arXiv subjects

Sam Powers

Publications and source records attributed to Sam Powers.

9 recordsLinked to original sources

arXiv2026

Emergent Space, Time, and Lorentz Symmetry from Binary Sequences

We construct an information theory framework in which the fundamental objects are binary sequences of length $n$, equipped with the bitwise XOR operation. The only physical observables are counts of XOR-generated symbol classes, while the exact locations of symbols are inaccessible. Averaging over those locations ultimately yields the Minkowski interval as an invariant object that maximizes (information) entropy. Correlations between two binary sequences are base-4 sequences that we label as ``events'', and events are connected with maps. The entire kinematic structure of Special Relativity is recovered under minimal assumptions, i.e. counts that represent space and time increments carry equal informational weight. The central claim is that Lorentz symmetry is the typical large-n behaviour of XOR counting. At finite $n$ the framework yields a discrete rapidity spectrum, a bound $\gamma_{\max} = O(\sqrt{n})$, and interval fluctuations of relative size $O(n^{-1/2})$, with standard special relativity recovered as $n \to \infty$. However, the light cone and one null coordinate are exact for every microscopic configuration, so the symmetry group is undeformed and dispersion relations are unmodified. Thus, the theory, though discrete, implies no Lorentz violation of the standard phenomenological kind. Ultra-high-energy cosmic rays already require $n \gtrsim 10^{23}$; interferometry excludes the variant in which $n$ scales linearly with system size, leaving a holographic area law. The most striking prediction of the framework concerns systems at the maximum of their information capacity, where $n$ is necessarily finite: black holes and de Sitter space. There the corrections are of order one within a Planck proper length of the horizon, regardless of the horizon's size, and the spacetime description fails altogether at the endpoint of black-hole evaporation, where $n$ itself is of order unity.

physics.gen-ph↗

arXiv2025

Basis Vector Metric: A Method for Robust Open-Ended State Change Detection

We test a new method, which we will abbreviate using the acronym BVM (Basis Vectors Method), in its ability to judge the state changes in images through using language embeddings. We used the MIT-States dataset, containing about 53,000 images, to gather all of our data, which has 225 nouns and 115 adjectives, with each noun having about 9 different adjectives, forming approximately 1000 noun-adjective pairs. For our first experiment, we test our method's ability to determine the state of each noun class separately against other metrics for comparison. These metrics are cosine similarity, dot product, product quantization, binary index, Naive Bayes, and a custom neural network. Among these metrics, we found that our proposed BVM performs the best in classifying the states for each noun. We then perform a second experiment where we try using BVM to determine if it can differentiate adjectives from one another for each adjective separately. We compared the abilities of BVM to differentiate adjectives against the proposed method the MIT-States paper suggests: using a logistic regression model. In the end, we did not find conclusive evidence that our BVM metric could perform better than the logistic regression model at discerning adjectives. Yet, we were able to find evidence for possible improvements to our method; this leads to the chance of increasing our method's accuracy through certain changes in our methodologies.

cs.CL↗

arXiv2023

An event centric approach to modeling quantum systems

Event centric approaches to modeling physics have gained traction in recent decades. In this work, we present a first principles approach to this idea, which assumes nothing but the existence of causal networks of events and their relationships. The modeling elements we employ consist solely of classical bits, or the abstract symbols $0$ and $1$. Using sequences of these symbols, we model primitive elements of causal networks consisting of two causally connected events. By introducing an epistemic constraint on observers, we derive a statistical picture of these network elements, leading to the emergence of non-determinism and the subsequent derivation of a quantum theory. We then apply this event centric framework to three physical scenarios involving spin, including a Bell test. Comparing the resulting predictions to non-relativistic quantum mechanics, we find good agreement, including a violation of the CHSH inequality. More broadly, the results presented here highlight this novel framework's explanatory and predictive power. When coupled with recent advancements in event centric approaches to modeling spacetime, we argue that this framework may provide some insight into the issue of quantum gravity.

physics.gen-ph↗

arXiv2023

Evaluating Continual Learning on a Home Robot

Robots in home environments need to be able to learn new skills continuously as data becomes available, becoming ever more capable over time while using as little real-world data as possible. However, traditional robot learning approaches typically assume large amounts of iid data, which is inconsistent with this goal. In contrast, continual learning methods like CLEAR and SANE allow autonomous agents to learn off of a stream of non-iid samples; they, however, have not previously been demonstrated on real robotics platforms. In this work, we show how continual learning methods can be adapted for use on a real, low-cost home robot, and in particular look at the case where we have extremely small numbers of examples, in a task-id-free setting. Specifically, we propose SANER, a method for continuously learning a library of skills, and ABIP (Attention-Based Interaction Policies) as the backbone to support it. We learn four sequential kitchen tasks on a low-cost home robot, using only a handful of demonstrations per task.

cs.RO↗

arXiv2023

Spatial-Language Attention Policies for Efficient Robot Learning

Despite great strides in language-guided manipulation, existing work has been constrained to table-top settings. Table-tops allow for perfect and consistent camera angles, properties are that do not hold in mobile manipulation. Task plans that involve moving around the environment must be robust to egocentric views and changes in the plane and angle of grasp. A further challenge is ensuring this is all true while still being able to learn skills efficiently from limited data. We propose Spatial-Language Attention Policies (SLAP) as a solution. SLAP uses three-dimensional tokens as the input representation to train a single multi-task, language-conditioned action prediction policy. Our method shows an 80% success rate in the real world across eight tasks with a single model, and a 47.5% success rate when unseen clutter and unseen object configurations are introduced, even with only a handful of examples per task. This represents an improvement of 30% over prior work (20% given unseen distractors and configurations). We see a 4x improvement over baseline in mobile manipulation setting. In addition, we show how SLAPs robustness allows us to execute Task Plans from open-vocabulary instructions using a large language model for multi-step mobile manipulation. For videos, see the website: https://robotslap.github.io

cs.RO↗

arXiv2023

A statistical model for quantum spin and photon number states

The most irreducible way to represent information is a sequence of two symbols. In this paper, we construct quantum states using this basic building block. Specifically, we show that the probabilities that arise in quantum theory can be reduced to counting more fundamental ontic states, which we interpret as event networks and model using sequences of 0's and 1's. A completely self contained formalism is developed for the purpose of organizing and counting these ontic states, which employs the finite cyclic group $\mathbb{Z}_2 = \{0, 1\}$, basic set theory, and combinatorics. This formalism is then used to calculate probability distributions associated with particles of arbitrary spin interacting with sequences of two rotated Stern-Gerlach detectors. A central ingredient of this construction is the rule which converts the abstract counts labelling an operation into the physical angle of rotation it represents. We show that this rule is not linear in the counts, but is instead fixed by the half-angle law $\tan(\theta_{ab}/2)=\tilde{B}_{map}/\tilde{A}_{map}$, which is the unique assignment consistent with the composition of successive rotations. These calculations are compared with the predictions of non-relativistic quantum mechanics and shown to agree exactly in the limit of large sequence length $n$, with finite $n$ corrections which vanish as $O(1/n)$ and with no free or fitted parameters. The residual deviation at finite $n$ does not lead to violations of relevant no-go theorems, such as Bell's inequalities, the Kochen-Specker theorem, or the PBR theorem. The proposed model is then extended to an optical system involving photon number states passing through a beam splitter. Leveraging recent advancements in high precision experiments on these systems, we then propose a means of testing the new model using a tabletop experiment.

quant-ph↗

arXiv2022

Self-Activating Neural Ensembles for Continual Reinforcement Learning

The ability for an agent to continuously learn new skills without catastrophically forgetting existing knowledge is of critical importance for the development of generally intelligent agents. Most methods devised to address this problem depend heavily on well-defined task boundaries, and thus depend on human supervision. Our task-agnostic method, Self-Activating Neural Ensembles (SANE), uses a modular architecture designed to avoid catastrophic forgetting without making any such assumptions. At the beginning of each trajectory, a module in the SANE ensemble is activated to determine the agent's next policy. During training, new modules are created as needed and only activated modules are updated to ensure that unused modules remain unchanged. This system enables our method to retain and leverage old skills, while growing and learning new ones. We demonstrate our approach on visually rich procedurally generated environments.

cs.LG↗

arXiv2021

An alternative formalism for modeling spin

We present an alternative formalism for modeling spin. The ontological elements of this formalism are base-2 sequences of length $n$. The machinery necessary to model physics is then developed by considering correlations between base-2 sequences. Upon choosing a reference base-2 sequence, a relational system of numbers can be defined, which we interpret as quantum numbers. Based on the properties of these relational quantum numbers, the selection rules governing interacting spin systems are derived from first principles. A tool for calculating the associated probabilities, which are the squared Clebsch-Gordan coefficients in quantum mechanics, is also presented. The resulting model offers a vivid information theoretic picture of spin and interacting spin systems. Importantly, this model is developed without making any assumptions about the nature of space-time, which presents an interesting opportunity to study emergent space-time models.

quant-ph↗

arXiv2021

CORA: Benchmarks, Baselines, and Metrics as a Platform for Continual Reinforcement Learning Agents

Progress in continual reinforcement learning has been limited due to several barriers to entry: missing code, high compute requirements, and a lack of suitable benchmarks. In this work, we present CORA, a platform for Continual Reinforcement Learning Agents that provides benchmarks, baselines, and metrics in a single code package. The benchmarks we provide are designed to evaluate different aspects of the continual RL challenge, such as catastrophic forgetting, plasticity, ability to generalize, and sample-efficient learning. Three of the benchmarks utilize video game environments (Atari, Procgen, NetHack). The fourth benchmark, CHORES, consists of four different task sequences in a visually realistic home simulator, drawn from a diverse set of task and scene parameters. To compare continual RL methods on these benchmarks, we prepare three metrics in CORA: Continual Evaluation, Isolated Forgetting, and Zero-Shot Forward Transfer. Finally, CORA includes a set of performant, open-source baselines of existing algorithms for researchers to use and expand on. We release CORA and hope that the continual RL community can benefit from our contributions, to accelerate the development of new continual RL algorithms.

cs.LG↗