$ \newcommand{\undefined}{} \newcommand{\hfill}{} \newcommand{\qedhere}{\square} \newcommand{\qed}{\square} \newcommand{\ensuremath}[1]{#1} \newcommand{\bit}{\{0,1\}} \newcommand{\Bit}{\{-1,1\}} \newcommand{\Stab}{\mathbf{Stab}} \newcommand{\NS}{\mathbf{NS}} \newcommand{\ba}{\mathbf{a}} \newcommand{\bc}{\mathbf{c}} \newcommand{\bd}{\mathbf{d}} \newcommand{\be}{\mathbf{e}} \newcommand{\bh}{\mathbf{h}} \newcommand{\br}{\mathbf{r}} \newcommand{\bs}{\mathbf{s}} \newcommand{\bx}{\mathbf{x}} \newcommand{\by}{\mathbf{y}} \newcommand{\bz}{\mathbf{z}} \newcommand{\Var}{\mathbf{Var}} \newcommand{\dist}{\text{dist}} \newcommand{\norm}[1]{\\|#1\\|} \newcommand{\etal} \newcommand{\ie} \newcommand{\eg} \newcommand{\cf} \newcommand{\rank}{\text{rank}} \newcommand{\tr}{\text{tr}} \newcommand{\mor}{\text{Mor}} \newcommand{\hom}{\text{Hom}} \newcommand{\id}{\text{id}} \newcommand{\obj}{\text{obj}} \newcommand{\pr}{\text{pr}} \newcommand{\ker}{\text{ker}} \newcommand{\coker}{\text{coker}} \newcommand{\im}{\text{im}} \newcommand{\vol}{\text{vol}} \newcommand{\disc}{\text{disc}} \newcommand{\bbA}{\mathbb A} \newcommand{\bbB}{\mathbb B} \newcommand{\bbC}{\mathbb C} \newcommand{\bbD}{\mathbb D} \newcommand{\bbE}{\mathbb E} \newcommand{\bbF}{\mathbb F} \newcommand{\bbG}{\mathbb G} \newcommand{\bbH}{\mathbb H} \newcommand{\bbI}{\mathbb I} \newcommand{\bbJ}{\mathbb J} \newcommand{\bbK}{\mathbb K} \newcommand{\bbL}{\mathbb L} \newcommand{\bbM}{\mathbb M} \newcommand{\bbN}{\mathbb N} \newcommand{\bbO}{\mathbb O} \newcommand{\bbP}{\mathbb P} \newcommand{\bbQ}{\mathbb Q} \newcommand{\bbR}{\mathbb R} \newcommand{\bbS}{\mathbb S} \newcommand{\bbT}{\mathbb T} \newcommand{\bbU}{\mathbb U} \newcommand{\bbV}{\mathbb V} \newcommand{\bbW}{\mathbb W} \newcommand{\bbX}{\mathbb X} \newcommand{\bbY}{\mathbb Y} \newcommand{\bbZ}{\mathbb Z} \newcommand{\sA}{\mathscr A} \newcommand{\sB}{\mathscr B} \newcommand{\sC}{\mathscr C} \newcommand{\sD}{\mathscr D} \newcommand{\sE}{\mathscr E} \newcommand{\sF}{\mathscr F} \newcommand{\sG}{\mathscr G} \newcommand{\sH}{\mathscr H} \newcommand{\sI}{\mathscr I} \newcommand{\sJ}{\mathscr J} \newcommand{\sK}{\mathscr K} \newcommand{\sL}{\mathscr L} \newcommand{\sM}{\mathscr M} \newcommand{\sN}{\mathscr N} \newcommand{\sO}{\mathscr O} \newcommand{\sP}{\mathscr P} \newcommand{\sQ}{\mathscr Q} \newcommand{\sR}{\mathscr R} \newcommand{\sS}{\mathscr S} \newcommand{\sT}{\mathscr T} \newcommand{\sU}{\mathscr U} \newcommand{\sV}{\mathscr V} \newcommand{\sW}{\mathscr W} \newcommand{\sX}{\mathscr X} \newcommand{\sY}{\mathscr Y} \newcommand{\sZ}{\mathscr Z} \newcommand{\sfA}{\mathsf A} \newcommand{\sfB}{\mathsf B} \newcommand{\sfC}{\mathsf C} \newcommand{\sfD}{\mathsf D} \newcommand{\sfE}{\mathsf E} \newcommand{\sfF}{\mathsf F} \newcommand{\sfG}{\mathsf G} \newcommand{\sfH}{\mathsf H} \newcommand{\sfI}{\mathsf I} \newcommand{\sfJ}{\mathsf J} \newcommand{\sfK}{\mathsf K} \newcommand{\sfL}{\mathsf L} \newcommand{\sfM}{\mathsf M} \newcommand{\sfN}{\mathsf N} \newcommand{\sfO}{\mathsf O} \newcommand{\sfP}{\mathsf P} \newcommand{\sfQ}{\mathsf Q} \newcommand{\sfR}{\mathsf R} \newcommand{\sfS}{\mathsf S} \newcommand{\sfT}{\mathsf T} \newcommand{\sfU}{\mathsf U} \newcommand{\sfV}{\mathsf V} \newcommand{\sfW}{\mathsf W} \newcommand{\sfX}{\mathsf X} \newcommand{\sfY}{\mathsf Y} \newcommand{\sfZ}{\mathsf Z} \newcommand{\cA}{\mathcal A} \newcommand{\cB}{\mathcal B} \newcommand{\cC}{\mathcal C} \newcommand{\cD}{\mathcal D} \newcommand{\cE}{\mathcal E} \newcommand{\cF}{\mathcal F} \newcommand{\cG}{\mathcal G} \newcommand{\cH}{\mathcal H} \newcommand{\cI}{\mathcal I} \newcommand{\cJ}{\mathcal J} \newcommand{\cK}{\mathcal K} \newcommand{\cL}{\mathcal L} \newcommand{\cM}{\mathcal M} \newcommand{\cN}{\mathcal N} \newcommand{\cO}{\mathcal O} \newcommand{\cP}{\mathcal P} \newcommand{\cQ}{\mathcal Q} \newcommand{\cR}{\mathcal R} \newcommand{\cS}{\mathcal S} \newcommand{\cT}{\mathcal T} \newcommand{\cU}{\mathcal U} \newcommand{\cV}{\mathcal V} \newcommand{\cW}{\mathcal W} \newcommand{\cX}{\mathcal X} \newcommand{\cY}{\mathcal Y} \newcommand{\cZ}{\mathcal Z} \newcommand{\bfA}{\mathbf A} \newcommand{\bfB}{\mathbf B} \newcommand{\bfC}{\mathbf C} \newcommand{\bfD}{\mathbf D} \newcommand{\bfE}{\mathbf E} \newcommand{\bfF}{\mathbf F} \newcommand{\bfG}{\mathbf G} \newcommand{\bfH}{\mathbf H} \newcommand{\bfI}{\mathbf I} \newcommand{\bfJ}{\mathbf J} \newcommand{\bfK}{\mathbf K} \newcommand{\bfL}{\mathbf L} \newcommand{\bfM}{\mathbf M} \newcommand{\bfN}{\mathbf N} \newcommand{\bfO}{\mathbf O} \newcommand{\bfP}{\mathbf P} \newcommand{\bfQ}{\mathbf Q} \newcommand{\bfR}{\mathbf R} \newcommand{\bfS}{\mathbf S} \newcommand{\bfT}{\mathbf T} \newcommand{\bfU}{\mathbf U} \newcommand{\bfV}{\mathbf V} \newcommand{\bfW}{\mathbf W} \newcommand{\bfX}{\mathbf X} \newcommand{\bfY}{\mathbf Y} \newcommand{\bfZ}{\mathbf Z} \newcommand{\rmA}{\mathrm A} \newcommand{\rmB}{\mathrm B} \newcommand{\rmC}{\mathrm C} \newcommand{\rmD}{\mathrm D} \newcommand{\rmE}{\mathrm E} \newcommand{\rmF}{\mathrm F} \newcommand{\rmG}{\mathrm G} \newcommand{\rmH}{\mathrm H} \newcommand{\rmI}{\mathrm I} \newcommand{\rmJ}{\mathrm J} \newcommand{\rmK}{\mathrm K} \newcommand{\rmL}{\mathrm L} \newcommand{\rmM}{\mathrm M} \newcommand{\rmN}{\mathrm N} \newcommand{\rmO}{\mathrm O} \newcommand{\rmP}{\mathrm P} \newcommand{\rmQ}{\mathrm Q} \newcommand{\rmR}{\mathrm R} \newcommand{\rmS}{\mathrm S} \newcommand{\rmT}{\mathrm T} \newcommand{\rmU}{\mathrm U} \newcommand{\rmV}{\mathrm V} \newcommand{\rmW}{\mathrm W} \newcommand{\rmX}{\mathrm X} \newcommand{\rmY}{\mathrm Y} \newcommand{\rmZ}{\mathrm Z} \newcommand{\bb}{\mathbf{b}} \newcommand{\bv}{\mathbf{v}} \newcommand{\bw}{\mathbf{w}} \newcommand{\bx}{\mathbf{x}} \newcommand{\by}{\mathbf{y}} \newcommand{\bz}{\mathbf{z}} \newcommand{\paren}[1]{( #1 )} \newcommand{\Paren}[1]{\left( #1 \right)} \newcommand{\bigparen}[1]{\bigl( #1 \bigr)} \newcommand{\Bigparen}[1]{\Bigl( #1 \Bigr)} \newcommand{\biggparen}[1]{\biggl( #1 \biggr)} \newcommand{\Biggparen}[1]{\Biggl( #1 \Biggr)} \newcommand{\abs}[1]{\lvert #1 \rvert} \newcommand{\Abs}[1]{\left\lvert #1 \right\rvert} \newcommand{\bigabs}[1]{\bigl\lvert #1 \bigr\rvert} \newcommand{\Bigabs}[1]{\Bigl\lvert #1 \Bigr\rvert} \newcommand{\biggabs}[1]{\biggl\lvert #1 \biggr\rvert} \newcommand{\Biggabs}[1]{\Biggl\lvert #1 \Biggr\rvert} \newcommand{\card}[1]{\left| #1 \right|} \newcommand{\Card}[1]{\left\lvert #1 \right\rvert} \newcommand{\bigcard}[1]{\bigl\lvert #1 \bigr\rvert} \newcommand{\Bigcard}[1]{\Bigl\lvert #1 \Bigr\rvert} \newcommand{\biggcard}[1]{\biggl\lvert #1 \biggr\rvert} \newcommand{\Biggcard}[1]{\Biggl\lvert #1 \Biggr\rvert} \newcommand{\norm}[1]{\lVert #1 \rVert} \newcommand{\Norm}[1]{\left\lVert #1 \right\rVert} \newcommand{\bignorm}[1]{\bigl\lVert #1 \bigr\rVert} \newcommand{\Bignorm}[1]{\Bigl\lVert #1 \Bigr\rVert} \newcommand{\biggnorm}[1]{\biggl\lVert #1 \biggr\rVert} \newcommand{\Biggnorm}[1]{\Biggl\lVert #1 \Biggr\rVert} \newcommand{\iprod}[1]{\langle #1 \rangle} \newcommand{\Iprod}[1]{\left\langle #1 \right\rangle} \newcommand{\bigiprod}[1]{\bigl\langle #1 \bigr\rangle} \newcommand{\Bigiprod}[1]{\Bigl\langle #1 \Bigr\rangle} \newcommand{\biggiprod}[1]{\biggl\langle #1 \biggr\rangle} \newcommand{\Biggiprod}[1]{\Biggl\langle #1 \Biggr\rangle} \newcommand{\set}[1]{\lbrace #1 \rbrace} \newcommand{\Set}[1]{\left\lbrace #1 \right\rbrace} \newcommand{\bigset}[1]{\bigl\lbrace #1 \bigr\rbrace} \newcommand{\Bigset}[1]{\Bigl\lbrace #1 \Bigr\rbrace} \newcommand{\biggset}[1]{\biggl\lbrace #1 \biggr\rbrace} \newcommand{\Biggset}[1]{\Biggl\lbrace #1 \Biggr\rbrace} \newcommand{\bracket}[1]{\lbrack #1 \rbrack} \newcommand{\Bracket}[1]{\left\lbrack #1 \right\rbrack} \newcommand{\bigbracket}[1]{\bigl\lbrack #1 \bigr\rbrack} \newcommand{\Bigbracket}[1]{\Bigl\lbrack #1 \Bigr\rbrack} \newcommand{\biggbracket}[1]{\biggl\lbrack #1 \biggr\rbrack} \newcommand{\Biggbracket}[1]{\Biggl\lbrack #1 \Biggr\rbrack} \newcommand{\ucorner}[1]{\ulcorner #1 \urcorner} \newcommand{\Ucorner}[1]{\left\ulcorner #1 \right\urcorner} \newcommand{\bigucorner}[1]{\bigl\ulcorner #1 \bigr\urcorner} \newcommand{\Bigucorner}[1]{\Bigl\ulcorner #1 \Bigr\urcorner} \newcommand{\biggucorner}[1]{\biggl\ulcorner #1 \biggr\urcorner} \newcommand{\Biggucorner}[1]{\Biggl\ulcorner #1 \Biggr\urcorner} \newcommand{\ceil}[1]{\lceil #1 \rceil} \newcommand{\Ceil}[1]{\left\lceil #1 \right\rceil} \newcommand{\bigceil}[1]{\bigl\lceil #1 \bigr\rceil} \newcommand{\Bigceil}[1]{\Bigl\lceil #1 \Bigr\rceil} \newcommand{\biggceil}[1]{\biggl\lceil #1 \biggr\rceil} \newcommand{\Biggceil}[1]{\Biggl\lceil #1 \Biggr\rceil} \newcommand{\floor}[1]{\lfloor #1 \rfloor} \newcommand{\Floor}[1]{\left\lfloor #1 \right\rfloor} \newcommand{\bigfloor}[1]{\bigl\lfloor #1 \bigr\rfloor} \newcommand{\Bigfloor}[1]{\Bigl\lfloor #1 \Bigr\rfloor} \newcommand{\biggfloor}[1]{\biggl\lfloor #1 \biggr\rfloor} \newcommand{\Biggfloor}[1]{\Biggl\lfloor #1 \Biggr\rfloor} \newcommand{\lcorner}[1]{\llcorner #1 \lrcorner} \newcommand{\Lcorner}[1]{\left\llcorner #1 \right\lrcorner} \newcommand{\biglcorner}[1]{\bigl\llcorner #1 \bigr\lrcorner} \newcommand{\Biglcorner}[1]{\Bigl\llcorner #1 \Bigr\lrcorner} \newcommand{\bigglcorner}[1]{\biggl\llcorner #1 \biggr\lrcorner} \newcommand{\Bigglcorner}[1]{\Biggl\llcorner #1 \Biggr\lrcorner} \newcommand{\ket}[1]{| #1 \rangle} \newcommand{\bra}[1]{\langle #1 |} \newcommand{\braket}[2]{\langle #1 | #2 \rangle} \newcommand{\ketbra}[1]{| #1 \rangle\langle #1 |} \newcommand{\e}{\varepsilon} \newcommand{\eps}{\varepsilon} \newcommand{\from}{\colon} \newcommand{\super}[2]{#1^{(#2)}} \newcommand{\varsuper}[2]{#1^{\scriptscriptstyle (#2)}} \newcommand{\tensor}{\otimes} \newcommand{\eset}{\emptyset} \newcommand{\sse}{\subseteq} \newcommand{\sst}{\substack} \newcommand{\ot}{\otimes} \newcommand{\Esst}[1]{\bbE_{\substack{#1}}} \newcommand{\vbig}{\vphantom{\bigoplus}} \newcommand{\seteq}{\mathrel{\mathop:}=} \newcommand{\defeq}{\stackrel{\mathrm{def}}=} \newcommand{\Mid}{\mathrel{}\middle|\mathrel{}} \newcommand{\Ind}{\mathbf 1} \newcommand{\bits}{\{0,1\}} \newcommand{\sbits}{\{\pm 1\}} \newcommand{\R}{\mathbb R} \newcommand{\Rnn}{\R_{\ge 0}} \newcommand{\N}{\mathbb N} \newcommand{\Z}{\mathbb Z} \newcommand{\Q}{\mathbb Q} \newcommand{\C}{\mathbb C} \newcommand{\A}{\mathbb A} \newcommand{\Real}{\mathbb R} \newcommand{\mper}{\,.} \newcommand{\mcom}{\,,} \DeclareMathOperator{\Id}{Id} \DeclareMathOperator{\cone}{cone} \DeclareMathOperator{\vol}{vol} \DeclareMathOperator{\val}{val} \DeclareMathOperator{\opt}{opt} \DeclareMathOperator{\Opt}{Opt} \DeclareMathOperator{\Val}{Val} \DeclareMathOperator{\LP}{LP} \DeclareMathOperator{\SDP}{SDP} \DeclareMathOperator{\Tr}{Tr} \DeclareMathOperator{\Inf}{Inf} \DeclareMathOperator{\size}{size} \DeclareMathOperator{\poly}{poly} \DeclareMathOperator{\polylog}{polylog} \DeclareMathOperator{\min}{min} \DeclareMathOperator{\max}{max} \DeclareMathOperator{\argmax}{arg\,max} \DeclareMathOperator{\argmin}{arg\,min} \DeclareMathOperator{\qpoly}{qpoly} \DeclareMathOperator{\qqpoly}{qqpoly} \DeclareMathOperator{\conv}{conv} \DeclareMathOperator{\Conv}{Conv} \DeclareMathOperator{\supp}{supp} \DeclareMathOperator{\sign}{sign} \DeclareMathOperator{\perm}{perm} \DeclareMathOperator{\mspan}{span} \DeclareMathOperator{\mrank}{rank} \DeclareMathOperator{\E}{\mathbb E} \DeclareMathOperator{\pE}{\tilde{\mathbb E}} \DeclareMathOperator{\Pr}{\mathbb P} \DeclareMathOperator{\Span}{Span} \DeclareMathOperator{\Cone}{Cone} \DeclareMathOperator{\junta}{junta} \DeclareMathOperator{\NSS}{NSS} \DeclareMathOperator{\SA}{SA} \DeclareMathOperator{\SOS}{SOS} \DeclareMathOperator{\Stab}{\mathbf Stab} \DeclareMathOperator{\Det}{\textbf{Det}} \DeclareMathOperator{\Perm}{\textbf{Perm}} \DeclareMathOperator{\Sym}{\textbf{Sym}} \DeclareMathOperator{\Pow}{\textbf{Pow}} \DeclareMathOperator{\Gal}{\textbf{Gal}} \DeclareMathOperator{\Aut}{\textbf{Aut}} \newcommand{\iprod}[1]{\langle #1 \rangle} \newcommand{\cE}{\mathcal{E}} \newcommand{\E}{\mathbb{E}} \newcommand{\pE}{\tilde{\mathbb{E}}} \newcommand{\N}{\mathbb{N}} \renewcommand{\P}{\mathcal{P}} \notag $
$ \newcommand{\sleq}{\ensuremath{\preceq}} \newcommand{\sgeq}{\ensuremath{\succeq}} \newcommand{\diag}{\ensuremath{\mathrm{diag}}} \newcommand{\support}{\ensuremath{\mathrm{support}}} \newcommand{\zo}{\ensuremath{\{0,1\}}} \newcommand{\pmo}{\ensuremath{\{\pm 1\}}} \newcommand{\uppersos}{\ensuremath{\overline{\mathrm{sos}}}} \newcommand{\lambdamax}{\ensuremath{\lambda_{\mathrm{max}}}} \newcommand{\rank}{\ensuremath{\mathrm{rank}}} \newcommand{\Mslow}{\ensuremath{M_{\mathrm{slow}}}} \newcommand{\Mfast}{\ensuremath{M_{\mathrm{fast}}}} \newcommand{\Mdiag}{\ensuremath{M_{\mathrm{diag}}}} \newcommand{\Mcross}{\ensuremath{M_{\mathrm{cross}}}} \newcommand{\eqdef}{\ensuremath{ =^{def}}} \newcommand{\threshold}{\ensuremath{\mathrm{threshold}}} \newcommand{\vbls}{\ensuremath{\mathrm{vbls}}} \newcommand{\cons}{\ensuremath{\mathrm{cons}}} \newcommand{\edges}{\ensuremath{\mathrm{edges}}} \newcommand{\cl}{\ensuremath{\mathrm{cl}}} \newcommand{\xor}{\ensuremath{\oplus}} \newcommand{\1}{\ensuremath{\mathrm{1}}} \notag $
$ \newcommand{\transpose}[1]{\ensuremath{#1{}^{\mkern-2mu\intercal}}} \newcommand{\dyad}[1]{\ensuremath{#1#1{}^{\mkern-2mu\intercal}}} \newcommand{\nchoose}[1]{\ensuremath} \newcommand{\generated}[1]{\ensuremath{\langle #1 \rangle}} \notag $
$ \newcommand{\eqdef}{\mathbin{\stackrel{\rm def}{=}}} \newcommand{\R} % real numbers \newcommand{\N}} % natural numbers \newcommand{\Z} % integers \newcommand{\F} % a field \newcommand{\Q} % the rationals \newcommand{\C}{\mathbb{C}} % the complexes \newcommand{\poly}} \newcommand{\polylog}} \newcommand{\loglog}}} \newcommand{\zo}{\{0,1\}} \newcommand{\suchthat} \newcommand{\pr}[1]{\Pr\left[#1\right]} \newcommand{\deffont}{\em} \newcommand{\getsr}{\mathbin{\stackrel{\mbox{\tiny R}}{\gets}}} \newcommand{\Exp}{\mathop{\mathrm E}\displaylimits} % expectation \newcommand{\Var}{\mathop{\mathrm Var}\displaylimits} % variance \newcommand{\xor}{\oplus} \newcommand{\GF}{\mathrm{GF}} \newcommand{\eps}{\varepsilon} \notag $
$ \newcommand{\class}[1]{\mathbf{#1}} \newcommand{\coclass}[1]{\mathbf{co\mbox{-}#1}} % and their complements \newcommand{\BPP}{\class{BPP}} \newcommand{\NP}{\class{NP}} \newcommand{\RP}{\class{RP}} \newcommand{\coRP}{\coclass{RP}} \newcommand{\ZPP}{\class{ZPP}} \newcommand{\BQP}{\class{BQP}} \newcommand{\FP}{\class{FP}} \newcommand{\QP}{\class{QuasiP}} \newcommand{\VF}{\class{VF}} \newcommand{\VBP}{\class{VBP}} \newcommand{\VP}{\class{VP}} \newcommand{\VNP}{\class{VNP}} \newcommand{\RNC}{\class{RNC}} \newcommand{\RL}{\class{RL}} \newcommand{\BPL}{\class{BPL}} \newcommand{\coRL}{\coclass{RL}} \newcommand{\IP}{\class{IP}} \newcommand{\AM}{\class{AM}} \newcommand{\MA}{\class{MA}} \newcommand{\QMA}{\class{QMA}} \newcommand{\SBP}{\class{SBP}} \newcommand{\coAM}{\class{coAM}} \newcommand{\coMA}{\class{coMA}} \renewcommand{\P}{\class{P}} \newcommand\prBPP{\class{prBPP}} \newcommand\prRP{\class{prRP}} \newcommand\prP{\class{prP}} \newcommand{\Ppoly}{\class{P/poly}} \newcommand{\NPpoly}{\class{NP/poly}} \newcommand{\coNPpoly}{\class{coNP/poly}} \newcommand{\DTIME}{\class{DTIME}} \newcommand{\TIME}{\class{TIME}} \newcommand{\SIZE}{\class{SIZE}} \newcommand{\SPACE}{\class{SPACE}} \newcommand{\ETIME}{\class{E}} \newcommand{\BPTIME}{\class{BPTIME}} \newcommand{\RPTIME}{\class{RPTIME}} \newcommand{\ZPTIME}{\class{ZPTIME}} \newcommand{\EXP}{\class{EXP}} \newcommand{\ZPEXP}{\class{ZPEXP}} \newcommand{\RPEXP}{\class{RPEXP}} \newcommand{\BPEXP}{\class{BPEXP}} \newcommand{\SUBEXP}{\class{SUBEXP}} \newcommand{\NTIME}{\class{NTIME}} \newcommand{\NL}{\class{NL}} \renewcommand{\L}{\class{L}} \newcommand{\NQP}{\class{NQP}} \newcommand{\NEXP}{\class{NEXP}} \newcommand{\coNEXP}{\coclass{NEXP}} \newcommand{\NPSPACE}{\class{NPSPACE}} \newcommand{\PSPACE}{\class{PSPACE}} \newcommand{\NSPACE}{\class{NSPACE}} \newcommand{\coNSPACE}{\coclass{NSPACE}} \newcommand{\coL}{\coclass{L}} \newcommand{\coP}{\coclass{P}} \newcommand{\coNP}{\coclass{NP}} \newcommand{\coNL}{\coclass{NL}} \newcommand{\coNPSPACE}{\coclass{NPSPACE}} \newcommand{\APSPACE}{\class{APSPACE}} \newcommand{\LINSPACE}{\class{LINSPACE}} \newcommand{\qP}{\class{\tilde{P}}} \newcommand{\PH}{\class{PH}} \newcommand{\EXPSPACE}{\class{EXPSPACE}} \newcommand{\SigmaTIME}[1]{\class{\Sigma_{#1}TIME}} \newcommand{\PiTIME}[1]{\class{\Pi_{#1}TIME}} \newcommand{\SigmaP}[1]{\class{\Sigma_{#1}P}} \newcommand{\PiP}[1]{\class{\Pi_{#1}P}} \newcommand{\DeltaP}[1]{\class{\Delta_{#1}P}} \newcommand{\ATIME}{\class{ATIME}} \newcommand{\ASPACE}{\class{ASPACE}} \newcommand{\AP}{\class{AP}} \newcommand{\AL}{\class{AL}} \newcommand{\APSPACE}{\class{APSPACE}} \newcommand{\VNC}[1]{\class{VNC^{#1}}} \newcommand{\NC}[1]{\class{NC^{#1}}} \newcommand{\AC}[1]{\class{AC^{#1}}} \newcommand{\ACC}[1]{\class{ACC^{#1}}} \newcommand{\TC}[1]{\class{TC^{#1}}} \newcommand{\ShP}{\class{\# P}} \newcommand{\PaP}{\class{\oplus P}} \newcommand{\PCP}{\class{PCP}} \newcommand{\kMIP}[1]{\class{#1\mbox{-}MIP}} \newcommand{\MIP}{\class{MIP}} $
$ \newcommand{\textprob}[1]{\text{#1}} \newcommand{\mathprob}[1]{\textbf{#1}} \newcommand{\Satisfiability}{\textprob{Satisfiability}} \newcommand{\SAT}{\textprob{SAT}} \newcommand{\TSAT}{\textprob{3SAT}} \newcommand{\USAT}{\textprob{USAT}} \newcommand{\UNSAT}{\textprob{UNSAT}} \newcommand{\QPSAT}{\textprob{QPSAT}} \newcommand{\TQBF}{\textprob{TQBF}} \newcommand{\LinProg}{\textprob{Linear Programming}} \newcommand{\LP}{\mathprob{LP}} \newcommand{\Factor}{\textprob{Factoring}} \newcommand{\CircVal}{\textprob{Circuit Value}} \newcommand{\CVAL}{\mathprob{CVAL}} \newcommand{\CircSat}{\textprob{Circuit Satisfiability}} \newcommand{\CSAT}{\textprob{CSAT}} \newcommand{\CycleCovers}{\textprob{Cycle Covers}} \newcommand{\MonCircVal}{\textprob{Monotone Circuit Value}} \newcommand{\Reachability}{\textprob{Reachability}} \newcommand{\Unreachability}{\textprob{Unreachability}} \newcommand{\RCH}{\mathprob{RCH}} \newcommand{\BddHalt}{\textprob{Bounded Halting}} \newcommand{\BH}{\mathprob{BH}} \newcommand{\DiscreteLog}{\textprob{Discrete Log}} \newcommand{\REE}{\mathprob{REE}} \newcommand{\QBF}{\mathprob{QBF}} \newcommand{\MCSP}{\mathprob{MCSP}} \newcommand{\GGEO}{\mathprob{GGEO}} \newcommand{\CKTMIN}{\mathprob{CKT-MIN}} \newcommand{\MINCKT}{\mathprob{MIN-CKT}} \newcommand{\IdentityTest}{\textprob{Identity Testing}} \newcommand{\Majority}{\textprob{Majority}} \newcommand{\CountIndSets}{\textprob{\#Independent Sets}} \newcommand{\Parity}{\textprob{Parity}} \newcommand{\Clique}{\textprob{Clique}} \newcommand{\CountCycles}{\textprob{#Cycles}} \newcommand{\CountPerfMatchings}{\textprob{\#Perfect Matchings}} \newcommand{\CountMatchings}{\textprob{\#Matchings}} \newcommand{\CountMatch}{\mathprob{\#Matchings}} \newcommand{\ECSAT}{\mathprob{E#SAT}} \newcommand{\ShSAT}{\mathprob{#SAT}} \newcommand{\ShTSAT}{\mathprob{#3SAT}} \newcommand{\HamCycle}{\textprob{Hamiltonian Cycle}} \newcommand{\Permanent}{\textprob{Permanent}} \newcommand{\ModPermanent}{\textprob{Modular Permanent}} \newcommand{\GraphNoniso}{\textprob{Graph Nonisomorphism}} \newcommand{\GI}{\mathprob{GI}} \newcommand{\GNI}{\mathprob{GNI}} \newcommand{\GraphIso}{\textprob{Graph Isomorphism}} \newcommand{\QuantBoolForm}{\textprob{Quantified Boolean Formulae}} \newcommand{\GenGeography}{\textprob{Generalized Geography}} \newcommand{\MAXTSAT}{\mathprob{Max3SAT}} \newcommand{\GapMaxTSAT}{\mathprob{GapMax3SAT}} \newcommand{\ELIN}{\mathprob{E3LIN2}} \newcommand{\CSP}{\mathprob{CSP}} \newcommand{\Lin}{\mathprob{Lin}} \newcommand{\ONE}{\mathbf{ONE}} \newcommand{\ZERO}{\mathbf{ZERO}} \newcommand{\yes} \newcommand{\no} $
Back to Arithmetic Circuits
Back to notes

Depth reduction

80%

Formulas to logarithmic depth formulas

Let $p$ be an $n$-variate and degree $d$ polynomial computed by an arithmetic formula of size $s$, then $p$ can also be computed by an arithmetic formula of size $O(\poly(s))$ and depth $O(\log s)$.

The key observation is that a formula is simply a tree and hence at any vertex of the formula, we can naturally define a sub-formula. This allows us to perform induction.

General circuits to poly-logarithmic depth and depth-4 circuits

We have seen depth reduction for formula before by induction on the depth of the sub formulas. Since a formula is a tree, the induction is natural and don’t have too many technicalities. It turns out that the depth reduction for arithmetic branching programs is not so difficult either and have the same $O(\log d)$ result. However, for general arithmetic circuits, it does not seem to have such nice structure for depth reduction since a circuit could be arbitrary directed graph. Surprisingly, the following theorem shows that any arithmetic circuits can be reduced to log-depth circuits. Concretely, it implies $\VP=\VNC{2}$.

Let $p$ be an $n$-variate and degree $d$ polynomial computed by a homogeneous arithmetic circuit $C$ of size $s$. Then, there exists a circuit $C'$ that computes $p$ and satisfies the following properties.
1. $\text{size}(C')=\poly(s,n,d)$ and $\text{depth}(C')=\log d$.
2. The product gates have bounded fan-in and the sum gates have unbounded fan-in.
3. For any product fate $u$ in $C'$ and $u'$ be its input. $\text{deg}(u')\leq\text{deg}(u)/2$.

From [VSBR83], we can further reduce general circuits to depth four circuits of polynomial-size.

Let $p$ be an $n$-variate and degree $d$ polynomial computed by a homogeneous arithmetic circuit $C$ of size $s$. Then, there exists a $\Sigma\Pi\Sigma\Pi$ circuit $C'$ of size $s^{O(\sqrt{d})}$ that computes $p$. Specifically, the fan-in of the product gates is $O(\sqrt{d})$.

Generalized gate quotient

Let $C$ be a circuit and $u,v$ be any two gates in $C$. The gate quotitient of $u$ with respect to $v$, denoted as $[u,v]$, is defined as follows.
1. If $u=v$, $[u,v]=1$.
2. If $v$ is not in the sub-circuit of $u$, then $[u,v]=0$.
3. If $u$ is a sum gate, i.e., $u=u_1+u_2$, then $[u,v]=[u_1,v]+[u_2,v]$.
4. If $u$ is a product gate, i.e., $u=u_L\cdot u_R$, then $[u,v]=[u_L]+[u_R,v]$.
Recall that $[u]$ is defined as the polynomial computed by $u$.
Let $C$ be a circuit and $m\in\N$. Define the $m$th front family as follows. \begin{equation} \mathcal{F}_m = \{v:\ v=u\cdot w,\ \text{deg}(v)\geq m,\ \text{deg}(u),\text{deg}(w)<m \}. \end{equation}
Let $C$ be a circuit and $u,v$ be any two gates in $C$ such that $[u,v]\neq0$. We have $\text{deg}([u,v])=\text{deg}(u)-\text{deg}(v)$.

The proof is based on induction on the degree of $u$. Note that it suffices to consider $u$ being a product gate. For the base case, $u$ is a product gate multiplying two gates that are either variable or constant. Clearly that the equality holds. Consider $u=u_L\cdot u_R$, by definition we have $[u,v]=[u_L]\cdot[u_R,v]$ and thus \begin{equation} \text{deg}([u,v]) = \text{deg}([u_L]\cdot[u_R,v]) = \text{deg}(u_L) + \text{deg}([u_R,v]). \end{equation} By the induction hypothesis, we have $\text{deg}([u_R,v])=\text{deg}(u_R)-\text{deg}(v)$. As $\text{deg}(u_L)+\text{deg}(u_R)=\text{deg}(u)$, we have $\text{deg}([u,v])=\text{deg}(u)-\text{deg}(v)$.
Let $m\in\N$, $C$ be a circuit, and $u,v$ be any two gates in $C$ such that $\text{deg}(u)\geq m>\text{deg}(v)$. We have \begin{equation} [u] = \sum_{w\in\mathcal{F}_m}[w]\cdot[u,w], \end{equation} and \begin{equation} [u,v] = \sum_{w\in\mathcal{F}_m}[u,w]\cdot[w,v]. \end{equation}

The proof is based on induction on the depth of $u$. The base case is then $u\in\mathcal{F}_m$. That is, $u=u_L\cdot u_R$ where $\text{deg}(u_L),\text{deg}(u_R)<m$. Note that for any $w\in\mathcal{F}_m\backslash\{u\}$, as $\text{deg}(w)\geq m$, $w$ definitely does not lie in the sub-circuit of $u$ and thus $[u,w]=0$. Namely, \begin{equation} [u] = [u]\cdot[u,u] = \sum_{w\in\mathcal{F_m}}[w]\cdot[u,w]. \end{equation} Similarly, \begin{equation} [u,v] = [u,u]\cdot[u,v] = \sum_{w\in\mathcal{F}_m}[u,w]\cdot[w,v]. \end{equation} Now, consider two cases: 1. $u=u_1+u_2$ and 2. $u=u_L\cdot u_R$ where $\text{deg}(u_R)\geq m$.
1. By induction hypothesis, both $u_1$ and $u_2$ have frontier decomposition and thus \begin{equation} [u]=[u_1]+[u_2]=\sum_{w\in\mathcal{F}_m}[w]\cdot[u_1,w]+\sum_{w\in\mathcal{F}_m}[w]\cdot[u_2,w]=\sum_{w\in\mathcal{F}_m}[w]\cdot[u,w], \end{equation} and \begin{align} [u,v]&=[u_1,v]+[u_2,v]=\sum_{w\in\mathcal{F}_m}[u_1,w]\cdot[w,v]+\sum_{w\in\mathcal{F}_m}[u_2,w]\cdot[w,v]\\ &=\sum_{w\in\mathcal{F}_m}[u,w]\cdot[w,v]. \end{align} 2. By induction hypothesis, $u_R$ has frontier decomposition and thus \begin{equation} [u]=[u_L]\cdot[u_R]=[u_L]\cdot\sum_{w\in\mathcal{F}_m}[w]\cdot[u_R,w]=\sum_{w\in\mathcal{F}_m}[w]\cdot[u,w], \end{equation} and \begin{equation} [u,v]=[u_L]\cdot[u_R,v]=[u_L]\cdot\sum_{w\in\mathcal{F}_m}[u_R,w]\cdot[w,v]=\sum_{w\in\mathcal{F}_m}[u,w]\cdot[w,v]. \end{equation}

Proof of [VSBR83]

The proof of the main theorem is then based on induction on the degree of each gates. Concretely, we are going to construct a series of circuits that $C_0,C_1,\dots,C_{\log d}$ such that for any $i\in\{0,1,\dots,\log d\}$, $C_i$ computes (i) $[u]$: $\forall u\in C$, $2^i<\text{deg}(u)\leq2^{i+1}$. (ii) $[u,v]$: $\forall u,v\in C$, $2^i<\text{deg}([u,v])\leq2^{i+1}$.

Also, $\text{depth}(C_i)=O(i)$ and $C_i$ satisfies the conditions stated in the theorem for $C’$. In the end, output $C’=C_{\log d}$.

One can first see that $C_0$ can be done with brute-force. Next, consider $i=1,2,\dots,\log d$.

(i) For any $u\in C$, $2^i<\text{deg}(u)\leq2^{i+1}$.

The goal is to construct a sum of product of lower level circuit, i.e., degree less than $\text{deg}([u])/2$, for $[u]$. Pick $m=\text{deg}(u)/2-1$. By the frontier decomposition lemma, we have \begin{equation} [u] = \sum_{w\in\mathcal{F}_m}[w]\cdot[u,w]. \end{equation} By the degree of gate quotitient lemma, we know that $\text{deg}([u,w])=\text{deg}([u])-\text{deg}([w])< m$. Thus, by induction hypothesis, we have a depth $O(i)$ circuit for $[u,w]$. However, $[w]$ the degree exceeds $m$ so we don’t have the guarantee of induction hypothesis. Nevertheless, as $w\in\mathcal{F}_m$, we know $w=w_L\cdot w_R$ where $\text{deg}(w_L),\text{deg}(w_R)<m$. As a result, $[u]$ can be constructed by a sum of three product of circuits from lower induction level. Note that as $\card{\mathcal{F}_m}\leq s$, the size of the new circuit for $[u]$ is still polynomial in $s$.

(ii) For any $u,v\in C$, $2^i<\text{deg}([u,v])\leq2^{i+1}$.

The goal is to construct a sum of product of lower level circuit, i.e., degree less than $\text{deg}([u,v])/2$, for $[u,v]$. Pick $m=\frac{\text{deg}(u)+\text{deg}(v)}{2}-1$. By the frontier decomposition lemma, we have \begin{equation} [u,v] = \sum_{w\in\mathcal{F}_m}[u,w]\cdot[w,v]. \end{equation} By the degree of gate quotitient lemma, we know that $\text{deg}(u)-\text{deg}(w)\leq\frac{\text{deg}(u)-\text{deg}(v)}{2}$. However, it is unclear about the degree of $[w,v]$ so we decompose $w$ by the definition of frontier and yield $[w,v]=[w_L]\cdot[w_R,v]$ where $\text{deg}(w_L),\text{deg}(w_R)<m$. Note that $\text{deg}([w_R,v])\leq\frac{\text{deg}(u)-\text{deg}(v)}{2}$. As a result, now we only need to deal with $[w_L]$.

Pick $m’=\frac{\text{deg}(u)-\text{deg}(v)}{2}$. By the frontier decomposition lemma, we have \begin{equation} [w_L] = \sum_{w’\in\mathcal{F}_{m’}}[w’]\cdot[w_L,w’]. \end{equation} By the degree of gate quotitient lemma, we know that $\text{deg}([w_L,w’])\leq\frac{\text{deg}(u)-\text{deg}(v)}{2}$ and by applying frontier decomposition lemma on $[w’]$, we finally get $[w’]=[w’_L]\cdot[w’_R]$ where $\text{deg}([w’_L]),\text{deg}([w’_R])\leq\frac{\text{deg}(u)-\text{deg}(v)}{2}$.

To sum up, we have \begin{equation} [u,v] = \sum_{w\in\mathcal{m},w’\in\mathcal{F}_{m’}}[u,w]\cdot[w’_L]\cdot[w’_R]\cdot[w_L,w’]\cdot[w_R,v]. \end{equation} That is, $[u,v]$ can be constructed by a sum of five product of circuits from lower induction level.

Proof of reduction to depth four circuits

From [VSBR83], we know that if a degree $d$ polynomial $p$ can be computed by a size $s$ circuit $C$, then there exists a circuit $C’$ that computes $p$ as follows. \begin{equation} C’= \sum_{i\in[s]}g_{i1}\cdot g_{i2}\cdots g_{i5}, \end{equation} where $g_{ij}$ is in the form of $\Sigma^s\Pi^{d/2}$, i.e., sum of product of polynomials of degree at most $d/2$. Now, the goal is to transform those $g_{ij}$ into $\Sigma\Pi^t$ where $t$ could be much less that $d/2$, say $O(\sqrt{d})$. A naive approach is replacing those $g_{ij}$ that consists some polynomials having degree greater than $t$. Concretely, suppose $g_{ij}=\sum_{i’}g’_{i’1}\cdot g’_{i’2}\cdots g’_{i5}$ and exists $j’\in[5]$ such that $\text{deg}(g’_{i’j’})>t$. We will replace $g_{ij}$ by expanding it with Theorem (VSBR83). Note that in each round, we will scan through all $i\in[s]$ in the first level and replace thouse which has a large degree component. For instance, say $g_{i5}$ has component of large degree for every $i\in[s]$, then we replace it by $g_{i5} = \sum_{i’}g’_{i’1}\cdots g’_{i’5}$ and get \begin{equation} C’= \sum_{i,i’\in[s]}g_{i1}\cdot g_{i2}\cdots g_{i4}\cdot g’_{i’1}\cdots g’_{i’5}. \end{equation}

Now, we are claiming that this replacing procedure can be done in $O(d/t)$ steps. Namely, after $O(d/t)$ rounds, the circuit will become in $\Sigma\Pi\Sigma\Pi^{t}$ form. Specifically, the number of summands would be $s^{O(d/t)}$ since each rouns increase at most $s$ summands.

The main reason is because the circuit is homogeneous. As a result, after the decomposition the total degree at the first level remains $d$. As in each round we only decompose those component that has degree larger than $t$, there exists at least one new component having degree at least $t/5$. Suppose we perform $r$ rounds, the total degree is at least $r\cdot t/5$, which should be less than $d$. Thus, we have $r=O(d/t)$.

Last but not least, let’s estimate the size of the resulting circuit. As we can brute-forcely generate all monomials of degree at most $t$, the last two levels have size at most $n^t=s^{(t)}$. Also, as we discussed before the total number of summands in the first level is at most $s^{O(t/d)}$ and thus the size of the resulting circuit is at most $s^{O(t+d/t)}$. By picking $t=O(\sqrt{d})$, we have a $\Sigma\Pi\Sigma\Pi$ circuit of size $s^{O(\sqrt{d})}$ for $p$.