MIP∗ = RE Zhengfeng Ji∗1, Anand Natarajan†2,3, Thomas Vidick‡3, John Wright§2,3,4, and Henry Yuen¶5 1Centre for Quantum Software and Information, University of Technology Sydney 2Institute for Quantum Information and Matter, California Institute of Technology 3Department of Computing and Mathematical Sciences, California Institute of Technology 4Department of Computer Science, University of Texas at Austin 5Department of Computer Science and Department of Mathematics, University of Toronto September 30, 2020 Abstract We show that the class MIP∗ of languages that can be decided by a classical verifier interacting with multiple all-powerful quantum provers sharing entanglement is equal to the class RE of recursively enumerable languages. Our proof builds upon the quantum low-degree test of (Natarajan and Vidick, FOCS 2018) and the classical low-individual degree test of (Ji, et al., 2020) by integrating recent devel- opments from (Natarajan and Wright, FOCS 2019) and combining them with the recursive compression framework of (Fitzsimons et al., STOC 2019). An immediate byproduct of our result is that there is an efficient reduction from the Halting Problem 1 to the problem of deciding whether a two-player nonlocal game has entangled value 1 or at most 2 . Using a known connection, undecidabilityof the entangled value implies a negative answer to Tsirelson’s problem: we show, by providing an explicit example, that the closure Cqa of the set of quantum tensor product correlations is strictly included in the set Cqc of quantum commuting correlations. Following work of (Fritz, Rev. Math. Phys. 2012) and (Junge et al., J. Math. Phys. 2011) our results provide a refutation of Connes’ embedding conjecture from the theory of von Neumann algebras. arXiv:2001.04383v2 [quant-ph] 29 Sep 2020 ∗Email: [email protected] †Email: [email protected] ‡Email: [email protected] §Email: [email protected] ¶Email: [email protected] 1 Contents 1 Introduction 5 1.1 Interactive proof systems . ........... 6 1.2 Statementofresult ............................... ........ 10 1.3 Consequences.................................... ...... 12 1.4 Openquestions ................................... ...... 14 2 Proof Overview 17 3 Preliminaries 24 3.1 Turingmachines .................................. ...... 24 3.2 Linearspaces .................................... ...... 25 3.3 Finitefields ..................................... ...... 27 3.3.1 Subfieldsandbases.............................. ..... 27 3.3.2 Bit string representations . .......... 30 3.4 Polynomials and the low-degree code . ............ 32 3.5 Linear spaces and registers . ........... 33 3.6 Measurements and observables . ........... 34 3.7 Generalized Pauli observables . ............ 34 4 Conditionally Linear Functions, Distributions, and Samplers 38 4.1 Conditionally linear functions and distributions . .................... 38 4.2 Conditionally linear samplers . ............. 43 5 Nonlocal Games 46 5.1 Gamesandstrategies .............................. ........ 46 5.2 Distancemeasures ................................ ....... 48 5.3 Normalformverifiers.... ...... ..... ...... ...... ... ........ 51 6 Types 53 6.1 Typed samplers, deciders, and verifiers . ............... 53 6.2 Graphdistributions .............................. ......... 55 6.3 Detyping typed verifiers . ......... 57 7 Classical and Quantum Low-degree Tests 60 7.1 The classical low-degree test . ............ 60 7.1.1 Thegame....................................... 60 7.1.2 Complexity of the classical low-degree test. ............... 64 7.2 TheMagicSquaregame .............................. ...... 65 7.3 ThePaulibasistest ............................... ........ 69 7.3.1 Thegame....................................... 69 7.3.2 Canonical parameters and complexity of the Pauli basistest ............. 77 2 8 Introspection Games 80 8.1 Overview ........................................ .... 80 8.2 The introspective verifier . ........... 81 8.2.1 Complexity of the introspective verifier . ............. 86 8.2.2 Introspection theorem . ....... 87 8.3 Completeness of the introspective verifier . ................ 88 8.3.1 Preliminarylemmas ...... ..... ...... ...... ..... ..... 88 8.3.2 Completeness of the introspective verifier . .............. 90 8.4 Soundness of the introspective verifier . ............... 92 8.4.1 ThePaulitwirl ................................. .... 92 8.4.2 Preliminarylemmas ...... ..... ...... ...... ..... ..... 98 8.4.3 Soundnessproof ................................ .... 100 9 Oracularization 109 9.1 Overview ........................................ .... 109 9.2 Oracularizing normal form verifiers . ............. 109 9.3 Completeness and complexity of the oracularized verifier................... 110 9.4 Soundness of the oracularized verifier . .............. 112 10 Answer Reduction 114 10.1 Circuit preliminaries . ........... 115 10.2 A Cook-Levin theorem for bounded deciders . .............. 115 10.3 A succinct 5SAT description for deciders . ............... 123 10.4 A PCP for normal form deciders . .......... 128 10.4.1 Preliminaries ................................ ...... 128 10.4.2 ThePCP ....................................... 129 10.5 A normal form verifier for the PCP . ........... 134 10.5.1 Parameters and notation . ........ 134 10.5.2 The answer-reduced verifier . ......... 134 10.5.3 Main theorem for answer reduction . .......... 136 10.6 Completeness of the answer-reduced verifier . ................. 138 10.7 Soundness of the answer-reduced verifier . ............... 140 11 Parallel Repetition 148 11.1 The anchoring transformation . ............. 148 11.2 Parallel repetition of anchored games . ................ 149 12 Gap-preserving Compression 152 12.1 Proof of Theorem 12.1 ..................................... 152 12.2 An MIP∗ protocol for the Halting problem . .... 158 12.3 An explicit separation . ........... 162 A Analysis of the Pauli basis test 164 A.1 Preliminaries ................................... ....... 164 A.2 Strategies ...................................... ...... 166 A.3 Expanding the Hilbert space and defining commuting observables .............. 170 3 A.4 Combining the X and Z measurements............................. 174 A.5 Applying the classical low-degree test . ............... 181 A.6 Pulling the X and Z measurementsapart. 189 4 1 Introduction 2 2 For integer n, k 2 define the quantum (spatial) correlation set C (n, k) as the subset of Rn k that ≥ qs contains all tuples (pabxy) representing families of bipartite distributions that can be locally generated in non-relativistic quantum mechanics. Formally, (p ) C (n, k) if and only if there exist separable abxy ∈ qs Hilbert spaces A and B, for every x 1,..., n (resp. y 1,..., n ), a collection of projec- x H H y ∈ { } ∈ { } tions Aa a 1,...,k on A (resp. Bb b 1,...,k on B) that sum to identity, and a state (unit vector) { } ∈{ } H { } ∈{ } H ψ A B such that ∈H ⊗H x y x, y 1,2,..., n , a, b 1,2,..., k , p = ψ∗ A B ψ . (1) ∀ ∈ { } ∀ ∈ { } abxy a ⊗ b Note that due to the normalization conditions on ψ and on Ax and By , for each x, y, (p ) is a { a } { b } abxy probability distribution on 1,2,..., k 2. By taking direct sums it is easy to see that the set C (n, k) is { } qs convex. Let C (n, k) denote its closure (it is known that C (n, k) = C (n, k), see [Slo19a]). qa qs 6 qa Our main result is that the family of sets Cqa(n, k) n,k N is extraordinarily complex, in the following { } ∈ computational sense. For any 0 < ε < 1 define the ε-weak membership problem for Cqa as the problem of 2 2 deciding, given n, k N and a point p = (p ) Rn k , whether p lies in C (n, k) or is ε-far from ∈ abxy ∈ qa it in ℓ1 distance, promised that one is the case. Then we show that for any given 0 < ε < 1 the ε-weak membership problem for Cqa cannot be solved by a Turing machine that halts with the correct answer on every input. We show this by directly reducing the Halting problem to the weak membership problem for Cqa: we show that for all 0 < ε < 1 and any Turing machine one can efficiently compute integers n, k N and a linear functional ℓ on Rn2k2 such that, whenever M halts it holds that ∈ M M sup ℓ (p) = 1, (2) M p Cqa(n,k) ∈ whereas if does not halt then M sup ℓ (p) 1 ε . (3) M ≤ − p Cqa(n,k) ∈ By standard results in convex optimization, this implies the aforementioned claim on the undecidability of the ε-weak membership problem for Cqa (for any 0 < ε < 1). Our result has interesting consequences for long-standing conjectures in quantum information theory and the theory of von Neumann algebras. Through a connection that follows from the work of Navascues, Piro- nio, and Acin [NPA08] the undecidability result implies a negative answer to Tsirelson’s problem [Tsi06]. Let Cqc(n, k) denote the set of quantum commuting correlations, which is the set of tuples (pabxy) arising from operators Ax and By acting on a single Hilbert space and a state ψ such that { a } { b } H ∈H x y x y x, y 1,..., n , a, b 1,..., k , p = ψ∗ A B ψ and A , B = 0. (4) ∀ ∈ { } ∀ ∈ { } abxy a b a b Then Tsirelson’s problem asks if, for all n, k, the sets Cqa(n, k)and Cqc(n, k) are equal. Using results 2 2 from [NPA08] we give integer n, k and an explicit linear function ℓ on Rn k such that 1 sup ℓ(p) = 1, but sup ℓ(p) , ≤ 2 p Cqc(n,k) p Cqa(n,k) ∈ ∈ which implies that C (n, k) = C (n, k). By an implication of Fritz [Fri12] and Junge et al. [JNP+11] qa 6 qc we further obtain that Connes’ Embedding Conjecture [Con76] is false; in other words, there exist type II1 5 von Neumann factors that do not embed