From 50e63d8234e8465eb37e057d08766eaf98a6e79d Mon Sep 17 00:00:00 2001
From: Johannes Schwab PI Sjors Scheres added 22022021
 <schwab@lg49.lmb.internal>
Date: Tue, 31 Oct 2023 15:17:00 +0000
Subject: [PATCH] continuation

---
 .../optimize_deformations.cpython-310.pyc     | Bin 23209 -> 23455 bytes
 .../deformations/optimize_deformations.py     |  75 +++++++++++-------
 2 files changed, 47 insertions(+), 28 deletions(-)

diff --git a/dynamight/deformations/__pycache__/optimize_deformations.cpython-310.pyc b/dynamight/deformations/__pycache__/optimize_deformations.cpython-310.pyc
index 5d60935056228480e4a0a3fbc0e1b31024afa1a5..7817e2225ee96a7f5732f408885bdfa240bcd4be 100644
GIT binary patch
delta 7255
zcma)Bdw5humcO?<Nhjn%5)xua!ow!YOAwF&A-seH^MJ@Rq#I}g6_5^e;Py>`aLNXn
zz$mt9xgZZ==pIEtMZlJo$L@?cJLB$*qYk<<8=Y@vcXW0f*U|Btb^qDH{Z-xDO^4tp
z`MU0@Q>V_UbLyN^r;5`T)Z?$JeOt4$`zZM5%$(WWZd1?qeIO%!M7g4b^E=vNjTx`o
zY?Ko#%W~TSc1_i1lY>UkXnHW1vMhx!<&E6M-FmursJEQvMee2%QI(~#{E&@Yp32+e
z-Elm-S5J>r(StNUYBOgRC~T}XcaAw16c5HJ0zH^ln*@J{iQ97<CP_Tpo@$zpW!9i|
zdun_W^jau3hdG0Gb2)rz4~+3jPXF>cT13?}kBvK;#h2-G*{kamTK25^glbK337qkH
zPO&{j8)szpEf0L!;UO2f*Vtw&%jfbHw8A@&7P9dnRaNM&NIknxe~4V(N~SQS)5aWY
zY{g1a-zoaiPN4BlV~%147M{Sq7fN9hqjrnOB=B(G!o%v{3#K=j0C_tmVf&^<ER`?U
zOWFO#Z~GQ5o*%Yv{a97FM^!pTz|#F91yMy+Fg}?)K^q{k<qIMUp~{qEg;hjtT)<O{
z6?|RZMXYoH)bK3AvP{_RS!7h_mJPT2>l>hI3jTsvzAs~}61Mxl+6;4uzefBhEi8=J
zl{?&?;x9C=<d#=2rj^*E#pH@C5$hng`Nevs1%)B+*86yKy_qd`%&cT%g#kSc_(ES#
z11*g#jVio}8Y4@M;d$l#8nGdL+4PW&H|ve$7O>HjcTd1YOIQK0>!95*?{LhHW4so|
z<#n?ev;>YS=voqV$=LPSu4Qm)m2XHuK7BcStAIwZnISuGrB=Q%LlJ+>Nuyb5Kcz(B
zn~d!ILif*^gXgdZxQ<^Ne(m^eX0!Mf@LCa7x|-wKw%Ed{T7l4%Mi(eFra<9ag}#Yi
z2i5a!_z`A$@LOr{{6S5#u~Qr}9+6gR@~*P%Bq86Ou<&Zj!sgc%VOC#$Vev}HY>upn
zDy&NEATV<vH-RzKLHz=${+T^?bg6eO50HZ?a@9W*cG1gda!`eprdhoT*_(v5Z3jl^
z>XzIe6g#XxOf!QjLa1b!jql()X(y|OuLUVc(`Y;Xf~H>=A&4pzqpM%u$QH@zVSOFV
z($l;SRui&)1BXcl@8?{&kETVu@RMmYNq#5w^3zoR)nu)($|`6vO0IhsbI>lDR-v#`
zd?!_iplvZ`7L;eMCsk?yos3NdMfW?jDc@#gN`s~B;WTX$by7}Cih%SD#?^u`g}z8D
z*0>KAHzC7Uq;JH?s2t{m$SD*?dl#!UrWY0lYNKk@2G<H%V^5f47Iwdd?dpL&kPvwQ
z_)YXUw&*Ddg#aCt(0)LNBy<4KVc)>pQ`TYnEu@lgHzw_t&>ld00BNj_!Ue)+ehNp`
z5+pf(#KJ#m;*CN53j+>qGO~k1OV@RHq0jX75|J?5$jRnXC!6<l3adY6D?&uk5%lZ&
zE9{|ADm+LRHcQNK%Ng@;nelY;4BwBW6f2^Hl)!fsb92gf5wjPHR#811*eSBRb&?31
z--<0WH{e#ZxfxYPeNkbcN@O03g|KmQe-G80Bgo9{3D}(kyF6`4_%@MUgumfC06E6x
z-Ha@Al#WupIEZji2?&cWR_;*vF-Y?S(5$3mBBe2B)W(o*5+X$klZoiI<8IHEUIFLu
z|0+4)=)TBSnT#e_-%jR0a5%Itk%Sh=ARAxG7Ge(<kw2Isvxh`Wm$rq|wOPpbx6S><
zF19$4*OT%;TWoxAM{#+&Hie}MYam;uiY)7C@TK2+&KSkTLjoO<4#ekk`Vg7nkQ*ql
z)&_4rc@x8&5_1aXSO%Uo7(Fdvw}hVuyp2u+`+|g?2lS%UK!FBF<yn`B^RDk2Tf4-F
zm3c_4%R_-)%bYBhiD4XulE~$qV<};NU0(AL5NgFCm$%xY!MB&dw|dFAX^KGn%thW}
z&LX8ntnf4Qd`9k1*Z#<yd@6oteAAJir{Z_UH<_mLmvl~(sgjmrf5y`|YNe^Hgo<bw
zJw=Dul2A53OLeS)Uc%YCn9dd`MGAazDxHxtr(mWy$;TV74#-<)QqS#0eMXPx;!?Jh
zrph{|lr_>B0l{k=SYd{Cjg95YX*nBf_Bv`U1B(6cVKV&@V{A!D;WlcGs&tYxno9ZB
zdR<0S$)E{tDiWN{{egMscpFR}rm1C5n!G{r%}7!7`QePMZyFa$%GG7Y2PKWE8UC44
zMt`Hev|pYpRIn_SpXBHGdHynw=>fxEIwa6XtDs!0+#VLtIkp^8sZ)Cq&Y==veSxh>
zjEg$yrtgTcPO3tN6q=(atGaS}C8>ljWh+Erk<Qyq?x;*0`~s}y5%Z>riT)*0ge{z@
zRa32GCZL|efPp)UV;zMh4#4eECDLllFDt)aqcR#JQhre7m*^5JO)Mygo^ztYFJWdC
zbD~(nI9-aS^2@-q7-!4Mig(Zz@yAxOrclZ<JO4-iPy9z1-D!MWR-Lz!E=SceMbL|a
z)f+UX4=fApLQ1=A<t4})ps-aa%U(k%wi@}Qk9NbN?xfdPDPt)0UK0>jGeMMUEh+pW
zU6jS{MSZ$=H{HpaVGwcjWSB<}47es_>*~)J(1OSwx)>P=Z6P%>DhfSIVSq8T!1MiH
z2`|$>-uT~vWn(k_vyu~^&&CvfG48#+47(X#QaNa3zqO%4=^<D7>-;DDpM7N#KWq_;
zyN)%0Jo>1WN98xpSbq%Rw;BOB3jsJDA%!4DAPU#cXAg(6apOD5f69Nx-_Rd3z8+N4
z5Xb!s|5yHV8YhklsVTJNzd*v<i{In)3%d6XXpJ=XCXMew^UDOyFKMhr^OhbmHVrO!
z-?&B1EbWHn$XVLZB)3IR574Nj)bTRHuqc?p-|_W{yD7pWOB*c`IDZF$^OwQJ0TC!|
zv^fzdS=xN*yMhvBK_V=&{7VyDY#2jPFK@fQvVuh1XG_^S86<6dGhG!Sh|=A`uUS!%
zrER7<UZ-bybNMhUKC%MZ4w}p7O62ffahn1`FDBF#**_ko0)IgU#Rc(&yJl%8sD|?H
zS9lF(EJn<O2I6bXr6egSXs=sJ1_YSQ_BvgYY_Ey0Wcv=)O18C_VzRvswzUbZZp3xL
z&d3XxwDJvVkQpFL`;t2OyYw!bFhJp3fEAJ409$zhw*+VlTOUMShI%er%jk8plc^$X
z@A@B5oMRhs36}Q-6tnL^FS9=-h)R+F+bxcgtxfG+Uwhll@`}HY+$-RPC|me-c!*VZ
zVIU4&*Pp;0AD612ip$2Y`M)EZbisGR^lRh{m-or2x>m(8<iJ(x#~9m(yd+OiqtWTK
zH+c;s;4E(14?j8JMHPqkAa%2e%(vFYT0=Gq+Q>GA?1+gE^h21vNXq(<|HhKl1Zf}k
zlJ;Q_X+wq<7}-Pfq{JaZoz5I>2zE4wY4X?=y+JG$`56$KO&hV|*~Z$TWmEm)z(P)J
zg`Ozvq~zc7-%+cz*5AeD-dr=B3zt?%Uuj$&T2$<Z?f!#aXX5*xbe`$>Za`l*{xY;O
zu&THDBdo_I^02g*VOotw2krd#`eAt}8s^a`aUUT&9HHOSN3a`eN&Ye9KaFeFXy&{_
zVQo=6|3fd#e^D&J+VLWR`jlUx7~5>Ye>9IA{t4a=aP;s$=||}kK_${M{V4&B3EHvg
zpYs2vE0JSS1;)GV%QUwVhm4O+#A4*ufqN%&T1Qo`_J1h7N)a|WTHpo-wpar>+TX;U
z$brAf_nxi#50FuJ;SHyQZDSNlrOw1V4wqw-keP6W#KLc(@n!eo@JJ{nG5<IVAXK)e
zVft0J-MDmD*}l8e93sylCyDO;5w_DzH3%kLr*Wl`*$TY}M2`iJ+LE~K0&S<9-6>{g
zA7{IP*&{J~1SW8tec#+eX+A6aCvkYx+|6$oeT?2uqTNf#N0udyv&T_P-Ry<vOj7R&
zq&qnlfH{t{E~KfOq+xn9srMwkmFU&8eR|Y%as7wz49iOZ8K#d2PYBu~&zOhUP3K1q
zALn1s_9yWOB;1Qyzjd;>(%Zth{zB;wYB(2{d-48v;M?E-NL=Lnv|C(^#rbN!|50a-
zFqdy|oQ;t=d`RFnWYVZ{@v9QzJLSBmzWM1e{&Gm-hVCca)q9Fhk}kLy;tA7uVR#AF
za9dg_?;t2+?2xc{;>ptvm+Fagt^tl6W{br`XOi`gB&?qvF%#^wH+M3=V-+D>xa2lk
z9vL4H4?s!Gp4Ly$dp$5m^%KU95hZ~<zazQ)(&eR&p2H>ApA_49{&H%dEr@LabyBg&
zOW5OltsVwLr5jeeM5$-iy4bgP@)3K81eET54nI+qo<tN?!yw{xKpQ~8`Pn*!9iyd&
z%6>qNXpfT%ZG_xtqx3e~AL3o}R8#@<3|=-(=vfFep%ch+CiEP=Z9yk#wFR9bW<jTE
zodrElUJL5`0<E>c7wO^1DdY6WlE9gxsV7wa=Y+<mk~GfKVl<S!gu~bY%}>*x31=f*
zGlx*A$-4OLG4T%7V-ycI=d985X4T9;*Pn+aD(Q1PmX4v%*vs<aMpoDX7PE+Zme^%h
zi7j(Egzq)8y7bjV!HrG0eZ8IUPwlbMW-<TDZ6wXIb;U)=!$OLZ(03L!S(NsHe1A<3
zk=B(|SQDO<udNWJo2aw!%ayg1z-pG=0qHN0$}Z4A@%XAL;tH`)38!xQg9*uNWd7Fu
zs=~j(<9(i9N?(ZDRL=Sb8fZlCDN`@Tetgdad)g=C@c%wl#$WG!EEX8sqN)`8blm&4
zw5Nd84jOMv_;lnU0Uj33L!vn%8gm#>ce>WR#<QWx-!j~$9Ws9Ny*=tPM)Sl+({>B8
zJ;q-r-j!*BqJJt@KIvI|T8DtQ8E;RSp}uJ3R$Ns(V%I7bsDp!oVM?p7+23t%+OV-Z
zRcrDwUw0bd_9ji+Vho-(E9RSaSXH+f=`+d;y8CoAYps6I`gJYM-D#Vfv<+<=jfFD~
zradAAQ0$8tV{9rJgJ<1miHx~s4O7dX#HcH`Z6j;an>}l^=1pzQzSUc`92Jw~%$*=G
zuFfttUYpG-z7o8)py{@6&{n%rs@q%FYdb{WR=2>Fy0O(?x1c+9%{pJxdhI1adsZ~(
zM04I)Sverv5VWbyn%1gmUmJTWhoxN<gO`j8mBTVF3-AgWqfeF7$fydd`;8-2@2Y!a
zzBymn)JUwkroXDb9HZJAHSMY(zG}Q#=TqN{O_|%Ij>`Hv&E7{%|2oZPw*3Wc&zEg7
zoUZ2EW085a=_P(>IecEX!{cGCt34j=CxYisMN?v=H4G~JNPyj<c|$Z`isr4@<c3kU
f@;ilrt^?7XvZ@tE5UezFO@c~7T5NCQK=uCt3bAx4

delta 6951
zcma)Bdw5humcO?<Z$cg<5JD0lHc?)JF(^^q@)8o>kC0BVjZ}biNOyAk5+Izig(jeA
zL&KFf!EOhMs3_Png3P+0&dl!4FpfH-V{pF_b!K!N8F%-a-9KjcoBdVYn@)$|r{wFp
zr%s)!bLzb6RPoH~>gn@prni6pGzI^z{^#`i;(sXTG9O4ySwBKiLIc`zqt&Tz*=#I1
zT9jU6E4681eHJCtXc|Wk1`=u$_$pq_oxCQbdP?=#z|LikfIWN%jgF{%l|I{O%q%XM
zL(6F%Eo65cPv^DzT=wGzg=$Z#dsK_gDR9QQ%zV3DJ8E3bEb;GdbCHv3*4kz%s~7Nx
z=ppxmw2a*yR8@s;56@#|`h0S_E11HRb_loDmaoM19ip$NgewKkJ(VTrD-o4VVD|(Q
z*u4?EwZ=rOQFGH8R{t(6+Or6dw^}4*Yxysh$XDxytlWrX4P3DxWZ$wwRk%x4+D7Xs
z?woLLL{Sxt-$$;14N!`EgnMCl5w>!FzQQI&Y<wBu$@vPtPWNI~I0V~pErwVoT;p17
z^vf<9W%J#r!d4UT7l8Pl)KVp6_Z`>-b%?)e{3t18*WNRhWRJ4jeXkijyJX%HdYEcl
zOUM~sD&(LV^9y;V1cV~*(9_&m?tYDS%>06t3T3&L;fwvcmeH#4@`%Fgs5-pd_&U2J
zX9WzB$)*Nvyk1{HH3BxO^6&Joq?Ife*aqxwq<a)*cVpZZ#_6tN(`aR`BIz1qbn(~?
zFxM(t0;g$;K_0yt4wXxz+4P{Dd&$e2Qx);Y95jv<<|qXUZ!z8-FtDb9CG!k6gX{Qh
z#IF^<P3!@_8EZKs%HI0!-8R}n30hR_l*Z*MG(K11Tf}~2y*66Jx8g^bS%Y7#F)e3U
z-Aq`DgU6#%ORL=vS$dL?=Qe2AWog*_I>QfJeffpPXMty3_>qXhW(x}fGY5QQ7~{P=
zbLta(pZ2ooD)$=hr(~u;>O`$x=tVE0`vWSpG}Y=&mc4Sw_82f?uQhV@xx&IaqZt7e
zAyiOn<J<Ul+Rh#{CgzSAlP2kzUQaXiBzH2K7qs0>2cHa3fXa<&xsL2oni^gUZ<tEu
z@>_0P$;lc2aJ;nOUJ2>ON_lp$WZFSfCo8NF-}1>K7}pv0fh9HTNtKdGSwrgHYnEmD
z!51^v+FF%RlGgQo`lKZ2#15T&MD3K(m>?j%!T9;Wv4f#zceoLv?uU9!$nfOp8!&jE
zI^R$S4fM~Cs1X}nC&(NSYZC22dpqmKK5b$5_QD>p7$s>FfN!FwV1k1Z3IaMLp{D^I
zme2t}M?6DsO<9TQH<3!hotU&!Lc0L%0_0&86v`F4^HM0H79ff7qZWRji8roeLrL1l
zi2fw4Q*RdAH+{K4B*YeSuu5uY3yvkQg&np$L<}88zhaQW76lXGA<aTfu^BaT#^Rf1
z9E+dfc~f$+B1Ulep=^4l{17%zVxyJDyLkietfWOqzYdYZt@A?YyDwJgzS77ZJkV7s
z5{X4CR2tvkMT^W4Bw8&7*Tlh2S4$ke1!}Ai9)<4^qztFqXS5F<Hbo`}>JVv&IU+WO
z6cJlj2hZ;wuW{+d^}#u7n&ZQVCWJT2lyQ&sZD+}F>||{pQpFM(GKFj@46%%Sfdm=i
z5-a`E2D;CXly8}!^H;LvvE&<<q}g(#BR{_+S?gknVvk5^i6RxcmU&WcJ`9X&`NjU$
za4X{KlwK^;5CWb2>uhkz&K?YNNX+xt|0=Ax38SYaTqEHZ0B@$#z`iJ<7XZB^cc6q)
zv`O+n>PZB^2+3Sqgj}vI{yt*fFT}(!PAo~}bhlZXFuzW>nW@FLx*?}~t3`uvABIhR
zc=V{Excgv4%3+Q?rBTTIlzE6EL8oZDjDmvvNksS4Y{KrTjP52cO<@I;M<eMV9cC+n
z{rM|Y$yU<KI1HE5E4fOZf+J`Oo#8L*WOxe-{3gwUJ~U_a?$lYxs%VNVCkojrIwK(X
zMKWYa)mm67Urno7so86*uv8>0+)F7shm30=<vMDhMw$Ux%OH2VpCwDvcq1xxl15W#
zfF*o&Ft;|5ck;9R9DkKZW#toJ<2Wf#h+>)y8gFxpng&cw?aQ9DLY3l~o}lOpLaAFC
zRkK<u(+g1{i4)6syf9a-F**vX6H|Pj>fHEy;i&AJSnH1(_Y@WTQ?+Z9g?-zx?`w3H
z)uIN}sWlIWPJytFz|N12i(;q8x5QXG%|RM8^>$5`W+n6*OJoCrHn>#<;%>DNiFDp-
zu8#5|nV*NQ9}@GX3KjeQCRx}*Y1)@mAr})+Z)U(ir{Zk>1D11!{ShVHXygnnxz9sI
zG+yNOfXXk@MOJ7Dl&0N=o@1iIFJfjTbD-kEI9-e+@;884XS570%72V5i9hCI4+j%!
z?ffVFpZJ>?-DY$SotIrpZ$#80MbL|?)4knLhZXq)NMvtVi3<FND69^()n(LEk09No
zX&n~z4!X<=*&5V!mjwiACWv~gF@ay83$kLppigtR(`}4FA>w>VHP06)us&$pJBTl(
zrQw})Av^}V1=sM{2=-YB1&pVqu3de~*)-q##>HVpd1=1S;`5@%#uR=*UvGRfEYCg&
z{*XL;OwQWiK<SEC_*?vK{*I?e;zurL-Hu`nAgMlXB~|&2h0qh0^Uv&dd&)ql3V$EK
z)&=|HTGPpY%74axu0LU18D3D;jr$k=ul(QWZgCvQePK`h7f64*@q3bfL3a(t9+A|3
zNoBoguEuDtQmIArEBz^>thmJAag(>CYaN!`_S4>ub5iv5M`>(aa(NZ8nHNaqzxJec
zJ0*glpLSKGbN*|Kx*D~*NJVd3)bG(187doTlL!^RjFo=cx6*yZ^l#&t`u-=jv@Mjb
zrCV|Go)srGGEUaZIN8WInM<PtPv-Aifsw8aqDo$=r@OQGNGn9rwc)gYFOWF@s6L^a
z07ey-4g}N_sG@gX#>IK@g~z6AvuHkL->2~Tn6VsT4;rwa-+PzBb1~9tZBjtis(ysA
z=)NzPeP4Vfsh29`vK5$OF58M_O|is`j@VvI17<<>0$U}MK)SY%+W80c0h=&H;hTXK
z!CZx$(g4^^n^|K3<rm5}f4cU3pBtK49ozT*4J2pT23%%k-H)R6H`tfipA<kv$bZ|%
zdp_w~J-aotfZ`t_7k9%S!D~=h86P;ZCkAmn$9*WE;-2t3{xNdM9{A8+`W-Tb(~a75
zor(i18MmN+#@Ie&BYCzO1tT2x)$aYqj1l>D23+H5_W{`+q%L+ZYg%Vxo}kTwysSBB
zM{Hcv4`KFV$?FsT?-sAs;Py!$ZlCnxc4<V3<o5XphoirC85Yc7E%G21y<sd7rUb-h
zQZuBTX^bi<^5fPhv!GWGg6j~;_4oW!@>+6!+Rd{^UicDbfpaZ+MR_MGZ}Yy1quNX7
zn1;L2b$UyFKSSS#eiQoQ>!ItWt&n-I(`Uv{OUnFeUyIyyKRhh;x!=?(jSJZM=lWrJ
zZi;IRPECpX0$Jn;eNJCMuP8bBA0XpV+_c6q$6$r&5j+25AIzU9>Sr7A*nl?>eu<*2
z)q=k?4<h~*o&|6q@jvUw=qo`bk~RG~28|Ec;S^u<ztE-d@rV-2(5gNCOjG0NX=3+6
z7-?Fp-p(A>0hXm5BR?xdBxPvGaP4f9HIShl7j}~a$K_MbX1xQMcL$zlwy-v~HJC{4
zu_qdmW92y5af^iDH_>?d_gR4l6JqmEutyO+{v=Gl!u-b8JBs?BV2@ep21yVYM;WRH
z7!22B7+EGp^d1mBw*9y*Zm9s~?T{<&uvTgZW~apL6d3;rw#zi2)SY!i#I5kSY0B>y
zeS$uWquouo+FEKj!Jb4J*3)~6&cyYCNPlw74_%yKdyuwzNcYk&<9Z?bRjfCU?bE|%
z4CoQKh2<iE_L~kO+#p~JKWm<JJ<)-(zkB_8%!pgVAM-8pe(z*)fwzS+eXmJpP(zux
z%8Pfsr@#MQkHn46OFKnaibL3Z*Q53fp)Swx?q!UtqlyvzeaNIzP4-zn>;=B4hO)E~
za^8XOeiH0{5hQUl_Y&^tz1hcc7poZJY0_vKT>u$wNh|ps1Z9*R5*m*^W!m9Vy>X7U
zz_G(@xp=^g)Bcf|_EW-U_We<hCA%#)D~LOqG^6E_-2>vuCvLH$`bqjtFU&DuersaH
z^OQf!cPhRz)8&1TP7PMLFD}5deD6>dZAO3#sGagfa)Pb1wHr_zieNnYh$_%5gJHk!
z3?yucER^DY4nI+_o<cNjg<`}Jg4T~(^qU5S9j8@>$~veT?Fn+CeTHh#hUs0j5xiS=
zMifBL;^oqWegtMFbP~zVgr1{!E$9@jv7k<37W6zdSkP&5TTtc;w9W!wr1jxWqj7A3
z|E1%J&#C-|wc~8x++I6gN+Z}A9Ma-S^*r4m9H4N~3>pD_iIVx{4)G4vYZMPQuUMmJ
z&4QZ$RX+_)RM21X7&@N5Vdvz-jV!+X>{W}nclBLlRo^n9gZXYVwM(Ck)#5P1txNH2
z-^pHvHtYQ8EjZ1z?d@hIPYlUQLO+<*$D-=@XZybD#nQU#3T?uZvbFycC7r0h@XL}l
zRzhfimTl=Bkp2Rx>@^xH9#>UG+%A?Wp~RMU6Oz}>0b9ycg@233`D|P_zZEsBoOPat
z8f(TCsjo-3jl0Kg{G@cZIwv~k?nYJJ8hy3wLtD}T{MU$f(AabD*JBO|@UUnWi{^-E
z%wa%XDO&wnS5uv@ag<FvWSpG1OMTWDTmD2+yCB<XTq?i4p9zY7XY?=SC+$gF1l(pE
zoIG89(fDZc6?IGW$dsk3kvVNzKW)2MHNoqt_eB>@JEE#>#+TDeZr3&obXr@z=JmNY
zG&I(CC2gwHnl?6f+3T8`yOPkf)@erZj6+FWtiCn+=8W++b*ph><^+phbmXj&sv0tu
z%pR1Hrh-NiZUeQkcXqz9XEvKWM4%-ndy}@tnJ}-_yFq(QaJE%0btX1@eU(eQ64y3(
z>NaR+1e=#d^NMKB8sp{+>AzplCe~}3SJVD(te7)0>2)!9!3Y4nD8M(2b8~*I8pis$
zAE>*dlOOzVoBB+&tYVO=o{P?}oUbNb7F3swqYFIhOVQMYb!t-9KL#WWP>tU^2S?8?
zDorV9fs~^bb|t%9%)7?r(%up?-Vsfn@xN8W27WHUk45t{(fo&Ku14=#HP%)#Tx{?|
ZAi5GB@nUacDebZ#=xLgx+nhty{|8$KEJ^?X

diff --git a/dynamight/deformations/optimize_deformations.py b/dynamight/deformations/optimize_deformations.py
index b4d7fc5..ab93f49 100644
--- a/dynamight/deformations/optimize_deformations.py
+++ b/dynamight/deformations/optimize_deformations.py
@@ -176,7 +176,7 @@ def optimize_deformations(
                 particle_dataset, val_indices)
             lambda_regularization_half1 = cp['regularization_parameter_h1']
             lambda_regularization_half1 = cp['regularization_parameter_h2']
-            n_warmup_epochs = 0
+            n_warmup_epochs = n_warmup_epochs
             half1_indices = inds_half1
             half2_indices = inds_half2
             print('continuing training from a given checkpoint file')
@@ -283,6 +283,25 @@ def optimize_deformations(
             encoder_half1, encoder_half2, decoder_half1, decoder_half2 = load_models(
                 checkpoint_file, device, box_size, n_classes
             )
+            decoder_half1.model_positions = torch.nn.Parameter(
+                initial_points.to(device), requires_grad=True)
+            decoder_half2.model_positions = torch.nn.Parameter(
+                initial_points.to(device), requires_grad=True)
+            decoder_half1.amp = torch.nn.Parameter(
+                50 * torch.ones(n_classes, n_points).to(device), requires_grad=False
+            )
+            decoder_half1.ampvar = torch.nn.Parameter(
+                torch.randn(n_classes, n_points).to(device), requires_grad=True
+            )
+            decoder_half2.amp = torch.nn.Parameter(
+                50 * torch.ones(n_classes, n_points).to(device), requires_grad=False
+            )
+            decoder_half2.ampvar = torch.nn.Parameter(
+                torch.randn(n_classes, n_points).to(device), requires_grad=True
+            )
+            decoder_half1.n_points = n_points
+            decoder_half2.n_points = n_points
+
         else:
             encoder_half1 = HetEncoder(box_size, latent_dim, 1).to(device)
             encoder_half2 = HetEncoder(box_size, latent_dim, 1).to(device)
@@ -305,33 +324,33 @@ def optimize_deformations(
             decoder_half1 = DisplacementDecoder(**decoder_kwargs).to(device)
             decoder_half2 = DisplacementDecoder(**decoder_kwargs).to(device)
 
-            if initialization_mode == ConsensusInitializationMode.MAP:
-                with mrcfile.open(initial_model) as mrc:
-                    Ivol = torch.tensor(mrc.data)
-                    fits = False
-                    while fits == False:
-                        try:
-                            for decoder in (decoder_half1, decoder_half2):
-                                decoder.initialize_physical_parameters(
-                                    reference_volume=Ivol)
-                                summ.add_figure("Data/cons_points_z_half1",
-                                                tensor_scatter(decoder_half1.model_positions[:, 0],
-                                                               decoder_half1.model_positions[:, 1], c=torch.ones(decoder_half1.model_positions.shape[0]), s=3), -1)
-                                summ.add_figure("Data/cons_points_z_half2",
-                                                tensor_scatter(decoder_half2.model_positions[:, 0],
-                                                               decoder_half2.model_positions[:, 1], c=torch.ones(decoder_half2.model_positions.shape[0]), s=3), -1)
-                            fits = True
-                            print('consensus gaussian models initialized')
-                            torch.cuda.empty_cache()
-                        except Exception as error:
-                            torch.cuda.empty_cache()
-                            print(
-                                'volume too large: change size of output volumes. (If you want the original box size for the output volumes use a bigger gpu.', error)
-                            Ivol = torch.nn.functional.avg_pool3d(
-                                Ivol[None, None], (2, 2, 2))
-                            Ivol = Ivol[0, 0]
-                            decoder_half1.vol_box = decoder_half1.vol_box//2
-                            decoder_half2.vol_box = decoder_half2.vol_box//2
+        if initialization_mode == ConsensusInitializationMode.MAP:
+            with mrcfile.open(initial_model) as mrc:
+                Ivol = torch.tensor(mrc.data)
+                fits = False
+                while fits == False:
+                    try:
+                        for decoder in (decoder_half1, decoder_half2):
+                            decoder.initialize_physical_parameters(
+                                reference_volume=Ivol)
+                            summ.add_figure("Data/cons_points_z_half1",
+                                            tensor_scatter(decoder_half1.model_positions[:, 0],
+                                                           decoder_half1.model_positions[:, 1], c=torch.ones(decoder_half1.model_positions.shape[0]), s=3), -1)
+                            summ.add_figure("Data/cons_points_z_half2",
+                                            tensor_scatter(decoder_half2.model_positions[:, 0],
+                                                           decoder_half2.model_positions[:, 1], c=torch.ones(decoder_half2.model_positions.shape[0]), s=3), -1)
+                        fits = True
+                        print('consensus gaussian models initialized')
+                        torch.cuda.empty_cache()
+                    except Exception as error:
+                        torch.cuda.empty_cache()
+                        print(
+                            'volume too large: change size of output volumes. (If you want the original box size for the output volumes use a bigger gpu.', error)
+                        Ivol = torch.nn.functional.avg_pool3d(
+                            Ivol[None, None], (2, 2, 2))
+                        Ivol = Ivol[0, 0]
+                        decoder_half1.vol_box = decoder_half1.vol_box//2
+                        decoder_half2.vol_box = decoder_half2.vol_box//2
 
             if mask_file:
                 with mrcfile.open(mask_file) as mrc: