Verifying the fully kinematic nature of the cosmic microwave background (CMB) dipole is of fundamental importance in cosmology. In the standard cosmological model with the Friedman-Lemaitre-Robertson-Walker (FLRW) metric from the inflationary expansion the CMB dipole should be entirely kinematic. Any non-kinematic CMB dipole component would thus reflect the preinflationary structure of spacetime probing the extent of the FLRW applicability. Cosmic backgrounds from galaxies after the matter-radiation decoupling, should have kinematic dipole component identical in velocity with the CMB kinematic dipole. Comparing the two can lead to isolating the CMB non-kinematic dipole. It was recently proposed that such measurement can be done using the near-IR cosmic infrared background (CIB) measured with the currently operating Euclid telescope, and later with Roman. The proposed method reconstructs the resolved CIB, the Integrated Galaxy Light (IGL), from Euclid's Wide Survey and probes its dipole, with a kinematic component amplified over that of the CMB by the Compton-Getting effect. The amplification coupled with the extensive galaxy samples forming the IGL would determine the CIB dipole with an overwhelming signal/noise, isolating its direction to sub-degree accuracy. We develop details of the method for Euclid's Wide Survey in 4 bands spanning 0.6 to 2 mic. We isolate the systematic and other uncertainties and present methodologies to minimize them, after confining the sample to the magnitude range with negligible IGL/CIB dipole from galaxy clustering. These include the required star-galaxy separation, accounting for the extinction correction dipole using the method newly developed here achieving total separation, accounting for the Earth's orbital motion and other systematic effects. (Abridged)

Language models (LMs) may memorize personally identifiable information (PII) from training data, enabling adversaries to extract it during inference. Existing defense mechanisms such as differential privacy (DP) reduce this leakage, but incur large drops in utility. Based on a comprehensive study using circuit discovery to identify the computational circuits responsible PII leakage in LMs, we hypothesize that specific PII leakage circuits in LMs should be responsible for this behavior. Therefore, we propose PATCH (Privacy-Aware Targeted Circuit PatcHing), a novel approach that first identifies and subsequently directly edits PII circuits to reduce leakage. PATCH achieves better privacy-utility trade-off than existing defenses, e.g., reducing recall of PII leakage from LMs by up to 65%. Finally, PATCH can be combined with DP to reduce recall of residual leakage of an LM to as low as 0.01%. Our analysis shows that PII leakage circuits persist even after the application of existing defense mechanisms. In contrast, PATCH can effectively mitigate their impact.

Before executing a quantum algorithm, one must first decompose the algorithm into machine-level instructions compatible with the architecture of the quantum computer, a process known as quantum compiling. There are many different quantum circuit decompositions for the same algorithm but it is desirable to compile leaner circuits. A fundamentally important cost metric is the $T$ count -- the number of $T$ gates in a circuit. For the single qubit case, optimal compiling is essentially a solved problem. However, multi-qubit compiling is a harder problem with optimal algorithms requiring classical runtime exponential in the number of qubits. Here, we present and compare several efficient quantum compilers for multi-qubit Clifford + $T$ circuits. We implemented our compilers in C++ and benchmarked them on random circuits, from which we determine that our TODD compiler yields the lowest $T$ counts on average. We also benchmarked TODD on a library of reversible logic circuits that appear in quantum algorithms and found that it reduced the $T$ count for 97\% of the circuits with an average $T$-count saving of 20\% when compared against the best of all previous circuit decompositions.

Autoformalisation, the task of expressing informal mathematical statements in formal language, is often viewed as a direct translation process. This, however, disregards a critical preceding step: conjecturing. Many mathematical problems cannot be formalised directly without first conjecturing a conclusion such as an explicit answer, or a specific bound. Since Large Language Models (LLMs) already struggle with autoformalisation, and the evaluation of their conjecturing ability is limited and often entangled within autoformalisation or proof, it is particularly challenging to understand its effect. To address this gap, we augment existing datasets to create ConjectureBench, and redesign the evaluation framework and metric specifically to measure the conjecturing capabilities of LLMs both as a distinct task and within the autoformalisation pipeline. Our evaluation of foundational models, including GPT-4.1 and DeepSeek-V3.1, reveals that their autoformalisation performance is substantially overestimated when the conjecture is accounted for during evaluation. However, the conjecture should not be assumed to be provided. We design an inference-time method, Lean-FIRe to improve conjecturing and autoformalisation, which, to the best of our knowledge, achieves the first successful end-to-end autoformalisation of 13 PutnamBench problems with GPT-4.1 and 7 with DeepSeek-V3.1. We demonstrate that while LLMs possess the requisite knowledge to generate accurate conjectures, improving autoformalisation performance requires treating conjecturing as an independent task, and investigating further how to correctly integrate it within autoformalisation. Finally, we provide forward-looking guidance to steer future research toward improving conjecturing, an overlooked step of formal mathematical reasoning.

Laws and their interpretations, legal arguments and agreements\ are typically expressed in writing, leading to the production of vast corpora of legal text. Their analysis, which is at the center of legal practice, becomes increasingly elaborate as these collections grow in size. Natural language understanding (NLU) technologies can be a valuable tool to support legal practitioners in these endeavors. Their usefulness, however, largely depends on whether current state-of-the-art models can generalize across various tasks in the legal domain. To answer this currently open question, we introduce the Legal General Language Understanding Evaluation (LexGLUE) benchmark, a collection of datasets for evaluating model performance across a diverse set of legal NLU tasks in a standardized way. We also provide an evaluation and analysis of several generic and legal-oriented models demonstrating that the latter consistently offer performance improvements across multiple tasks.

Neuro-symbolic NLP methods aim to leverage the complementary strengths of large language models and formal logical solvers. However, current approaches are mostly static in nature, i.e., the integration of a target solver is predetermined at design time, hindering the ability to employ diverse formal inference strategies. To address this, we introduce an adaptive, multi-paradigm, neuro-symbolic inference framework that: (1) automatically identifies formal reasoning strategies from problems expressed in natural language; and (2) dynamically selects and applies specialized formal logical solvers via autoformalization interfaces. Extensive experiments on individual and multi-paradigm reasoning tasks support the following conclusions: LLMs are effective at predicting the necessary formal reasoning strategies with an accuracy above 90 percent. This enables flexible integration with formal logical solvers, resulting in our framework outperforming competing baselines by 27 percent and 6 percent compared to GPT-4o and DeepSeek-V3.1, respectively. Moreover, adaptive reasoning can even positively impact pure LLM methods, yielding gains of 10, 5, and 6 percent on zero-shot, CoT, and symbolic CoT settings with GPT-4o. Finally, although smaller models struggle with adaptive neuro-symbolic reasoning, post-training offers a viable path to improvement. Overall, this work establishes the foundations for adaptive LLM-symbolic reasoning, offering a path forward for unifying material and formal inferences on heterogeneous reasoning challenges.

Large-scale quantum computers have the potential to hold computational capabilities beyond conventional computers for certain problems. However, the physical qubits within a quantum computer are prone to noise and decoherence, which must be corrected in order to perform reliable, fault-tolerant quantum computations. Quantum Error Correction (QEC) provides the path for realizing such computations. QEC continuously generates a continuous stream of data that decoders must process at the rate it is received, which can be as fast as 1 MHz in superconducting quantum computers. A little known fact of QEC is that if the decoder infrastructure cannot keep up, a data backlog problem is encountered and the quantum computer runs exponentially slower. Today's leading approaches to quantum error correction are not scalable as existing decoders typically run slower as the problem size is increased, inevitably hitting the backlog problem. That is: the current leading proposal for fault-tolerant quantum computation is not scalable. Here, we show how to parallelize decoding to achieve almost arbitrary speed, removing this roadblock to scalability. Our parallelization requires some classical feed forward decisions to be delayed, leading to a slow-down of the logical clock speed. However, the slow-down is now only polynomial in code size, averting the exponential slowdown. We numerically demonstrate our parallel decoder for the surface code, showing no noticeable reduction in logical fidelity compared to previous decoders and demonstrating the parallelization speedup.