles $\solw$ described in Condition~\ref{con:g.Omega} are local minimizers of the Hamiltonian $\Ham_{V=0}(\psi)$ restricted to the spheres \begin{equation} \{\psi\in\Hone: N(\psi)=\mm\} \end{equation} for $m>0$ (see \cite{Grillakis+Shatah+Strauss1987}, Thm 3). Hence they are critical points of the functional \begin{equation}\label{eq:EW} \Ew(\psi):= \frac{1}{2}\int |\nabla \psi|^2 + \freq|\psi|^2 \diff^d x- F(\psi), \end{equation} where $\freq = \freq(m)$ is a Lagrange multiplier, and \eqref{eq:sp2} is just the Euler-Lagrange equation for $\Ew(\psi)$. Observe that the functional $\Ew(\psi)$ also arises as $(t_{1}-t_{0})\Ew(\phi)=\Sn_{V=0}(\phi \exp{\i \freq t})$ for any $\phi(x)$, where $\Sn_{V}$ is the action for \eref{eq:sp1}: \begin{equation} \Sn_{V}(\psi) :=\int_{t_{0}}^{t_{1}} \left ( \frac{1}{2}\int_{\mathbb{R}^{d}} \IM \dot{\psi}\bar{\psi}\diff^{d} x+\Ham_{V}(\psi) \right) \diff t. \end{equation} The functional $\Ew(\psi)$ will play an important role in our approach. We will use it as a Lyapunov functional in estimating the fluctuations $w$. Finally, we note that the operator $\LL$ that appears in Condition~\ref{con:g.spec} is the Hessian of $\Ew(\psi)$ at $\solw$: $\LL:=\Ew''(\solw)$. \section{Symmetries, zero modes, and the manifold of solitary waves} \label{sec:man} In this section we introduce the manifold of solitary waves which is obtained by applying the generalized symmetry transforms (see below) to a fixed solitary wave. The tangent space to the manifold is introduced, and its inherited symplectic form is derived. Furthermore we prove the key fact that the inherited symplectic form is non-degenerate. Starting from this section we will often use the abbreviation $\sol\equiv \solw$. \Eref{eq:sp1} with $V\equiv 0$ is invariant under spatial translations $\Tr$, gauge transformations $\Rot$, and Galilean transformations $\Bo$, where \begin{align} \Tr &:\psi(x,t) \mapsto \psi(x-a,t) \; , && \Rot : \psi(x,t)\mapsto \exp{\i\gamma}\psi(x,t) \label{eq:T1} \; , \\ \Bo &: \psi(x,t)\mapsto \exp{\i(\frac{1}{2}\vip\cdot x - \frac{1}{4}|\vip|^2 t) } \psi(x-\vip t,t)\; . \label{eq:T2} \end{align} (Transformations \eqref{eq:T1}--\eqref{eq:T2} map solutions of eq.~\eqref{eq:sp1} with $V=0$ into solutions of \eqref{eq:sp1} with $V=0$.) The symmetries $\Tr$, $\Rot$, $\Bo$, have associated conserved quantities: field momentum, mass and 'center of mass motion', (\cf \cite{Frohlich+Tsai+Yau2002}), \begin{equation} \int \i \bar{\psi}\nabla\psi \diff^d x, \ \ \int |\psi|^2 \diff^d x, \ \ \int \bar{\psi} (x+2\i t \nabla)\psi \diff^d x. \end{equation} For $f(\psi)=|\psi|^{2\spow}\psi$ and $V=0$, \eref{eq:sp1} is also invariant under the scaling transformation \begin{align} \Sc : \psi(x,t) \mapsto \freq^{\frac{1}{2\spow}} \psi(\sqrt{\freq} x,\freq t)\; . \end{align} When the external potential is introduced into the problem, the translational and Galilean invariance are broken. In particular, conservation of the field momentum is replaced by the following `Newton's law' (Ehrenfest's theorem) \begin{equation}\label{eq:Ehrenfest} \partial_t \dotp{\psi}{-\i \nabla \psi} = - \dotp{\psi}{(\nabla V)\psi}, \end{equation} which plays an important role in our analysis, and which is proved in Appendix~\ref{app:Ehrenfest}. Let $\Ph:\psi(x)\mapsto \exp{\frac{\i}{2}\vip\cdot x}\psi(x)$, be the boost transform. We introduce the combined symmetry transformations $\TB$: \begin{equation} p(x)\mapsto p_{a\vip \gamma} :=\TB p = \Tr \Ph\Rot p(x) = \exp{\i (\frac{1}{2}\vip \cdot (x-a)+\gamma)}p(x-a) \; . \end{equation} Let $\sol_{a\vip\freq\gamma}:=\TB \solw$. We define the manifold of solitary waves as \begin{equation} \Mf := \{\sol_{a\vip \gamma\freq} : a,\vip, \gamma,\freq \in \mathbb{R}^d\times \mathbb{R}^d\times [0,2\pi]\times I \} \; . \end{equation} The tangent space to this manifold at the solitary wave profile $\solw\in \Mf$ is given by \begin{equation} \set{T}_{\solw}\Mf = \Span(\za,\zg,\zt,\zn) \; , \end{equation} where \begin{align} \tr :=&\left.\nabla_a \Tr \solw \right|_{a=0} = -\nabla \solw\; , \label{eq:t1} \\ \gu :=& \left. \frac{\partial}{\partial \gamma} \Rot \solw\right|_{\gamma=0} = \i \solw\; ,\\ \ga := &\left. 2 \nabla_\vip \Bo \solw \right|_{\vip=0,t=0} = \i x\solw \; , \end{align} and \begin{equation}\label{eq:t4} \sa := \partial_{\freq} \solw \; . \end{equation} Symmetries of \eqref{eq:sp1} which are broken by the solitary wave solution $\solw\exp{\i \freq t}$ lead to zero modes and ``symplectically associated zero modes'' of the Hessian $\LL$. Namely we have for $\sol=\solw$ \begin{equation} \LL\tr = 0 \; , \ \mbox{and} \ \LL\gu = 0 \; , \label{eq:trgu} \end{equation} and \begin{equation} \LL\ga = 2 \i \tr \, , \ \ \mbox{and} \ \LL\sa = \i \gu \; . \label{eq:gasa} \end{equation} The functions $\ga$ and $\sa$ are zero modes for the operator $(\i \LL)^2$, \ie \begin{equation} \i \LL(\i \LL\ga) = 0 \; , \ \ \mbox{and} \ \i \LL(\i \LL\sa)= 0 \; . \end{equation} The relations \eqref{eq:trgu} are proved by taking the derivatives of the equation $\Ew'(\Tr\Rot\solw)$ $=0$ with respect to the parameters $a$ and $\gamma$, at $a=0$ and $\gamma=0$, and similarly for \eqref{eq:gasa}. We have shown above that \begin{equation} \TM \subset \gN(\J \LL ) \; , \end{equation} where \begin{equation} \gN(\J \LL) := \Span\{ \bigcup_{n\geq 1} \Null{(\J \LL)^n}\} \; . \end{equation} In what follows we denote $\Par:=\{a,\vip,\gamma,\freq\}$ and $\eta_\Par:=\eta_{a\vip\gamma\freq}$. The manifold $\Mf$ inherits a symplectic form from $(\Hone,\omega)$. This symplectic form is determined by the operator \begin{equation}\label{eq:iom} \piom:=\J^{-1}\upharpoonright_{\set{T}_{\sol_{\Par}}\Mf}\equiv\Pro J^{-1}\Pro\upharpoonright_{\set{T}_{\sol_{\Par}}\Mf}, \end{equation} where $\Pro:\set{T}_{\sol_{\Par}}\Hone\mapsto \set{T}_{\sol_{\Par}}\Mf$ is the $\Ltwo$-orthogonal projection onto $\set{T}_{\sol_{\Par}}\Mf$. Namely $\rfrm{u}{v}:=\dotp{u}{\piom v}$. The key fact here is that this symplectic form is non-degenerate, \ie the operator $\J^{-1}\upharpoonright_{\set{T}_{\sol_{\Par}}\Mf}$, is invertible $\forall \sol_{\Par}\in\Mf$, as shown in Lemma~\ref{lem:Oinv} below. Define \begin{equation}\label{eq:mw} \mw:=\frac{1}{2}\int |\solw|^2 \diff^d x. \end{equation} \begin{lemma}\label{lem:Oinv} If $m'(\freq)>0$, then $\piom$ is invertible. \end{lemma} Note that $m'(\freq) > 0$ is exactly the assumption for stability in Section~\ref{sec:res}, condition~\ref{con:g.stab}. \begin{proof} We prove that $\iom$ is invertible by showing that its matrix $\Miom:=(\dotp{z_{j}}{\J^{-1}z_{k}})$ in the basis $\{z_{1},\ldots,z_{2d+2}\} \equiv \{\tr,\ga,\gu,\sa\}$ in $\TM$ is invertible. We compute %the %determinant in a basis is non-zero. The tangent vectors, to the %manifold $\Mf$, $\{\tr,\ga,\gu,\sa\}$ are a basis to $\TM\subset %\set{T}_\sol \Gamma$. We compute the $\iom$ matrix elements %as $\rdot{\J^{-1} z}{z'}$. Note that %$\rdot{\J^{-1} z}{z'} = -\rdot{z}{\J^{-1} z'}=\rdot{z}{\J z'}$, %for all $z,z'=\{\tr,\ga,\gu,\sa\}$. %Here %\begin{equation} % \gu := -\J \sol \; , \ \ % (\tr)_j := -\partial_{x_j} \sol \; , \ \ % (\ga)_j := -\J x_j \sol \; , \ \ % \sa := \partial_\freq \sol % \; . %\end{equation} %Now %\begin{align} % \dotp{\J^{-1} z}{z} &= 0 \; , \ \ \forall z \in \TM \; , \\ % \dotp{\J^{-1} z_k}{z_j} & = -\dotp{\J^{-1} z_j}{z_k} \; , \ \ % \forall z_k,z_j \in \TM \; , \\ % \dotp{\J^{-1}(\tr)_k}{\gu} &= -\int %(\partial_{x_k}\sol) \sol \diff^d x= 0 \; , \\ % \dotp{\J^{-1}(\tr)_k}{(\ga)_j} & = % -\int (\partial_{x_k}\sol) x_j \sol \diff^d x= % \frac{\delta_{kj}}{2}\int \sol^2 \diff^d x\; , \\ % \dotp{\J^{-1}\gu}{\sa} &= - % \int \sol\partial_\freq \sol \diff^d x= - % \frac{1}{2} \partial_\freq\int \sol^2 \diff^d x % \; , \\ % \dotp{\J^{-1}\sa}{(\ga)_j} &= % \int (\partial_\freq \sol) x_j \sol \diff^d x= % \frac{1}{2}\partial_\freq \int x_j \sol^2 \diff^d x =0\; , %\end{align} %where we have used the assumption that $\sol(x)=\sol(|x|)$ %Hence \begin{equation}\label{eq:eiom} \left.\iom\right|_{\{z_{k}\}} = \begin{pmatrix} 0 & -\mm\bone & 0 & 0 \\ \mm\bone & 0 & 0 & 0 \\ 0 & 0 & 0 & \mmp \\ 0 & 0 & -\mmp & 0 \end{pmatrix}, \end{equation} This matrix is invertible provided $\mmp\neq 0$ ($m>0$ always). Since $\Omega^{-1}_{\sol_\Par}$ is related to $\iom$ by a similarity transform (see \eqref{eq:dk}), it is invertible as well. \qed \end{proof} \begin{corollary}\label{cor:nondeg} $\forall z\in\TM$, there exists at least one element, $z'\in \TM$, such that $\stfrm_\sol(z,z')\neq 0$. \end{corollary} \begin{proof} This follows directly from the non-degeneracy of $\iom$.\qed \end{proof} \begin{remark} The real form of the vectors \eqref{eq:t1}--\eqref{eq:t4} is \begin{equation} (-\nabla \sol , 0 ), \ \ (0 , \sol ), \ \ ( 0 , x \sol ), \ \ (\partial_\freq \sol , 0 ). \end{equation} We abuse notation and denote both the real and complex representations of the vectors as $-\nabla \sol$, $-\J \sol$, $-\J x \sol$ and $\partial_\freq \sol$, where for the real representation we interpret $\sol$ as $(\sol,0)$, and $\J$ is of the form \eqref{eq:Jreal}, whereas in the complex notation $\J=\i^{-1}$. \end{remark} \begin{remark} For the special case with $f(\psi)=\lambda |\psi|^{2\spow}\psi$, \begin{equation} \sa :=\left.\partial_{\freq} \Sc\sol \right|_{\freq=1,t=0}= \frac{1}{2}(\frac{1}{\spow} + x\cdot \nabla )\sol = \partial_\freq \sol_{\freq} \; . \end{equation} \end{remark} \begin{remark} Equation \eqref{eq:sp1} with $V=0$ and with a nonlinearity of the form \eqref{eq:non}, where $W$ is spherically symmetric has also rotational symmetry: $\psi(x,t)\mapsto \psi(Rx,t)$, $R\in \RotSpace(d)$. This symmetry does not play a role in our analysis since we consider only spherically symmetric solitary wave solutions of \eqref{eq:sp2}. \end{remark} \begin{remark} Note that for $f(\psi)=\lambda |\psi|^{2\spow}\psi$ we can put the $\freq$ dependence on the same footing as the transformation induced for $\vip$, $a$ and $\gamma$. Indeed, define \begin{equation} \Tb = \TB \circ \Sc. \end{equation} Then $\sol_{a\vip\gamma\freq}=\Tb \sol_1$. \end{remark} \section{Skew orthogonal decomposition} \label{sec:on} In this section we introduce the skew orthogonal decomposition of a solution $\psi$ along the manifold $\Mf$ and in the skew orthogonal direction (see \cite{Arnold1989}), and show that this uniquely defines the solitary wave solution parameters appearing in the decomposition. We recall the notation $\Par = \{\gamma, a, \vip, \freq\}$ so that $\solp := \sol_{a \vip \gamma\freq}$. Define the $\delta$-neighbor\-hood \begin{equation} U_{\delta} = \{\psi\in \Hone:\inf_{\Par\in \setPar_0} \nrmHo{\psi-\sol_\Par} \leq \delta\} \; \end{equation} of the manifold $\Mf':=\{\solp:\Par \in \setPar_0\}$, where \begin{equation} \setPar_0:= \mathbb{R}^d\times\mathbb{R}^d\times [0,2\pi)\times I_0. \end{equation} Here $I_0$ is any bounded closed interval contained in $I\backslash \partial I$, (for the definition of $I$ Condition~\ref{con:g.Omega}). The main result of this section is \begin{proposition}\label{prop:IFT} Let $\psi \in U_{\delta}$. For $\delta>0$ sufficiently small, there exists a unique $\Par = \Par(\psi)\in \C{1}(U_{\delta},\setPar)$ such that \begin{equation} \tfrm{\psi-\sol_{\Par}}{z}=0 \ \text{\ie}\ \rdot{\psi-\sol_{\Par}}{\J^{-1} z} = 0 \; , \forall z \in \mathrm{T}_{\sol_{\Par}}\Mf \; . \label{eq:od3} \end{equation} \end{proposition} \begin{proof} We will use the following notation: \begin{eqnarray} \Par:=\{\Par_1,...,\Par_{2d+2}\}:=\{a,\vip,\gamma,\freq\},\\ \{z_{\freq,1},...,z_{\freq,2d+2}\}:= \{\tr, \ga, \gu, \sa\}, \end{eqnarray} and \begin{equation}\label{eq:dk} \partial_k = \partial_{\Par_k},\ k=d+1,...,2d+2,\ \partial_k = \partial_{\Par_{k}}+\frac{1}{2}\Par_{k+d} \partial_{\Par_{2d+1}},\ k=1,...,d. \end{equation} Denote $z_{\Par,j}:=\TB z_{\freq,j}$ so that $\partial_k \solp = z_{\Par,k}=\TB z_{\freq,k}$. Clearly $\{z_{\Par,j}\}$ is a basis in $\mathrm{T}_{\sol_{\Par}}\Mf$. We use the implicit function theorem for the map $G:\Hone\times \setPar \mapsto \mathbb{R}^{2d+2}$, defined by \begin{equation} G_j(\psi,\Par) := \rdot{\psi-\sol_\Par}{\J^{-1} z_{\Par,j}} \ \ \forall j=1,...,2d+2. \end{equation} We verify that ($i$) $G\in \C{1}$, ($ii$) $G(\sol_{\Par_0},\Par_0) = 0$ for any $\Par_0 \in \setPar$ and ($iii$) $\left.\partial_\Par G(\sol_{\Par_0},\Par)\right|_{\Par=\Par_0}$ is invertible. $G$ is in $\C{1}$ in $\Par$, since so are $\sol_\Par$ and $z_{\Par,j}$, and $G$ is $\C{1}$ in $\psi$ since it is linear in $\psi$. So ($i$) follows. ($ii$) follows directly from the definition of $G$. To prove ($iii$), we use \eqref{eq:dk} to compute, \begin{equation} \left. \partial_{k} G_j(\sol_{\Par_0},\Par)\right|_{\Par=\Par_0} = -\rdot{z_{\Par_0,k}}{\J^{-1} z_{\Par_0,j}} = -\rdot{z_{\freq_0,k}}{\J^{-1} z_{\freq_0,k}} , \end{equation} where in the last equality we used that $\TB$ is symplectic (see Remark~\ref{rem:uni} below). With the definition of $\iom$ in Section \ref{sec:man} we find \begin{equation} \left. \partial_\Par G(\sol_{\Par_0},\Par)\right|_{\Par=\Par_0} = \left.-\iom\right|_{\{z_{\freq_{0},k}\}_{k=1}^{2d+2}}. \end{equation} Here we use the notation $\left.\iom\right|_{\{z_{\freq_{0},k}\}_{k=1}^{2d+2}}$ to denote the matrix of $\iom$ in the basis $\{z_{\freq_{0},k}\}_{k=1}^{2d+2}$. Thus $\left. \partial_\Par G(\sol_{\Par_0},\Par)\right|_{\Par=\Par_0}$ is invertible for all $\Par_{0}$ by Lemma~\ref{lem:Oinv}. This shows ($iii$). With properties ($i$)--($iii$) the implicit function theorem implies that there exists a unique $\C{1}$ map $\Par = \Par(\psi)$, satisfying $G(\psi,\Par(\psi))=0$ in a neighborhood, $\set{V}_{\Par_0}$, of $\sol_{\Par_0}$. Now take $\Par_0=\{0,0,0,\freq_0\}$, with $\freq_0\in I$. Denote $\Mm:=\{a,\vip,\gamma\}$ so that $\mathcal{S}_\Mm:=\TB$. Then for all $\Mt$ the map $\tilde{\Par}$, defined on $\mathcal{S}_{\Mt}\set{V}_{\Par_0}$, \begin{equation} \tilde{\Par}(\psi):= \Mt\circ \Par(\mathcal{S}_\Mm^{-1}\psi) \end{equation} where $\Mt\circ \{\Mm,\freq\}= \{\Mt\circ \Mm,\freq\}$ and $\Mt\circ\Mm$ is defined by $\mathcal{S}_{\Mt}\circ \mathcal{S}_\Mm=:\mathcal{S}_{\Mt\circ\Mm}$, solves equation \eqref{eq:od3} for any $\psi\in \mathcal{S}_{\Mt}\set{V}_{\Par_0}$. Since the neighborhood $ \mathcal{S}_{\Mt}\set{V}_{\Par_0}$ with $\Mt\in \mathbb{R}^d\times\mathbb{R}^d\times [0,2\pi)$ and $\freq_0\in I$ covers the neighborhood $U_\delta$ the statement of the proposition follows. \qed \end{proof} So if we know that for a given initial condition \eqref{eq:sp1} has a $\set{C}(\mathbb{R},\Hone(\mathbb{R}^d))\cap \C{1}(\mathbb{R},\Sob{-1}(\mathbb{R}^{d}))$ solution $\psi=\psi(t)$ which, for times $0\leq t\leq T$, stays in the neighborhood $U_\delta$, then by Proposition~\ref{prop:IFT} the solitary wave solution parameters $\Par$ trace out a unique $\C{1}$ trajectory $\Par(\psi(t))$. We make the choice $\epsE\ll \delta$. \begin{remark}\label{rem:uni} The operator $\TB$ is a symplectic, or canonical, operator; \ie it leaves $\rdot{\cdot}{\J^{-1}\cdot}$, unchanged: $\rdot{\TB w}{\J^{-1}\TB q}=\rdot{w}{\J^{-1}q}$. This follows from the fact that $[\J,\TB]=0$. Since $z \in \set{T}_{\TB \solw}\Mf$ implies that $\TB^{-1}z\in \set{T}_{\solw}\Mf$, we can write \eqref{eq:od3} in the form \begin{equation} \rdot{\TB^{-1}\psi-\solw}{\J^{-1} z} = 0 \; , \ \forall z \in \set{T}_{\solw} \Mf \; . \label{eq:od4} \end{equation} where $\solw$ and $z$ depend on $\freq(t)$. \end{remark} \section{Equation of motion in the moving frame} \label{sec:mf} Given a solution $\psi$ to \eqref{eq:sp1}, we define the parameterization $\{\Par,w\}$ for it by the equations \begin{equation}\label{w} \psi -\sol_{\Par} \bot \J^{-1}\set{T}_{\sol_{\Par}}\Mf\ \text{and}\ w:=\TB^{-1}(\psi-\sol_{\Par}). \end{equation} By Proposition~\ref{prop:IFT}, this parameterization is well defined as long as $\psi \in U_{\delta}$. In this section we find the equation for the parameters $\{\sigma,w\}$. To this end we use the equation for the function $u$, defined by \begin{equation} u := \TB^{-1}\psi. \label{u} \end{equation} We introduce the anti-self-adjoint generators \begin{equation} \lk_{j} = \partial_{x_j}, \ \ \lk_{d+j} = -\J x_j, \ \ \lk_{2d+1}=-\J,\ \ \lk_{2d+2}=\partial_\freq,\ \ j=1,...,d \label{eq:gen} \end{equation} and the corresponding coefficients \begin{equation} \alpha_{j} = \dot{a}_j- \vip_j, \ \ \alpha_{d+j} = -\frac{1}{2}\dot{\vip}_j - \partial_{x_j} V(a), \ \ j=1,...,d, \label{eq:mu} \end{equation} \begin{equation} \alpha_{2d+1}=\freq-\frac{1}{4}\vip^2+\frac{1}{2}\dot{a}\cdot \vip -V(a)-\dot{\gamma}, \ \ \alpha_{2d+2}=-\dot{\freq}. \label{eq:nu} \end{equation} Let \begin{equation} \Lal = \sum_{j=1}^{2d + 2} \alpha_j \lk_j \ \ \mbox{and}\ \ \Sal = \sum_{j=1}^{2d + 1} \alpha_j \lk_j. \end{equation} The main result in this section is \begin{lemma} If $\psi$ satisfies \eqref{eq:sp1} then $u$ satisfies \begin{equation} \dot{u} = \J \big((-\Laplace +\freq)u - f(u) \big) + \Sal u + \J \VR u, \label{eq:ee104b} \end{equation} where \begin{equation} \VR:= V(x+a) - V(a) - \nabla V(a)\cdot x = \Oh((\eps x)^2), \label{eq:VR} \end{equation} \end{lemma} \begin{proof} For this proof, we will use complex notation, \ie $\J = \i^{-1}$, and in particular we note that $[\J,\TB]=0$. Let $\psi_a(x)=\psi(x+a)$, and let $\phi=\frac{1}{2}\vip\cdot x + \gamma$. With this notation $u=\exp{-\i\phi}\psi_a$. Differentiating \eqref{u}, we find \begin{equation} \i \dot{u} = \exp{-\i \phi}(-\Laplace + \freq + V_a)\psi_a - f(\psi_a) + \i \exp{-\i \phi} \dot{a}\cdot \nabla \psi_a + \\ (\frac{1}{2}\dot{\vip}\cdot x+ \dot{\gamma}-\freq)u \; , \label{eq:sp7} \end{equation} where $V_a(x)=V(x+a)$ and $\exp{-\i\phi}f(\psi_a)=f(u)$, by Condition~\ref{con:g.sym}. We rewrite the potential, using \eqref{eq:VR}, in the form \begin{equation}\label{eq:VVVV} V(x+a)=V(a)+\nabla V(a)\cdot x+ \VR(x). \end{equation} Now use \begin{equation} \exp{-\i \phi} \nabla \psi_a = \nabla(\exp{-\i \phi}\psi_a) + \i \exp{-\i \phi}\psi_a \nabla{\phi} \end{equation} and \begin{align} \exp{-\i \phi}\Laplace \psi_a &= \Laplace(\exp{-\i\phi}\psi_a) + \i \nabla \phi \cdot \nabla (\exp{-\i \phi}\psi_a) + \i \exp{-\i \phi}\nabla \phi \cdot \nabla \psi_a \\ & = \Laplace(\exp{-\i\phi}\psi_a) + 2\i \nabla \phi \cdot \nabla(\exp{-\i\phi}\psi_a) - |\nabla \phi|^2 \exp{-\i\phi}\psi_a \; , \end{align} where $\phi = \frac{1}{2}\vip \cdot x + \gamma$, to conclude \eqref{eq:ee104b}. Note that the parameters and their time derivatives, as well as $V(a)$ and $\nabla V(a)$ are collected into $\Ass$.\qed \end{proof} We now re-parameterize the non-linear Schr\"odinger equation into separate equations for $\Par$ and $w$. Recalling that $u=\sol+w$, $\partial_t \sol = \dot{\freq}\partial_\freq \sol$, $\Ew'(\sol)=0$, and $\Ew'(\sol+w) = \LL w + \NII{w}$, where $\LL=\Ew''(\sol)$, we can rewrite \eqref{eq:ee104b} as \begin{equation}\label{eq:mellan} \dot{\freq} \partial_\freq \sol + \dot{w} = \J\LL w + \J \NII{w} + \Sal (\sol+w) +\J \VR (\sol+w) \end{equation} We collect the linear terms acting on $w$ as $\Le w$, where \begin{equation}\label{eq:Le} \Le := \LL + \VR + \J^{-1}\Sal, \end{equation} and the remaining terms into a source term \begin{equation}\label{eq:f} q(\Par) := \Lal \sol + \J\VR\sol, \end{equation} to obtain \begin{equation} \dot{w} = \J \Le w+ \NII{w} + q(\Par). \end{equation} The equations for the parameters are obtained using the skew orthogonality condition and \eqref{eq:mellan}. Let $z\in \TM$. Upon recalling that $\dotp{\J z}{\J \LL w}=0$, and so \begin{equation} 0=\partial_t \dotp{\J z}{w}= \dotp{\J z}{\dot{w}} + \dot{\freq}\dotp{\J \partial_\freq z}{w}, \end{equation} we find \begin{equation}\nonumber \dot{\freq}\dotp{\J z}{\partial_\freq \sol} - \Ass \cdot \dotp{\J z}{\lss \sol} = \dot{\freq}\dotp{\J \partial_\freq z}{w}+ \dotp{z}{\NII{w}}+\Ass \cdot \dotp{\J z}{\lss w} + \dotp{z}{\VR (\sol+w)}. \end{equation} Recall that $\partial_\freq \sol = \lk_{2d+2}\sol$, so the left-hand side is $\sum_j \dotp{\J z}{\lk_j\sol}\alpha_j=\sum_j \dotp{\J z}{z_j}\alpha_j$, where we used $\lk_j\sol=z_j$. Now let $z$ be one of the basis vectors, $z=z_k$. Then the inner product on the left-hand side coincides with the definition of $(\iom)_{kj}$. Furthermore, using $\adjoint{\lss}=-\lss$, and $[\lss,\J]=0$, we combine $\Ass\cdot\dotp{\J z}{\lss w}$ and and $\dot{\freq}\dotp{\J\partial_\freq z}{w}$, into $\alpha \cdot \dotp{\lk z}{\J w}$. The result is \begin{equation}\label{eq:inter} \sum_{j=1}^{2d+2}(\Omega^{-1})_{kj}\alpha_j = \dotp{z_k}{\NII{w}+\VR (w+\sol)} + \alpha\cdot \dotp{\lk z_k}{\J w} . \end{equation} Replacing $\alpha$ by the explicit expression, \eqref{eq:inter} reads, for $k=2d+1$ and $k=1,..,d$, \begin{align} \dot{\freq}&=(\mm'(\freq))^{-1} \left( \dotp{\sol}{\J \NII{w}+\J\VR w} - \alpha\cdot\dotp{\lk \sol}{w}\right), \\ \frac{1}{2}\dot{v}_k & = -\partial_{x_k}V(a) + (\mm(\freq))^{-1}\left( \dotp{\partial_k\sol}{\NII{w}+\VR w}- \alpha\cdot\dotp{\lk \partial_k \sol}{\J w} \right. \nonumber \\ & \quad \left. +\dotp{\partial_k \sol}{\VR \sol} \right), \end{align} where $\mm(\freq):=2^{-1}\nrm{\sol}^{2}$, and we have used $\dotp{\J \sol}{\VR\sol}=0$. For $k=d+1,...,2d$, $k=2d+2$, we use the expressions for $\dot{\freq}$ and $\dot{v}_k$ obtained above to find \begin{align} \dot{a}_k&=\vip_k+ (\mm(\freq))^{-1}\left[\dotp{x_k \sol}{\J \NII{w}+\J \VR w}+\alpha\cdot\dotp{\lk x_k \sol}{w} \right] , \\ \dot{\gamma} & = \freq-\frac{1}{4}\vip^2+\frac{1}{2}\dot{a}\cdot \vip- V(a)- (\mm'(\freq))^{-1}\left[\dotp{\partial_\freq \sol}{\NII{w}+ \VR w} \right. \nonumber \\ & \left. \quad - \alpha\cdot\dotp{\lk \partial_\freq \sol}{\J w} + \dotp{\partial_\freq \sol}{\VR \sol} \right], \nonumber \end{align} where we used $\dotp{x_k \sol}{\J \VR \sol}=0$, and $\sol(x)=\sol(|x|)$ so that $\dotp{x\sol}{\sol}=0$. Furthermore, observe that all terms containing $w$ and $\VR$ are of higher order. We abbreviate this as follows \begin{equation}\label{eq:end} \dot{\Par} = X(\Par) - \delta X(\Par, w). \end{equation} For the estimates used later, we note that $\dot{\Par_j}-X_j(\Par)=\alpha_j$. We formalize the above calculation in the following Proposition: \begin{proposition}\label{prop:aw} (1) The parameters $\Par$ and the fluctuation $w$ (defined in \eqref{w}) satisfy the equations \begin{equation} \dot{\Par}= X(\Par)-\delta X(\Par,w), \label{eq:Aj} \end{equation} and \begin{equation} \dot{w} = \J \Le w + \J \NII{w}+ q(\Par) \label{eq:dw}. \end{equation} Here $\Le$, $q(\Par)$ and $\NII{w}$ are given by \eqref{eq:Le}, \eqref{eq:f} and \eqref{eq:N} respectively, and (with $\om$ defined in Lemma~\ref{lem:Oinv}) \begin{equation}\label{eq:dX} \delta X_j(\Par ,w)=\sum_{k=1}^{2d+2}(\om)_{jk} [\rdot{z_k}{ \NII{w}+\VR (w+\sol)}+\alpha\cdot \dotp{\lk z_k}{\J w }], \end{equation} $\forall j=1,...,2d+2$, with $\{z_k\}_{k=1}^{2d+2}:=\{\tr, \ga,\gu, \sa\}$, and $\VR$ given by \eqref{eq:VR}. (2) The vector field $\delta X$ satisfies the following estimate for $\nrmHo{w}\leq 1$: \begin{equation} \delta X = \Oh(\anrm{\alpha}\nrm{w} + \eps^2 + \nrmHo{w}^2) \; , \label{eq:OhX} \end{equation} where $\anrm{\alpha} := \max_{j=1,...,2d+2} |\alpha_j|$. \end{proposition} \begin{proof} (1) is the result of the calculation done in \eqref{eq:mellan}--\eqref{eq:end}. In particular \eqref{eq:dX} follows from \eqref{eq:inter}. (2) Estimate \eqref{eq:OhX} follows directly from \eqref{eq:dX}, together with the facts that $\nrm{\VR z_k}=\Oh(\eps^2)$ and $\nrmD{\NII{w}} \leq c\nrmHo{w}^2$ for $\nrmHo{w}\leq 1$ (see \eqref{eq:N}).\qed \end{proof} The goal is to show that $\sup_{t\in (0,T)}|\delta X|=\Oh(\eps^2+\epsE^2)$ and $\sup_{t\in(0,T)} \nrmHo{w}=\Oh(\eps+\epsE)$, for some $T=\Oh(1/(\eps+\epsE^{2}))$. \section{Approximate Conservation of a Lyapunov functional} \label{sec:Lf} In this section, we show that the Lyapunov functional $\Ew(u)-\Ew(\sol_{\freq})$, is approximately conserved. Recall $\sol=\solw$ is the solitary wave profile (see Section~\ref{sec:intro}) and $u$ is the solution $\psi$ of \eref{eq:sp1}, transformed to the moving frame: $u:=\TB^{-1}\psi$ ($\TB$ is defined in Section~\ref{sec:man}). We use the skew-orthogonal decomposition of $u$ (see Section~\ref{sec:on}): $u:=\TB^{-1}\psi = \sol + w$, with \begin{equation}\label{eq:skew} \stfrm_\sol(w,z)=0\ \ \text{for all}\ z\in \TM, \end{equation} provided $\psi\in U_\delta$. The main result of this section is the following \begin{proposition}\label{prop:EE} Let $\psi\in U_\delta$ solve \eref{eq:sp1} and let $u$, $w$ and $\sol$ be defined as above. Then \begin{equation}\label{eq:EE} \partial_t (\Ew(u)-\Ew(\solw)) = \Oh(|\alpha|\nrmHo{w}^2 + \eps^2 \nrmHo{w}+\eps \nrmHo{w}^2). \end{equation} \end{proposition} To prove this proposition we use the following \begin{lemma}\label{lem:tEw} Let $u$ be defined as above. Then \begin{equation} \partial_t \Ew(u) = \frac{1}{2}\dot{\freq}\nrm{u}^2 -\dotp{(\frac{1}{2}\dot{\vip}+\nabla V_a)\i u}{\nabla u}. \end{equation} \end{lemma} \begin{proof}[of Proposition~\ref{prop:EE}] We first recall that $\sol$ is a critical point to $\Ew(u)$, thus \begin{equation} \partial_t \Ew(\sol) = \frac{1}{2}\dot{\freq}\nrm{\sol}^2. \end{equation} Using Lemma~\ref{lem:tEw}, we find \begin{equation}\label{eq:estEE} \partial_t (\Ew(u)-\Ew(\sol)) = A - B. \end{equation} where $A:=2^{-1}\dot{\freq}(\nrm{u}^2-\nrm{\sol}^2)$ and $B:=\dotp{(2^{-1}\dot{\vip}+\nabla V_a)\i u}{\nabla u}$. First we use the decomposition $u=\sol+w$, the condition $0=\dotp{\i \gu}{w}=\dotp{\sol}{w}$ (from \eqref{eq:skew}), and the estimate $|\dot{\freq}|=|\alpha_{2d+2}| \leq \anrm{\alpha}$ to obtain \begin{equation}\label{eq:A} A=\frac{1}{2}\dot{\freq}\nrm{w}^2 = \Oh(|\alpha|\nrm{w}^2). \end{equation} For the term $B$, recall that $0=\dotp{\i \tr}{w}=\dotp{\i \nabla\sol}{w}$, and furthermore that $\dotp{\i q\sol}{\nabla \sol}=0$ for any real-valued function $q\in \Lp{\infty}$. Then \begin{equation} B = \dotp{(\frac{1}{2}\dot{\vip}+\nabla V_a)\i w}{\nabla w} + \dotp{(\nabla V_a)\i \sol}{\nabla w}+\dotp{(\nabla V_a)\i w}{\nabla \sol}. \end{equation} Now, we use that $\nabla V(a)\cdot \dotp{\i w}{\nabla \sol}=0=\nabla V(a) \cdot\dotp{\i \sol}{\nabla w}$, to obtain \begin{multline} B = (\frac{1}{2}\dot{\vip}+\nabla V(a))\cdot\dotp{\i w}{\nabla w} + \dotp{(\nabla V_a - \nabla V(a))\i w}{\nabla w}+\\ \dotp{(\nabla V_a-\nabla V(a))\i \sol}{\nabla w}+\dotp{(\nabla V_a-\nabla V(a))\i w}{\nabla \sol}. \end{multline} Recall that $-\alpha_{d+j}:=\frac{1}{2}\dot{\vip}_j+\partial_j V(a)$ (see \eqref{eq:nu}). Since $|\alpha_j|\leq\anrm{\alpha}$, the first term on the right hand side is $\Oh(\anrm{\alpha}\nrmHo{w}^2)$. Since $\nabla V_a = \Oh(\eps)$ the second term on the right-hand side is $\Oh(\eps\nrmHo{w}^2)$. Finally due to $\nabla V_a-\nabla V(a)=\Oh(\eps^2|x|)$ and $|x|\sol, |x|\nabla \sol\in \Ltwo$, (Condition~\ref{con:g.Omega}) the third and fourth terms are $\Oh(\eps^2\nrmHo{w})$. Collecting these estimates we arrive at \begin{equation}\label{eq:B} B = \Oh(\anrm{\alpha}\nrmHo{w}^2 + \eps^2 \nrmHo{w}+\eps \nrmHo{w}^2). \end{equation} Relations \eqref{eq:estEE}, \eqref{eq:A} and \eqref{eq:B} imply \eqref{eq:EE}. \qed \end{proof} It remains to prove Lemma~\ref{lem:tEw}. To this end we note the following \begin{lemma}\label{lem:REL} Let $\psi$ be a solution to \eref{eq:sp1}. Then \begin{equation}\label{eq:btr} \partial_t \frac{1}{2}\int V |\psi|^2 = \dotp{(\nabla V)\i \psi}{\nabla\psi}. \end{equation} \end{lemma} \begin{proof} \eqref{eq:btr} is obtained by integrating the relation \begin{equation} \partial_t (V|\psi|^2) =\i \nabla \cdot (V\bar\psi \nabla \psi -V \psi \nabla \bar\psi) - \i (\nabla V)\cdot(\bar\psi\nabla\psi-\psi\nabla\bar\psi), \end{equation} which follows from the nonlinear Schr\"odinger equation \eqref{eq:sp1}. \qed \end{proof} \begin{proof}[of Lemma~\ref{lem:tEw}] Note that the identity \begin{equation}\label{eq:eid} \Ham_V(\TB^{-1}\psi) + \frac{1}{2}\freq \nrm{\TB^{-1}\psi}^2 - \frac{1}{2}\int V |\TB^{-1}\psi|^2 = \Ew(u). \end{equation} holds for all $u=\TB^{-1}\psi$. We observe the following relations \begin{equation} \nrm{\TB^{-1}\psi}^2 = \nrm{\psi}^2, \ \ \int V |\TB^{-1}\psi|^2 = \int V_{-a} |\psi|^2 \end{equation} and \begin{equation} 2\Ham_V(\TB^{-1}\psi) = 2\Ham_V(\psi) + \frac{1}{4}\vip^2 \nrm{\psi}^2 - \vip \cdot \dotp{\i \psi}{\nabla \psi} + \int (V_{-a} -V)|\psi|^2 \end{equation} Inserting the above relations into \eqref{eq:eid} we find \begin{equation} 2\Ham_V(\psi)+ (\frac{1}{4}\vip^2 + \freq) \nrm{\psi}^2- \vip \cdot \dotp{\i \psi}{\nabla \psi} - \int V |\psi|^2 = 2\Ew(u). \end{equation} We now take the time derivative of the above relation. Using the fact that $\Ham_V(\psi)$ and $\nrm{\psi}^2$ are conserved quantities, and using Lemma~\ref{lem:REL} and Ehrenfest's theorem \eqref{eq:Ehrenfest} we find \begin{equation} (\frac{1}{2}\dot{\vip}\cdot \vip+\dot{\freq})\nrm{\psi}^2 - \dot{\vip}\cdot\dotp{\i\psi}{\nabla \psi} +\vip\cdot\dotp{\nabla V\psi}{\psi} - 2\dotp{(\nabla V)\i \psi}{\nabla \psi}=2\partial_t \Ew(u). \end{equation} Collecting terms of the form $(\frac{1}{2}\dot{\vip}+\nabla V)$ and replacing $\psi$ with $\TB u$ gives \begin{equation} \dot{\freq}\nrm{u}^2+ \dotp{\i (\dot{\vip}+2\nabla V_a)\cdot \nabla u}{u} =2\partial_t \Ew(u). \end{equation} By observing that $\dotp{q\i\nabla u}{u}=-\dotp{q\i u}{\nabla u}$ for real $q\in\Lp{\infty}$, we arrive at the result of the lemma. \qed \end{proof} \section{Lower bound on the Lyapunov functional} \label{sec:lL} \begin{proposition}\label{prop:Ebbl} Let $\sol$ and $w$ be as in Proposition~\ref{prop:EE}. Then there exist constants $\rho>0$ and $c>0$ independent of $\eps$ and $\epsE$ such that for $\nrmHo{w}\leq 1$, \begin{equation}\label{eq:Ebbl} |\Ew(\sol+w) - \Ew(\sol)| \geq \frac{\rho}{2}\nrmHo{w}^2 - c \nrmHo{w}^3 \; . \end{equation} \end{proposition} \begin{proof} We expand $\Ew(u)$ around $\sol$. Using that $\sol$ is a critical point to $\Ew$ ($\Ew'(\sol)=0$), we write \begin{equation}\label{eq:exp} \Ew(\sol+w) - \Ew(\sol) = \frac{1}{2} \rdot{w}{\LL w} + \RIII{w} \; , \end{equation} where, recall, $\LL:= \Ew''(\sol)$ and where $\RIII{w}$ is defined in \eqref{eq:RIII}. From condition~\ref{con:g.ham} and \eqref{R} we have for $\nrmHo{w}\leq 1$ \begin{equation} |\RIII{w}|\leq c \nrmHo{w}^3 \; . \label{eq:R} \end{equation} Let $\set{Y}_\sol:=\{w\in\Hone(\mathbb{R}^d): \stfrm_\eta(w,z)=0, \forall z\in \TM, \nrmHo{w}=1\}$. It is shown in Appendix \ref{app:BBL} (see \cite{Weinstein1986}) that \begin{equation}\label{eq:rho-eq} \rho:=\inf_{w\in \set{Y}_\sol}\dotp{w}{\LL w}>0. \end{equation} Hence, for $w$ that satisfy \eqref{eq:skew} we have the following coercivity estimate \begin{equation}\label{eq:rhoX} \dotp{w}{\LL w} \geq \rho \nrmHo{w}^2. \end{equation} Using this estimate and the bound \eqref{eq:R} on $R_{\sol}$ in \eqref{eq:exp} we arrive at \eqref{eq:Ebbl}. \qed \end{proof} \begin{remark} Since $\sigma_{\mathrm{ess}}(\LL)=[\freq,\infty)$, we have that $\rho\leq \freq$. We expect that for a wide class of nonlinearities $\rho\geq c\freq$ for some constant $c>0$. \end{remark} \section{Upper bound on $\nrmHo{w}$ and proof of the main result} \label{sec:MT} In this section we prove the main result of the paper, by providing an upper bound on $\nrmHo{w}$. To achieve this, we use both the approximate conservation of $\Ew$ and the lower bound on the Lyapunov functional. For vector functions $s\mapsto w(s)\in \Hone$, and $s\mapsto \alpha(s)\in \mathbb{R}^{2d+2}$ we introduce the norms $\WW := \sup_{s\in [0,t]} \nrmHo{w(s)}$ and $\Anrm{\alpha} := \sup_{s\leq t}\anrm{\alpha(s)}$. We state the main result of this section \begin{proposition}\label{prop:wup} Assume $\eps$, $\epsE$ are sufficiently small. There are constants $c,c'<\infty$, independent of $\eps$ and $\epsE$ such that for $t\leq c(\eps+ \epsE^2)^{-1}$, \begin{align} \nrmHo{w}&\leq c'(\eps + \epsE), \label{eq:wupA}\\ \Anrm{\alpha} &\leq c'(\eps^2 + \epsE^2), \label{eq:wupB} \end{align} where $w=\TB^{-1}\psi-\solw$ and $\alpha_j=\dot\Par_j-X_j(\sigma)$ (see \eqref{eq:Aj}). \end{proposition} Denote \begin{equation}\label{eq:tri} \triangle \En := \En_{\freq_0}(u_0)-\En_{\freq_0}(\sol_{\freq_0}) \; , \end{equation} where $u_0:=\left.\TB^{-1}\psi\right|_{t=0}$. We begin with two simple auxiliary lemmas. \begin{lemma}\label{lem:dE} There exists a constant $c>0$ independent of $\eps$ and $\epsE$ such that $|\triangle \En|$ satisfies the inequality \begin{equation}\label{est:DE} |\triangle \En|\leq c \epsE^2. \end{equation} \end{lemma} \begin{proof} To estimate $\triangle \En$, we use the fact that $\DE_{\freq_0}(\sol_{\freq_0}) = 0$ to obtain \begin{align} 2\triangle \En &= 2(\En_{\freq_0}(\sol_{\freq_0} + w_0) - \En_{\freq_0}(\sol_{\freq_0}) - \rdot{\DE_{\freq_0}(\sol_{\freq_0})}{w_0}) \nonumber \\& = \nrm{\nabla w_0}^2 + \freq_{0}\nrm{w_0}^2 -\RII{w_{0}}. \label{eq:pmt1} \end{align} Since $\nrmHo{w_0}\leq 1$, the estimate \eqref{R} gives \begin{equation} \nrm{\RII{w_0}} \leq c \nrmHo{w_0}^{2} \; . \end{equation} This together with the identity \eqref{eq:pmt1} and the fact that $\nrmHo{w_{0}}\leq \epsE<1$ gives the estimate \eqref{est:DE}.\qed \end{proof} \begin{lemma}\label{lem:wb1} Let $\rho>0$ be the coercivity constant given in \eqref{eq:rho-eq}. There exists a constant $c$ independent of $\eps$ and $\epsE$, such that for $\nrmHo{w}\leq 1$ \begin{equation}\label{eq:L7} \rho \WW^{2} \leq c\epsE^{2} + ct(\eps^{2}\WW+(\eps+\Anrm{\alpha})\WW^{2}) + c\WW^{3} \end{equation} \end{lemma} \begin{proof} Using the approximate conservation of the Lyapunov functional (the time integral of \eqref{eq:EE}), and pulling $\sup$ of norms out of the time integral we obtain \begin{equation} |\Ew(\solw+w) - \Ew(\solw)| \leq |\triangle \En|+ c t (\eps^2 \WW + (\eps+\Anrm{\alpha}) \WW^2) \label{eq:Ettt} \end{equation} where $|\triangle \En|$ is defined in \eqref{eq:tri}. Now we substitute \eqref{eq:Ettt} into the lower bound \eqref{eq:Ebbl}, use the initial condition estimate \eqref{est:DE} and take the $\sup_{s\in[0,t]}$ of the resulting expression to obtain \eqref{eq:L7}.\qed \end{proof} \begin{proof}[of Proposition~\ref{prop:wup}] Using the triangle inequality we derive from \eqref{eq:L7} that the function $X(t):=\WW$ satisfies the equation \begin{equation}\label{e} 0\leq \rho \eps^{2}+ c \epsE^{2} - \rho X^{2}+ c X^{3} \end{equation} for times $c t (\eps + |\alpha|)\leq \rho/4$, provided $X\leq 1$. The graph of the right-hand side of \eqref{e} is shown in Figure \ref{fig:tt}. \psfrag{X}{$X$} \psfrag{p}{$p$} \psfrag{q}{$q$} \psfrag{Q}{$q_0$} \psfrag{1}{$1$} \begin{figure} \centering \includegraphics{ngamma2.eps} \caption{\small Schematic graph of the left-hand side of \eqref{e}.} \label{fig:tt} \end{figure} Thus for $\rho \eps^{2}+c \epsE^{2}0$, implies by a Perron-Frobenius argument (see \cite{Reed+Simon+4}) that \begin{equation} L_2\geq 0\ \text{and}\ \Null{L_2}= \mathbb{C}\sol. \label{eq:nL2} \end{equation} Thus it remains to analyze the operator $L_1$. To begin with, we observe that since $\LL$ has exactly one negative eigenvalue, relation \eqref{eq:nL2} implies that $L_1$ has exactly one negative eigenvalue (the same as $\LL$). First we consider the case $d=1$. Then the zero mode, $\sol'$, of $L_1$ has exactly one zero (at $x=0)$ and consequently (by Sturm-Liouville theory) $L_1$ has exactly one negative eigenvalue, as we concluded above from general considerations. (Remember that the lowest eigenvalue in our case is simple, and that the nonlinearity is local) \begin{theorem}[\cite{Weinstein1985}] Let $d=1$. Then $\Null{L_1}=\mathbb{C}\sol'$. \end{theorem} \begin{proof} The proof follows Weinstein \cite{Weinstein1985}. We know one solution, $\sol'$. The fact that $\sol''(0)\neq 0$ allow us to choose the first linearly independent solution to be $w_1 = \sol'/\sol''(0)$. Then $w_1(0)=0$, $w_1'(0)=1$. Consider a second linearly independent solution, $w_2$, with $w_2(0)=1$. Since the Wronskian \begin{equation} W[w_1,w_2] = w_1w_2'-w_1'w_2\; \end{equation} is constant with respect to $x$, and since \begin{equation} W[w_1,w_2](x=0) = w_1(0) w'_2(0) - w_1'(0) w_2(0) = -w_2(0) = -1 \; , \end{equation} we have $W[w_1,w_2]=-1$ for all $x$, and therefore \begin{equation} \sol'(x)w_2'(x) - \sol''(x)w_2(x) = -\sol''(0) >0 \; . \end{equation} The last equation can be rewritten as \begin{equation} (\frac{w_2}{\sol'})' = \frac{-\sol''(0)}{(\sol')^2}. \end{equation} Now, for $\epsilon>0$ and $x>0$, \begin{equation} w_2(x) = w_2(\epsilon)\frac{\sol'(x)}{\sol'(\epsilon)} - \sol'(x)\sol''(0)\int_\epsilon^{x} \frac{1}{(\sol'(\tilde{r}))^2} \diff \tilde{r} . \end{equation} Since $\sol'(x)<0$ for $x>0$, $\sol''(0)<0$, and $|\sol'(x)|\leq C\exp{-c|x|}$, we have \begin{equation} w_2(x) \leq w_2(\epsilon)\frac{\sol'(x)}{\sol'(\epsilon)} - \sol'(x)\sol''(0)\int_\epsilon^{x} \frac{\diff s}{C^2 \exp{-2cs}} \rightarrow -\infty \ \ \text{as} \ \ x\rightarrow \infty \; . \end{equation} Hence $w_2\notin \Null{L_1}$ and we are done.\qed \end{proof} Next, we consider the case $d\geq 2$. We use the assumption that $\sol$ is spherically symmetric. In this case the operator $L_1$ (and also $L_2$ and $\LL$) is spherically symmetric; \ie it commutes with the action of the rotation group $\RotSpace(d)$ on $\Ltwo(\mathbb{R}^{d})$. As a result it can be decomposed into a direct sum of ordinary differential operators corresponding to the eigenfunction expansion of the Laplacian $\Laplace_{\set{S}^{d-1}}$ on the sphere $\set{S}^{d-1}$. Denote the orthonormal eigenfunctions of $-\Laplace_{\set{S}^{d-1}}$ corresponding to the eigenvalues $\lambda_k=k(d-2+k)$ by $Y_k(\theta)$, $\theta\in \set{S}^{d-1}$. Then the operator $L_1$ can we written as the direct sum \begin{equation} L_1 = \oplus A_{\freq,k} \end{equation} acting on the direct sum $\Ltwo(\mathbb{R}^d,\set{d}^{d}x)=\bigoplus_{k=0}^{\infty} \Ltwo(\mathbb{R}_{+},r^{d-1}\set{d}r)\otimes Y_k(\theta)$, where the operators $A_{\freq,k}$ are defined on $\Ltwo(\mathbb{R}_{+},r^{d-1}\set{d}r)$ by \begin{equation} A_{\freq,k} = -\Laplace_r + \freq + V_k(r), \end{equation} and where $\Laplace_r = \partial_r^2 + (d-1)r^{-1}\partial_r$ is the radial Laplacian in $\mathbb{R}^d$ and $V_k(r):=-f'(\sol)(r)+\lambda_k r^{-2}$. Clearly $A_{\freq,k}^{*}=A_{\freq,k}$ and $\sigma_{\mathrm{ess}}(A_{\freq,k})=[\freq,\infty)$. Now observe that \begin{equation} (\partial_{x_j}\sol)(x) = \hat{x}_j\sol'(r)\in \Ltwo(\mathbb{R}_{+},r^{d-1}\set{d}r)\otimes Y_1(\theta), \end{equation} where $\hat{x}=x|x|^{-1}\in \set{S}^{d-1}$. Hence \begin{equation} \sol'\in \Null{A_{\freq,1}}. \end{equation} Since $\sol'(r)<0$ for $r>0$, we have by an extension of the Perron-Frobenius (or Sturm-Liouville) theory (see \cite{OS1}) that zero is the lowest eigenvalue of $A_{\freq,1}$ and is simple. Thus \begin{equation} A_{\freq,1} \geq 0\ \text{and}\ \Null{A_{\freq,1}}=\mathbb{C}\sol'. \end{equation} Next, since for $k\geq 2$ \begin{equation} A_{\freq,k} - A_{\freq,1} = \frac{\lambda_k-\lambda_1}{r^2}>0, \end{equation} we conclude that $A_{\freq,k}>0$ for $k\geq 2$. \begin{theorem} $\Null{A_{\freq,0}}$ is trivial provided $f(\psi)=h(|\psi|^{2})\psi$ with $h$ satisfying \begin{equation}\label{eq:mic} h'(r)+h''(r)r>0, \ r>0. \end{equation} \end{theorem} \begin{proof} Assume there exists $\xi\in\Ltwo$ such that $\Am\xi = 0$. Then $0$ is the second eigenvalue of $\Am$, and so the corresponding eigenfunction $\xi$ has exactly one zero in $(0,\infty)$, say at $r_{0}$. Observe the following properties \begin{enumerate} \item \label{pA1} $\xi\bot\Range{\Am}$; \item \label{pA2} $\Am \sol = 2h'(\sol^{2})\sol^{3}$; \item \label{pA3} $\Am \partial_{\freq}\sol = -\sol$. \end{enumerate} Properties~\ref{pA2} and~\ref{pA3} follow from the equation \begin{equation} (-\Laplace_{r} + \freq)\sol - f(\sol) = 0 \end{equation} for $\sol$. Properties~\ref{pA1}--\ref{pA3} imply the relation \begin{equation}\label{eq:sstar} \dotp{\xi}{(h'(\sol^{2})\sol^{2}-\alpha)\sol}=0, \ \forall \alpha\in \mathbb{R}. \end{equation} Since $\sol$ is monotonically decreasing from some $\sol(0)>0$ at $0$ to $0$ as $r\rightarrow\infty$ (see \eg \cite{Berestycki+LionsI1983}), and since $h'(s)s$ is monotonically increasing function of $s$ by condition \eqref{eq:mic}, we can choose $\alpha$ such that the monotonically decreasing function $h'((\sol(r))^2)\sol^{2}(r)-\alpha$ has a zero exactly at $r_{0}$. In that case the left hand side of \eqref{eq:sstar} is non-zero, which leads to a contradiction. Thus the equation $\Am \xi = 0$ has no nontrivial solutions in $\Ltwo$. \qed \end{proof} Finally we present the conditions under which McLeod's \cite{McLeod1993} uniqueness proof of positive solitons implies that $\Am$ has trivial null space in the case $d>1$. There exists $\alpha>0$ such that \begin{align} \freq s - sh(s^2)>0,&\ \text{for}\ 00,&\ \text{when}\ s=\alpha, \nonumber \end{align} and for each $S>\alpha$, $\exists \lambda=\lambda(S)\in\set{C}((\alpha,\infty)$, $\mathbb{R}_+)$ such that \begin{equation} K(s,\lambda)\geq 0\ \text{for} \ s\in(0,S), \ \ K(s,\lambda)\leq 0\ \text{for}\ s\in (S,\infty), \end{equation} where \begin{equation} K(s,\lambda)=\freq s + \lambda s^3 h'(s^2) -sh(s^2). \nonumber \end{equation} \section{Coercivity of $\LL$} \label{app:BBL} The goal of this appendix is to prove the following result, essentially due to \cite{Weinstein1986}. \begin{proposition}\label{prop:LL} There is $\rho' > 0$ such that if $w$ satisfies $\tfrm{w}{z}=0$ $\forall z\in \TM$, then \begin{equation} \dotp{w}{\LL w}\geq \rho' \nrmHo{w}^2. \end{equation} \end{proposition} \begin{proof} We break the proof into three steps. The proof utilizes the fact that $\LL$ has exactly one non-degenerate negative eigenvalue and the assumption \ref{con:g.spec}; \ie that $\Null{\LL} = \Span\{(0 , \sol), (\partial_{x_j} \sol, 0 ), j=1,...,d\}. $ \begin{step} Let $\set{X}_1=\{w\in \Hone: \nrm{w}=1\; , \dotp{(\sol,0)}{w}=0\}$. Then \begin{equation} \inf_{w\in \set{X}_1} \rdot{w}{\LL w}=0 \; .\label{eq:L9} \end{equation} \end{step} \begin{proof} Let $\alpha:=\inf_{w\in \set{X}_1}\dotp{w}{\LL w}$, (see \eqref{eq:Lm}). Clearly $\nu\leq \alpha\leq 0$, where $\nu<0$ is the negative eigenvalue of $\LL$ (see Appendix~\ref{app:spec}). That $\alpha\leq 0$ is clear, as $w=(0,\sol)/\nrm{\sol}\in \set{X}_1$ yields $\dotp{w}{\LL w}=0$. Moreover $\alpha\neq \nu$. Indeed if $\alpha=\nu$ then the minimizer, $v$, of \eqref{eq:L9} would be an eigenfunction of $\LL$ corresponding to the smallest eigenvalue, $\nu$. Since $(\sol,0)\bot v$, $(\sol,0)\bot \Null{\LL}$ and since $\nu$ is the only negative eigenvalue of $\LL$ (see Proposition~\ref{prop:neg}), we conclude that $(\sol,0)$ is in the spectral subspace of $\LL$ corresponding to the interval $[\delta,\infty)$ for some $\delta>0$. Therefore $\LL^{-1}(\sol,0)$ is well defined and $\dotp{(\sol,0)}{\LL^{-1}(\sol,0)}>0$. On the other hand the equation $\LL (\partial_\freq \sol,0)=-(\sol,0)$ implies that \begin{equation}\label{eq:D3} \dotp{(\sol,0)}{\LL^{-1}(\sol,0)}=-\mm'(\freq)<0 \end{equation} by Condition \ref{con:g.stab} which contradicts $\dotp{(\sol,0)}{\LL(\sol,0)}>0$. Hence $\alpha=\nu$ is impossible. To show that $\alpha=0$ we use the Euler-Lagrange equations corresponding to \eqref{eq:L9} \begin{equation} \LL w = \alpha w + \beta (\sol,0) \end{equation} where $\alpha$ and $\beta$ are the Lagrange multipliers corresponding to $\nrm{w}=1$ and $\dotp{(\sol,0)}{w}=0$ respectively. Assume $\nu<\alpha<0$. If $\beta=0$, then $\alpha$ would be a negative eigenvalue in $(\nu,0)$ which contradicts that $\nu$ is the only negative eigenvalue. Thus $\beta\neq 0$. Given $\nu<\alpha<0$, $\beta\neq 0$, we can solve the Euler-Lagrange equation as \begin{equation} w= \beta(\LL-\alpha)^{-1}(\sol,0). \end{equation} The inner product of the equation above with $(\sol,0)$, the orthogonality relation $\rdot{w}{(\sol,0)}=0$, and $\beta\neq 0$, give \begin{equation} 0=\rdot{(\sol,0)}{(\LL+|\alpha|)^{-1}(\sol,0)}=:q(|\alpha|). \label{eq:g0} \end{equation} $q(\lambda)$ is analytic in $\lambda\in(0,|\nu|)$, and hence differentiable. Moreover, it is monotonically decreasing, since \begin{equation} q'(\lambda) = -\rdot{(\sol,0)}{(\LL+\lambda)^{-2} (\sol,0)} = -\nrm{(\LL+\lambda)^{-1}(\sol,0)}^2<0. \end{equation} Furthermore by \eqref{eq:D3} $q(0)=\rdot{(\sol,0)}{\LL^{-1}(\sol,0)} <0$. Thus $q(|\alpha|)\neq 0$, for $\alpha\in (\nu,0)$, which contradicts \eqref{eq:g0}. Hence $\alpha=0$.\qed \end{proof} \begin{step}\label{lem:LLp} Let $\set{X}:= \{w\in \Hone(\mathbb{R}^d,\mathbb{C}): \nrm{w}=1,\ \tfrm{w}{z}=0,\ \forall z\in \TM\}$. Then \begin{equation} \inf_{w\in \set{X}} \rdot{w}{\LL w}>0 \; .\label{eq:LL9} \end{equation} \end{step} \begin{proof} The Euler-Lagrange equation corresponding to \eqref{eq:LL9} is \begin{equation} \LL w = \alpha w + \sum_k \gamma_k \J z_k \end{equation} where $\{z_k\}$ is a basis for $\TM$. Here $\alpha$ and $\{\gamma_k\}$ are the Lagrange multipliers corresponding to the constraints $\nrm{w}=1$ and $\tfrm{w}{z_{k}}=0$ $\forall k$ respectively. Note that $\alpha=\dotp{w}{\LL w}$, and that $\set{X}\subset \set{X}_1$, hence $\alpha\geq 0$. Assume that $\alpha=0$, and that one $\gamma_j\neq 0$. Then for at least one $z_k\in \TM$, we have \begin{equation} \dotp{z_k}{\LL w} = \gamma_j (\om)_{jk} \neq 0, \end{equation} for some $k$, which contradicts $\dotp{z_{k}}{\LL w}=\dotp{\LL z_k}{w}=0$ $\forall k$. Here we have used that $\det\om\neq 0$, and that $z_k$ is either a zero-eigenfunction or an associated zero-mode for $\LL$. Thus either $\alpha>0$ or $\gamma_j=0$. Consider the latter case. In this case \begin{equation} \LL w = 0. \end{equation} which implies that $w\in \Null{\LL}$. Since $\Null{\LL}\subset \TM$, the relation $\tfrm{w}{z_k}=0$ for all $z_k\in \TM$, contradicts the non-degeneracy of $\iom$ (see Corollary~\ref{cor:nondeg}). Thus $\alpha>0$.\qed \end{proof} \noindent {\bf Step 3. End of Proof.} \Eref{eq:LL9} implies that there exists a $\rho''>0$ such that \begin{equation} \rdot{w}{\LL w}\geq \rho'' \nrm{w}^2\; , \label{eq:cL} \end{equation} for some $\rho''=\rho''(\freq)$. To improve the coercivity from $\Ltwo$ to $\Hone$, we let $0<\delta<1$, and estimate $\dotp{w}{\LL w}$ using \eqref{eq:cL} as \begin{equation} (1-\delta)\rho''\nrm{w}^2 + \delta \rdot{w}{\LL w} \leq \rdot{w}{\LL w} \; . \end{equation} Upon using the explicit form of $\LL$ we find that \begin{equation} \rdot{w}{\LL w} \geq \nrm{\nabla w}^2 - C_\freq\nrm{w}^2, \end{equation} where \begin{equation} C_\freq = \sup_x (\freq + |f'(\sol)|). \end{equation} The last two estimates with $\delta := \rho''(1+ \rho'' + C_\freq)^{-1}$ imply \begin{equation} \rdot{w}{\LL w} \geq \rho'\nrmHo{w}^2 \; , \end{equation} where $\rho' = \rho''(1+\rho''+C_\freq)^{-1}$. This concludes the proof of Proposition~\ref{prop:LL}.\qed \end{proof} %\bibliography{stein} %\bibliographystyle{habbrv} \newcommand{\SortNoop}[1]{} \begin{thebibliography}{10} \bibitem{Adachi2002} S.~Adachi. \newblock A positive solution of a nonhomogeneous elliptic equation in $\mathbb{R}^{N}$ with {$G$}-invariant nonlinearity. \newblock {\em Comm. PDE.}, 27(1\&2):1--22, 2002, doi:10.1081/PDE-120002781. \bibitem{Arnold1989} V.~I. Arnol'd. \newblock {\em Mathematical methods of classical mechanics}. \newblock Number~60 in Graduate Texts in Mathematics. Springer-Verlag, New York, second edition, 1989. \bibitem{Berestycki+Gallouet+Kavian1983} H.~Berestycki, T.~Gallouet, and O.~Kavian. \newblock {\'E}quations de champs scalaires euclidiens non lin\'eaires dans le plan. \newblock {\em C. R. Acad. Sci. Paris S\'er. I Math.}, 297(5):307--310, 1983. \bibitem{Berestycki+LionsI1983} H.~Berestycki and P.-L. Lions. \newblock Nonlinear scalar field equations. {I}. {E}xistence of a ground state. \newblock {\em Arch Rational Mech. Anal.}, 82(4):313--345, 1983. \bibitem{Berestycki+LionsII1983} H.~Berestycki and P.-L. Lions. \newblock Nonlinear scalar field equations. {II. E}xistence of infinitely many solutions. \newblock {\em Arch Rational Mech. Anal.}, 82(4):347--375, 1983. \bibitem{Berestycki+Lions+Peletier1981} H.~Berestycki, P.-L. Lions, and L.~A. Peletier. \newblock An {ODE} approach to the existence of positive solutions for semilinear problems in {$\mathbb{R}^{N}$}. \newblock {\em Indiana Univ. Math. J.}, 30(1):141--157, 1981. \bibitem{Bronski+Jerrard2000} J.~C. Bronski and R.~L. Jerrard. \newblock Soliton dynamics in a potential. \newblock {\em Math. Res. Lett.}, 7(2-3):329--342, 2000. \bibitem{BP92} V.~S. Buslaev and G.~S. Perel'man. \newblock Scattering for the nonlinear {S}chr{\"o}dinger equation: states that are close to a soliton. \newblock {\em Algebra i Analiz}, 4(6):63--102, 1992. \bibitem{Buslaev+Perelman1995} V.~S. Buslaev and G.~S. Perel'man. \newblock On the stability of solitary waves for nonlinear {S}chr\"{o}dinger equations. \newblock {\em Amer. Math. Soc. Transl. Ser.}, 2(164):74--98, 1995. \bibitem{Buslaev+Sulem2002} V.~S. Buslaev and C.~Sulem. \newblock On asymptotic stability of solitary waves for nonlinear {S}chr{\"o}dinger equations. \newblock {\em Ann. IHP. Analyse Nonlin\'eaire}, 20:419--475, 2003, doi:10.1016/S0294-1449(02)00018-5. \bibitem{Cazenave1996} T.~Cazenave. \newblock {\em An introduction to nonlinear {S}chr\"{o}dinger equations}. \newblock Number~26 in Textos de M\'etodos Matem\'aticos. Instituto de Matematica - UFRJ, Rio de Janeiro, RJ, third edition, 1996. \bibitem{Cazenave+Lions82} T.~Cazenave and P.-L. Lions. \newblock Orbital stability of standing waves for some nonlinear {S}chr{\"o}dinger equations. \newblock {\em Comm. Math. Phys.}, 85(4):549--561, 1982. \bibitem{Cuccagna2001} S.~Cuccagna. \newblock Stabilization of solutions to nonlinear {S}chr\"{o}dinger equations. \newblock {\em Comm. Pure Appl. Math.}, 54(9):1110--1145, 2001, doi:10.1002/cpa.1018. \bibitem{Cuccagna2002} S.~Cuccagna. \newblock Asymptotic stability of the ground states of the nonlinear {S}chr{\"o}dinger equation. \newblock {\em Rend. Istit. Mat. Univ. Trieste}, 32(suppl. 1):105--118, 2002. \bibitem{Derks+Groesen} G.~Derks and E.~van Groesen. \newblock Energy propagation in dissipative systems. {P}art {II}: {C}entrovelocity for nonlinear wave equations. \newblock {\em Wave Motion}, 15:159--172, 1992. \bibitem{Frohlich+Tsai+Yau2000} J.~Fr\"{o}hlich, T.-P. Tsai, and H.-T. Yau. \newblock On a classical limit of quantum theory and the non-linear {H}artree equation. \newblock {\em Geom. Funct. Anal.}, Special Volume, Part I:57--78, 2000. \bibitem{Frohlich+Tsai+Yau2002} J.~Fr\"{o}hlich, T.-P. Tsai, and H.-T. Yau. \newblock On the point-particle ({N}ewtonian) limit of the non-linear {H}artree equation. \newblock {\em Comm. Math. Phys.}, 225(2):223--274, 2002, doi:10.1007/s002200100579. \bibitem{Ginibre+Velo1979} J.~Ginibre and G.~Velo. \newblock On a class of nonlinear {S}chr\"{o}dinger equations. {I,II}. \newblock {\em J. Func. Anal.}, 32:1--71, 1979. \bibitem{Ginibre+Velo1980} J.~Ginibre and G.~Velo. \newblock On a class of nonlinear {S}chr{\"o}dinger equation with nonlocal interaction. \newblock {\em Math. Z.}, 170(2):109--136, 1980. \bibitem{Grillakis+Shatah+Strauss1987} M.~Grillakis, H.~Shatah, and W.~Strauss. \newblock Stability theory of solitary waves in the presence of symmetry. {I}. \newblock {\em J. Funct. Anal.}, 74(1):160--197, 1987. \bibitem{Grillakis+Shatah+Strauss1990} M.~Grillakis, H.~Shatah, and W.~Strauss. \newblock Stability theory of solitary waves in the presence of symmetry. {II}. \newblock {\em J. Funct. Anal.}, 94(2):308--348, 1990. \bibitem{Groesen+Mainardi} E.~S. Groesen and F.~Mainardi. \newblock Energy propagation in dissipative systems. {P}art~{I}: {C}entrovelocity for linear systems. \newblock {\em Wave Motion}, 11:201--209, 1989. \bibitem{GS} S.~Gustafson and I.~M. Sigal. \newblock Dynamics of magnetic vortices. \newblock preprint, 2003. \bibitem{Jones+Kupper1986} C.~K. R.~T. Jones and T.~K\"{u}pper. \newblock On the infinitely many solutions of a semilinear elliptic equation. \newblock {\em SIAM J. Math. Anal.}, 17(4):803--835, 1986. \bibitem{Kato1987} T.~Kato. \newblock On nonlinear {S}chr\"{o}dinger equations. \newblock {\em Ann. IHP. Physique Th\'eorique}, 46:113--129, 1987. \bibitem{Li+Li1993} C.~Li and Y.~Y. Li. \newblock Nonautonomous nonlinear scalar field equations in {$\mathbb{R}\sp 2$}. \newblock {\em J. Diff. Eqn.}, 103(2):421--436, 1993. \bibitem{Li1990} Y.~Y. Li. \newblock Nonautonomous nonlinear scalar field equations. \newblock {\em Indiana Univ. Math. J.}, 39(2):283--301, 1990. \bibitem{Lions1986b} P.-L. Lions. \newblock On positive solutions of semilinear elliptic equations in unbounded domains. nonlinear diffusion equations and their equilibrium states, {II} ({B}erkeley, {CA, 1986}. \newblock {\em Math. Sci. Res. Inst. Publ.}, 13:85--122, 1988. \bibitem{McLeod1993} K.~Mc{L}eod. \newblock Uniqueness of positive radial solutions of {$\Laplace u + f(u)=0$} in {$\mathbb{R}^n$}, {II}. \newblock {\em Am. Math. Soc.}, 339(2):495--505, 1993. \bibitem{McLeod+Serrin1987} K.~Mc{L}eod and J.~Serrin. \newblock Uniqueness of positive radial solutions of {$\Laplace u + f(u)=0$} in {$\mathbb{R}^n$}. \newblock {\em Arch. Rational Mech. Anal.}, 99(2):115--145, 1987. \bibitem{OS1} Y.~N. Ovchinnikov and I.~M. Sigal. \newblock Ginzburg-{L}andau equation. {I}. {S}tatic vortices. \newblock In {\em Partial differential equations and their applications (Toronto, ON, 1995)}, number~12 in CRM Proc. Lecture Notes, pages 199--220. Amer. Math. Soc., Providence, RI, 1997. \bibitem{Peletier+Serrin1983} L.~A. Peletier and J.~Serrin. \newblock Uniqueness of positive solutions of semilinear equations in {$\mathbb{R}^{n}$}. \newblock {\em Arch. Rational Mech. Anal.}, 81(2):181--197, 1983. \bibitem{Pelinovsky1996} D.~E. Pelinovsky, V.~V. Afanasjev, and Y.~S. Kivshar. \newblock Nonlinear theory of oscillating, decaying, and collapsing solitons in the general nonlinear {S}chr{\"o}dinger equation. \newblock {\em Phys. Rev. E}, 53(2):1940--53, 1996, doi:10.1103/PhysRevE.53.1940. \bibitem{Pelinovsky+Grimshaw1997} D.~E. Pelinovsky and R.~H.~J. Grimshaw. \newblock Asymptotic methods in soliton stability theory. \newblock In L.~Debnath and S.~R. Choudhury, editors, {\em Nonlinear instability analysis}, volume~12, pages 245--312. Comput. Mech., Southampton, 1997. \bibitem{Perelman2001} G.~S. Perel'man. \newblock Preprint, 2001. \bibitem{Reed+Simon+4} M.~Reed and B.~Simon. \newblock {\em Methods of Modern Mathematical Physics. {IV A}nalysis of Operators}. \newblock Academic Press, New York, 1978. \bibitem{RSS} I.~Rodnianski, W.~Schlag, and A.~Soffer. \newblock Asymptotic stability of $n$-soliton states of {NLS}. \newblock preprint, 2003, ArXiv:math.AP/0309114. \bibitem{Soffer+Weinstein1990} A.~Soffer\SortNoop{A} and M.~I. Weinstein. \newblock Multichannel nonlinear scattering for nonintegrable equations. \newblock {\em Comm. Math. Phys.}, 133(1):119--146, 1990. \bibitem{Soffer+Weinstein1992} A.~Soffer\SortNoop{B} and M.~I. Weinstein. \newblock Multichannel nonlinear scattering for non{\-}integrable equations {II}. {T}he case of anisotropic potentials and data. \newblock {\em J. Differential Equations}, 98(2):376--390, 1992. \bibitem{Soffer+Weinstein2003} A.~Soffer\SortNoop{C} and M.~I. Weinstein. \newblock Selection of the ground state for nonlinear {S}chr\"odinger equations. \newblock Preprint, 2003, ArXiv:nlin.PS/0308020. \bibitem{Strauss1977} W.~A. Strauss. \newblock Existence of solitary waves in higher dimensions. \newblock {\em Comm. Math. Phys.}, 55(2):149--162, 1977. \bibitem{Stuart2001} D.~M.~A. Stuart. \newblock Modulation approach to stability of non-topological solitions in semilinear wave equations. \newblock {\em J. Math. Pures Appl.}, 80(1):51--83, 2001, doi:10.1016/S0021-7824(00)01189-2. \bibitem{Sulem1999} C.~Sulem and P.-L. Sulem. \newblock {\em The Nonlinear {S}chr{\"{o}}dinger Equation. Self-Focusing and Wave Collapse}. \newblock Number 139 in Applied Mathematical Sciences. Springer, New York, 1999. \bibitem{Tsai+Yau2002} T.-P. Tsai and H.-T. Yau. \newblock Asymptotic dynamics of nonlinear {S}chr\"{o}dinger equations: resonance-dominated and dispersion-dominated solutions. \newblock {\em Comm. Pure Appl. Math.}, 55(2):153--216, 2002, doi:10.1002/cpa.3012. \bibitem{Tsai+Yau2002b} T.-P. Tsai and H.-T. Yau. \newblock Relaxation of excited states in nonlinear {S}chr\"odinger equations. \newblock {\em Int. Math. Res. Not.}, 2002(31):1629--1673, 2002, doi:10.1155/S1073792802201063. \bibitem{Tsai+Yau2002c} T.-P. Tsai and H.-T. Yau. \newblock Stable directions for excited states of nonlinear {S}chr{\"o}dinger equations. \newblock {\em Comm. PDE}, 27(11-12):2363--2402, 2002, doi:10.1081/PDE-120016161. \bibitem{Weinstein1985} M.~I. Weinstein. \newblock Modulational stability of ground states of nonlinear {S}chr\"{o}dinger equations. \newblock {\em SIAM J. Math. Anal.}, 16(3):472--491, 1985. \bibitem{Weinstein1986} M.~I. Weinstein. \newblock Lyapunov stability of ground states of nonlinear dispersive evolution equations. \newblock {\em Comm. Pure Appl. Math.}, XXXIX:51--68, 1986. \end{thebibliography} \end{document} ---------------0309260838251--