Quantum Foundations

In this talk, I will argue and practically illustrate that insights in quantum information, concretely coming from the tensor network representations of quantum many-body states, can help in devising better privacy-preserving machine learning algorithms. In the first part, I will show that standard neural networks are vulnerable to a type of privacy leak that involves global properties of the data used for training, thus being a priori resistant to standard protection mechanisms. In the second, I will show that tensor networks, when used as machine learning architectures, are invulnerable to this vulnerability. The proof of the resilience is based on the existence of canonical forms for such architectures. Given the growing expertise in training tensor networks and the recent interest in tensor-based reformulations of popular machine learning architectures, these results imply that one may not have to be forced to make a choice between accuracy in prediction and ensuring the privacy of the information processed when using machine learning on sensitive data.

---

Zoom link

Decision-making under uncertainty and causal thinking are fundamental aspects of intelligent reasoning. Decision-making has been well studied when the available information is considered at the associative (probabilistic) level. The classical Theorems of von Neumann-Morgenstern and Savage provide a formal criterion for rational choice using associative information: maximize expected utility. There is an ongoing debate around the origin of probabilities involved in such calculation. In this work, we will show how the probabilities for decision-making can be grounded in causal models by considering decision problems in which the available actions and consequences are causally connected. In this setting, actions are regarded as an intervention over a causal model. Then, we extend a previous causal decision-making result, which relies on a known causal model, to the case in which the causal mechanism that controls some environment is unknown to a rational decision-maker. In this way, action-outcome probabilities can be grounded in causal models in known and unknown cases. Finally, as an application, we extend the well-known concept of Nash Equilibrium to the case in which the players of a strategic game consider causal information.

---

Zoom link

Nonclassical causal modeling was developed in order to explain violations of Bell inequalities while adhering to relativistic causal structure and faithfulness -- that is, avoiding fine-tuned causal explanations. Recently, a no-go theorem stronger than Bell's theorem has been derived, based on extensions of Wigner's friend thought experiment: the Local Friendliness (LF) no-go theorem. Here we show that the LF no-go theorem poses formidable challenges for the field of causal modeling, even when nonclassical and/or cyclic causal explanations are considered. We first recast the LF inequalities, one of the key elements of the LF no-go theorem, as special cases of monogamy relations stemming from a statistical marginal problem; we then further recast LF inequalities as causal compatibility inequalities stemming from a nonclassical causal marginal problem, for a causal structure implied by well-motivated causal-metaphysical assumptions. We find that the LF inequalities emerge from the causal modeling perspective even when allowing the latent causes of observed events to admit post-quantum descriptions, such as Generalised Probabilistic Theories (GPT) or even more exotic theories. We further prove that no nonclassical causal model can explain violations of LF inequalities without violating the No Fine-Tuning principle. Finally, we note that these obstacles cannot be overcome even if one were to appeal to cyclic causal models.

--- 

Zoom link

The principle of information causality, proposed as a generalization of no signaling principle, has efficiently been applied to outcast beyond quantum correlations as unphysical. In this talk, we show that this principle, when utilized properly, can provide physical rationale toward structural derivation of multipartite quantum systems. In accordance with the no signaling condition, the state and effect spaces of a composite system can allow different possible mathematical descriptions, even when descriptions for the individual systems are assumed to be quantum. While in one extreme, namely, the maximal tensor product composition, the state space becomes quite exotic and permits composite states that are not allowed in quantum theory, the other extreme - minimal tensor product composition - contains only separable states, and the resulting theory allows only Bell local correlation. As we show, none of these compositions is commensurate with information causality, and hence cannot be the bona-fide description of nature. Information causality therefore promises an information-theoretical derivation of self duality of the state and effect cones for composite quantum systems.

---

Zoom link

The field of indefinite order in quantum theory was born from an attempt to construct a theory of quantum gravity, where the first step is to construct a generalized quantum theory in which events could have an indefinite order [1]. It is expected that such a theory would lead us more naturally to the construction of a quantum gravity theory. One way to explore this topic operationally is to consider that two agents Alice and Bob apply operations A and B on a given target system and that quantum mechanics holds locally for each agent [2].  The quantum switch is the simplest example of a task with indefinite order, where the order of operations applied by two agents on a target system is entangled with the state of a quantum control system. In particular, in the gravitational quantum switch, the order of these operations is entangled with the state of a quantum spacetime [3].

In this talk, I will present a recent result, where we propose a distinct protocol for performing a gravitational quantum switch [4]. One of the agents crosses the interior region of massive spherical shells in a superposition of different radii and becomes entangled with their geometry. This entanglement is used as a resource to control the order of operation in the implementation of the quantum switch. Novel features of the protocol include: i) the superposition of nonisometric geometries; ii) the existence of a region with a definite geometry; iii) the fact that the agent that experiences the superposition of geometries is in free fall, preventing information on the global geometry to be obtained by this agent.

[1] Hardy, J. Phys. A: Math. Theor. 40, 3081 (2007);
[2] Chiribella, D’Ariano, Perinotti, Valiron, PRA 88, 022318 (2013); Oreshkov, Costa, Brukner, Nat. Commun. 3, 1092 (2012).
[3] Zych, Costa, Pikovski, Brukner, Nat. Commun. 10, 3772 (2019).
[4] Móller, Sahdo, Yokomizo, arXiv:2306.10984 (2023).

---

Zoom link https://pitp.zoom.us/j/96568644174?pwd=ak9qM0Y1R29WdXRXREIxdTFtcDhYUT09

The theory of quantum computation has brought us rapid technological developments, together with remarkable improvements in how we understand quantum theory. In this talk, I shall describe the foundations of a theoretical programme to extend the quantum theory of computation beyond quantum theory itself, based on the recently proposed constructor theory. I will then explain a recent application of this new approach to the problem of witnessing quantum effects in gravity.

---

Zoom link https://pitp.zoom.us/j/99012897226?pwd=U2hFQlU3TWxkSDIwZ1E5OHNPMVJhUT09

Network nonlocality, and more specifically, triangle network nonlocality, is a basic feature of modern causal modelling when going beyond Bell scenarios. However, despite the apparent simplicity of the problems one may formulate, relatively little is known due to the hardness of certifying nonlocality in networks. In this talk, I will describe a motivating example of a quantum triangle distribution, the Elegant Joint Measurement due to Nicolas Gisin, that is strongly believed to be nonlocal even in the presence of experimental noise. I will then present the ongoing effort to produce a computer-assisted proof of nonlocality for this distribution, thereby developing a toolkit to tackle general nonlocality problems. This effort is based on the inflation technique for causal inference, but taken to higher levels than what was generally considered tractable. This is made possible by a number of optimization techniques, involving symmetry reductions, branch-and-bound optimization, and most importantly, the use of a Frank-Wolfe algorithm to bypass the need to call a standard linear program solver.

---

Zoom link https://pitp.zoom.us/j/97499052021?pwd=R1EyU2pmc1hFSzJ1UEpJQ1h0RnQzdz09

In the usual operational picture, operations are represented by boxes having inputs and outputs. Further, we usually consider the causally simple case where the inputs are prior to the outputs for each such operation. In this talk (motivated by an attempt to formulate an operational probabilistic field theory) I will consider what I call the "causally complex" situation. Operations are represented by circles. These circles have wires going in and out. Each such wire can represent an input and an output. Further, each operation will have a causal diagram associated with it. The causal structure can be more complicated than the simple case. These circles can be joined together to create new operations. I will discuss conditions on these causally complex operations so that we have positivity (probabilities are non-negative) and causality (to be understood in a time symmetric manner). I will also discuss how these properties compose when we join causally complex operations. Causally complex operations are related to objects in the causaloid formalism as well as to quantum combs.

---

Zoom link https://pitp.zoom.us/j/99425886198?pwd=ODR0VVFzQUJHeER4OVJ2cEo3cVdDQT09

We will present a teleportation protocol which uses the properties of superoscillatory wavefunctions. Then we will briefly introduce a different protocol, the port based teleportation. We will use it to study teleportation over infinitesimally small distances, where the vacuum of a quantum field serves as the source of entanglement. We find that the resulting motion is equivalent to a quantum teleportation-induced Brownian motion. Purifying the interactions, from measurements to unitary operations, leads to motion described by Schrodinger’s equation. Therefore, we find that the notions of teleportation and continuous quantum and classical motion are unified in this sense.

--- 

Zoom link https://pitp.zoom.us/j/93504138534?pwd=VUFPSXpRd2g0TVNxVTNlY3Rxa1ZjUT09

Out of the many lessons quantum mechanics seems to teach us, one is that it seems there are things we cannot experimentally have access to. There is, indeed, a fundamental limit to our ability to experimentally explore the world. In this work we accept this lesson as a fact and we build a general theory based on this principle. We start by assuming the existence of statements whose truth value is not experimentally accessible. That is, there is no way, not even in theory, to directly test if these statements are true or false. We further develop a theory in which experimentally accessible statements are a union of a fixed minimum number of inaccessible statements. For example, the value of truth of the statements a and b is not accessible, but the value of truth of the statement “a or b" is accessible. We do not directly assume probability theory, we exclusively define experimentally accessible and inaccessible statements and build on these notions using the rules of classical logic. We find that an interesting structure emerges. Developing this theory, we relax the logical structure, naturally obtaining a derivation of a constrained quasi-probabilistic theory rich in structure that we name theory of inaccessible information. Surprisingly, the simplest model of theory of inaccessible information is the qubit in quantum mechanics. Along the path for the construction of this theory, we characterise and study a family of multiplicative information measures that we call inaccessibility measures. arXiv:https://arxiv.org/abs/2305.05734

---

Zoom link https://pitp.zoom.us/j/91350754706?pwd=V1dVdGM3Zk9MNkp4VlpCYUoxbXg3UT09

We report a deterministic and exact protocol to reverse any unknown qubit-unitary and qubit-encoding isometry operations. To avoid known no-go results on universal deterministic exact unitary inversion, we consider the most general class of protocols transforming unknown unitary operations within the quantum circuit model, where the input unitary operation is called multiple times in sequence and fixed quantum circuits are inserted between the calls. In the proposed protocol, the input qubit-unitary operation is called 4 times to achieve the inverse operation, and the output state in an auxiliary system can be reused as a catalyst state in another run of the unitary inversion. This protocol only applies only for qubit-unitary operations, but we extend this protocol to any qubit-encoding isometry operations. We also present the simplification of the semidefinite programming for searching the optimal deterministic unitary inversion protocol for an arbitrary dimension presented by M. T. Quintino and D. Ebler [Quantum 6, 679 (2022)]. We show a method to reduce the large search space representing all possible protocols, which provides a useful tool for analyzing higher-order quantum transformations for unitary operations.

---

Zoom link https://pitp.zoom.us/j/92900413520?pwd=a1JqU1IzMVdSRGQreWJIbEFCT2hWUT09

Supermaps are higher-order transformations taking maps as input.  We consider quantum algorithms implementing supermaps for the input given by unknown Hamiltonian dynamics, which can be regarded as infinitely divisible unitary operations.  We first show a quantum algorithm that approximately but universally transforms black-box Hamiltonian dynamics into controlled Hamiltonian dynamics utilizing a higher-order transformation called neutralization.  Then, we present another universal algorithm that efficiently simulates linear transformations of any Hamiltonian consisting of a polynomial number of terms in system size, using only controlled-Pauli gates and time-correlated randomness.  This algorithm for implementing Hamiltonian supermaps is an instance of quantum functional programming, where the desired function is specified as a concatenation of higher-order quantum transformations. As examples, we demonstrate the simulation of negative time-evolution and time-reversal, and perform a Hamiltonian learning task.

References:

Q. Dong, S. Nakayama, A. Soeda and M. Murao, arXiv:1911.01645v3
T. Odake, Hlér Kristjánsson, A. Soeda M. Murao, arXiv:2303.09788

---

Zoom Link: https://pitp.zoom.us/j/94278362588?pwd=MGszYk9uN1A3K1RTOVhYSGpkL1FQdz09