The goal of this paper is to reformulate the conjectural Ihara lemma for
class="mathmlsrc">class="formulatext stixSupport mathImg" data-mathURL="/science?_ob=MathURL&_method=retrieve&_eid=1-s2.0-S0022314X16000433&_mathId=si1.gif&_user=111111111&_pii=S0022314X16000433&_rdoc=1&_issn=0022314X&md5=392e90231cd5286a84121fffc96b1854" title="Click to view the MathML source">U(n)class="mathContainer hidden">class="mathCode"> in terms of the local Langlands correspondence in families
class="mathmlsrc">title="View the MathML source" class="mathImg" data-mathURL="/science?_ob=MathURL&_method=retrieve&_eid=1-s2.0-S0022314X16000433&_mathId=si2.gif&_user=111111111&_pii=S0022314X16000433&_rdoc=1&_issn=0022314X&md5=4db273d3c250d6e134037768dd9f2dfd">
class="imgLazyJSB inlineImage" height="16" width="37" alt="View the MathML source" title="View the MathML source" src="/sd/grey_pxl.gif" data-inlimgeid="1-s2.0-S0022314X16000433-si2.gif">class="mathContainer hidden">class="mathCode">, as developed by Emerton and Helm. The reformulation roughly takes the following form. Suppose we are given an irr
educible mod
ℓ Galois representation
class="mathmlsrc">title="View the MathML source" class="mathImg" data-mathURL="/science?_ob=MathURL&_method=retrieve&_eid=1-s2.0-S0022314X16000433&_mathId=si3.gif&_user=111111111&_pii=S0022314X16000433&_rdoc=1&_issn=0022314X&md5=77fb4cdb8c69cf27157a9375eddc1e75">
class="imgLazyJSB inlineImage" height="13" width="49" alt="View the MathML source" style="margin-top: -5px; vertical-align: middle" title="View the MathML source" src="/sd/grey_pxl.gif" data-inlimgeid="1-s2.0-S0022314X16000433-si3.gif">class="mathContainer hidden">class="mathCode">, which is modular of full level (and small weight), and a finite set of places Σ, none of which divides
ℓ . Then
class="mathmlsrc">title="View the MathML source" class="mathImg" data-mathURL="/science?_ob=MathURL&_method=retrieve&_eid=1-s2.0-S0022314X16000433&_mathId=si4.gif&_user=111111111&_pii=S0022314X16000433&_rdoc=1&_issn=0022314X&md5=640b9544b714ea12822b7857a0b0c9e9">
class="imgLazyJSB inlineImage" height="16" width="51" alt="View the MathML source" title="View the MathML source" src="/sd/grey_pxl.gif" data-inlimgeid="1-s2.0-S0022314X16000433-si4.gif">class="mathContainer hidden">class="mathCode"> exists, and has a global realization as a localized module of
ℓ -integral algebraic modular forms, where
class="mathmlsrc">class="formulatext stixSupport mathImg" data-mathURL="/science?_ob=MathURL&_method=retrieve&_eid=1-s2.0-S0022314X16000433&_mathId=si5.gif&_user=111111111&_pii=S0022314X16000433&_rdoc=1&_issn=0022314X&md5=c22b2d4357c3e38fe978e7a5b6b81504" title="Click to view the MathML source">rmclass="mathContainer hidden">class="mathCode"> is the universal deformation of
class="mathmlsrc">title="View the MathML source" class="mathImg" data-mathURL="/science?_ob=MathURL&_method=retrieve&_eid=1-s2.0-S0022314X16000433&_mathId=si6.gif&_user=111111111&_pii=S0022314X16000433&_rdoc=1&_issn=0022314X&md5=20de4f55003f3b101e20ffdd961d4557">
class="imgLazyJSB inlineImage" height="10" width="10" alt="View the MathML source" style="margin-top: -5px; vertical-align: middle" title="View the MathML source" src="/sd/grey_pxl.gif" data-inlimgeid="1-s2.0-S0022314X16000433-si6.gif">class="mathContainer hidden">class="mathCode"> of type Σ. This is unconditional for
class="mathmlsrc">class="formulatext stixSupport mathImg" data-mathURL="/science?_ob=MathURL&_method=retrieve&_eid=1-s2.0-S0022314X16000433&_mathId=si7.gif&_user=111111111&_pii=S0022314X16000433&_rdoc=1&_issn=0022314X&md5=94606eeb6cb21582f2b69bebb041e14b" title="Click to view the MathML source">n=2class="mathContainer hidden">class="mathCode">, where Ihara's lemma is an almost trivial consequence of the strong approximation theorem. We emphasize that throughout we will work with
banal primes
class="mathmlsrc">class="formulatext stixSupport mathImg" data-mathURL="/science?_ob=MathURL&_method=retrieve&_eid=1-s2.0-S0022314X16000433&_mathId=si8.gif&_user=111111111&_pii=S0022314X16000433&_rdoc=1&_issn=0022314X&md5=63f405044537b871c554682dd55d2359" title="Click to view the MathML source">ℓ>nclass="mathContainer hidden">class="mathCode">. That is, those for which
class="mathmlsrc">class="formulatext stixSupport mathImg" data-mathURL="/science?_ob=MathURL&_method=retrieve&_eid=1-s2.0-S0022314X16000433&_mathId=si9.gif&_user=111111111&_pii=S0022314X16000433&_rdoc=1&_issn=0022314X&md5=b1a6e0d97f50814c1348a6a2108fa435" title="Click to view the MathML source">#k(v)i≢1class="mathContainer hidden">class="mathCode"> (mod
ℓ ), where
class="mathmlsrc">class="formulatext stixSupport mathImg" data-mathURL="/science?_ob=MathURL&_method=retrieve&_eid=1-s2.0-S0022314X16000433&_mathId=si10.gif&_user=111111111&_pii=S0022314X16000433&_rdoc=1&_issn=0022314X&md5=2570ba4573ebde15cd9b464c73999962" title="Click to view the MathML source">1≤i≤nclass="mathContainer hidden">class="mathCode">, for
class="mathmlsrc">class="formulatext stixSupport mathImg" data-mathURL="/science?_ob=MathURL&_method=retrieve&_eid=1-s2.0-S0022314X16000433&_mathId=si11.gif&_user=111111111&_pii=S0022314X16000433&_rdoc=1&_issn=0022314X&md5=3b25ac4cc3996761bd83282bd4c982e6" title="Click to view the MathML source">v∈Σclass="mathContainer hidden">class="mathCode">. A weakening of this assumption (quasi-banal) is ubiquitous in
[CHT] (particularly in their Chapter 5 on the non-minimal case and Ihara), where the mod
ℓ representation theory of
class="mathmlsrc">title="View the MathML source" class="mathImg" data-mathURL="/science?_ob=MathURL&_method=retrieve&_eid=1-s2.0-S0022314X16000433&_mathId=si12.gif&_user=111111111&_pii=S0022314X16000433&_rdoc=1&_issn=0022314X&md5=a09edde11e99d9e7c38df1a7be95e6af">
class="imgLazyJSB inlineImage" height="16" width="62" alt="View the MathML source" title="View the MathML source" src="/sd/grey_pxl.gif" data-inlimgeid="1-s2.0-S0022314X16000433-si12.gif">class="mathContainer hidden">class="mathCode"> plays a pivotal role. We will need actual banality, to ensure the functors we define are exact. Banality also removes many of the subtleties in the Emerton–Helm correspondence. For example, in yet unpublished work
[Hel13], Helm shows that local Langlands in families do exist in the banal case, although we do not use his result. We should stress that our main result has been common knowledge among experts for some time, but its proof has never appeared in print.