Sharedwww / padictwist.dviOpen in CoCalc
����;� TeX output 1999.10.11:1434������y�����?�����Wv��D��tG�G�cmr17�Twisting�7to��qv�ercon�v�ergen�t����g�G�cmmi12�p�-adic�eigenforms�����7��X�Qffcmr12�(Still��/v���ery�preliminary)��������|�f�X�Qcmr12�Rob�S�ert��F.�Coleman����������0William��A.�Stein�������������Octob�S�er��11,�1999��+č��>�*��N�ffcmbx12�1��VL�Twists�ffcan�b�s3e�appro���ximated�������>�,�"V

cmbx10�Theorem��T1.1.�������+�':

cmti10�L��}'et��f�
�b>

cmmi10�F�J��b�e�a�form�of�tame�level��N����and�let��!��[ٟ�^��	0e�rcmmi7�r�����b�e�a�p�ower�of�the����>T��;�eichmul���ler�%char��}'acter��!�r��at��p�,�7�with��K�y

cmr10�(�N���;���p�)���=�1�.�#OThen�%the�twist��F����
!",�

cmsy10�
����!��[ٟ�^��r����c�an�b�e����>appr��}'oximate�d���arbitr�arily�closely�by�eigenforms�of�level��N��.������>Pr��}'o�of.���]UI�First��w���e�lose�nothing�b�y�starting�with�an�old�form��F�J)�on��X����ٓ�Rcmr7�1��|s�(�N�p�),���since�if����>w���e�'�appro�ximate�its�t�wist�b�y�a�form�whose�slop�Ge�is�large,�1but�small�with�resp�ect����>to�o1its�w���eigh�t,�u�it�o1is�congruen���t�to�the�form�it�came�from�on��X����1��|s�(�N��).��[Supp�Gose�the����>w���eigh�t���of��F�Q|�is��k�>��and�its�slop�Ge�is���	z�.�OOLet��F�c��(�q�[ٲ)��=��������u

cmex10�P�����a����n��q~�q����^��n���D�b�Ge���the��q��-expansion�of��F�c��.����>Let�)z�A��b�Ge�a�p�ositiv���e�in�teger.�c)Then�if��m��is�an�in�teger�sucien�tly�small��p�-adically����>there�jrexists�an�o���v�ercon�v�ergen�t�jreigenform��G��of�w���eigh�t�jr�k�����c�m��and�slop�Ge���s�suc���h�that����>�G�(�q�[ٲ)�����F�c��(�q��)�q�(�mo�Gd����p���^��A�����).�P=If��k��8��o��m����0,��then��F����1��C��=����G��^��m�O!�cmsy7��k�+B�+1���L�G��is�an�o���v�ercon�v�ergen�t����>eigenform��form�of�w���eigh�t���k����1��	h��:=�습m�����k�e>�+�2�and�slop�Ge������1��	h��:=�습�!�+��k����1������1.�N�If����>�F����1��|s�(�q�[ٲ)��=������P�����b����n��q~�q����^��n���W�,�UUthen������ӵb����n��8�����n������k�����Zcmr5�1��� ��1����a����n��qn�(�mo�Gd����p������A�����)�:����>�It�CGfollo���ws�that�if��B�ø�is�large�there�exists�an�eigenform��G����1�����of�w�eigh�t��k����1��S��׀�p���^��B����r����>�and��|slop�Ge������1��q�suc���h�that��G����1��|s�(�q�[ٲ)�����F����1���(�q�[ٲ)�q�(�mo�Gd����p���^��A�����).�R=No���w��|if��k����1�� �����p���^��B����r� <���0,��then��
�:��>�F����2��C��:=�����G��^��p����r�O
�\cmmi5�B��� �r�7��k���1��� �+1��'F��G����1���w�is�an�o���v�ercon�v�ergen�t�eigenform�of�w���eigh�t��k����2���:=���p���^��B����r��]���@�k����1��"��+�2����>and�UUslop�Ge������2��C��:=�������1���S�+�8�k����2�����1.�q�If�UU�F����2��|s�(�q�[ٲ)��=������P�����c����n��q~�q����^��n���W�,�UUthen��>���)�c����n��8�����n������k���2��� ��1����b����n�����n������p����r�B��� �r��x�a����n��qn�(�mo�Gd����p������A�����)�:����M�It�"follo���ws�that�if��v�p��is�sucen�tly�small��p�-adically��*�,�!�then�there�exists�a�classical����>eigenform����F����3��Zq�of�w���eigh�t����k����2��Ƨ�+�J4�v�"�=���k��˲+��v��
�+��p���^��B����r��Q���m��and�slop�Ge������2��Zq�suc���h�that��F����3��|s�(�q�[ٲ)�������>�F����2��|s�(�q�[ٲ)�q�(�mo�Gd����p���^��A�����).���It�qfollo���ws�from�the�ab�Go�v�e�that�if��B�u�>��SA��then��F����3��텲is�congruen�t����>to�UUthe�t���wist�of��F���b�y�the��r�G�-th�p�o���w�er�UUT��*�eic�hm�uller�c�haracter�mo�Gdulo��p���^��A�����.����+��ff����d�ff�Y��ff����ff����!č��>�2��VL�Some�ffn���umerical�exp�s3erimen�ts�����>�W��*�e�*�ha���v�e�exp�Gerimen�tally�in�v�estigated�the�question�of�appro�ximation�of�t�wists�of����>nite�UUslop�Ge�forms.������1����*�y�����?������M�One�w���a�y�to�mak�e�\appro�ximate"�precise�and�falsiable,�%is�to�ask�the�follo�w-����>ing:��let�8�n�����1�b�Ge�an�in���teger;���do�es�there�exist�a�newform��g�"�2���S���:�k�+B�+�'�(�p����n���l�)���_�(����0��|s�(�N�p�);����Z����p���R�)����>suc���h�>that��g�"����f�����^��"��	̃�(�mo�Gd����p���^��n��q~�)?�V(By�this�congruence,��w�e�mean�that��ord���{$���p��v�(�g�����f�����^��"��Zò)��=�����>min���N�����i��1��\;n�f��ord���x���p��8�(�a����i��TL�(�g�[ٲ)�8���a����i���(�f�����^��"��Zò))�g����n:�)���6���>�5��N� cmbx12�2.1��\�A��false�conjecture��uT��>�In�ܝthis�section�w���e�state�a�v�ery�general,��o�and�]very�false�,�conjecture.��Let�ܝ�p��b�G e�a����>prime�UUand�let��N�lp�b�G e�a�p�ositiv���e�in�teger�not�divisible�b�y��p�.������>�Conjecture��T2.1�(F�� �alse�conjecture).����l��L��}'et����f�脸2����S����k��됲(����0��|s�(�N��)�;����Z����p���R�)��b�e�a� nite�slop�e����>newform,��and��µ"��a�char��}'acter�mo�d��p��such�that��"���^��2��Ŷ�=�IC1�.�q'Then�for�al���l��n����1��ther��}'e����>exists���an�eigenform��g�"�2���S���:�k�+B�+�'�(�p����n���l�)�� �_�(����0��|s�(�N�p�);����Z����p���R�)��such�that��g�����f�����^��"�� ̃�(�mo�G d����p���^��n��q~�)�.�����>�2.2��\�Numerical��exp�erimen��ts��uT��>�In�;�the�remainder�of�this�pap�G er�w���e�rep�ort�on�n���umerical�exp�erimen���ts�designed�to����>gather�[�evidence�ab�G out�Conjecture�2.1.��LThese�computations�w���ere�carried�out�in����>�6�-� cmcsc10�Ma�� gma�UU�(cf.�[�Can��� ]).����MW��*�e�UUp�G erform�the�follo���wing�exp�erimen���ts:��' ����H|���WSection�ŗ2.3.�Let��f����=��/�b�G e�the�unique�cusp�form�of�w���eigh�t�ŗ12.�F��*�or�eac���h����W�p�i��=�2�;����3�;��5�;��7�;��11�Ptconsider�the�t���wist��f�����^��"�� �7�b�y�either�the�trivial�or�quadratic����Wmo�G d�k��p��c���haracter.���W��*�e�compute��ord���䝟��p����(�g�NN���u�f�����^��"��Zò)�for�eac�h�rational�newform����W�g�"�2���S���:�k�+B�+�'�(�p����n���l�)�� �_�(����0��|s�(�N�p�);����Z����p���R�),��Zfor�W�the� rst�few�v��q�alues�of��n�.� 4In�fact,��ord���@���p�����(�g��H��=o�f�����^��"��Zò)����Ww���as��Ucomputed�using�only�17�terms�of�the��q�[ٲ-expansion,��whic�h�is�probably����Wenough,�UUbut�w���e�ha�v�e��not��pro�v�ed�this.������H|���WSection���2.3.1.�I�Let��f��r�b�G e�the�w���eigh�t-2���newform�corresp�onding�to�the�elliptic����Wcurv���e�UU�X����0��|s�(11).�q�F��*�or��p���=�2�;����3�;��5�UUw�e�p�G erform�the�same�computation�as�for�.��uT��MSince,�in���this�exp�G erimen���t,�w�e���only�t���wist��f��R�b�y�the�trivial�or�quadratic�c�har-����>acter,�E1w���e�A(adopt�the�notation��f�����^��(1)��,�and��f�����^��(��1)��Q-�for�these�t�wists,�E1resp�G ectiv�ely��*�.�k Th�us����>�f�����^��(1)��%Y�is�UUthe��p�-depriv��q�ation�of��f���,�and��f����^��(��1)��eZ�is�the�quadratic�t���wist.������>�R��}'emark���2.2.���xk��The�UUnewforms�at�lev���el��N�p��all�ha�v�e�slop�G e�(�k��w��8�2)�=�2.���6���>�2.3��\�Twisting�������>�Consider�UUthe�w���eigh�t-12�UUcuspform�;�the��q�[ٲ-expansion�of��is��qˍ��Op|����a��=�����s��q���������Y����͟�(1�8���q��[ٟ����n���W�)������24�����#ݍ���a�Ӳ=�����s��q�����8�24�q��[ٟ����2��,�+�252�q��[ٟ����3�����1472�q��[ٟ����4���+�4830�q��[ٟ����5�����6048�q��[ٟ����6�����16744�q��[ٟ����7���+���������8׵:������>�Let��Եp��b�G e�a�prime�and�let��"�uA�:�(�Z�=p�Z�)���^����%�!�f�1�g��Բb�e�a�c���haracter;��then��"���^��2���=�uA1,���so����>���^��"���Ƹ2����S����12��x�(����0��|s�(9)).�� As��koutlined�in�Section�2.2,��qw���e�compute�eac�h�of�the�rational����>newforms�UU�g�"�2���S���:�12+�'�(�p����n���l�)��#%��(����0��|s�(�p�)�;����Z����p���R�)�;��and�then��ord����;���p��m��(���^��"�����8�g�[ٲ).����MThe��computations�are�giv���en�in�the�follo�wing�tables.�sZF��*�or�example,�+Rthe�4th����>line�^�of�the� rst�table�b�G elo���w�sa�ys�that�when��n��I�=�4�^�there�are�t�w�o��Z����2��|s�-rational��� ���2����)�y�����?������>�newforms�HH�g����1��Ļ�and��g����2��|s�;���and�that����^��(1)��u��\�g����1���3�(�mo�G d���2���^��5���)�and����^��(1)��u��\�g����2���3�(�mo�G d���2���^��5���).�� ��>The�UUnon-�Z����p���R�-rational�newforms�in��S���:�12+�'�(�p����n���l�)��#%��(����0��|s�(�p�)�;�������_�fe����Z�������p�� Q�)�are�not�listed.��d$��]BEigenforms�UUon�����0��|s�(2)�congruen���t�mo�Gd�2�to�the�2-depriv��q�ation�of�.��4������������ow�n��Ο���z�ff�����p�ord���������2���em�(���^��(1)���U��8�g����i��TL�)����ov��fft!�fd���1��g����ff���"�none������2��g����ff���"�3,�UU3������3��g����ff���"�4������4��g����ff���"�5,�UU5������5��g����ff���"�6,�UU6������6��g����ff���"�7,�UU7,�7,�7������<ԥ��wEigenforms�UUon�����0��|s�(3)�congruen���t�mo�Gd�3�to�t�wists�of���(�����������bc3�n��Ο���z�ff�����cвord����ܶ���3���Y)�(���^��(1)���U��8�g����i��TL�)�-�ԟ���z�ff�����Pord����6���3��u��(���^��(��1)��5V��8�g����i��TL�)���\c2��ff9��fd���1��g����ff���"�1,�UU1,�1�^�e����ff����1,�UU1,�1������2��g����ff���"�1,�UU2,�1�^�e����ff����2,�UU1,�2������3��g����ff���"�3,�UU1,�3,�3,�1�Hw����ff����1,�UU3,�1,�1,�3������4��g����ff���"�4,�UU4,�1,�1,�1,�3,�4,�4,�1,�1,�1��͟���ff����1,�UU1,�4,�4,�3,�1,�1,�1,�4,�4,�3������0ԥ��wEigenforms�UUon�����0��|s�(5)�congruen���t�mo�Gd�5�to�t�wists�of������������D�n��Ο���z�ff����j��ord���xy����5��|���(���^��(1)���U��8�g����i��TL�),��
UVord����<���5�� J��(���^��(��1)��5V���g����i���)���>��ff�9��fd���1��g����ff���"�[1,�UU0,�1,�0,�0],�[�0,�1,�0,�1,�1]������2��g����ff���"�[2,�UU1,�1,�0,�0,�2,�1,�1,�0,�0,�0],�[0,�0,�0,�1,�2,�0,�0,�0,�1,�1,�2]������3��g����ff���"�[2,�UU2,�2,�2,�1,�1,�1,�1,�1,�0,�0,�0,�0,�0,�0,�0,�0,�0,�2,�2,�2,�2,�1,�1,�1,�1,�1,�0,�0,�0,�0,�0,�0,�0,�0,�0,�0],�������i����ff���"�[�UU0,�0,�0,�0,�0,�0,�0,�0,�0,�1,�1,�1,�1,�1,�3,�2,�2,�2,�0,�0,�0,�0,�0,�0,�0,�0,�0,�1,�1,�1,�1,�1,�1,�3,�2,�2,�2�]������0ԥ��ulcEigenforms�UUon�����0��|s�(7)�congruen���t�mo�Gd�7�to�t�wists�of���"�����������Ec0�n��Ο���z�ff����kcͲord���yܳ���7��~Y&�(���^��(1)���U��8�g����i��TL�),��
UVord����<���7�� J��(���^��(��1)��5V���g����i���)���?c/��ffV9��fd���1��g����ff���"�[1,�UU1,�0,�0,�1,�1,�0,�0,�0],�[1,�1,�0,�0,�1,�1,�0,�0,�0]������2��g����ff���"�[1,�UU2,�1,�1,�1,�1,�1,�0,�0,�0,�0,�0,�0,�1,�2,�1,�1,�1,�1,�1,�0,�0,�0,�0,�0,�0,�0],�������i����ff���"�[1,�UU1,�2,�1,�1,�1,�1,�0,�0,�0,�0,�0,�0,�1,�1,�2,�1,�1,�1,�1,�0,�0,�0,�0,�0,�0,�0]������*ԥ��plbEigenforms�UUon�����0��|s�(11)�congruen���t�mo�Gd�11�to�t�wists�of��������������t
�n��Ο���z�ff�����}�ord�����c���11���I�(���^��(1)���U��8�g����i��TL�),��
UVord����<���11��$G"�(���^��(��1)��5V���g����i���)���n ߟ�ff��B�fd���1��g���� ff���"�[�UU0,�0,�0,�0,�0,�0,�0,�0,�0,�0,�0,�0,�0,�1,�0,�0,�0,�0�],�������i���� ff���"�[�UU0,�0,�1,�0,�0,�0,�0,�0,�0,�0,�1,�0,�0,�0,�0,�0,�0,�0�]������&Q���>�2.3.1��a#�Twisting��Tthe�w��9eigh�t-�2��T�newform�on��X����0��|s�(11)��uT��>Let���f�'�b�G e�the�form�corresp�onding�to�the�elliptic�curv���e�of�conductor�11.��jFix�a����>prime��=�p�,���and�let��"��b�G e�the�trivial�c���haracter�mo�dulo��p�.��~W��*�e�list�the�congruences����>b�G et���w�een�UU�f�����^��"����and�the��Z����2��|s�-rational�newforms�in��S���:�2+�'�(�p����n���l�)�� )B�(����0���(�p�8���11)).����MF��*�or�(�example,�]�the�2nd�line�of�the�table�b�G elo���w�sa�ys�that�when��n�'\�=�2�(�there����>are� three��Z����2��|s�-rational�newforms��g����1���;���g����2���;�g����3���;� zand� that��f�����^��(1)��� ����g����1���3�(�mo�G d���3),��f�����^��(1)���6��g����2���� ����3����!�y�����?������>�(�mo�G d���3),�$Dand���f�����^��(1)�������g����3���3�(�mo�Gd��3).�bmThe�3rd�line�sa���ys�there�are�v�e��Z����2��|s�-rational����>newforms�UUin��S���:�2+�'�(2����3��� �)���(����0��|s�(22)),�and�the�rst,��g�[ٲ,�satises��f�����^��(1)�������g�͙�(�mo�Gd���3���^��2���).��N8��]Eigenforms�UUon�����0��|s�(22)�congruen���t�mo�Gd�2�to�the�2-depriv��q�ation�of��f��.��������������n��Ο���z�ff����ݞL�ord����2���2��𓥲(�f�����^��(1)����8�g����i��TL�)��������ffqĤ�fd���1��g����ff���"�none������2��g����ff���"�1,�UU0,�1������3��g����ff���"�2,�UU1,�0,�1,�2������4��g����ff���"�1,�UU1,�3,�3,�0������5��g����ff���"�4,�UU1,�0,�4,�0,�1,�0����������4����1?�y�����?������u�F�Eigenforms�UUon�����0��|s�(33)�congruen���t�mo�Gd�3�to�t�wists�of��f��"�����������Dn��Ο���z�ff����j��ord���xy����3��|���(�f�����^��(1)����8�g����i��TL�)�g������z�ff���>��ff�V�fd���1��g����ff���"�0,�UU1,�1,�0��"1����ff���٫\1,�UU0,�0,�0������2��g����ff���"�1,�UU0,�0,�1,�0,�0�t�K����ff���٫\0,�UU2,�1,�0,�1,�0������3��g����ff���"�0,�UU0,�0,�0,�1,�1,�1,�0,�1,�1,�1,�0,�0,�0,�0,�0��͟���ff���٫\3,�UU1,�1,�0,�0,�0,�0,�0,�0,�0,�0,�0,�2,�1,�1,�1������,����u�FEigenforms�UUon�����0��|s�(55)�congruen���t�mo�Gd�5�to�t�wists�of��f�������������Dn��Ο���z�ff����j��ord���xy����5��|���(�f�����^��(��1)��H��8�g����i��TL�)�as�����z�ff����!�]ord���0\$C���5��4���(�f�����^��(��1)��H��8�g����i��TL�)���>��ff�V�fd���1��g����ff���"�1,�UU0��[����ff���٫\0,�UU0������2��g����ff���"�1,�UU0,�0,�1,�0,�0,�0,�0,�0,�0,�0,�1,�0,�0,�0,�1��͟���ff���٫\0,�UU2,�1,�0,�0,�0,�0,�0,�0,�0,�0,�0,�1,�1,�1,�0������&n����>�2.3.2��a#�Twisting��T�X����0��|s�(14)��Ì��R2�Eigenforms�UUon�����0��|s�(3�8���14)�UUcongruen���t�mo�Gd�3�to�t�wists�of��h���2��S����2��|s�(����0���(14))������������њ�n��Ο���z�ff������7�ord����K���3���ǐ�(�h���^��(1)���U��8�g����i��TL�)��͟���z�ff����Pord����6���3��U��(�h���^��(��1)��5V��8�g����i��TL�)����љ��ff�\ϟfd���1��g����ff���"�0,�UU1�?>q����ff���w��0,�UU0������2��g����ff���"�0,�UU1,�0,�0,�0,�0�̥����ff���w��0,�UU0,�2,�0,�2,�0������*D��>�References�������>�[Can]���b��J.�UUCannon,��The����ma��gma��c��}'omputational�algebr�a�system,����b���9��<x

cmtt10�http://www.maths.usyd.edu.au:8000/u/magma/�.������>[CM98]���b��R.��Coleman�and�B.�Mazur,�	��The��eigencurve�,�Galois��represen���tations�in����b��arithmetic�hDalgebraic�geometry�(Durham,���1996),�Cam���bridge�hDUniv.�Press,����b��Cam���bridge,�UU1998,�pp.�1{113.������>[Shi94]���b��G.���Shim���ura,��e�Intr��}'o�duction�to�the�arithmetic�the��}'ory�of�automorphic�func-����b��tions�,���Princeton���Univ���ersit�y�Press,���Princeton,�NJ,�1994,�Reprin���t�of�the����b��1971�UUoriginal,�Kan�Memorial�Lectures,�1.������5����4����;�y��9��<x

cmtt10�6�-�

cmcsc10�5��N�cmbx12�,�"V

cmbx10�+�':

cmti10�*��N�ffcmbx12�X�Qffcmr12���g�G�cmmi12�X�Qcmr12�D��tG�G�cmr17�
!",�

cmsy10�O!�cmsy7�
�b>

cmmi10�	0e�rcmmi7�O
�\cmmi5�K�y

cmr10�ٓ�Rcmr7���Zcmr5���u

cmex10�>:�������