Uiuc 1996-2008
Uiuc 1996-2008
"
3\T36m36V:5 OT5ÌNj3\ä|khVJ_5Wi)3£!3FQ3@m3@V:5GN8Mh 5jiJkh5khYG3MDH)5ÌkOTV)3_H:ÓNj8OQ5Ì]èi:OTVAX3\[AkD]F5PTÓ MDVA3%Mhz5jiA3
_`OTXhOQ5ÌNÎMh Fe|5jiA3_`ONj5GkhVJ]\3H365R8363@VkhV:Óß5P8M3\T36m36V:5ÌN%Mh Ò836YG3kh5%T3 kDNj55WiAYG363D5WiA36V5jiA3
!
Nj365ÌN 8MD:_ HJ3 _`ON jMhOQV568khVJ_ 838MD:_ iJkSD3 ~ Ø_:36VAMD5POTVAX²5jiA3]6khYF_`OTVkäOT5RÓ<MhkNj365 %
Av 4khVJ_Ù5Wi)3@YG3\-MDYG3 íü vwb hü vÒa :e8i:JN6^Oä Jéa :¯5WiA36YG3
mJNj5+H3l5P8M3\T36m36V:5ÌNùOT V ?|i)M/Nj3l_`ONj5GkhVJ]\3O¥Nkh5LmM/Nj58b:e
#%$'&)(+*-,/. " 0 1¯3657A\·6$#%H3lYG3 k"V:AmHJ36YFN8éâa/AkhVJ_Q3@5
v<mOTVZO[
T \ a{^U`_ Ôa8a
*]
a{
v<mOTVZO[ *]
B WC BD _ <a8a
aI{
³±{1¥C ³ED \ a{?
"!#
~ ~ eUW
b W ge+WV
³1hjAeWV <G ³ h ³ Í
~¿aW ~ Ù b ~¿aW
3
äT9<FS9 Ì 3 Ì
5
3 2 4 ge+WV
ÜhÜ a ³ h 2354 £
6 aW
D¨KNFS9Ü 3 hXH'FSJ³ h ÜÊ©eÜ hW5d8v 7 DH'FHfqN;E9GBAFH';KNF/M§DE9<BEKN9 D[®KN;EIJ9|BA9 VH D¨KNF7M;E9GBAQD<ZS;EI79
DE9<BEKN9 D%9;:354
AeWV Ú 6 aWVvK¯DsVFoV9<BAMV9<F?;[\K;IDEP7Q qF#EWaWV"hqFªb?Z$R?¸ ;EI79$Hf¸VqNsVBDE9<BEKN9 D
9µCSH'FD¨KNsVFr3sVBTqNFWTaÜgd\äT9<FS9;EI79DE9 éVP79<FS9/ Ü 0 sVF?oV9GBAMV9 D®[\KN;EIqtKNQK;qFªb?d
!
" #$%&
')(+*-,/.1032 465
798;:=<?>6:&@BAC:ED3F?GHDJIKGLD3MNON&N-PRQ!SL:UT$V@W>6XYB:&SZ8\[]8_^W:C`B@6:U^SL:UAC>6SZ8a[cbd:CVceYWe
D3F)fhg3G DJiKjIkf gklnDWimoDJi p mfrqWGCg3GLsWG&NON&NutHN
vw[c@x^yTzd:&@6:&SZT$VoQ1PdSLX>WV]T{Q1PdS{DJi6|
}*B.~w=*B5
3:&CifgdKDJi6|-:zR[cbd:&@SL:UAZ>-SLSL:O@xAZ:)6:&@HTRd:U8k;6:_Q1PdSLX
F fhg3G iKj\I f i lm p mfrqWGCg3GHsWG&N&N&NtHN
;:&SZTR[c@-z;W[]8kSL:UAC>6SLSL:&@BAC::PdY6LT$[c@JQ1PdSmfhg3GHsWG&N&N&NUG
i3oI
ZifhCi3oIl p mgdt/f&&WfhHFl fg£l sg m p m¤gdt/f sg m M sg ml¥g3N
Z¡ Fw¢
¦k:&@BAC: oI
g g M g
D i f Ci f¨§ s m s m©lgUª p mfhg3GLsWGON&N&NutHN
')(+*-,/.1032 «B5
¬bKT$Vc>BTR;: i®¡ I s ¯I Q1PdSmfg3GLsWG&NON&N
}*B.~w=*B5 ¢
3:&° p± tf i®¡ F ± |-:O@r°x² p± tyf iZ¡ I ± ¯I £8;P³-: zR[cbd:&@´8;>-X []8©:=<?>BT$VP°B² p sµtH|
¶_@·iKj;I6:Pd;6:&SEBTR@x^o8;>6XX[c@-z;6:zd:OPdX:O;S;[]A8;¢ :OS;[:=8)[@¸-:^W:C`B@W[[cPd@PRQ£° p± tEzR[bd:=8° p± tEf
p± gdt; p± gdtHxTR@B^y^¹[»º¼:&SL:&@W[½TR[c@-zJ[]8Q1>6@BAC[Pd@:PdY6LT$[c@
i iKjI gdt
° ² p± tf p± gdt p ml¥gdt ± gdt M p± N
p¾±
¦k:&@BAC: i
s oI f´° ² sµtf ml¥gdts i ¸s iKj\I l¥g f mgdts i l¥g3N
Z¡ IW¢ p p p
')(+*-,/.1032 ¿o5
Àk[cbd:&@nT©@6Pd@6:&XÁ6e`B@J[c:8;:&ÂÃPRQSL:=T$V@W>6XYB:&SZ8&?Vc:&ÅÄ Â)t£^W:&@6Pd;:-:XyT$Ƶ[cXyT$Vo:ZV:OX:O@W_PRQ
 |v-PdSmfg3GLsWGON&N&NdVc:&£° p mwt/Yx:_-:{8;>-XÇPRQ!Ä p Â)tLW£6:&SLp :ÂÈSL>6@B8PRbd:&S;6:s i Ég@6Pd@-ÊË:&XÁ-e
8;>6YB8;:&L8PRQ/6:8;:&EÌRg3GLsWG&N&NON&GLmÍd|Àk[cbd:TR@:CƹÁWV»[½AL[Q1PdSLX>WV]T{Q1PdS° p mwtL|
} *B.~w=*B5
³:BTUbd:° p mwtEf iZ¡ I p tLw£6:&SL: p t^W:&@6Pd;:=8);6:@¹>-XYx:&S£PRQ£@6Pd@-ÊË:&XÁ-ey8;>6YB8;:&L8_PRQ
ÌRg3GLsWG&NON&N&GLmÍ £6P38;:XT$Æ3¢[cXyT$Vo¢ :CVc:&X:&@W[]8 |¢ Vc:UTRS;Vced p gdtfg368a[c@xAZ:;6:8;:&ÌRg3Í[½8£;6: Pd@WVce
8;:&PRQ_ÁBP38a[ca[cbd:[c@W;:&zd:OSZ8[c¥XT$Æ3[cXyT$V:Z¢ V:OX:O@Wg3|v6PdS sW;6:y8;:&H8ACPd>6@W;:=^nY¹e p t
TRSL::CÆJTdACaVce6P38;:PRQ{;6:Q1PdSLX ¨f¨ÂI ¸Ì ¯ÍdI £6:&SL:yÂ)¢ I[]8T p ÁBP38L8a[cYJVce:&XÁ-e6t8;>6YB8;:&o¢ PRI Q
ÌRg3GLsWG&NON&N&G g3Íd|?[c@BAC:;6:&SL:TRSL::CÆJTdACaVce·s ¢ 8;>BA®·8;:&H8)ÂIRx[c)Q1PRV»VcPR8 ;BTR p t_f s
Q1PdS£:UTdA® ¢ |¦k:O@xAZ: i ¢
¢ ° p mwtf s ¯I f p mgdt¾s i l¥g3G
Z¡ I6¢
YWe6: SL:U8;>WVckPRQ!;6:EÁ-SL:Ob?[cPd>B8kÁ6SLPdYWV:OXÉ|
')(+*-,/.1032 o5
o:&_° p± t£Yx:ETÁBPRVe3@6PdX[½T$VPRQ^J:OzdSL:&:m³8;>BALBTR_° p tf p l¥gdt/Q1PdS f´qWGCg3G&NON&N&GLm|vw[@B^
° p ml¥gdtL| ¢ ¢ ¢ ¢
}*B.~w=*B5
3:& p¾± t/f p± l¥gdt° p± t/ ± |6:& @ p¾± t[]8_TÁBPRVceµ@6PdX[]T$VPRQ/^W:&zdSL:&:Emlg£W[]A®yxTd8SLP3PdL8£TR
:UTdALPRQ/6:mlg@W>6XYB:&SZ8/qWGCg3G&N&NON&GLm|¦k:&@BAC: p± tX>x8;/YB: PRQ/-:Q1PdSLX
i
p t p¾± t
f p± t
Z¡ F ¢
Q1PdS£8;PdX:_AZPd@x8;HTR@WK|µ:O;a[c@6z ± fg_[@y;6:E^W:C`B@J[c[Pd@PRQ p± t:_PdY6LT$[c@ p gdt/f 3g |e p t
[ckQ1PRV»VcPR8k;BTR i
g)f p gdtf p )g t/f p gdt iKjI p mlgdtN
Z¡ F ¢
¦k:&@BAC: f p )gdt iKjI p mlrgdt c|3>6YB8;a[c;>6a[c@6z;W[]8)bKT$Vc>6:[@¹;P p tTR@B^ 8;:&;[@6z ± f mh
l g3o:
PdY6LT$[c@ i iKjI
mp ©l¥gdt/ f p ml¥g t/f p mgdl¥t gdt p ml¥gdt Wf p gdt iKjI G
Z¡ F ¢ p
£J[]AL[cXÁWV»[c:U8 iKj\I
° p ml¥gdtf p m©lm©gdtwlls ml¥g f m©lgkm©l lp )s gdt N
')(+*-,/.1032 B5
o:& ± IKG ± MdG&N&N&UN G ± iYx:mSL:UT$V@W>6XYB:&SZ8/8LTRa[]8aQ1ed[@6z
i i
frqWG ± fg3N
Z¡ I ± Z¡ I! !
"SLPKbd:;BTR ## i ##
## ± ## g g N
# I #%$ s sdm
Z¡ ¢
} *B.~w=*B5
-:zR[cbd:&@ ACPd@B^µ[ca[cPd@B8_Pd@ ± c[ XÁJVceBTR&WQ1PdS_TR@WeSL:UT$V@W>6XYx:OSo
## i ## ## i ##
## ± ## f ## g #
# Z¡ I # # Z¡ I ± § Wª ##
¢ i ¢ ## ## i ## ##
i # g # # g #
$ I! ± ! XyZ¡ T$IÆ # # f XyZ¡ T$IÆ # # N
Z¡ ¢ ¢
TRd[c@-zfÃgdKs)l¥gdKsdmB;6:XyT$Ƶ[cX>6XbKT$Vc>6:PRQ gd p gdKs)lrgdKsdmwt [½8TRLT$[c@6:U^³TR fÃg
TR@B^ f´mÉTR@B^:=<?>BT$V!;PgdKs)gdKsdm|6:E^W:U8a[cSL:U^©! [c@6:U¢ <d>BT$Vu[ce;6:&@Q1PRV»VcPR! 8&| ¢
¢
')(+*-,/.1032 o5
3>6Á-ÁBP38;: m f D I D M N&N&NZD I []8-:n^J:=A®[cXT$V{SL:&Á6SL:U8;:&@WHTRa[cPd@ PRQTR@¥[c@W;:Ozd:&S m9ACPd@B8a[]8;a[c@6z¥PRQ
:CÆJTdACaVce³g @6Pd@6Ê&:OSLP^µ[czR[cL8D¤ÌRg3GHsWG&N&N&N&GWÍd|{36PR´;BTR_mÉ[]8k:C[c-:OS ^µ[cbd[½8a[cYWVc:YWeg W6PdS
A&TR@³YB:ALBTR@6zd:U^;P TR@[c@W;:&zd:OS ;BTR[½8E^µ[bd[]8a[cYJVc:yYWeÉg YWeSL:&ÁWV]TdA®[c@6z 8;PdX:d-Y6>6@6Pd T$V»VËPRQ
;6:Å^¹[czR[cL8DYWeqW|
}*B.~w=*B5
o:&moF fÈqTR@x^Q1PdS fÃg3GLsWG&NON&N&GCg Vc:&Em ^W:&@6Pd;:-:@¹>-XYx:&SPdY6LT$[c@6:U^YWeSL:&ÁWV]TdA®[c@-zyT$V»V
Y6>6_;6: `BSZ8; ^µ[czR[cL8¢ PRQm³YWe-:^µ[czR[cqWW[Ë| :d|cwm fDWIODµM/N&N&NHD qdqN&N&NHqW| e;6:ÁW[czd:&Pd@ -PRVc:
Á6S;[c@xAL[ÁWVc:dWP ¢ PRQ¼;6:g )@W>6XYB:&SZ8wmoFKGLmwIKG&N&N&N&GHmwIRW8LTUem TR@B^m [c; M IX>B8;wYB:
ACPd@6zdSL>6:&@WXP¹^W>WVPg W|Q1PRV»VcPR8_;BTR£;6:^µ[»ºo:&SL:O@xAZ: m ¸m []8_^µ[bd[]8a[cYJVc¢ :YWe¢ g W|kPR
m m []8;6:@W>6XYx:OSPdY6HT$[@6:U^Q1SLPdX mYWeSL:&ÁWV]TdA®[c@-z;6:_`xSZ8; ITR@B^;6:_V]Td8; p g k MRt
^µ[czR[H8)YWeyqW|_d[c@xAZ: M IUx:EBT$bd: Il p g ) M$t g WB8;P ¢ @6Pd_T$V»V/PRQ-:g ^¹[czR¢ [cL8
TRSL:SL:&ÁWV]TdAC:U^YWe O:&SL¢ P6|W¢ >B8&µ6:)@¹>-X¢ Yx:&S m ¸m ¢ BTd8;6: SL:=<?>W[cSL:U^Á6SLPdÁB:&SLa[c:U8&|
!
" $#%&(')
* +.-0/13254 607
,
8:9);=<?>A@B9DCBEF;G9H;GIB9JKCL;G9)MN9)OQPDRSENTU9)TV;W;GE$X YZ=[]\0EFOW9_^a`FbcaRK9FdB<0egfh<5i=fkj`FCB@l<5m=fn<LopfrqsTUJKCBPD9
X jpfkjNd X q,fkjNtvuBjt%t%twd X x fyjNt{z x t%t%t{d0`FCB@ X ufrq5Z{|~}F`FRKa`;G9;UIB9TUBb
f <j ep <5j i%%% <ej GG t
q
-1]:.3-7Ò}9)OUTU9D?B9DCBPD9sEN:;GIB9,OG9)5BJKOU9D@¨²EFOGb¾P%`FC&¡9¢QOGJ;U;G9DC&`FT
ºeG»He¬°·j®ºi)»QiD°·j(t%t%t¶°·j®ºí§eG»=í§e%d:¢=IB9)OG99%`FP¬IºDWJSTs`©aRSE?P FENpj¦T`FCB@9%`FP¬I» D`©aRSE5PCFEN
°vT)d¢=JK;GI9DbcB;ÝaRSE5PCF5T~¡9DJSCBMª`FRSRKE.¢g9D@Åd`FCB@Ù;GIB9©TUBb ENQ;UIB9RK9DCBMF;GIBT~EN=;UIB9©aRKE5PCF·TºDH`FCB@
» D=JKT,YÚ£qB?ªZô EFOG9DE®9)O%d:;GIB9TU9)5B9DCBPD9JSTBCBJS5B9)RK@B9);G9)OUb$JSCB9D@ª? ;GIB9¨[ÁqB? qN|ÀÁ;U·caRS9Ú[Ýj.|
[¥É.eDî¬ÊæeDî¬É)iæî¬ÊDi.î%t%t%t_î¬Éí§eDî¬Ê=íeG|¢IB9_OG9É D`FCB@Ê D§@B9DCBEF;G9p;GIB9HCLBb9)OWENÛ9DRK9Db$9DC;GTJKC$;GIB9paRSE?P F5T
ºDW`FCB@Ö» DVdÅOG9)TVc9DP_;GJK9DRíZ$§e 1ENCL9_OGTU9DRdè`FCÚ;G·caRK9EN;GIB9²EFOGb [¶j.|¢J;GICBENCBCB9DM`;GJK9~JKCL;G9)MN9)OGT
É D§`FCB@lÊ D§TG`;GJKTUÍ·JKCBM ³ D µ e []É D Ê DÁ|fØ[3Y£qB?&|@B9);G9_OGb$JSCB9)TW`sTU9D5B9)CBPD9,ENÅ;GIB9pOU9D5BJOG9D@A;5c9Z
åH9DCBPD9;GIB9CLBb9)OENaTUBP¬ITU9D?B9DCBPD9DTJST9D?a`FR?;GEH;GIB9CLBb9)OENB¢W`%·TENCB9PD`FC¢QOGJK;U9WqY=£ ?`FT
`pTUBbyEN0qB? qN|!CBENCBCB9DM`;UJK9JSC;G9DMN9_OGTDdF¢=JK;GIEFOG@B9_O;¬`BF9DCJSC;GE`FPDP)ENBCL;%ZÔWIB9RÍ`;V;G9)O!cBOUEFaRS9DbyJKT
9D?BJK`FRS9DC;;GEPDENBC;GJSCBM;GIB9CLBb~¡9)OENB¢W`®5TENP¬IBE?ENTUJSCBM,qY£$?Ä@BENC?·;GTÍOGENbØqB? q`OGJS9_;GJS9DT)d
`s¢9DRKRÀ F5CBE.¢=CPDENbaJSCa`;GEFOGJS`FRcBOGEFaRK9DbÒ¢=IBENTU9`FCBTV¢9)O=JSTMNJ9DCL$;GIB9HaJKCBENb$JÍ`FRPDE59PDJS9)CL;
¢=JK;GIÇ<f []YÖ>Ní£ãe qB?¨| [¥qB? qN|£ñj fàY j¨`FCB@çÉlf [ÁqB? qN|Q£njfËqB? jNZåH9DCBPD9
[]YîC?&|f ií§e Z
x
!#"$"%"&'(" )+* , -.*
/ 0 123
46587:9;=<?> @
A?BDesL`CEM&CGFI|+HK}~J#LNubMPOQLSy6RUTW}~V:azRDX8HKVZBEYWYWMY[V:L`R]MP\^OQLpOQRt_`JbTad0cIegfhRtJjikXelihL`JjYWcIYWJmYW_N0HjnpRDo2p_NesF8HfhxEJegL`fDMPelHKOQesLqPlL`JvMEtJm_`&JbaJ[nrDelHKLSesitLNJjcIJm_`Hvu$OwiExzy
{ 7U;Z8=7U JmL2p|SOwiEx%¡Fw_¢£¥¤IV¦?VWYWYWYV`a§wsJmL2G\¨R:TI¨RDX]&©W©W©`R]\ªn«¬L`MEJCPelcIJjFIiEMEFIsJ
CEH`Ow_`³¨esiEJvjesCP_`JmsJw³´§:OwL el{ iEF%xEJmFI_µ®LBD`CªMEFIJjHKi&J¯xEa$egfDelHKel¤¯FIi¶elih]L`Jj¶cIJ[a_`nHj§:·vH`JjO°iE&jJ¯G¹OwiE¸zxGp+± { RtegL`TMZ|+!}'Rtu+X~y¯©W}~©W©KaZRD²[q§Eel³#HºBEO#HK³#LBElJjsO°L`fhegCPJ(lJvL`MEFIJ
an
46587:9;=<?> »
¼¨OwiEJmLNLGxkF6¼#L`y6MEGÆÇJºJµHNOqel±ÉxERvCªJºFI0FIelihG½Ls0JjeliEiE¾mHNcwes²NL`xEÊIM¨Jµ¦?R:nOq§to2LNOw_`_`iEelF°Owx$fhiEJ(lcIJ[lLNLºJ¬MPÀDOQMPLÃOWfDOwesiEiExcvÀDHNelÄ2xEªJjHJFIxEPJmsÅPJjiEiEJmcwx¶L`MEOwHiPRUOwVPlFI½8cIV:FI¾¿BEn HN ghJmn L ÀDJ[ÁLºGÀJ¬ÆÇL`MEJ³¨xEelesiHKL±ÈÀEOwiEÁtVmjÀEJÃjÂ_VmNFIÀD³Ä²
À}ÌË ±Íyµ¸ÎRD²j±Íyµ¸d½j²m±Íy¸d¾m²NÊIyhY
Ò 7UME;JZOQ8_`=JW7UOwH FI®LÏ`MEJ¨J¯HK_NL`JjBEHNBExDs/LNeliEL`MEc%J¨LNMDOQ_`_NJjJWJO&iEFIJL`MELNJ%_NeÂLNOw_`iEeÂOwcIiElJmcIHpsJwOQn$_`J¨Ðv±Ó_¤8O ÊI{ ¦I²KlRPesiEÀDJjÁhH¯§®±ÔÂ_`¤8FIÊI³Ñ¦I²K½°¼dÀDÃmL`§®F±ÓL`¤8MEÊIJ%¦I²KL`¾°MDÀD_NÄjJjn J$Ò fhM]JmBE_NLNHeljL`JjMEHmJ n
OQ_`JWO+FI®L`MEJ(Fw_`elcIesiPOwGLN_`elOwiEcIlJ(elHbÕwÖ×lØ{ Õ¿ÙÖ
¤ ±ÉRv0½0¾m²D³¨eliG±ÚÀEÁtVmÀEÃjVmÀDIJ¹ÛywÀªY
¦
Üvi&L`MEJ¯FwLNMEJm_qMPOwiEx§Pt·vJ[_`FIiÝÞH¡Fw_`³#BEÂOD§DL`MEJ6OQ_`JWO+FI®L`MEJ¯Fw_NelcIeliPOwGLK_`eÂOwiEcIsJ(esH
Ë yt±Íy¸dR?²j±Íyµ¸Î½j²j±Íyµ¸Î¾j²[Y
Ò M]BEHj§ªyQÀ} Ë y]±Íyµ¸ÎRD²j±Éyº¸d½m²j±Íyº¸Î¾m²[§ { MEesMelHºJjß?BEegfwOwsJjihLL`F+L`MEJ6OwHNHNJm_NLNJjx^eliEJjß]BPOwlegLÔhn
46587:9;=<?> à
|JmÑL%á'á±ÓOw¤8iE²vxÎâãâG±ÓJ¨¤8²[Ln¶{ A?eljBDJ´CECªjFIFIihHNJ¨L`esi]L`MPBEOQFILBEHN|ãgÎxEesá䪱=JmèG_`²bJjih¥L`eÂOQâP±=sèGJ¶²S¡B¡FEwiE_mL`|ãelFIiEH#è~FIi忤¶|DVOw¤jiEç x.{ L`esMPLNM!OQL#á᪱Ééæ|hͲ±êè²bâ±É||h²¡Fw _
|#~èÛ¤In¬A?MEF { L`MPOQL ë T Xì ë T XGì
á é ±=èG² èk} â é ±=è² èíY
¤
á éÞ7U;~Z8|%=7UOwiExká^ihL`JjcwâG_§ OQLNelJ6iEcMPO°tfh(J C:OQ_NL`HíOwiExBEHNeliEcL`MEJ¬OwHNHKBE³$CEL`elFIiEHá±Ó¤8²â±Ó¤8² á±É|h²â±|h² |D§
ë T { X[ì T ë T ë T
á é ±êè² èÛá±=èG²Ná é ±=èG² ¸ ë á±=èG²Ná éÞ ±=èG² è^Û¸ ë â±êè²Ná éæ ±=è² ì è ì
T T
̸â±êè²Ná é ±=èG² T â é ±=èG²Ná é ±=èG² ì è â é ±=èG²Ná é ±êè² ì è
lHKFE§D] ºOwBEMhGÝæHqeliEJmß?BPOwlegLÚh§
ë T â é ±=èG²Ná é ±=èG² ì è X } ë T â é ±=èG² Xmì è ë T á é ±êè² X[ì è .V
Ò M]BEHj§
ë T á é ±=è² Xmì è X } ë T â é ±=èG² Xmì è ë T á é ±=èG² Xmì è .V
T
OwiExxEegfDelxEesiEc+t á é ±=è² X ì èkcIegfhJjHµL`MEJ6_`JmHNBEsLWn
46587:9;=<?>
o2iEF%_NF8_`fhOQJbL`elL`FIMPiPOQOwLSGes_`RF]:FwL`§Hj½¿n §OwiExã¾bOQ_`J#F?xExkCGFIHNegL`esfhJbesitL`JmcIJm_`Hj§L`MEJjikL`MEJ#CGFIgDiEFI³¨elOw®Rhè X ~½è$~¾MPOwH
Jjß]7UBP;OQZL`8el=FI7Ui´Rh è X A?/BDCE½CªèvFIkHNJ¯¾µL`MPOQ|DLºn èÒ MEJjuIi/ÊI±y¯Q²es2HqR]O#u X _OQkL`el½[FIuwiPyíOwkHK¾jFIy lX BDL`esFI|Di&nelA?iesiElF j{ JqJjL`HKMELJqL`Â_Jm_`Ow³$mL`HºesFIFIi¨íL`uIMEÊIJ(ypCªelH2FIelsi´?iEsFFI³¨JmeÂHKOwL
L`elHJm_NF?³¨xExHj§P§]OQLNLµMEJj³¨i¨FIL HKLºFbFIiEFIJSªL`FIME®uJq#LNMDOwiE_`JjxJpy6L`JmW_NOw³¨iHGFIJSi$JmLNfhMEJjJpinlJjÂL¬ZJ[FIrE¹±OwmQ²L`2s¶OQ_`FIJiEJmJSfhFIJjíui#OwiEOwx¨iEx^L`MEy6JqesHºLNMEJ[esfh_`x$JjiLNJmOw_`iE³1xesL`H¬MEF?J6xEFwx{ LN§]MEHNJmF _
L`OQME_`JbJjeg_vF]xEHNxBE³§OwiEWOwx^iE{ L`iEMEFwJjLSes_pJjß?HNBEBP³ Owí|DOwchn Oweli^ªjFwOwL`iEM^iEu¨FwLvOwGiEx/JbJjy#ß]OQBP_`OwJíF]L`F¶xEx|D§Un L`ÒMEJmMtikBEHjOw§GlíL`MELNMDJb_`JjJjß]J¯BPL`OQJ[L`_`el³¨FIikHqRhFIèiX LN'MEJb½lè$JjÂL'vFI¾vº±ÌQ²|
xEF?JmHqiEFwLMPO°fhJ¯O+_OQL`esFIiPOwG_`F?FwLjn
46587:9;=<?>
fQOwlBPOQLNJ(L`MEJ(eliDÅPiEegL`J¯HKJm_`elJmH
T¦ ¸!¤ ¦ T ¦ X ¸Î¤ ¦ X ¦ µ¸!¤ ¦ ¦ ¸!¤ ¦ ©W©W©
7U;Z8=7U J[L(á±=èG²¬ \ ±=è X ¸dè XX ² T §ªHNF¨X L`"MP! OQLqT L`MEJcIesfhJmi^HNJm_`esJjHvelHSá±Ô¤8ÊI¦I²[n Ï _`esL`esiEc
JWelihOwL`F$M.O%L`JmcI_NJj³ FI³¨eliãJ[LN_`L`elMESelH6HNJ[HN_`Jmel_`JjesHmJj§ H#OwJ6HcIOJmL CE_NF?xEBEmLè ±Ó¤v¸ è ² OwiExdJrDC:OwiExEeliEcL`MEJ¨HNJjmFIiEx!êOw[L`Fw_
{
á±=èG² \# è X # è X "! $ \&# % è X QX "! $ \&# % è X 'sX $ T)( Y
$ $ $
¦
F?i mjLNBDME_`J(HblJ[OwrEHKLµOwmHNL`Jms_NãelJjFIHjiE§EjL`MEJ¶JSOwJ[Hbr?HNCGBEFIiEM!JjihOwL`iÎHqJ[¦ rD\ Cª±Í¦FIµiEJj~ihLW¤8n²¬·vOQ_`JjJviECªjJ¨FIHNLNesL`MEegJ¨fhJvÂOwelihHKLL`JjHNcIJmJm_`_NesHjJj§:HbOweliEHxJmß?JWBPOwOwM¶2LNCGF FIHNegL`es fhJSesitè L` JmcIJm _
èGÊE±Ó¤µ¸.è²[§:OwiExHNF¶á±=èG² èÊE±Ô¤¸.èG²[n Ò MEJ(fwOwsBEJ6FIíL`MEJ6cIegfhJji&HKBE³ esHL`MEJm_NJj¡Fw_`Jbá±Ô¤8ÊI¦I²Ì¤In
T
åeli¶i$HNMEOwF sL`J[el_`iEiPcEOQ§?L`tesfh´JºelHNiEFIxEsBEBD[L`L`elFIelFIii±§?¡FILNMPBEOQiELx$L`tME$JvÐ(HNBEOW³ fDelx%FILÐS`BEMEJmJpGÅEJm_`_jHÓ§ Lµ6aOwL`BEJmHN_`ME³¨e HqFILF°hNME§tJvOwiEcIx esfh6Jji¶JmiHNJmA?_NelJjMEHJme es{ HvJ°¤2²¸ÎjFI±ÉiE¦ XHNesHKL`¸ H
¤8² T n2{ A?eliEjJ¯LNMEelHºJ[rDCE_`JmHNHNelFIiL`JjiExEHLNF&¤OwHºa §EL`MEJ(HNBE³ FIíL`MEJ6HKJm_`elJmHqelHºJjß]BPOwLNF&¤Inæç
46587:9;=<?>
JmLvá±=aZ²xEJmiEFwL`J¯L`MEJ(i]BE³bªJm_pFI2¤IÝÞHºeli&L`MEJSPeliPOQ_KJrDC:OwiEHNesFIiFI®an fQOwlBPOQL`JSL`MEJ(HNBE³
# a¹±áa$±aZ² ¤8² Y
\T
¤7U0;ZsFI8c =7UaZ²[nvHNesJmiEL&c¨L`MEJ¯xEJj_`iEJjFwlOQL`L`J^elFIL`iEMEHvJ^áHN±ÍBE¦Q³ aZ²FI vL`áME±JkaZ²cIOwesfhiEJmxãiáHK±ÍJm¦Q_`a$elJmHz± ¤8{ ²¬MEÌesMá±Íj¦QFIaZih²fhJm_`cI¤(Jj¥H¶HNáesiE±aZjJã²á±¤baZ²/OwiE} x
HKCPsesLNL`esiEXc%L`MEJ(HNJm_NelJjHelihL`F$F]xEx^OwiEx&JmfhJmi&C:OQ_NL`Hj§ { J6FwELOweli
. \# T ±Í¦QaZ²já±Í±Í¦Q¦Qa%aZ² ¤8² \# ±Í¦Qa$á±Í¦Q¤8a%²j±Í¦Qa%¤8²0¦I²
\# T ±Í¦QaZ²já±Í¦Q±aZa%² ¤8² \# ±Í¦Qa$á±aZ¤8²j²±Í¦Qa%¤ 0¦I²
\# ±Í¦Qa%¤8²j¤ ±Í¦Qa%0¦I² Î \# T ±Í¦QaZ²já±Í¦Q±aZa%² ¤8² ±É¦Qa¨á¤8²j±aZ±É¦Q² a$ ¦I²
# Ó
± (
¸ 8
¤ ² \
¥¸ \ T a \# T ±É¦QaZá²j±±a%aZ² ¤8² sFIc¦º ¦¤ Y
·vJjiEjJQ§G.sFIc¦º0ÊI¦?§:OwiExHNF´dÛ¦lFIc¦(lFIc En
UIUC UNDERGRADUATE MATH CONTEST
APRIL 21, 2001, 10 am – 1 pm
SOLUTIONS
Problem 1
Given a positive integer n, let n1 be the sum of digits (in decimal) of n, n2 the sum of digits
of n1 , n3 the sum of digits of n2 , etc. The sequence {ni } eventually becomes constant, and
equal to a single digit number. Call this number f (n). For example, f (1999) = 1 since for
n = 1999, n1 = 28, n2 = 10, n3 = n4 = · · · = 1. How many positive integers n ≤ 2001 are
there for which f (n) = 9?
Solution. Since an integer is divisible by 9 if and only if its sum of digits is divisible by
9, the numbers n with f (n) = 9 are exactly the multiples of 9. Since 2001 = 9 · 222 + 3,
there are 222 such numbers below 2001.
Problem 2
Let x, y, and z be nonzero real numbers satisfying
1 1 1 1
+ + = .
x y z x+y+z
Show that xn + y n + z n = (x + y + z)n for any odd integer n.
Solution. From the given relation one obtains, after clearing denominators and simpli-
fying, (x + y)(x + z)(y + z) = 0. Hence x = −y, x = −z, or y = −z. In the first case,
xn + y n = 0 for odd n, and so xn + y n + z n = (x + y + z)n . The other cases are analogous.
Problem 3
Suppose that an equilateral triangle is given in the plane, with none of its sides vertical.
Let m1 , m2 , m3 denote the slopes of the three sides. Show that
m1 m2 + m2 m3 + m3 m1 = −3.
Solution. Let A, B, and C be the vertices of the triangle, labelled so that the path
ABCA is counter-clockwise, and let m1 , m2 , m3 denote the slopes of the three sides AB,
BC, and AC, respectively. Without loss of generality, we may assume that the vertex A
is located at the origin. Then m1 = tan θ, m2 = tan(θ + 2π/3), and m3 = tan(θ + π/3),
where θ is the angle between the positive x axis and AB. (Note that θ ≥ 0 if the triangle
lies entirely in the first quadrant; θ < 0 if the triangle extends into the fourth quadrant.)
Using the identity tan(x + y) = (tan x + tan y)/(1 − tan x tan y), we get, with T = tan θ,
√ √ √ √
T (T − 3) (T − 3) (T + 3) T (T + 3)
m1 m2 = √ , m 2 m3 = √ · √ , m 3 m1 = √ .
1+T 3 1+T 3 1−T 3 1−T 3
0
Adding these three terms and simplifying gives m1 m2 +m2 m3 +m3 m1 = −3, independently
of the value of T (and θ).
Problem 4
Let x1 ≥ x2 ≥ · · · ≥ xn > 0 be real numbers. Prove that
x1 x2 xn−1 xn x2 x3 xn x1
+ + ··· + + ≤ + + ··· + + .
x2 x3 xn x1 x1 x2 xn−1 xn
Qn−1
Solution. Set qi = xi /xi+1 . Then x1 /xn = i=1 qi , so the inequality to be proved can
be written as
n−1
X n−1
Y 1 n−1
X 1 n−1 Y
qi + ≤ + qi ,
i=1 i=1
q i i=1
q i i=1
or equivalently
n−1
X n−1 n−1
1 Y Y 1
(∗) qi − − qi + ≤ 0.
i=1
qi i=1 i=1
qi
Let f (q1 , . . . , qn−1 ) denote the function on the left of (∗). The hypothesis that the xi are
non-increasing implies that qi ≥ 1 for all i. Since f (1, . . . , 1) = 0, to prove (∗) it therefore
suffices to show that the partial derivatives of f are ≤ 0 when qi ≥ 1 for all i. This is
indeed the case: we have
∂f 1 Y 1 Y 1
=1+ 2 − qj − 2
∂qi qi qi qj
j6=i j6=i
1 Y
= 1+ 2 1− qj ≤ 0,
qi
j6=i
Q
since j6=i qj ≥ 1. (When n = 2, this last product is empty, but in that case the sums and
products on the left of (∗) reduce to a single term (corresponding to i = 1), and a direct
computation shows that the derivative of f with respect to q1 is equal to zero, so the last
inequality remains valid for this case.)
Problem 5
Suppose that q(x) is a polynomial satisfying the differential equation
d
7 {xq(x)} = 3q(x) + 4q(x + 1), −∞ < x < ∞.
dx
1
Solution. The left-hand side of the given equation is 7xq 0 (x) + 7q(x), so the equation
simplifies to
Z x+1
0
(∗) 7xq (x) = −4q(x) + 4q(x + 1) = 4 q 0 (t)dt.
x
The left-hand side of (∗) is zero at x = 0, so by the mean value theorem for integrals
(which can be applied here since q(x) is a polynomial and hence has continuous derivatives
of all orders) there exists a number x1 ∈ (0, 1) with q 0 (x1 ) = 0. Setting x = x1 in (∗), we
obtain a number x2 ∈ (x1 , x1 + 1) with q 0 (x2 ) = 0. Repeating this process, we obtain an
infinite sequence x1 < x2 < · · · of values x at which q 0 (x) = 0. Since q 0 is a polynomial, q 0
must be identically zero. Hence q is constant.
Alternative solution: The above solution was the one we had in mind when posing the
problem. However, all students who correctly solved the problem, did so via the following
approach (or a variant of it): Write q(x) = an xn + an−1 xn−1 + · · · + a0 , where an 6= 0.
Then the left side of the differential equation is a polynomial of degree n with leading term
7an (n + 1)xn , while the right-hand side has leading term 7an xn . Equating the coefficients
of those terms, we obtain 7an (n + 1) = 7an ; since an 6= 0, this can only hold when n = 0,
i.e., when q(x) is constant.
Problem 6
P2n
Evaluate the sum k=n nk 2−k .
Solution. Let S(n) denote the given sum. We claim that S(n) = 1 for all n. Since
S(1) = 1, it suffices to show
that S(n
+ 1) = S(n) for all n. Writing k = n + 1 + h and
using the identity n+1+h
n+1 = n+1+h
h = n+h
h + n+h
h−1 for h ≥ 1, we have
n+1
X n+1 n+1
n+1 n + 1 + h −h X n + h −h X n + h −h
2 S(n + 1) = 2 = 2 + 2
h h h−1
h=0 h=0 h=1
n
X n+h n+1
−h 2n + 1 −n−1 X n + 1 + h −h−1 2n + 2 −n−2
= 2 + 2 + 2 − 2
h n+1 h n+1
h=0 h=0
2n + 1 1 2n + 2
n n
= 2 S(n) + 2 S(n + 1) + − 2−n−1 .
n+1 2 n+1
2
UIUC UNDERGRADUATE MATH CONTEST
APRIL 13, 2002, 10 am – 1 pm
SOLUTIONS
Problem 1
Without any numerical calculations, determine which of the two numbers eπ and π e is
larger.
Solution. Let a = eπ and b = π e . We will show that a > b. Since ln a = π ln e = π and
ln b = e ln π, and since taking logarithms preserves inequalities, we see that a > b holds
if and only if (∗) (ln e)/e > (ln π)/π. Now consider the function f (x) = (ln x)/x. We
have f 0 (x) = (1 − ln x)/x2 , so f is decreasing for x > e, and since π > e, it follows that
f (π) < f (e) which is equivalent to (∗).
Problem 2
Let 0ABC be a tetrahedron with three right angles at the point O. Let SA be the area
of the face opposite to the point A, i.e., the area of the triangle OBC, and define SB , SC ,
2 2 2 2
and SO analogously. Prove that SO = SA + SB + SC .
Solution. Place the tetrahedron so that its vertices are located at O = (0, 0, 0), A =
(a, 0, 0), B = (0, b, 0), and C = (0, 0, c). Clearly SA = bc/2, SB = ac/2, and SC = ab/2.
Moreover, the area SO of the triangle ABC is 1/2 times the area of the parallelogram
determined by the vectors AB = (−a, b, 0) and AC = (−a, 0, c), which in turn is given by
the magnitude of the cross product of these two vectors. Computing this cross product
gives (bc, ac, ab), so
1
S02 = (1/2)2 k(bc, ac, ab)k2 = (bc)2 + (ac)2 + (ab)2 = SA
2 2 2
+ SB + SC .
4
Problem 3
Let θn = arctan n. Prove that, for n = 1, 2, . . .,
1
θn+1 − θn < .
n2 +n
Solution. Using the fact that arctan x has derivative 1/(1 + x2 ), we obtain
Z n+1
dx
θn+1 − θn = arctan(n + 1) − arctan n =
n 1 + x2
Z n+1
dx 1 1 1
< 2
= − = .
n x n n+1 n(n + 1)
Problem 4
Determine, with proof, whether the (double) series
X m mn
,
n
(∗)
(∗) n = 2, 3, . . . , m = 1, 2, . . . , n − 1
converges.
Solution. Split the range (∗) into the subranges (I) n = 2, 3, . . . , m ≤ n/2 and (II)
n = 2, 3, . . . , n/2 < m ≤ n − 1. It suffices to show that the series taken over each of these
two ranges converge.
In the range (I), we have (m/n)mn ≤ (1/2)mn . Summing this upper bound first over
n (from n = 2m to infinity) gives a geometric series P with sum (1/2)m(2m) /(1 − (1/2)m )
∞
which is at most (1/2) (with a lot to spare). Since n=1 (1/2)m converges, so does the
m
series (∗) over the subrange (I). To deal with the second subrange, we set h = n − m, so
that 1 ≤ h < n/2 in the range (II). Using the bounds (m/n) = (1 − h/n) ≤ e−h/n and
mn ≥ n2 /2, we obtain (m/n)mn ≤ exp{− nh · 12 n2 } = exp{− hn 2 }. Summing the last term
2
over n, from n = 2h to infinity, gives again a geometric series, with sum e−h /(1 − e−h/2 )
which is at most e−h (1 − e−1/2 )−1 . Summing the latter bound, from h = 1 to infinity,
we again obtain a convergent geometric series. Hence the series over the subrange (II)
converges as well.
Problem 5
Let a1 = 2, a2 = 4, a3 = 8, and for n ≥ 4 define an to be last digit of the sum of the
preceding three terms in the sequence. Thus the first few terms of this sequence of digits
are (in concatenated form) 248468828 . . .. Determine, with proof, whether or not the string
2002 occurs somewhere in this sequence.
Solution. First note that the sequence can be continued backwards in a unique manner
by setting an−1 = an+2 − an+1 − an mod 10. Doing so, one finds that the first four
terms prior to the given terms are 2, 0, 0, and 2. Thus, the string 2002 occurs in the
extended sequence. To show that it also occurs in the original sequence (i.e., to the right
of 2484 . . .), note that the sequence is uniquely determined, backwards and forwards, by
any three consecutive digits in the sequence. Since there are finitely many possibilities for
such triples of consecutive digits, one such triple must occur again in the sequence, and the
sequence is therefore is periodic (in both directions). In particular, any string that occurs
somewhere in the extended sequence, occurs infinitely often and arbitrarily far out along
the given sequence. Hence 2002 does occur in this sequence.
Problem 6
Call a set of integers A double-free if it does not contain two elements a and a0 with a0 = 2a.
Determine, with proof, the size of the largest double-free subset of the set {1, 2, . . . , 256}.
Solution. We will show that the maximal cardinality sought is 171. To prove that
the cardinality cannot exceed 171, suppose A ⊂ {1, 2, . . . , 256} is double-free. Given any
element a ∈ A, let a0 denote the odd part of a, so that a = a0 2i with a0 odd and i a
nonnegative integer. For each odd integer m, let Am denote the set of a ∈ A with a0 = m.
The sets Am , m = 1, 3, . . . , 255 partition the set A, so |A| = |A1 | + |A3 | + · · · + |A255 |. To
obtain an upper bound for |A| we consider |Am | for different ranges of m.
If (1) 128 < m ≤ 256, then there can be at most one a ∈ A with a0 = m, namely
a = m. Thus, the sum over |Am | for m in the range (1) is at most equal to the number of
odd m in this range, i.e., 64.
If (2) 64 < m ≤ 128, then any a ≤ 256 with a0 = m must be of the form a = m or
a = 2m, but because of the double-free condition at most one of these can belong to a.
Hence |Am | ≤ 1 for m in the range (2), and the sum of |Am | over such m is at most 32.
If (3) 32 < m ≤ 64, then a0 = m implies that a = m2i with i = 0, 1, or 2, but
the double-free condition again implies that at most two of these can belong to A. Hence
|Am | ≤ 2 in the range (3), and the sum of |Am | over m in this range is at most 16 · 2 = 32.
Similarly, considering the ranges (4) 16 < m ≤ 32, (5) 8 < m ≤ 16, (6) 4 < m ≤ 8,
(7) 2 < m ≤ 4 (i.e., m = 3), and (8) m = 1, we see that |Am | is at most 2 in the range (4),
3 in the ranges (5) and (6), 4 in the range (7), and 5 in the range 8, and the corresponding
sums over |Am | are bounded by 8 · 2 = 16, 4 · 3 = 12, 2 · 3 = 6, 1 · 4 = 4, and 1 · 5 = 5,
respectively. Adding up these bounds, we obtain
|A| ≤ 64 + 32 + 32 + 16 + 12 + 6 + 4 + 5 = 171.
To show that this bound can be achieved, take A to be the set of integers n ≤ 256 that
are of the form a0 2i with a0 odd and i = 0, 2, . . .. In this case, it is easy to check that the
inequalities for |Am | in the above argument become equalities, and so we have |A| = 171.
UIUC UNDERGRADUATE MATH CONTEST
April 12, 2003
Solutions
1. Let
99
z }| {
N = 9 + 99 + 999 + · · · + 99 . . . 9 .
Determine the sum of digits of N .
Solution. The answer is 99. To see this, evaluate N explicitly as
follows:
100
z }| {
N = (10 − 1) + (100 − 1) + · · · + (10 . . . 0 −1)
99 97
z }| { z }| {
= 11 . . . 1 0 − 99 = 11 . . . 1 011.
2. Evaluate
1 1 1
+ + + ···
1·2·3 3·4·5 5·6·7
Solution. Let S be the sum of the given series (which P is easily seen
to be convergent, e.g., by comparing it with the series ∞
n=1 n −2 ). We
Letting UN = N
P PN
n=1 1/(2n) and VN = n=1 1/(2n − 1) denote the
partial sums over the even resp. odd terms in the harmonic series, we
can write the last expression as
1 1 1 1
SN = VN − 2UN + VN − 1 + = VN −UN − + .
2 2N + 1 2 2(2N + 1)
But
2N
X (−1)n+1
VN − UN = ,
n
n=1
which converges to ln 2 as N → ∞. Hence S = limN →∞ SN = ln 2 −
1/2, as claimed.
1
Now, Si /2i is the arithmetic mean of the 2i numbers an , 2i ≤ n < 2i+1 .
By the arithmetic-harmonic mean inequality, this is at least equal to
P −1
the harmonic mean of these numbers, namely 2i 2i ≤n<2i+1 1/an .
Hence,
22i X 1
≤ ,
Si i i+1
an
2 ≤n<2
Solution. The answer is n = 166. To see this, suppose first that P (x)
is a polynomial of degree n satisfying the three conditions (a), (b),
and (c). Consider the polynomial Q(x) = P (x) − x11 . Then Q(x) has
degree at most n, and condition (a) implies that Q(x) has a root at
2
each of the numbers k = 1, 2, . . . , n. It follows that Q(x) is of the form
Q(x) = C(x − 1)(x − 2) . . . (x − n) for some constant C. To determine
C, we use condition (c), which implies
3
UIUC UNDERGRADUATE MATH CONTEST
April 17, 2004
Solutions
4. Let f (x) = a1 sin x + a2 sin 2x + · · · + an sin nx, where the ak are real
numbers. Suppose that f (x) satisfies |f (x)| ≤ | sin x| for all real x.
Show that |a1 + 2a2 + · · · + nan | ≤ 1.
Solution. We have f 0 (x) = nk=1 kak cos kx, and so f 0 (0) = nk=1 kak .
P P
Thus, the claim is equivalent to |f 0 (0)| ≤ 1. Now, by the defini-
tion of the derivative, we have f 0 (0) = limx→0 f (x)/x, and since
|f (x)| ≤ | sin x| ≤ |x| for all x, the inequality |f 0 (0)| ≤ 1 follows.
5. Let
n h i
X n
f (n) = ,
k
k=1
where [x] denotes the greatest integer ≤ x, and let g(n) = (−1)f (n) .
Find, with proof, g(2004).
Solution. Note that g(n) = 1 or g(n) = −1 depending on whether
f (n) is even or odd. Since, for each k, [n/k] counts the number of
positive integers m for which km ≤ n, the function f (n) is equal to
the number of pairs (k, m) of positive integers that satisfy km ≤ n.
Among these pairs, the number of those with k 6= m is even since we
1
can pair up (k, m) with (m, k). Hence, modulo 2, f (n) is congruent
to the number of remaining pairs in the above count, i.e., those of the
√
form (k, k) with k ≤ n. Clearly, there are [ n] such k, so we have
√ √
f (n) = (−1)[ n] .
f (n) ≡ [ n] modulo 2, and therefore g(n) √ = (−1)
Since 442 = 1936 and 452 = 2025, we have [ 2004] = 44, so g(2004) =
(−1)44 = 1.
6. Find, with proof, all functions f (x) that are defined for real numbers
x with |x| < 1, continuous at x = 0, and which satisfy
f (x)
f (0) = 1, f (x2 ) = (|x| < 1).
1+x
Solution. First note that the function f (x) = 1/(1 − x) satisfies the
given conditions. We will show that this is the only solution. Suppose
f (x) is a solution, and let g(x) = (1 − x)f (x). Note first that since,
by assumption, f (x) is continuous at x = 0, g(x) is also continuous at
x = 0. From the given relation for f (x) we obtain g(0) = f (0) = 1,
and for |x| < 1,
(1 − x2 )f (x)
g(x2 ) = (1 − x2 )f (x2 ) = = g(x).
1+x
Iterating this identity, we get
n
g(x) = g(x2 )
n
for any positive integer n. Since, for |x| < 1, x2 tends to 0 as n → ∞
and g is continuous at 0, it follows that
n
g(x) = lim g(x2 ) = g(0) = 1
n→∞
for all x with |x| < 1. Hence f (x) = g(x)/(1 − x) = 1/(1 − x), so the
function 1/(1 − x) is indeed the only solution.
2
UIUC UNDERGRADUATE MATH CONTEST
April 16, 2005
Solutions
a1 a2 + a2 a3 + · · · + an−1 an + an a1 = 0
Rπ
2. Evaluate the integral I = 0 ln(sin x)dx.
Solution. We will show that I = −2π ln 2.
Using the identity sin x = 2 sin(x/2) cos(x/2) we can write I as
Z π Z π Z π
I= ln 2dx + ln sin(x/2)dx + ln cos(x/2)dx
0 0 0
Z π/2 Z π/2
= π ln 2 + 2 ln sin ydy + 2 ln cos ydy
0 0
= π ln 2 + 2I1 + 2I2 ,
4. Prove that, for any real numbers x and y in the interval (0, 1),
x + y x+y
≤ xx y y .
2
1
Solution. Let f (x) = log xx = x log x. Taking logarithms and divid-
ing both sides by 2, the given inequality is equivalent to
1 1
f (x + y) ≤ (f (x) + f (y)). (1)
2 2
Now note that f 0 (x) = log x + 1, and f 00 (x) = 1/x > 0 for positive
x, so the function f (x) is convex (i.e., concave up) for x > 0. Since
any convex function satisfies (1), the result follows. (The fact that (1)
holds for any convex function f is easily seen by sketching the graph
of a convex function and comparing the value of the function at the
midpoint of an interval [x, y], with the average of the values of the
function at the two end points x and y.)
converges or diverges.
√
Solution. We show that the series diverges. Note that sin x ≤ − 3/2
whenever x falls into one of the intervals
Each of these intervals has length π/3 > 1 and the gap between two
successive intervals has length < (5/3)π < 6. Hence, among any 7
consecutive integers n at least one must fall into one
√ of the intervals
Ik ; for this value of n we have 1.8 + sin n < 1.8 − 3/2 < 1, so the
corresponding term in the above series is greater than 1/n. Therefore
the above series is bounded from below by
∞ 7m+7 ∞ ∞
X X 1 X 1 1X 1
≥ = = ∞,
n1.8+sin n 7m + 7 7 m
m=0 n=7m+1 m=0 m=1
2
is a perfect square if and only if the exponents αIj , j = 1, . . . , m, are
all even, i.e., if and only if the system of congruences
X
αij ≡ 0 mod 2, j = 1, . . . , m (1)
i∈I
Thus, we need to show that, if m < n, then the latter system has a
solution i ∈ {0, 1}, with i 6= 0.
To this end we consider first the system
n
X
αij xi = 0, j = 1, . . . , m, (3)
i=1
3
UIUC UNDERGRADUATE MATH CONTEST
April 8, 2006
Solutions
1. Determine,
√ without numerical
√ calculations, which of the two numbers
√ 2006 √ 2005
2005 and 2006 is larger.
√
√ 2006
Solution. We will show that √ the first √
of the two numbers, 2005 ,
is the larger one. Taking the ( 2005 · 2006)-th root √
of the two given
√ 1/ 2005
numbers,√ this amounts to showing that 2005 is larger than
√ 1/ 2006
2006 . This will follow provided we can show that the√func-
tion √f (x) = x1/x is decreasing in an interval that includes 2005
and 2006. To do this, we compute the derivative of f (x): Writing
f (x) = exp{(ln x)/x}, we have, by the chain rule,
0 ln x x(1/x) − (ln x) · 1 1 − ln x
f (x) = exp 2
= x1/x .
x x x2
1
Solution. We claim that the given series is equal to 1/2005.
More generally, let
∞ n N n
X x2 X x2
S(x) = , SN (x) = .
n=0
1 − x2n+1 n=0
1 − x2n+1
Note that setting x = 1/2006 in S(x) gives the series of the problem.
We will show that, for any x with 0 < x < 1,
x
S(x) = , (∗)
1−x
and so, in particular, S(1/2006) = (1/2006)(1 − 1/2006) = 1/2005, as
claimed.
We give two proofs for (∗).
Note that all terms in this double series are positive, so we can rear-
range the terms in this series. Since each positive integer has a unique
decomposition as a power of 2 times an odd positive integer, the expo-
nents 2n (1 + 2k) in the latter double series correspond in a one-to-one
manner to the positive integers, and the latter double sum, and hence
S(x), is therefore exactly ∞ m
P
m=1 x = x/(1 − x). This again gives (∗).
2
4. For any positive integer n, define a sequence {nk }∞ k=0 as follows: Set
n0 = n, and for each k ≥ 1, let nk be the sum of the (decimal)
digits of nk−1 . For example, for n = 1729 we get the sequence
1729, 19, 10, 1, 1, 1, . . . . In general, for any given starting value n, the
resulting sequence {nk } eventually stabilizes at a single digit value.
Let f (n) denote this value; for example, f (1729) = 1. Determine
f (22006 ).
Solution. The key to this problem is the fact (which underlies the
well-known disibility test by 9) that the sum of the decimal digits of a
number has the same remainder modulo 9 as the number itself. Thus,
each of the numbers nk in the given sequence has the same remainder
modulo 9. In particular, f (n) has the same remainder modulo 9 as
n, and since f (n) must be among {1, 2, . . . , 9}, the remainder of n
modulo 9 determines f (n) uniquely. Thus, it remains to determine the
remainder of 22006 upon division by 9. This is easy using congruence
calculus: We have 26 = 64 ≡ 1 mod 9, and so
Hence f (22006 ) = 4.
3
components. Since we have 2007 > 1024 such vectors, by the pigeon-
hole principle two of these must have the same parity combination.
Denoting the indices of these two vectors by m and n (ordered so that
0 ≤ m < n ≤ 2006), we then have that Pn /Pm is a perfect square, as
claimed.
4
U OF I UNDERGRADUATE MATH CONTEST
April 14, 2007
Solutions
1. Let
f (n) = (12 + 1)1! + (22 + 1)2! + · · · + (n2 + 1)n!.
Find a simple general formula for f (n).
Solution. We will show by induction that
(∗) f (n) = n(n + 1)!
for n = 1, 2, . . . . For n = 1, (∗) holds trivially. Let now n ≥ 1, and assume that (∗)
holds for this value of n. Then
f (n + 1) = f (n) + ((n + 1)2 + 1)(n + 1)!
= n(n + 1)! + (n2 + 2n + 2)(n + 1)!
= (n2 + 3n + 2)(n + 1)!
= (n + 1)(n + 2)(n + 1)! = (n + 1)(n + 2)!,
which proves (∗) with n + 1 in place of n, completing the induction.
Since n is odd, the terms k n and (−k)n cancel each other out. Of the remaining
terms, all divisible by n2 , so the assertion follows.
3. For any positive integer k let f1 (k) denote the sum of the squares of the digits of
k (when written in decimal), and for n ≥ 2 define fn (k) iteratively by fn (k) =
f1 (fn−1 (k)). Find f2007 (2006).
Solution. Starting from k = 2006 and iterating the map “sum of squares of
the digits” we obtain the chain 2006 → 40 → 16 → 37 → 58 → 89 → 145 →
42 → 20 → 4 → 16, after which the sequence repeats itself, with period 8. Thus,
f1 (2006) = 40, f2 (2006) = 16, etc., and fn+8 (2006) = fn (2006) for all integers n ≥ 1.
Since 2007 = 8 · 250 + 7, it follows that follows that f2007 (2006) = f7 (2006) = 42.
4. Determine, with proof, whether the series
∞
1 n
X
e− 1+
n
n=1
converges.
Solution. We claim that the series diverges. To P show this, we first derive a bound for
ln(1 + x), using the Taylor expansion ln(1 + x) = ∞ n=1 (−x)
n+1 /n. For 0 < x < 1 the
latter series is an alternating series with decreasing terms, so the successive partial
sums of this series alternately undershoot and overshoot the limit, ln(1 + x). In
particular, we have, for 0 < x < 1,
x2 x3 x2 x2 x2
ln(1 + x) ≤ x − + ≤x− + =x− .
2 3 2 3 6
It follows that, for n ≥ 2,
1 n
1
1+ = exp n ln 1 + }
n n
1 1
≤ exp n − = e1−1/(6n) .
n 6n2
Hence, if an denotes the n-th term of the given series, we have the lower bound
1 n
1−1/(6n) −1/(6n)
an = e − 1 + ≥e−e =e 1−e
n
By another application of the alternating series properties, we see that, for 0 ≤ x ≤ 1,
∞
X (−x)k x2 x
e−x = ≤1−x+ ≤1− .
k! 2 2
k=0
1
Hence an ≥ 1/(2n) for n ≥ 2, and comparison with the harmonic series yields the
divergence of ∞
P
n=1 n .
a
where |Pi Pj | denotes the distance between Pi and Pj . In other words, S is the sum
of the squares of the pairwise distances between the points P1 , . . . , P12 . Determine,
with proof, the largest possible value of S among all choices of the points P1 , . . . , P12
on the unit circle.
Solution. We represent the points Pi by complex numbers zi . Points on the unit
circle correspond to complex numbers of absolute value 1, and the distance between
two such points is the absolute value of the difference between the corresponding
complex numbers. We thus can write
X
S = S(z1 , . . . , z12 ) = |zi − zj |2
1≤i<j≤12
and the problem reduces to that of maximizing this function subject to the condition
that |zi | = 1 for all i.
Since |zi − zi | = 0 and |zi − zj | = |zj − zi |, if we extend the above sum over 1 ≤ i <
j ≤ 12 to all pairs of indices (i, j), with 1 ≤ i, j ≤ 12, we count each summand twice.
Thus, S is exactly equal to half this extended sum, i.e.,
12 12
1 XX
S= |zi − zj |2
2
i=1 j=1
Using the assumption that |zi | = 1, we can expand the summands into
Thus, S ≤ 122 = 144 for any choice of the numbers zi (subject to |zi | = 1), and
S = 144 whenever
X12
zi = 0.
i=1
2
The latter condition can be achieved, for example, by choosing the first 6 points,
z1 , . . . , z6 , arbitrarily on the unit circle, and letting the remaining 6 points be the
points located diametrically opposite the 6 chosen points, i.e., z7 = −z1 , . . . , z12 =
−z6 . Hence 144 is the maximal value of S.
Thus the value of an−1 , along with that of the (constant) tuple a2007 = (a2007 , . . . , a1 ),
completely determines that of an , hence, by induction, the values of an+k and an+k
for all k ≥ 0.
Since the numbers an are bounded positive integers, there are only finitely many
possible values for the tuple an . By the pigeonhole principle, it therefore follows that
there exist positive integers n < m with an = am . In view of the above remark, this
implies an+k = am+k for all integers k ≥ 0. Thus, an is eventually periodic with
period m − n.
Note. In the original version of this problem, the assumption that the an be positive
was missing. This assumption ensures that the denominator in (1) is positive, so the
given recurrence can be written in the above form. Ben Kaduk constructed a example
showing that, i if the an are allowed to be 0, the conclusion need not not hold.
3
U OF I UNDERGRADUATE MATH CONTEST
March 8, 2008
Solutions
1. Does there exist a multiple of 2008 whose decimal representation involves only a single
digit (such as 11111 or 22222222)?
Solution. The answer is yes; specifically, we will show that there exists a multiple
of 2008 of the form 888 . . . 8.
Given a digit d ∈ {1, 2, . . . , 9}, let Nd,k be the number whose decimal representation
consists of k digits d. Note that
k−1
X d(10k − 1)
Nd,k = d 10i = .
9
i=0
Thus, a given positive integer m has a multiple of this form if and only if the con-
gruence (1) d(10k − 1) ≡ 0 mod 9m has a solution k. We apply this with d = 8
and m = 2008. Then (1) is equivalent to (2) 10k − 1 ≡ 0 mod 9(2008/8) = 9 · 251.
Since 10k ≡ 1k = 1 mod 9 for any positive integer k, (2) is equivalent to (3) 10k ≡
1 mod 251. Now, 251 is prime, so by Fermat’s Theorem, we have 10251−1 ≡ 1 mod 251.
Thus, (3) holds for k = 250, and so the number N8,250 = |88 {z
. . . 8} is divisible by 2008.
250
R1
2. What is the maximal value of the integral 0 f (x)x2008 dx among all nonnegative
R1
continuous functions f on the interval [0, 1] satisfying 0 f (x)2 dx = 1?
R1
Solution. Let I = 0 f (x)x2008 dx be the integral whose maximum we seek. Apply-
ing the integralRversion of Cauchy-Schwarz with the functions f (x) and x2008 , we get,
1
for any f with 0 f (x)2 dx = 1,
Z 1 Z 1
2 2 1
I ≤ f (x) dx x4016 dx = .
0 0 4017
√
Thus, I ≤√1/ 4017. Moreover, this upper
R1 bound is achieved by√taking f (x) = cx2008 ,
with c = 4017 (so that f satisfies 0 f (x)2 dx = 1). Thus, 1/ 4017 is the maximal
value for I.
|f (x + y) − f (x − y) − y| ≤ y 2
for all x, y ∈ R.
Solution. Any function of the form fc (x) = x/2 + c, where c is a constant, satisfies
fc (x + y) − fc (x − y) − y ≡ 0 for all x and y, and hence trivially satisfies the above
inequality. We will show that these are the only such functions.
Suppose f (x) is solution to the given inequality. Set g(x) = f (x) − x/2. Then
g(x + y) − g(x − y)
lim = lim y = 0,
y→0 y y→0
5. Let n be a positive integer, and denote by Sn the set of all permutations of {1, 2, . . . , n}.
Given a permutation σ ∈ Sn , define its perturbation index P (σ) as
i.e., P (σ) denotes the number of elements in {1, . . . , n} that are “perturbed” by σ,
in the sense of being mapped to a different element. Find the average perturbation
index of a permutation in Sn , i.e.,
1 X
P (σ).
#Sn
σ∈Sn
1
Solution. We have
X n
X X
P (σ) = 1
σ∈Sn σ∈Sn k=1
σ(k)6=k
n
X X
= 1
k=1 σ∈Sn
σ(k)6=k
Xn
= #{σ ∈ Sn : σ(k) 6= k}
k=1
Xn
= (n! − #{σ ∈ Sn : σ(k) = k})
k=1
Xn
= (n! − (n − 1)!)
k=1
= (n − 1)n!,
since, for any k, there are exactly (n − 1)! permutations in Sn that fix k. Since
#Sn = n!, it follows that the average perturbation index is n − 1.
6. Let A be a collection of 100 distinct, nonempty subsets of the set {0, 1, . . . , 9}. Show
that there exist two (distinct) sets A, A0 ∈ A whose symmetric difference has at
most two elements. (The symmetric difference of two sets A and A0 is defined as
the set of elements that are in one of the two sets, but not in both, i.e., the set
(A ∪ A0 ) \ (A ∩ A0 ).)
Solution. Let A∆B denote the symmetric difference of two sets A and B, and let
d(A, B) = |A∆B| denote the number of elements in A∆B. It is easy to see that the
function d satisfies the triangle inequality:
Now, let A be a collection of subsets of {0, 1, . . . , 9} with |A| = 100. In the above
terminology, we need to show that if A has at least 100 elements then it contains two
elements, A and A0 , such that d(A, A0 ) ≤ 2.
Given A ∈ A, define a “neighborhood” of A by
i.e., U(A) consists all nonempty subsets of {0, 1, . . . , 9} whose symmetric difference
with A has at most 1 element. We are going to estimate the sum of the cardinalities
of these “neighborhoods”, X
S= |U(A)| .
A∈A
2
To this end, note that U(A) consists of the following sets: (i) the set A itself, (ii) any
proper nonempty subset of A obtained by removing exactly one element from A, and
(iii) any set obtained by adding to A exactly one element from {0, 1, . . . , 9}.
If A has k elements with k ≥ 2, then there are exactly k sets of type type (ii), and
(10 − k) sets of type (iii), so U(A) has exactly 1 + k + (10 − k) = 11 elements. If A
has 1 element, then there is no set of type (ii) (since removing the single element of
A would leave an empty set), so in that case U(A) has 10 elements. Setting
A1 := {A ∈ A : |A| = 1}, A2 := A \ A1
we therefore have
by our assumption |A| = 100 and the trivial bound |A1 | ≤ 10, since there are 10 one
element subsets of {0, 1, . . . , 9}.
On the other hand, if the sets U(A), A ∈ A, were pairwise disjoint, we would have
[
(3) S= U(A)
A∈A
≤ |{B ⊂ {0, 1, . . . , 9} : B 6= ∅}|
= 210 − 1 = 1023,
contradicting the lower bound (2). Thus, two of these sets, say U(A) and U(A0 ),
must have a nonempty intersection. By the definition of the neighborhoods U this
means that there exists a nonempty subset B ⊂ {0, 1, . . . , 9} such that d(B, A) ≤ 1
and d(B, A0 ) ≤ 1. By the triangle inequality (1), this implies