International Association
for Cryptologic Research

This paper studies the quantum computational complexity of the discrete logarithm (DL) and related group-theoretic problems in the context of ``generic algorithms''---that is, algorithms that do not exploit any properties of the group encoding. We establish the quantum generic group model and hybrid classical-quantum generic group model as quantum and hybrid analogs of their classical counterpart. This model counts the number of group operations of the underlying cyclic group $G$ as a complexity measure. Shor's algorithm for the discrete logarithm problem and related algorithms can be described in this model and make $O(\log |G|)$ group operations in their basic form. We show the quantum complexity lower bounds and (almost) matching algorithms of the discrete logarithm and related problems in these models. * We prove that any quantum DL algorithm in the quantum generic group model must make $\Omega(\log |G|)$ depth of group operation queries. This shows that Shor's algorithm that makes $O(\log |G|)$ group operations is asymptotically optimal among the generic quantum algorithms, even considering parallel algorithms. * We observe that some (known) variations of Shor's algorithm can take advantage of classical computations to reduce the number and depth of quantum group operations. We show that these variants are optimal among generic hybrid algorithms up to constant multiplicative factors: Any generic hybrid quantum-classical DL algorithm with a total number of (classical or quantum) group operations $Q$ must make $\Omega(\log |G|/\log Q)$ quantum group operations of depth $\Omega(\log\log |G| - \log\log Q)$. * When the quantum memory can only store $t$ group elements and use quantum random access classical memory (QRACM) of $r$ group elements, any generic hybrid quantum-classical algorithm must make either $\Omega(\sqrt{|G|})$ group operation queries in total or $\Omega(\log |G|/\log (tr))$ quantum group operation queries. In particular, classical queries cannot reduce the number of quantum queries beyond $\Omega(\log |G|/\log (tr))$. As a side contribution, we show a multiple discrete logarithm problem admits a better algorithm than solving each instance one by one, refuting a strong form of the quantum annoying property suggested in the context of password-authenticated key exchange protocol.

Recently, Aaronson et al. (arXiv:2009.07450) showed that detecting interference between two orthogonal states is as hard as swapping these states. While their original motivation was from quantum gravity, we show its applications in quantum cryptography. 1. We construct the first public key encryption scheme from cryptographic non-abelian group actions. Interestingly, ciphertexts of our scheme are quantum even if messages are classical. This resolves an open question posed by Ji et al. (TCC ’19). We construct the scheme through a new abstraction called swap-trapdoor function pairs, which may be of independent interest. 2. We give a simple and efficient compiler that converts the flavor of quantum bit commitments. More precisely, for any prefix X, Y ∈ {computationally,statistically,perfectly}, if the base scheme is X-hiding and Y-binding, then the resulting scheme is Y-hiding and X-binding. Our compiler calls the base scheme only once. Previously, all known compilers call the base schemes polynomially many times (Crépeau et al., Eurocrypt ’01 and Yan, Asiacrypt ’22). For the security proof of the conversion, we generalize the result of Aaronson et al. by considering quantum auxiliary inputs.

We present a new cryptanalytic algorithm on obfuscations based on GGH15 multilinear map. Our algorithm, statistical zeroizing attack, directly distinguishes two distributions from obfuscation while it follows the zeroizing attack paradigm, that is, it uses evaluations of zeros of obfuscated programs.Our attack breaks the recent indistinguishability obfuscation candidate suggested by Chen et al. (CRYPTO’18) for the optimal parameter settings. More precisely, we show that there are two functionally equivalent branching programs whose CVW obfuscations can be efficiently distinguished by computing the sample variance of evaluations.This statistical attack gives a new perspective on the security of the indistinguishability obfuscations: we should consider the shape of the distributions of evaluation of obfuscation to ensure security.In other words, while most of the previous (weak) security proofs have been studied with respect to algebraic attack model or ideal model, our attack shows that this algebraic security is not enough to achieve indistinguishability obfuscation. In particular, we show that the obfuscation scheme suggested by Bartusek et al. (TCC’18) does not achieve the desired security in a certain parameter regime, in which their algebraic security proof still holds.The correctness of statistical zeroizing attacks holds under a mild assumption on the preimage sampling algorithm with a lattice trapdoor. We experimentally verify this assumption for implemented obfuscation by Halevi et al. (ACM CCS’17).

We initiate a systematic study of pseudorandom functions (PRFs) that are computable by simple matrix branching programs; we refer to these objects as “matrix PRFs”. Matrix PRFs are attractive due to their simplicity, strong connections to complexity theory and group theory, and recent applications in program obfuscation.Our main results are:We present constructions of matrix PRFs based on the conjectured hardness of computational problems pertaining to matrix products.We show that any matrix PRF that is computable by a read-c, width w branching program can be broken in time poly$$(w^c)$$; this means that any matrix PRF based on constant-width matrices must read each input bit $$\omega (\log (\lambda ))$$ times. Along the way, we simplify the “tensor switching lemmas” introduced in previous IO attacks.We show that a subclass of the candidate local-PRG proposed by Barak et al. [Eurocrypt 2018] can be broken using simple matrix algebra.We show that augmenting the CVW18 IO candidate with a matrix PRF provably immunizes the candidate against all known algebraic and statistical zeroizing attacks, as captured by a new and simple adversarial model.

The random oracle model (ROM) is an idealized model where hash functions are modeled as random functions that are only accessible as oracles. Although the ROM has been used for proving many cryptographic schemes, it has (at least) two problems. First, the ROM does not capture quantum adversaries. Second, it does not capture non-uniform adversaries that perform preprocessings. To deal with these problems, Boneh et al. (Asiacrypt’11) proposed using the quantum ROM (QROM) to argue post-quantum security, and Unruh (CRYPTO’07) proposed the ROM with auxiliary input (ROM-AI) to argue security against preprocessing attacks. However, to the best of our knowledge, no work has dealt with the above two problems simultaneously.In this paper, we consider a model that we call the QROM with (classical) auxiliary input (QROM-AI) that deals with the above two problems simultaneously and study security of cryptographic primitives in the model. That is, we give security bounds for one-way functions, pseudorandom generators, (post-quantum) pseudorandom functions, and (post-quantum) message authentication codes in the QROM-AI.We also study security bounds in the presence of quantum auxiliary inputs. In other words, we show a security bound for one-wayness of random permutations (instead of random functions) in the presence of quantum auxiliary inputs. This resolves an open problem posed by Nayebi et al. (QIC’15). In a context of complexity theory, this implies $$ \mathsf {NP}\cap \mathsf {coNP} \not \subseteq \mathsf {BQP/qpoly}$$ relative to a random permutation oracle, which also answers an open problem posed by Aaronson (ToC’05).

In this paper, we propose cryptanalyses of all existing indistinguishability obfuscation (iO) candidates based on branching programs (BP) over GGH13 multilinear map for all recommended parameter settings. To achieve this, we introduce two novel techniques, program converting using NTRU-solver and matrix zeroizing, which can be applied to a wide range of obfuscation constructions and BPs compared to previous attacks. We then prove that, for the suggested parameters, the existing general-purpose BP obfuscations over GGH13 do not have the desired security. Especially, the first candidate indistinguishability obfuscation with input-unpartitionable branching programs (FOCS 2013) and the recent BP obfuscation (TCC 2016) are not secure against our attack when they use the GGH13 with recommended parameters. Previously, there has been no known polynomial time attack for these cases.Our attack shows that the lattice dimension of GGH13 must be set much larger than previous thought in order to maintain security. More precisely, the underlying lattice dimension of GGH13 should be set to $$n=\tilde{\varTheta }( \kappa ^2 \lambda )$$n=Θ~(κ2λ) to rule out attacks from the subfield algorithm for NTRU where $$\kappa $$κ is the multilinearity level and $$\lambda $$λ the security parameter.