Sharedwww / nsf / project_description.dviOpen in CoCalc
����;� TeX output 2004.03.31:1814��������������������vI�����5��#�K�'F
C�G�
cmbxti10�Wil��liam�)ZA.�Stein�H��Pr���oje�ct�)ZDescription��vˎ�����'�8�)m#�R

cmss10�(617)�UU308-0144�'OE�*m#�R
�3
cmss10�w��[email protected]�rva�rd.edu�http://mo�M�dula�r.fas.ha�rva�rd.edu�����'�8��ff��������V���������'�8���N�ffcmbx12�1��?�Bac���kground��q���'�8�K�`y
�3
cmr10�The���prop�M�osed�pro���ject�re
ects�the�in��!terpla�y���of�abstract�theory�with�explicit�mac��!hine��
����'�8computation,��fas�illustrated�b��!y�the�follo�wing�quote�of�Bry�an�Birc�h�[�Bir71��	f]:��\ݍ�B�<I�tw��!an�t��to�describ�M�e�some�computations�undertak��!en�b�y�m�yself�and�Swin-����B�<nerton-Dy��!er���on�EDSA�C���b�y�whic�h�w�e�ha�v�e�calculated�the�zeta-functions����B�<of�Y�certain�elliptic�curv��!es.���As�a�result�of�these�computations�w�e�ha�v�e����B�<found�K7an�analogue�for�an�elliptic�curv��!e�of�the�T��eamaga�w�a�n�um�b�M�er�of�an����B�<algebraic�Ygroup;��{and�conjectures�(due�to�ourselv��!es,���due�to�T��eate,�and����B�<due��fto�others)�ha��!v�e��fproliferated.����8�8The��nPI��[is�primarily�in��!terested�in�ab�M�elian�v��dDarieties�attac�hed�to�mo�M�dular�forms����'�8via�~0Shim��!ura's�construction�[�Shi73��(�],��"whic�h�w�e�no�w�recall.�e;Let���b>
�3
cmmi10�f���=��rK��ɖ���u
�3
cmex10�P���m�a���z��2cmmi8�n���P�q��d�����n��
��b�M�e�a����'�8w��!eigh�t�f�2�newform�on����z�|{Ycmr8�1����(�N�1��).�ȑThen��f��?�corresp�M�onds�to�a�dieren��!tial�on�the�mo�dular����'�8curv��!e�;�X���z�1����(�N�1��),��whic�h�is�a�curv�e�whose�ane�p�M�oin�ts�o�v�er��"�"V
�3
cmbx10�C��corresp�M�ond�to�isomor-����'�8phism�mclasses�of�pairs�(�E���;��1P��V�),�x�where��E���is�an�elliptic�curv��!e�and��P����!",�
�3
cmsy10�2�
��E��is�a�p�M�oin��!t�of����'�8order��f�N�1��.���W��ee�view�the�Hec��!k�e��falgebra���Q�����T�
��=��Z�[�T���z�1����;��1T���z�2���;�T���z�3���;��:�:�:�����]����'�8as��a�subring�of�the�endomorphism�ring�of�the�Jacobian��J���z�1����(�N�1��)�of��X���z�1���(�N�1��).�+=Let��I��Ȯ�f���
����'�8�b�M�e�ethe�k��!ernel�of�the�homomorphism��T�HQ�!��Z�[�a���z�1����;��1a���z�2���;�a���z�3���]�ethat�sends��T���z�n��

P�to��a���z�n���P�,���and����'�8attac��!h��fto��f��"�the�quotien�t�����"��A��Ȯ�f���ƹ=�
��J���z�1����(�N�1��)�=I��Ȯ�f��w�J���z�1���(�N��)�:������'�8�Then�HǾA��Ȯ�f��	��is�a�simple�ab�M�elian�v��dDariet��!y�o�v�er��Q��of�dimension�equal�to�the�degree�of����'�8the�X�eld��Q�(�a���z�1����;��1a���z�2���;�a���z�3���;��:�:�:�����)�X�generated�b��!y�the�co�M�ecien�ts�of��f�-��.���W��ee�also�sometimes����'�8consider��a�similar�construction�with��J���z�1����(�N�1��)�replaced�b��!y�the�Jacobian��J���z�0���(�N�1��)�of����'�8the���mo�M�dular�curv��!e��X���z�0����(�N�1��)�that�parametrizes�isomorphism�classes�of�pairs�(�E���;��1C�ȁ�),����'�8where��f�C�n�is�a�cyclic�subgroup�of��E�G��of�order��N�1��.����'�8�Denition��1�(Mo�Y�dular�ab�elian�v��L�ariet��ty).��7�A�;��%�':
�3
cmti10�newform�`ab��p�elian�variety�;�is�an����'�8ab�M�elian�چv��dDariet��!y�o�v�er��Q��of�the�form��A��Ȯ�f��w�.�z<An�ab�M�elian�v��dDariet�y�o�v�er�a�n�um�b�M�er�eld�is����'�8a��f�mo��p�dular���ab�elian�varieties��f�if�it�is�a�quotien��!t�of��J���z�1����(�N�1��)�for�some��N��.����8�8Ov��!er��Q�,�*gnewform�ab�M�elian�v��dDarieties�are�simple�and�ev�ery�mo�M�dular�ab�elian�v��dDa-����'�8riet��!y�Qjis�isogenous�to�a�pro�M�duct�of�copies�of�newform�ab�elian�v��dDarieties.���Newform����'�8ab�M�elian��fv��dDarieties�are�t��!ypically�not�absolutely�simple.����8�8Newform��	ab�M�elian�v��dDarieties��A��Ȯ�f��v(�are�imp�ortan��!t.��F��eor�example,the�celebrated�mo�d-����'�8ularit��!y���theorem�of�C.�Breuil,���B.�Conrad,�F.�Diamond,�R.�T��ea��!ylor,�and�A.�Wiles����'�8[�BCDT01��*�7]��Uasserts�that�ev��!ery�elliptic�curv�e�o�v�er��Q��is�isogenous�to�some��A��Ȯ�f��w�.��Also,����'�8J-P��e.��Serre�conjectures�that,�5qup�to�t��!wist,�ev�ery��t�w�o-dimensional�o�M�dd�irreducible����'�8mo�M�d��f�p��Galois�represen��!tation�app�ears�in�the�torsion�p�oin��!ts�on�some��A��Ȯ�f��w�.����8�8Muc��!h�ǰof�this�researc�h�prop�M�osal�is�inspired�b�y�the�follo�wing�sp�M�ecial�case�of�the����'�8Birc��!h��fand�Swinnerton-Dy�er�conjecture:�����������C�1���������*���������������fb�����5��#�K�Wil��liam�)ZA.�Stein�H��Pr���oje�ct�)ZDescription��vˎ�����'�8�(617)�UU308-0144�'OE�w��[email protected]�rva�rd.edu�http://mo�M�dula�r.fas.ha�rva�rd.edu�����'�8��ff��������V��������'�8�Conjecture��2�(BSD��Conjecture�(sp�Y�ecial�case)).��g�L��p�et��m�A��b�e�a�mo�dular�ab�elian��
����'�8variety���over��Q�.��������4�A1.����B�<�L�(�A;��1�1)�
�=�0����if�and�only�if��A�(�Q�)��is�innite.��
������4�A2.����B�<If���L�(�A;��1�1)�
��6�=�0�,�then���������������G�L�(�A;��1�1)����G�㦉p "�ߟ
���	߹
����A�������t�=���������ݕ��ɖ�Q��$?�c���z�p��6I��n�#��+�hV1
�3
wncyr10�X�����(�A�)��=ڟ�3�p kp[�
��#�A�(�Q�)���z�tor��
�/��n�#�A����0��K�cmsy8�_��*��(�Q�)���z�tor������p�h�;��-I��B�<�wher��p�e���the�obje�cts�and�notation�in�this�formula�ar�e�discusse�d�b�elow.�����8�8�Here���L�(�A;��1s�)�is�the��L�-series�attac��!hed�to��A�,��2whic�h�is�en�tire�b�M�ecause��A��is�mo�dular,����'�8so��ӾL�(�A;��1�1)�mak��!es�sense.��The�real�v�olume�
����A��
c��is�the�measure�of��A�(�R�)�with�resp�M�ect����'�8to�uFa�basis�of�dieren��!tials�for�the�N��"���Deron�mo�M�del�of��A�.�J~F��eor�eac�h�prime��p�cq�j��N�1��,���the����'�8in��!teger��c���z�p���`�=��#����A;p��
���(�F���z�p���]�)�is�the��T��)amagawa�X�numb��p�er��of��A��at��p�,�;�where�����A;p�����denotes����'�8the���comp�M�onen��!t�group�of�the�N��"���Deron�mo�del�of��A��at��p�.�pKThe�dual�of��A��is�denoted����'�8�A�����_��*��,���and��in�the�conjecture��A�(�Q�)���z�tor��Dݹand��A�����_���(�Q�)���z�tor��Dݹare�the�torsion�subgroups.��zThe����'�8�Shafar��p�evich-T��)ate���gr�oup��f�of��A��is��植��|���X���js�(�A�)�
�=��Ker������������0���ύ���@��� ��H���z��1����(�Q�;��1A�)��!����v������M���
2L���p�1���A��H���z��1���(�Q���z�p���]�;��1A�)�������1���ύA������Ѿ;��楍�'�8�whic��!h��Ais�a�group�that�measures�the�failure�of�a�lo�M�cal-to-global�principle.�}mWhen����'�8�L�(�A;��1�1)����6�=�0,��Kato���pro��!v�ed�in�[�Kat��@]�that���X��NT�(�A�)�and��A�(�Q�)�are�nite,��so�#��X�����(�A�)����'�8mak��!es��fsense�and�one�implication�of�part�1�of�the�conjecture�is�kno�wn.��&׍�'�8�R��p�emark�̹3�.�,�The���general�Birc��!h�and�Swinnerton-Dy�er�conjecture�(see�[�T��eat66����,��Lan91����])����'�8is��za�conjecture�ab�M�out�an��!y�ab�elian�v��dDariet��!y��A��o�v�er�a�global�eld��K�ȁ�.��It�asserts�that����'�8the�order�of�v��dDanishing�of��L�(�A;��1s�)�at��s�
��=�1�equals�the�free�rank�of��A�(�K�ȁ�),�#�and�giv��!es�a����'�8form��!ula�u�for�the�leading�co�M�ecien�t�of�the�T��ea�ylor�expansion�of��L�(�A;��1s�)�ab�M�out��s�
��=�1.����8�8The���rest�of�this�prop�M�osal�is�divided�in��!to�t�w�o�parts.��3The�rst�is�ab�M�out�com-����'�8puting���with�mo�M�dular�forms�and�ab�elian�v��dDarieties,��Pand�making�the�results�of�these����'�8computations�_a��!v��dDailable�to�the�mathematical�comm�unit�y��e.��The�second�is�ab�M�out�vis-����'�8ibilit��!y��^of�Mordell-W��eeil�and�Shafarevic�h-T��eate�groups,��[the�ultimate�goal�b�M�eing�to����'�8obtain��frelationships�b�M�et��!w�een��fparts�1�and�2�of�Conjecture�2.��!�%���'�8�2��?�Computing�ffwith�mo�s3dular�forms��q���'�8�The�N�PI�N�prop�M�oses�to�con��!tin�ue�N�dev�eloping�algorithms�and�making�a�v��dDailable�to�M�ols�for����'�8computing���with�mo�M�dular�forms,���mo�dular�ab�elian�v��dDarieties,���and�motiv��!es�attac�hed�to����'�8mo�M�dular�=�forms.��This�includes�nishing�a�ma���jor�new��-�-�
�3
cmcsc10�Ma���gma��[�BCP97��"]�pac��!k��dDage�for����'�8computing�-directly�with�mo�M�dular�ab�elian�v��dDarieties�o��!v�er�-n�um�b�er�-elds,�=^extending����'�8the��~Mo�M�dular�F��eorms�Database�[�Ste03a���%],��and�searc��!hing�for�algorithms�for�computing����'�8the��quan��!tities�app�M�earing�in�Conjecture�2�and�in�the�Blo�c��!h-Kato�conjecture�for����'�8mo�M�dular��fmotiv��!es.�����������C�2�������������������������fb�����5��#�K�Wil��liam�)ZA.�Stein�H��Pr���oje�ct�)ZDescription��vˎ�����'�8�(617)�UU308-0144�'OE�w��[email protected]�rva�rd.edu�http://mo�M�dula�r.fas.ha�rva�rd.edu�����'�8��ff��������V���������'�8�/��N�cmbx12�2.1��FZ8The��Mo�`dular�F���orms�Database������'�8�The��jMo�M�dular�F��eorms�Database�[�Ste03a���%]�is�a�freely-a��!v��dDailable�collection�of�data�ab�out��
����'�8ob���jects��attac��!hed�to�cuspidal�mo�M�dular�forms.���It�is�analogous�to�Sloane's�tables�of����'�8in��!teger�;�sequences,�`�and�extends�Cremona's�tables�[�Cre��S]�to�dimension�bigger�than����'�8one��Cand�w��!eigh�t��Cbigger�than�t��!w�o.��'Cremona's��Ctables�con��!tain�more�rened�data�ab�M�out����'�8elliptic�-�curv��!es�than�[�Ste03a���%],�O�but�the�PI�-�in�tends�to�w�ork�with�Cremona�to�mak�e����'�8the��fmo�M�dular�forms�database�a�sup�erset�of�[�Cre��S].����8�8The��database�is�used�w��!orld-wide�b�y�prominen�t�n�um�b�M�er�theorists,�_�including����'�8Noam���Elkies,���Matthias�Flac��!h,�Dorian�Goldfeld,�Benedict�Gross,�Ken�Ono,�and����'�8Don��fZagier.����8�8The�b.PI�a�prop�M�oses�to�greatly�expand�the�database.�5A�ma���jor�c��!hallenge�is�that����'�8data��ab�M�out�mo�dular�ab�elian�v��dDarieties�of�large�dimension�tak��!es�a�h�uge�amoun�t�of����'�8space���to�store.���F��eor�example,���the�database�curren��!tly�o�M�ccupies�40GB���disk�space.����'�8He�9prop�M�oses�to�nd�and�implemen��!t�a�b�etter�metho�d�for�storing�information�ab�out����'�8mo�M�dular��ab�elian�v��dDarieties�so�that�the�database�will�b�e�more�useful.��AHe�has�found�a����'�8metho�M�d��whereb��!y�a�certain�eigen�v�ector�is�computed�b�y�the�database�serv�er,���whic�h����'�8ma��!y�Ѽ(or�ma�y�not!)���enable�storing�co�M�ecien�ts�of�mo�M�dular�forms�far�more�ecien�tly;����'�8ho��!w�ev�er,��fhe�has�not�y��!et�tried�to�implemen�t�it�and�study�its�prop�M�erties.����8�8The�U=PI�T�prop�M�oses�to�impro��!v�e�U=the�usabilit��!y�of�the�database.�	�cIt�is�curren�tly����'�8implemen��!ted���using�a�P�ostgreSQL���database�coupled�with�a�Python�w�eb�in�terface.����'�8T��eo�r�sp�M�eed�access�and�impro��!v�e�r�eciency�,��
he�is�considering�rewriting�k��!ey�p�M�ortions����'�8of��/the�database�using�MySQL��and�PHP��e.�He�hop�M�es�to�rewrite�k��!ey�p�ortions�of�the����'�8database���in�resp�M�onse�to�user�feedbac��!k�that�he�has�receiv�ed.���The�database�curren�tly����'�8runs�aon�a�three-y��!ear-old�933Mhz�P�en�tium�I�M�I�I,�awhic�h�has�unduly�limited�disk�space����'�8and�Z�no�osite�bac��!kup,�i�so�the�PI�Zlis�requesting�a�p�M�o�w�erful�mo�M�dern�computer�with�a����'�8large��fhard�driv��!e�arra�y�and�external�hard�driv�es�for�osite�bac�kups.������'�8�2.1.1��NIM�0�"V

cmbx10�A��9GMA�2��pac��tk��L�age�for�mo�Y�dular�ab�elian�v��L�arieties����'�8�The��PI's�soft��!w�are��is�published�as�part�of�the�non-commercial��Ma���gma��computer����'�8algebra�Ԭsystem.�h�The�core�of��Ma���gma��is�dev��!elop�M�ed�b�y�a�group�of�academics�at����'�8the�ݾUniv��!ersit�y�of�Sydney��e,��who�are�supp�M�orted�mostly�b�y�gran�t�money��e.����Ma���gma��is����'�8considered���b��!y�man�y�to�b�M�e�the�most�comprehensiv�e�to�M�ol�for�researc�h�in�n�um�b�M�er����'�8theory��e,��nite�=group�theory�,��and�cryptograph��!y�,��and�is�widely�distributed.��cThe�PI����'�8has�5�already�written�o��!v�er�5�400�pages�(26000�lines)�of�mo�M�dular�forms�co�de�and�ex-����'�8tensiv��!e�`�do�M�cumen�tation�that�is�distributed�with��Ma���gma�,�n�and�in�tends�to�\publish"����'�8future��fw��!ork�in��Ma���gma�.����8�8As�9^men��!tioned�ab�M�o�v�e,�^an�ab�M�elian�v��dDariet�y��A��o�v�er�a�n�um�b�M�er�eld��K�߹is��mo��p�dular����'�8�if��it�is�a�quotien��!t�of��J���z�1����(�N�1��)�for�some��N��.���Mo�M�dular�ab�elian�v��dDarieties�w��!ere�stud-����'�8ied�$�in��!tensiv�ely�b�y�Ken�Rib�M�et,�>�Barry�Mazur,�and�others�during�recen��!t�decades,�and����'�8studying�Vthem�is�p�M�opular�b�ecause�results�ab�out�them�often�yield�surprising�insigh��!t����'�8in��!to�n�um�b�M�er�theoretic�questions.�G�Computation�with�mo�dular�ab�elian�v��dDarieties�is�����������C�3���������!F���������������fb�����5��#�K�Wil��liam�)ZA.�Stein�H��Pr���oje�ct�)ZDescription��vˎ�����'�8�(617)�UU308-0144�'OE�w��[email protected]�rva�rd.edu�http://mo�M�dula�r.fas.ha�rva�rd.edu�����'�8��ff��������V��������'�8�attractiv��!e�drb�M�ecause�they�are�m�uc�h�easier�to�describ�M�e�than�arbitrary�ab�elian�v��dDari-��
����'�8eties,��*and��their��L�-functions�are�reasonably�w��!ell�understo�M�o�d��when��K�i��is�an�ab�M�elian����'�8extension��fof��Q�.����8�8The�4#PI�4recen��!tly�designed�and�partially�implemen�ted�a�general�system�for�com-����'�8puting��ywith�mo�M�dular�ab�elian�v��dDarieties�o��!v�er��yn�um�b�er��yelds.��He�hop�es�to�dev��!elop����'�8and���rene�sev��!eral�crucial�comp�M�onen�ts�of�the�system.��3F��eor�example,���three�ma���jor����'�8problems���arose,��dand�the�PI���in��!tends�to�resolv�e�them�in�order�to�ha�v�e�a�completely����'�8satisfactory��fsystem�for�computing�with�mo�M�dular�ab�elian�v��dDarieties.���������4�]1.����B�<�Given��{a�mo��p�dular�ab�elian�variety��A�,���eciently�c�ompute�the�endomorphism����B�<ring��G�End����(�A�)�G�as�a�ring�of�matric��p�es�acting�on��H���z�1����(�A;��1�Z�)�.�Tr�The���PI���has�found�a����B�<mo�M�dular�C�sym��!b�ols�solution�that�dra��!ws�on�w�ork�of�Rib�M�et�[�Rib80��"�]�and�Shim�ura����B�<[�Shi73��(�],�$but���it�is�to�M�o�slo��!w�to�b�e�really�useful�in�practice.��In�[�Mer94��$�],�$Merel����B�<uses��Herbrand�matrices�and�Manin�sym��!b�M�ols�to�giv�e�ecien�t�algorithms�for����B�<computing��,with�Hec��!k�e��,op�M�erators.�
�.The�PI���in��!tends�to�carry�o�v�er�Merel's����B�<metho�M�d��fto�giv��!e�an�ecien�t�algorithm�to�compute��End��D�(�A�).�������4�]2.����B�<�Given���-�End���(�A�)�Y��
��Q�,�� c��p�ompute��-an�iso�geny�de�c�omp�osition�of��A��as�a�pr�o�duct�of����B�<simple���ab��p�elian�varieties.���This�^�is�a�standard�and�dicult�problem�in�general,����B�<but�Lit�migh��!t�b�M�e�p�ossible�to�com��!bine�w�ork�of�Allan�Steel�on�his�\c�haracter-����B�<istic�
0�Meataxe"�with�sp�M�ecial�features�of�mo�dular�ab�elian�v��dDarieties�to�solv��!e����B�<it�ѩin�practice.�_�It�is�absolutely��essential��to�solv��!e�this�problem�in�order�to����B�<explicitly�#�en��!umerate�all�mo�M�dular�ab�elian�v��dDarieties�o��!v�er��#���,R�p 	u��Ӯ��Q������of�#�giv�en�lev�el��N�1��.����B�<Suc��!h�Nan�en�umeration�w�ould�b�M�e�a�ma���jor�step�to�w�ards�the�ultimate�p�M�ossi-����B�<ble���generalization�of�Cremona's�tables�[�Cre��S]�to�mo�M�dular�ab�elian�v��dDarieties.����B�<Computation��_of�a�decomp�M�osition�is�also�crucial�to�other�algorithms,�Se.g.,����B�<computing��fcomplemen��!ts�and�duals�of�ab�M�elian�sub�v��dDarieties.�������4�]3.����B�<�Given�ttwo�mo��p�dular�ab�elian�varieties�over�a�numb�er�eld��K�ȁ�,�(�de�cide�whether����B�<ther��p�e�Yis�an�isomorphism�b�etwe�en�them.���When��"the�endomorphism�ring�of����B�<eac��!h�'aab�M�elian�v��dDariet�y�is�kno�wn�and�b�M�oth�are�simple,�G�it�is�p�ossible�to�reduce����B�<this�Һproblem�to�the�solution�of�a�norm�equation,��whic��!h�has�b�M�een�studied����B�<extensiv��!ely��in�man�y�cases.�	FThis�problem�is�analogous�to�the�problem�of����B�<testing��isomorphism�for�mo�M�dules�o��!v�er��a�xed�ring,��wwhic��!h�has�b�een�solv��!ed����B�<with��Fm��!uc�h�eort�for�man�y�classes�of�rings.��}One�application�is�to�pro�ving����B�<that��fsp�M�ecic�ab�elian�v��dDarieties�can�not�b�e�principally�p�olarized.��R����'�8�2.2��FZ8An��Example:�The�Arithmetic�of��1��g�cmmi12�J���̼1����.X�Qcmr12�(�p�)������'�8�W��ee��dnish�b�M�e�describing�recen��!t�w�ork�of�the�PI��on�the�mo�M�dular�Jacobian��J���z�1����(�p�),����'�8where���p��is�a�prime,��that�w��!as�partly�inspired�b�y�computation.�i�The�follo�wing�con-����'�8jecture���generalizes�a�conjecture�of�Ogg,���whic��!h�asserts�that��J���z�0����(�p�)(�Q�)���z�tor���ùis�cyclic�of����'�8order��fthe�n��!umerator�of�(�p�n����1)�=�12,��fa�fact�that�Mazur�pro�v�ed�in�[�Maz77��TG].�����������C�4���������1Ġ��������������fb�����5��#�K�Wil��liam�)ZA.�Stein�H��Pr���oje�ct�)ZDescription��vˎ�����'�8�(617)�UU308-0144�'OE�w��[email protected]�rva�rd.edu�http://mo�M�dula�r.fas.ha�rva�rd.edu�����'�8��ff��������V��������'�8�Conjecture��Y4�(Stein).�Oz�L��p�et�pZ�p��b�e�a�prime.���The�torsion�sub�gr�oup�of��J���z�1����(�p�)(�Q�)��is��
����'�8the���gr��p�oup�gener�ate�d�by�the�cusps�on��X���z�1����(�p�)��that�lie�over��1�
�2��X���z�0���(�p�)�.������8�8�The�'�PI�'�giv��!es�signican�t�n�umerical�evidence�for�this�conjecture�in�[�CES03�� d�],�A9and����'�8cuspidal�>3subgroups�of��J���z�1����(�p�)�are�considered�in�detail�in�[�KL81��O{],�d&where,�e.g.,�they����'�8compute��forders�of�suc��!h�groups�in�terms�of�Bernoulli�n�um�b�M�ers.����8�8Mazur's��ipro�M�of�of�Ogg's�conjecture�for��J���z�0����(�p�)�is�deep,��though�the�pro�of�for�the����'�8o�M�dd��_part�of��J���z�0����(�p�)(�Q�)���z�tor����is�m��!uc�h��_easier.���The�PI��^in��!tends�to�explore�whether�or�not����'�8it�f�is�p�M�ossible�to�build�on�Mazur's�metho�d�and�pro��!v�e�f�results�to��!w�ards�f�Conjecture�4.����'�8The��oPI��jalso�in��!tends�to�dev�elop�his�computational�metho�M�ds�for�computing�torsion����'�8subgroups��fin�order�to�answ��!er,�at�least�conjecturally��e,�the�follo�wing�question.����'�8�Question�5.�0m�If��ھA��Ȯ�f�� ��is�a�quotien��!t�of��J���z�1����(�p�)�attac�hed�to�a�newform,��\is�the�natural�map����'�8�J���z�1����(�p�)(�Q�)���z�tor��W��!��K�A��Ȯ�f��w�(�Q�)���z�tor���ٹsurjectiv��!e?��lIs�5�the�pro�M�duct�of�the�orders�of�all��A��Ȯ�f���(�Q�)���z�tor�����'�8�o��!v�er��fall�classes�of�newforms��f��"�equal�to�#�J���z�1����(�p�)(�Q�)���z�tor��^C�?����8�8The��PI��conjectured�that�the�analogous�questions�for��J���z�0����(�p�)�should�ha��!v�e��\y�es"����'�8answ��!ers,�$�and�yin�[�Eme01�� d�]�M.�Emerton�pro�v�ed�this�conjecture.���There�he�also�pro�v�ed����'�8that���the�natural�map�from�the�comp�M�onen��!t�group�of��J���z�0����(�p�)�to�that�of��A��Ȯ�f����is�surjectiv�e.����'�8By�m�[�CES03�� d�],�y
the�comp�M�onen��!t�group�of��J���z�1����(�p�)�is�trivial,�whic��!h�suggests�the�follo�wing����'�8question.����'�8�Question��n6.���If�
@�A��Ȯ�f���_�is�a�quotien��!t�of��J���z�1����(�p�)�attac�hed�to�a�newform,�+�is�the�comp�M�onen�t����'�8group��fof��A��Ȯ�f��	��trivial?����8�8Ev��!en�z�assuming�the�full�BSD�z�conjecture�(and�a�conjecture�ab�M�out�a�Manin�con-����'�8stan��!t),�^the�"PI��has�not�y�et�pro�M�duced�enough�data�to�giv�e�a�conjectural�answ�er����'�8to���this�question.�BPHe�has�man��!y�examples�in�whic�h�the�conjecture�predicts�that����'�8either��I��X����(�A��Ȯ�f��w�)�I�is�non��!trivial�or�the�comp�M�onen�t�group�of��A��Ȯ�f��	��is�non�trivial.�țHe�and����'�8B.���P��!o�M�onen�form�ulated,���and�hop�M�e�to�carry�out,�a�strategy�to�decide�whic��!h�of�these����'�8t��!w�o��!is�non��!trivial�b�y�using�an�explicit�description�of��End��%�(��A��Ȯ�f��w�=���,R�p 	u��Ӯ��Q�����ù)�to�obtain�a�curv�e����'�8whose��aJacobian�is��A��Ȯ�f��w�.�%�Note�that�computing��End��\�(�A��Ȯ�f���=���,R�p 	u��Ӯ��Q���	u��)�in�general�is�the�second����'�8problem��fin�Section�2.1.1�ab�M�o��!v�e.��"�A���'�8�3��?�Visibilit���y��q���'�8�The�HGunderlying�motiv��dDation�for�this�part�of�the�prop�M�osal�is�to�pro��!v�e�HGimplications����'�8b�M�et��!w�een�MCthe�t��!w�o�MCparts�of�Conjecture�2,�_in�examples�and�ev��!en�tually�MCin�some�gener-����'�8alit��!y��e.�U�That��fis,��fw�e�link�information�ab�M�out�the�rst�part�of�the�BSD��\conjecture�for����'�8an��ab�M�elian�v��dDariet��!y��B���to�information�ab�out�the�second�part�of�the�conjecture�for�a����'�8related��{ab�M�elian�v��dDariet��!y��A�.��9Visibilit�y�pro�vides�a�conceptual�framew�ork�in�whic�h�to����'�8organize��four�ideas.�����������C�5���������Aˠ��������������fb�����5��#�K�Wil��liam�)ZA.�Stein�H��Pr���oje�ct�)ZDescription��vˎ�����'�8�(617)�UU308-0144�'OE�w��[email protected]�rva�rd.edu�http://mo�M�dula�r.fas.ha�rva�rd.edu�����'�8��ff��������V���������'�8�3.1��FZ8Computational��problems������'�8�Barry�n�Mazur�in��!tro�M�duced�visibilit�y�in�order�to�unify�v��dDarious�constructions�of�Shafarevic�h-��
����'�8T��eate��fgroups.��%	��'�8�Denition�2�7�(Visibilit��ty�of�Shafarevic�h-T���\ate�Groups).��Supp�M�ose��fthat��
���ҵo��
��:��A�,��,��!��J����'�8�is���an�inclusion�of�ab�M�elian�v��dDarieties�o��!v�er����Q�.�oThe��visible���sub��p�gr�oup����of�H�����1����(�Q�;��1A�)�with����'�8resp�M�ect��fto��J���is�����|�Vis����~m���J���6��H���z��1����(�Q�;��1A�)�
�:=��Ker����(H���z��1���(�Q�;��1A�)��!��H���z��1���(�Q�;��1J�
��))�:����'�8�The����visible�3Usub��p�gr�oup��of���X���n�(�A�)�in��J��w�is�the�in��!tersection�of���X���(�A�)�with��Vis���q����J��)йH�����1����(�Q�;��1A�);����'�8equiv��dDalen��!tly��e,��������Vis����On���J������X����1�(�A�)�
�:=��Ker����(��X�����(�A�)��!���X���;�(�J�
��))�:��%	��8�8�The�#�terminology�\visible"�arises�from�the�fact�that�if��x�
��2���X���;�(�A�)�#�is�visible�in��J�
��,����'�8then���a�principal�homogenous�space��X����corresp�M�onding�to��x��can�b�e�realized�as�a����'�8sub��!v��dDariet�y��fof��J�
��.����8�8Before��cdiscussing�theoretical�questions�ab�M�out�visibilit��!y��e,��w�e��cdescrib�e�computa-����'�8tional�5�evidence�for�the�Birc��!h�and�Swinnerton-Dy�er�conjecture�for�mo�M�dular�ab�elian����'�8v��dDarieties���(and�motiv��!es)�that�the�PI��^and�A.�Agashe�obtained�using�theorems�in-����'�8spired��b��!y�the�denition�of�visibilit�y��e.��MIn�[�AS02��>�],��the�PI��and�Agashe�pro�v�e�a�theorem����'�8that�;:mak��!es�it�p�M�ossible�to�use�ab�elian�v��dDarieties�of�p�ositiv��!e�rank�to�explicitly�con-����'�8struct���subgroups�of�Shafarevic��!h-T��eate�groups�of�other�ab�M�elian�v��dDarieties.�fThe�main����'�8theorem�[is�that�if��A��and��B����are�ab�M�elian�sub��!v��dDarieties�of�an�ab�elian�v��dDariet��!y��J�
��,�7Xand����'�8�B����[�p�]�
����A�,��fthen,�under�certain�h��!yp�M�othesis,�there�is�an�injection������}�B����(�Q�)�=pB��(�Q�)�
��,��,��!���Vis����n���J���U��X��'1�(�A�)�:����'�8�The�%pap�M�er�concludes�with�the�rst�ev��!er�example�of�an�ab�elian�v��dDariet��!y��A��Ȯ�f�����attac�hed����'�8to�x�a�newform,��-of�large�dimension�(20),�whose�Shafarevic��!h-T��eate�group�has�order����'�8that��fis�pro��!v��dDably�divisible�b�y�an�o�M�dd�prime�(5).����8�8The�l3PI�k�has�used�the�result�describ�M�ed�ab�o��!v�e�l3to�giv��!e�evidence�for�the�BSD����'�8conjecture��for�man��!y��A��*���J���z�0����(�N�1��),�	where��A��is�attac�hed�to�a�newform�of�lev�el��N��������'�8�2333.�e�The�ӳPI��fprop�M�oses�to�giv��!e�similar�evidence�using�visibilit�y�in��J���z�0����(�N�1�M��)�ӳfor����'�8small��!�M�1��.��qMore�precisely��e,���[�AS��K�]�describ�M�es�the�computation�of�an�o�dd�divisor�of�the����'�8BSD���conjectural���order�of���X�����(�A�)�for�o��!v�er���ten�thousand��A�
����J���z�0����(�N�1��)���with��L�(�A;��1�1)�
��6�=�0����'�8(these�Aare��al��Fl��simple��A��with��N�������2333�and��L�(�A;��1�1)��6�=�0).�6nF��eor�o��!v�er�Aa�h��!undred�of����'�8these,�the��divisor�of�the�conjectural�order�of���X�����(�A�)�is�divisible�b��!y�an�o�M�dd�prime;����'�8for���a�quarter�of�these�the�PI���and�Agashe�pro��!v�e���that�if��n��is�the�conjectural�divisor����'�8of�x�the�order�of���X��7W�(�A�),���then�there�are�at�least��n��elemen��!ts�of���X���(�A�)�that�are�visible����'�8in��f�J���z�0����(�N�1��).�����������C�6���������Q)���������������fb�����5��#�K�Wil��liam�)ZA.�Stein�H��Pr���oje�ct�)ZDescription��vˎ�����'�8�(617)�UU308-0144�'OE�w��[email protected]�rva�rd.edu�http://mo�M�dula�r.fas.ha�rva�rd.edu�����'�8��ff��������V��������8�8�The���PI���in��!tends�to�in�v�estigate�the�remaining�75%�of�the��A��with��n�
�>��1���b�y�consid-��
����'�8ering��7the�image�of��A��in��J���z�0����(�N�1�M��)��7for�small�in��!tegers��M�1��.���Information�ab�M�out�whic�h��M����'�8�to��.c��!ho�M�ose�can�b�e�extracted�from�Rib�et's�lev��!el�raising�theorem�(see�[�Rib90��"�]).�
6As�a����'�8test,�0[the��PI��recen��!tly�tried�the�rst�example�with�conjectural�o�M�dd���X���l�(�A�)�that�is�not����'�8visible�BNin��J���z�0����(�N�1��)�(this�is�an�18�dimensional�ab�M�elian�v��dDariet��!y��A��of�lev�el�551�suc�h�that����'�89����j��#��X�����(�A�)).��tHe��sho��!w�ed�in�[�Ste03b�� =�]�that�there�are�elemen�ts�of�order�3�in���X���,�(�A�)����'�8that�\�are�visible�in��J���z�0����(551�ۍ���2).��MSince�\�the�dimension�of��J���z�0���(�N�1�M��)�\�gro��!ws�v�ery�quic�kly��e,����'�8a�ruh��!uge�amoun�t�of�computer�memory�will�b�M�e�required�to�in�v�estigate�visibilit�y�at����'�8higher�Qzlev��!el.���F��eortunately�,�bvthe�PI�Qerecen��!tly�receiv�ed�a�gran�t�from�Sun�Microsystems����'�8for�	�a�$67,000�computer�that�con��!tains�22GB�	�of�con�tiguously�addressable�RAM�	�(the����'�8pro�M�cessors��are�relativ��!ely�slo�w�and�the�hard�driv�e�is�small,�6�making�this�computer����'�8less�v�suitable�as�a�platform�for�the�mo�M�dular�forms�database,��/whic��!h�requires�a�large����'�8hard��fdriv��!e�but�not�so�m�uc�h�RAM).����8�8Some��of�these�ideas�generalize�to�the�con��!text�of�Grothendiec�k�motiv�es,���whic�h����'�8A.���Sc��!holl�attac�hed�to�newforms�of�w�eigh�t�greater�than�t�w�o.��]N.�Dummigan,����'�8M.���W��eatkins,���and�the�PI���did�w��!ork�in�this�direction�in�[�D�WS03��$Y].��There�w�e�pro�v�e�a����'�8theorem��that�can�sometimes�b�M�e�used�to�deduce�the�existence�of�visible�Shafarevic��!h-����'�8T��eate���groups�in�motiv��!es�attac�hed�to�mo�M�dular�forms,��;assuming�a�conjecture�of�Beilin-����'�8son�G�ab�M�out�ranks�of�Cho��!w�groups.�¥Ho�w�ev�er,�pdw�e�giv�e�sev�eral�pages�of�tables�that����'�8suggest��that�Shafarevic��!h-T��eate�groups�of�mo�M�dular�motiv�es�of�lev�el��N���are�rarely����'�8visible�=�in�the�higher-w��!eigh�t�=�motivic�analogue�of��J���z�0����(�N�1��),�R�m��!uc�h�=�more�rarely�than�for����'�8w��!eigh�t�!�2.���Just�as�ab�M�o��!v�e,�<^the�!�question�remains�to�decide�whether�one�exp�ects�these����'�8groups��Nto�b�M�e�visible�in�the�analogue�of��J���z�0����(�N�1�M��)��Nfor�some�in��!teger��M�1��.�%�It�w�ould�b�M�e����'�8relativ��!ely��Nstraigh�tforw�ard�for�the�PI��Gto�do�computations�in�this�direction,�ʈand�he����'�8in��!tends��fto�do�so.����8�8Before�2mo��!ving�on�to�theoretical�questions�ab�M�out�visibilit�y��e,�IJw�e�pause�to�empha-����'�8size�7that�the�ab�M�o��!v�e�7computational�in��!v�estigations�7in�to�the�Birc�h�and�Swinnerton-����'�8Dy��!er��%conjecture�motiv��dDated�the�PI��"and�others�to�dev�elop�new�algorithms�for�com-����'�8puting��'with�mo�M�dular�ab�elian�v��dDarieties.�� F��eor�example,��in�[�CS01���],�B.�Conrad�and����'�8the�2LPI�2/use�Grothendiec��!k's�mono�M�drom�y�pairing�to�giv�e�an�algorithm�for�computing����'�8orders�p�of�comp�M�onen��!t�groups�of�certain�purely�toric�ab�elian�v��dDarieties.�<HThis�algo-����'�8rithm��mak��!es�it�practical�to�compute�comp�M�onen�t�groups�of�quotien�ts��A��Ȯ�f��	7'�of��J���z�0����(�N�1��)����'�8at�dQprimes��p��that�exactly�divide��N�1��.��Without�suc��!h�an�algorithm�it�w�ould�proba-����'�8bly���b�M�e�dicult�to�get�v��!ery�far�in�computational�in�v�estigations�in�to�the�Birc�h�and����'�8Swinnerton-Dy��!er�$�conjecture�for�ab�M�elian�v��dDarieties;��indeed,��
the�only�other�pap�er����'�8in�:�this�direction�is�[�FpS�����+��x�01��%`7],�_�whic��!h�restricts�to�the�case�of�Jacobians�of�gen�us�2����'�8curv��!es.�����������C�7���������_����������������fb�����5��#�K�Wil��liam�)ZA.�Stein�H��Pr���oje�ct�)ZDescription��vˎ�����'�8�(617)�UU308-0144�'OE�w��[email protected]�rva�rd.edu�http://mo�M�dula�r.fas.ha�rva�rd.edu�����'�8��ff��������V���������'�8�3.2��FZ8Theoretical��problems�������'�8�3.2.1��NIVisibilit��ty�2�at�higher�lev�el����'�8�Supp�M�ose�أ�A��Ȯ�f��O¹is�a�quotien��!t�of��J���z�1����(�N�1��)�attac�hed�to�a�newform�and�let��A�
��=��A������_��N���f���	5W���J���z�1����(�N�1��)��
����'�8b�M�e�zits�dual.���One�exp�ects�that�most�of���X����(�A�)�is��not��visible�in��J���z�1����(�N�1��).���The�follo��!wing����'�8conjecture��fthen�arises.������'�8�Conjecture���8�(Stein).�.��F��)or�9le��p�ach��x����2���X��Wt�(�A�)�,�M
ther�e�9lis�an�inte�ger��M�k�and�a�mor-����'�8phism���f�P4�:�"x�A��!��J���z�1����(�N�1�M��)�,��Yof�nite�de��p�gr�e�e�and�c�oprime�to�the�or�der�of��x�,��Ysuch�that����'�8the���image�of��x��in���X���~�(�f�-��(�A�))��is�visible�in��J���z�1����(�N�1�M��)�.����8�8�In�|n[�AS02��>�],��the�PI�|#pro��!v�ed�|nthat�if��x�
��2��H�����1����(�Q�;��1A�),�then�|nthere�is�an�ab�M�elian�v��dDariet��!y��B����'�8�and���an�inclusion���0�:��A��!��B�I��suc��!h���that��x��is�visible�in��B����;��moreo�v�er,��B�I��is�a�quotien�t����'�8of�=�J���z�1����(�N�1�M��)�for�some��M��.�	cThis�theorem�is�the�main�reason�wh��!y�the�PI�
�mak�es����'�8Conjecture��8.�YThe�PI�vhop�M�es�to�pro��!v�e��Conjecture�8�b��!y�understanding�the�precise����'�8relationship�(�b�M�et��!w�een��A�,�I�B����,�and��J���z�1����(�N�1�M��).�d+First�he�will�in��!v�estigate�explicitly�the����'�8example��fwith��N�<F�=�
�551�describ�M�ed�in�Section�3.1�ab�o��!v�e.����8�8A��more��analytical,�:�and�p�M�ossibly�deep�er,�:�approac��!h�to�Conjecture�8�is�to�assume����'�8the�xrank�statemen��!t�of�the�Birc�h�and�Swinnerton-Dy�er�conjecture�and�relate�when����'�8elemen��!ts�S1of���X��Ź(�A�)�b�M�ecoming�visible�at�lev�el��N�1�M��йto�when�there�is�a�congruence����'�8b�M�et��!w�een�/�f�1�and�a�newform��g�h¹of�lev��!el��N�1�M�5ιwith��L�(�g�d�;��1�1)�
�=�0.���Suc�h�/an�approac�h�leads����'�8one�bfto�try�to�form��!ulate�a�renemen�t�of�Rib�M�et's�lev�el�raising�theorem�that�includes����'�8a���statemen��!t�ab�M�out�the�b�eha��!vior�of�the�v��dDalue�at�1�of�the��L�-function�attac�hed�to����'�8the���form�at�higher�lev��!el.��_The�PI���in�tends�to�do�further�computations�in�the�hop�M�es����'�8of���nding�a�satisfactory�conjectural�renemen��!t�of�Rib�M�et's�theorem,���whic�h�he�then����'�8hop�M�es��fto�subsequen��!tly�pro�v�e.����8�8The��PI���also�prop�M�oses�to�in��!v�estigate��whether�there�is�an��M���that�is�minimal�with����'�8resp�M�ect�9�to�some�prop�ert��!y��e,�O�suc�h�9�that�ev��!ery�elemen�t�of���X���i�(�A�)�is�sim�ultaneously�visi-����'�8ble�e�in��J���z�1����(�N�1�M��).��CThis�e�is�w��!ell�w�orth�lo�M�oking�in�to,�r�since�the�pa�y�os�could�b�M�e�h�uge|����'�8the��"existence�of�suc��!h�an��M����w�ould�imply�niteness�of���X��W��(�A�),���since��Vis���+���J���(��X�����(�A�))�is����'�8alw��!a�ys�Z�nite.�ĞFiniteness�of���X��<�(�A�)�is�a�m��!ysterious�op�M�en�problem�when��L�(�A;��1�1)�
�=�0����'�8and��f�A��is�not�a�quotien��!t�of��J���z�0����(�N�1��)�with��ord���B��z�s�=1��$�+�L�(�A;��1s�)�
�=��dim��ھA�.������'�8�3.2.2��NIVisibilit��ty�2�of�Mordell-W���\eil�groups������'�8�The���Gross-Zagier�theorem�asserts�that�p�M�oin��!ts�on�elliptic�curv�es�of�rank�1�come����'�8from�/�Heegner�p�M�oin��!ts,�R*and�that�p�oin��!ts�on�curv�es�of�rank�bigger�than�one�do�not.����'�8It��Oseems�dicult�to�describ�M�e�where�p�oin��!ts�on�elliptic�curv�es�of�rank�bigger�than�1����'�8\come�from".�2�The�PI��in��!tro�M�duced�the�follo�wing�denition,�4�in�hop�M�es�of�ev�en�tually����'�8creating��fa�framew��!ork�for�giving�a�conjectural�explanation.����'�8�Denition�F�9�(Visibilit��ty�of�Mordell-W���\eil�Groups).�	��Supp�M�ose��that������:�(�J�5��!��A����'�8�is�z?a�surjectiv��!e�morphism�of�ab�M�elian�v��dDarieties�with�connected�k�ernel.�YhThe��visible�����������C�8���������	p���������������fb�����5��#�K�Wil��liam�)ZA.�Stein�H��Pr���oje�ct�)ZDescription��vˎ�����'�8�(617)�UU308-0144�'OE�w��[email protected]�rva�rd.edu�http://mo�M�dula�r.fas.ha�rva�rd.edu�����'�8��ff��������V��������'�8�quotient���of��A�(�Q�)��fwith�resp�M�ect�to��J���(and���d��)�is��������%�Vis��������z��J������(�A�(�Q�))�
�:=��Cok��!er��.(�J�
��(�Q�)��!��A�(�Q�))�:����8�8�Visibilit��!y�n�of�Mordell-W��eeil�groups�is�closely�connected�to�visibilit�y�of�Shafarevic�h-��
����'�8T��eate�Kgroups.��If��C��̹is�the�k��!ernel�of���g޹and�����:��y�A�(�Q�)��!��H�����1����(�Q�;��1C�ȁ�)�Kis�the�connecting����'�8homomorphism��fof�Galois�cohomology��e,�then���}�induces�an�isomorphism��������D���4;~��㼍���R������:��
�Vis����n��z��J���l�(�A�(�Q�))���������
������l���
��=��������Vis���,[���J��$Y�(H���z��1����(�Q�;��1C�ȁ�))�:����'�8�Note��fthat�this�implies��Vis���9-���'�J��+�(�A�(�Q�))�is�nite.���Let��������Vis����z����g��J��:j���g��6!D��wncyr8�X�����x�(�A�(�Q�))�
�:=���D����~��㼍����j��z���1������(�Vis����ǟ��J��wŹ(��X�����(�C�ȁ�)))�:����8�8�Though�x}w��!e�ha�v�e�in�tro�M�duced�nothing�fundamen�tally�new,��this�dieren�t�p�M�oin�t��
����'�8of�d�view�suggested�questions�that�seemed�unnatural�b�M�efore,��Bwhic��!h�inspired�the����'�8follo��!wing�8etheorem�and�conjecture�(the�pro�M�of�of�the�theorem�relies�on�[�Kat��@,��Rub98��"e�]����'�8and��f[�Roh84����]):������'�8�Theorem�t�10�(Stein).���L��p�et��A��b�e�an�el��Fliptic�curve.��rIf��x�j��2��A�(�Q�)���has�or�der��n��(set����'�8�n��k�=�0����if��x��has�innite�or��p�der),���then�for�every�divisor��d��of��n�,�ther��p�e�is�surje�ctive����'�8morphism���J�D�!�
��A�,��Owith�c��p�onne�cte�d���kernel,�such�that�the�image�of��x��in���Vis���b����'�J��G��(�A�(�Q�))����'�8�has���or��p�der��d�.����'�8�Conjecture�p�11�(Stein).���Supp��p�ose�ؾA��is�a�mo�dular�ab�elian�variety�and��x�e�2��A�(�Q�)����'�8�has���or��p�der��n�.��F��)or�every�divisor��d��of���n��ther�e�is�a�surje�ctive�morphism��J�D�!�
��A�,�îwith����'�8c��p�onne�cte�d�V�kernel,�t�such�that�the�image�of�*'�x��in���Vis���齟��'�J��λ�(�A�(�Q�))��lies�in���Vis�����'��齿J��7�����X���G��(�A�(�Q�))��and����'�8has���or��p�der��d�.����8�8�W��ee�y�no��!w�describ�M�e�partial�results�ab�out�this�conjecture�that�the�PI�ywpro��!v�ed�y�in����'�8[�Ste04��(�].�
9�Supp�M�ose�oľE�V�is�an�elliptic�curv��!e�o�v�er��Q��with�conductor��N�1��,��and�let��f����'�8�b�M�e�ϱthe�newform�attac��!hed�to��E����.�Y�Fix�a�prime��p��!����
�3
msbm10�-��2�N��П�ɖ�Q��/z�c���z�p��	��suc�h�ϱthat�the�Galois����'�8represen��!tation���fGal���h(���,R�p 	u��Ӯ��Q���	u��)�
��!���Aut����(�E����[�p�])��fis�surjectiv�e.����'�8�Conjecture��12�(Stein).���Ther��p�e���is�a�prime��`�
��-��N�	��and���a�surje�ctive�Dirichlet�char-����'�8acter����
��:�(�Z�=`�Z�)�������ʫ�!�����z�p���G�such�that������n���L�(�E���;��1;��1)�
��6�=�0���b��and���>e�a��Ȯ�`����(�E��)��6��`�n�+�1�
�&(�mo�M�d���1�p�)�:����8�8�According�4�to�Sarnak�and�Ko��!w�alski,�K�this�4�conjecture�do�M�es�not�seem�amenable�to��
����'�8standard�Aanalytic�a��!v�eraging�Aargumen�ts.��{The�PI�has�v�eried�this�conjecture�for�the����'�8elliptic���curv��!e�of�rank�1�and�conductor�37�and�all��p������25000.��In���almost�all�cases,����'�8the�,Csmallest��`����-��N�]�suc��!h�,Cthat��a��Ȯ�`����(�E����)����6��`��)�+�1���(�mo�M�d���1�p�)�,Cand��`������1�(�mo�M�d���1�p�)�,Csatises����'�8the��fconjecture.����8�8The��fPI�pro��!v�ed��fthe�follo��!wing�theorem�in�[�Ste04��(�].�����������C�9���������
�����������������fb�����5��#�K�Wil��liam�)ZA.�Stein�H��Pr���oje�ct�)ZDescription��vˎ�����'�8�(617)�UU308-0144�'OE�w��[email protected]�rva�rd.edu�http://mo�M�dula�r.fas.ha�rva�rd.edu�����'�8��ff��������V��������'�8�Theorem�:13�(Stein).���L��p�et�[t�E���b�e�an�el��Fliptic�curve�over��Q��and�supp�ose��p��and����ar�e��
����'�8as��Jin�Conje��p�ctur�e��J12�ab��p�ove.���Then�ther�e�is�an�exact�se�quenc�e��0�
��!��A��!��J�D�!��E��9�!��0����'�8�that���induc��p�es�an�exact�se�quenc�e������lRʹ0�
��!��E����(�Q�)�=pE��(�Q�)�
��!���X���;�(�A�)��!���X���(�J�
��)��!���X���(�E����)��!��0�:����'�8�In���p��p�articular,����}e��E����(�Q�)�=pE��(�Q�)���������
������l���
��=��������Vis����z���,[�J��:j��,[�X���(�9�(�E��(�Q�))���������
������l���
��=��������Vis���,[���J��$Y�(��X�����(�A�))�:����8�8�W��ee��dnish�this�researc��!h�prop�M�osal�b�y�explaining�ho�w�Theorem�13�ma�y�lead�to�a��
����'�8link��b�M�et��!w�een�the�t�w�o�parts�of�the�BSD��pConjecture�(Conjecture�2).��2Supp�M�ose��E���is����'�8an�%Selliptic�curv��!e�o�v�er��Q��and��L�(�E���;��1�1)�
�=�0.���Then�%Spart�1�of�Conjecture�2�asserts�that����'�8�E����(�Q�)���is�innite.���Under�our�h��!yp�M�othesis�that��L�(�E�;��1�1)���=�0,��ea���standard�argumen��!t����'�8sho��!ws��fthat���ȍ����������`\�L�(�A;��1�1)���`\�㦉p "�ߟ
���	߹
����A�������$��
��0�
�&(�mo�M�d���1�p�)�;��Oq��'�8�where�kL�A��is�as�in�Theorem�13.�,�If�part�2�of�Conjecture�2�w��!ere�true,���there�w�ould����'�8b�M�e�U}an�elemen��!t��x���2���X�����(�A�)�U}of�order��p��(the�pro�of�of�Theorem�13�rules�out�the����'�8p�M�ossibilit��!y��that��p��divides�a�T��eamaga�w�a�n�um�b�M�er).��CIf,���in�addition,��x��w�ere�visible�in����'�8�J�
��,���then����E����(�Q�)�w��!ould�b�M�e�innite,�since��E����(�Q�)�has�no�elemen��!ts�of�order��p�.�֕P�art�2�of����'�8Conjecture�?02�do�M�es�not�assert�that��x��is�visible�in��J�
��,�ebso�one�can�only�hop�e�that�a����'�8close�Vexamination�of�an�ev��!en�tual�Vpro�M�of�of�part�2�of�Conjecture�2�w��!ould�yield�some����'�8insigh��!t�P�in�to�whether�or�not��x��is�visible.��ZAlternativ�ely��e,�a�one�could�try�to�replace�the����'�8isomorphism��f�E����(�Q�)�=pE��(�Q�)���������
������l���
��=��������Vis���,[���J��$Y�(��X�����(�A�))�b��!y�an�isomorphism���������Sel��������z��(�p�)���)ʹ(�E����)���������
������l���
��=���������X��X(�(�A�)[�I����]����'�8where��`�I�]X�is�an�appropriate�ideal�in�the�ring��Z�[����z�p���]�]�of�endomorphism�of��A�.�n�Then��
����'�8an��Happropriate�renemen��!t�of�part�2�of�Conjecture�2�migh�t�imply�that���X��Oܹ(�A�)[�I����]����'�8con��!tains��an�elemen�t�of�order��p�,���whic�h�w�ould�imply�that�either��E����(�Q�)�is�innite�or�����'�8�X��4X̹(�E���=�Q�)[�p�]��fis�nonzero.����8�8One�Zcan�also�w��!ork�orthogonally�to�the�ab�M�o�v�e�approac�h�b�y�in�v�estigating�similar����'�8situations�h�coming�from�lev��!el�raising,�uwhere�isomorphisms�lik�e�the�ones�ab�M�o�v�e�ma�y����'�8arise.�̗The�KOPI�K$in��!tends�to�in�v�estigate�this�cluster�of�ideas�from�v��dDarious�directions����'�8in�G�hop�M�es�of�nding�a�new�p�ersp�ectiv��!e�on�where�p�oin��!ts�on�elliptic�curv�es�of�rank����'�8bigger��fthan�one�come�from�it.�����������10����������C���������������fb�����5��#�K�Wil��liam�)ZA.�Stein�H��Pr���oje�ct�)ZDescription��vˎ�����'�8�(617)�UU308-0144�'OE�w��[email protected]�rva�rd.edu�http://mo�M�dula�r.fas.ha�rva�rd.edu�����'�8��ff��������V��������'�8�References��q�����'�8�[AS]���^_A.�)Agashe�and�W.��MA.�Stein,�ۘ�Visible��Evidenc��p�e�for�the�Bir�ch�and��
����^_Swinnerton-Dyer�29Conje��p�ctur�e�for�Mo�dular�Ab�elian�V��)arieties�of�A��\nalytic����^_Rank���0,��fT��eo�app�M�ear�in�Math.�of�Computation.��������'�8[AS02]����^_�ff ٟ����K,���Visibility��Jof�Shafar��p�evich-Tate�gr�oups�of�ab�elian�varieties�,��J.����^_Num��!b�M�er��fTheory��T�97��(2002),�no.�1,�171{185.�MR�2003h:11070������'�8[BCDT01]���^_C.�ڮBreuil,���B.�Conrad,�F.�Diamond,�and�R.�T��ea��!ylor,��On�the�mo��p�dularity����^_of�}�el��Fliptic�curves�over��Q�:�/�wild�3-adic�exer��p�cises�,�n6J.�[email protected]�Math.�So�M�c.����^_�14��f�(2001),�no.�4,�843{939�(electronic).�MR�2002d:11058������'�8[BCP97]���^_W.��Bosma,��J.�Cannon,�and�C.�Pla��!y�oust,��The�:vMagma�algebr��p�a�system.�I.����^_The��2user�language�,���J.��[Sym��!b�M�olic�Comput.��24��(1997),�no.�3-4,�235{265,����^_Computational�halgebra�and�n��!um�b�M�er�htheory�(London,�t{1993).�MR�g�1�484����^_478������'�8[Bir71]���^_B.��MJ.���Birc��!h,��0�El��Fliptic���curves�over��Q�:���A���pr��p�o�gr�ess���r�ep�ort�,�1969���Num��!b�M�er����^_Theory��OInstitute�(Pro�M�c.�Symp�os.�Pure�Math.,�F�V��eol.�XX,�State�Univ.����^_New�fY��eork,��Ston��!y�Bro�M�ok,�N.Y.,�1969),�Amer.�Math.�So�M�c.,�Pro��!vidence,����^_R.I.,��f1971,�pp.�396{400.������'�8[CES03]���^_B.�v9Conrad,��S.�Edixho��!v�en,�and�v9W.��MA.�Stein,��J���z�1����(�p�)���Has�Conne��p�cte�d��Fib�ers�,����^_Do�M�cumen��!ta��fMathematica��8��(2003),�331{408.������'�8[Cre]���^_J.��ME.��fCremona,��El��Fliptic���curves�of�c��p�onductor���
��17000�,����^_�8��<x
�3
cmtt10�http://www.maths.nott.ac.uk/personal/jec/ftp/data/�.������'�8[CS01]���^_Brian�:gConrad�and�William�A.�Stein,�P�Comp��p�onent���gr�oups�of�pur�ely�toric����^_quotients�,�TMath.��Res.�Lett.��8��(2001),�no.�5-6,�745{766.�MR���2003f:11087������'�8[D��!WS03]���^_N.�;�Dummigan,�aM.�W��eatkins,�and�W.��MA.�Stein,��Constructing�tPElements����^_in���Shafar��p�evich-T��)ate�Gr�oups�of�Mo�dular�Motives�,�|�Num��!b�M�er�R
theory�and����^_algebraic��$geometry��e,��Sed.�b��!y�Miles�Reid�and�Alexei�Sk�orob�M�ogato�v���303����^_�(2003),��f91{118.������'�8[Eme01]���^_M.��GEmerton,����Optimal��>Quotients�of�Mo��p�dular�Jac�obians�,���preprin��!t��G(2001).������'�8[FpS�����+��x�01]���^_E.��MV.�YFFlynn,���F.�Lepr��"���Dev��!ost,�E.��MF.�Sc��!haefer,�W.��MA.�Stein,�M.�Stoll,�and����^_J.��ML.�$�W��eetherell,�>��Empiric��p�al�s�evidenc�e�for�the�Bir�ch�and�Swinnerton-Dyer����^_c��p�onje�ctur�es��for�mo��p�dular�Jac�obians�of�genus�2�curves�,��&Math.�w�Comp.��70����^_�(2001),��fno.�236,�1675{1697�(electronic).�MR�1�836�926������'�8[Kat]���^_K.�ȔKato,�� �p�-adic�
\Ho��p�dge�the�ory�and�values�of�zeta�functions�of�mo�dular����^_forms�,��fPreprin��!t,�244�pages.�����������11����������U���������������fb�����5��#�K�Wil��liam�)ZA.�Stein�H��Pr���oje�ct�)ZDescription��vˎ�����'�8�(617)�UU308-0144�'OE�w��[email protected]�rva�rd.edu�http://mo�M�dula�r.fas.ha�rva�rd.edu�����'�8��ff��������V����������'�8�[KL81]���^_D.��MS.�<?Kub�M�ert�and�S.�Lang,����Mo��p�dular�`Kunits�,�Grundlehren�<?der�Mathe-��
����^_matisc��!hen�=�Wissensc�haften�[F��eundamen�tal�Principles�of�Mathematical����^_Science],��fv��!ol.�244,�Springer-V��eerlag,�New�Y�ork,�1981.�MR�84h:12009��������'�8[Lan91]���^_S.�(�Lang,�I>�Numb��p�er�b�the�ory.�III�,�(�Springer-V��eerlag,�I>Berlin,�1991,�Diophan-����^_tine��fgeometry��e.�MR�93a:11048������'�8[Maz77]���^_B.�sMazur,����Mo��p�dular��3curves�and�the�Eisenstein�ide�al�,���Inst.�sHautes��p+��;j��Etudes����^_Sci.��fPubl.�Math.�(1977),�no.�47,�33{186�(1978).������'�8[Mer94]���^_L.��AMerel,����Universal�(lFourier�exp��p�ansions�of�mo�dular�forms�,���On��AArtin's����^_conjecture�5ifor�o�M�dd�2-dimensional�represen��!tations,�(Springer,�1994,����^_pp.��f59{94.������'�8[Rib80]���^_K.��MA.�e�Rib�M�et,����Twists��of�mo��p�dular�forms�and�endomorphisms�of�ab�elian����^_varieties�,��fMath.�Ann.��253��(1980),�no.�1,�43{62.�MR�82e:10043������'�8[Rib90]����^_�ff ٟ����K,�Y��R��p�aising�+Zthe�levels�of�mo�dular�r�epr�esentations�,�Y�S��"���Deminaire��de����^_Th��"���Deorie�*�des�Nom��!bres,�CPP�aris�*�1987{88,�Birkh���fauser�Boston,�Boston,�MA,����^_1990,��fpp.�259{271.������'�8[Roh84]���^_D.��ME.�p�Rohrlic��!h,����On���L�-functions�of�el��Fliptic�curves�and�cyclotomic�towers�,����^_In��!v�en�t.��fMath.��75��(1984),�no.�3,�409{423.�MR�86g:11038b������'�8[Rub98]���^_K.���Rubin,����Euler��systems�and�mo��p�dular�el��Fliptic�curves�,�Galois���represen-����^_tations�Tin�arithmetic�algebraic�geometry�(Durham,��1996),�Cam��!bridge����^_Univ.��fPress,�Cam��!bridge,�1998,�pp.�351{367.�MR�2001a:11106������'�8[Shi73]���^_G.���Shim��!ura,����On���the�factors�of�the�jac��p�obian�variety�of�a�mo�dular�func-����^_tion���eld�,��fJ.�Math.�So�M�c.�Japan��25��(1973),�no.�3,�523{544.������'�8[Ste03a]���^_W.��MA.��fStein,��The���Mo��p�dular�Forms�Datab�ase�,����^_�http://modular.fas.harvard.edu/Tables��f�(2003).������'�8[Ste03b]����^_�ff ٟ����K,�J�Studying���the�Bir��p�ch�and�Swinnerton-Dyer�Conje�ctur�e�for�Mo�d-����^_ular�oAb��p�elian�V��)arieties�Using�MA��\GMA�,�6T��eo�app�M�ear�in�J.�Cannon,�Y�ed.,����^_�Computational�J/Exp��p�eriments�in�A��\lgebr�a�and�Ge�ometry�,��Springer-V��eerlag����^_(2003).������'�8[Ste04]����^_�ff ٟ����K,����Shafar��p�evich-tate�}�gr�oups�of�nonsquar�e�or�der�,���Pro�M�ceedings�\rof����^_MCA��ȈV��f2002,�Progress�of�Mathematics�(2004),�277{289.������'�8[T��eat66]���^_J.��mT��eate,��n�On��the�c��p�onje�ctur�es��of�Bir��p�ch�and�Swinnerton-Dyer�and�a�ge�o-����^_metric�f�analo��p�g�,�N�S��"���Deminaire�-+Bourbaki,�V��eol.�9,�So�M�c.�Math.�F��erance,�P��!aris,����^_1965/66,��fpp.�Exp.�No.�306,�415{440.�����������12��������������;���%�g0	�8��<x
�3
cmtt10�6!D��wncyr8�1��g�cmmi12�0�"V

cmbx10�/��N�cmbx12�.X�Qcmr12�-�-�
�3
cmcsc10�+�hV1
�3
wncyr10�*m#�R
�3
cmss10�)m#�R

cmss10�'F
C�G�
cmbxti10�%�':
�3
cmti10�"�"V
�3
cmbx10����
�3
msbm10���u
�3
cmex10��K�cmsy8�!",�
�3
cmsy10��2cmmi8��b>
�3
cmmi10�|{Ycmr8���N�ffcmbx12�K�`y
�3
cmr10��&�����