We prove an exponential separation between zero-sided-error randomized and two-sided-error pseudodeterministic communication complexities of partial boolean functions. This qualitatively improves and simplifies the proof of the separation by Göös, Harms, Riazanov, Sofronova, Sokolov, and Yuan (STOC 2026), which had a two-sided-error randomized upper bound. Then, we generalize our technique to separate the communication complexity analogues of MA and NP^BPP.

In an influential paper, Erd\H{o}s and Selfridge introduced the Maker-Breaker game played on a hypergraph, or equivalently, on a monotone CNF. The players take turns assigning values to variables of their choosing, and Breaker's goal is to satisfy the CNF, while Maker's goal is to falsify it. The Erd\H{o}s-Selfridge Theorem says that the least number of clauses in any monotone CNF with $k$ literals per clause where Maker has a winning strategy is $\Theta(2^k)$. We study the analogous question when the CNF is not necessarily monotone. We prove bounds of $\Theta(\sqrt{2}\,^k)$ when Maker plays last, and $\Omega(1.5^k)$ and $O(r^k)$ when Breaker plays last, where $r=(1+\sqrt{5})/2\approx 1.618$ is the golden ratio.

The classic TQBF problem can be viewed as a game in which two players alternate turns assigning truth values to a CNF formula's variables in a prescribed order, and the winner is determined by whether the CNF gets satisfied. The complexity of deciding which player has a winning strategy in this game is well-understood: it is NL-complete for 2-CNFs and PSPACE-complete for 3-CNFs. We continue the study of the unordered variant of this game, in which each turn consists of picking any remaining variable and assigning it a truth value. The complexity of deciding who can win on a given CNF is less well-understood; prior work by the authors showed it is in L for 2-CNFs and PSPACE-complete for 5-CNFs. We conjecture it may be efficiently solvable on 3-CNFs, and we make progress in this direction by proving the problem is in P, indeed in L, for 3-CNFs with a certain restriction, namely that each width-3 clause has at least one variable that appears in no other clause. Another (incomparable) restriction of this problem was previously shown to be tractable by Kutz.

Unmanned aerial systems (UAS) are a remarkably capable technology, and their untold potential scales greatly when multiple units work towards a common objective. However, current wireless and networking architectures limit the capabilities and applications of using multiple UAS primarily to pre-programmed routines - suppressing the advantages of multiple unmanned and autonomous platforms. Ad-hoc communications can unlock the true potential of unmanned and autonomous systems by enabling efficient and direct peer-to-peer data exchange and removing the system's reliance on infrastructure - enabling them to operate anywhere. Unfortunately, ad-hoc communication is poorly supported by current wireless hardware and standards which are designed for maximized throughput and minimal latency over point-to-point links at the cost of much worse throughput and latency over multicast channels. Attempts to integrate multi-radio and cognitive radio technologies into the wireless ecosystem to increase device inter-connectivity have faced many difficulties because coordinating network resources and implementing a routing algorithm that can take all the involved variables into consideration is a very difficult task in host-centric networking. Our research uses Named Data Networking (NDN), a data-centric Internet architecture that uniquely identifies and retrieves data by name directly, to grant individual nodes the capability to make practical forwarding decisions without the added overhead of complex and centralized routing and coordination algorithms. Each node receives and gathers network information which then enables it to perform intelligent per-hop forwarding decisions toward names that identify data instead of addresses that identify hosts. By understanding what kind of data the node is handling, the requirements to successfully deliver that data, and its standing among other nodes, it can concurrently send data across multiple data-specific channels and radios to ensure quick, reliable, and non-interfering delivery. This will allow the network to effectively utilize the resources available to it and scale rapidly.

Magnetic sensing has applications in diverse technological domains including geology and geophysics, structural inspection, geospatial navigation, and vehicle detection and avoidance. At-range detection of large assets such as vehicles, specifically, can be executed using UAV-deployed magnetometers to safely determine the presence or identity of vehicles even if they are visually obscured. However, several challenges must be overcome and, in particular, the electromagnetic noise generated by the aircraft motors and other onboard electronics sets the signal-to-noise ratio and thereby limits the effective range of such an application. Here, we present UAV integration with a compact optically pumped magnetometer (QuSpin QTFM) and characterize its performance in vehicle sensing from the drone platform. We first characterize magnetic noise produced by the drone rotors to understand the detection limits it imposes on SNR. We then demonstrate vehicle detection at standoff distances of 5 and 10m. We utilize a volumetric magnetic field background subtraction method in order to enhance vehicle detection, and we characterize the impact of UAV velocity on field detection.

Infrared pilotage sensors enhance pilots' situational awareness, aiding in obstacle avoidance and providing visibility during night flights or under degraded visual environments. Pilotage with a rotorcraft tends to be more difficult than with fixed-wingcraft as rotorcraft generally fly at a lower altitude, which increases the angular velocities of the ground below and other features nearby with respect to the pilotage sensor. This increase in the scene's angular velocity increases motion blur in the imagery. This increased blur due to integration time, or time constant, lowers the system's Modulation Transfer Function (MTF), thus reducing the Target Task Performance (TTP) metric. This lower TTP also corresponds to a decrease in the system's pilotage performance. There has not been a straightforward method for including motion blur in the degradation of the TTP metric for pilotage performance. In this study, data is collected from a helicopter from different perspectives, using well-characterized cameras to capture the two levels of motion blur in pilotage imagery from two different looking angles. The imagery and Inertial Measurement Unit (IMU) data is then used to calculate the amount of angular motion and jitter in milliradians, which is then used to calculate a new degraded MTF and TTP metric. The addition of motion blur into the TTP metric for pilotage is essential in accurately predicting and evaluating the pilotage performance of systems on platforms that are moving quickly and have significant sensor time constant blur.

In a STOC 1976 paper, Schaefer proved that it is PSPACE-complete to determine the winner of the so-called Maker-Breaker game on a given set system, even when every set has size at most 11. Since then, there has been no improvement on this result. We prove that the game remains PSPACE-complete even when every set has size 6.

Small Unmanned Aerial Systems (sUAS) provide a versatile platform for covering large areas quickly. By adding sensors to these drones, imagery of large areas can be taken for a variety of applications. Traditionally, a fixed staring system or a gimballed sensor is used to take this imagery. Both options require a compromise between field of view (FOV), resolution, scanning speed and flight path to properly perform the task at hand. With more than one type of sensing, additional information can be collected about imaged environments. If more than one sensor is integrated onto the drone, a wide FOV can still be covered without a scanning gimbal and with higher resolution than a traditional wide FOV system. Presented is a multi-camera, multi-wavelength design approach based on a constraining ground sample distance (GSD) for a wide area coverage (WAC) system. A figure of merit (FoM) is created to quantify and compare the performance of the WAC systems in the visible (0.4-0.7um), short wave infrared (1.0-1.7um) and longwave infrared (8-14um) in both good and bad visibility conditions. The performance of three optimized and fabricated WAC systems are compared and tested. The testing results of the flown fabricated systems show that the design approach described delivers the expected results.

We provide a complete picture of the extent to which amplificationof success probability is possible for randomized algorithmshaving access to one NP oracle query, in the settings of two-sided, onesided,and zero-sided error. We generalize this picture to amplifyingone-query algorithms with q-query algorithms, and we show our inclusionsare tight for relativizing techniques.

Autonomous micromobility has been attracting the attention of researchers and practitioners in recent years. A key component of many micro-transport vehicles is the DC motor, a complex dynamical system that is continuous and non-linear. Learning to quickly control the DC motor in the presence of disturbances and uncertainties is desired for various applications that require robustness and stability. Techniques to accomplish this task usually rely on a mathematical system model, which is often insufficient to anticipate the effects of time-varying and interrelated sources of non-linearities. While some model-free approaches have been successful at the task, they rely on massive interactions with the system and are trained in specialized hardware in order to fit a highly parameterized controller. In this work, we learn to steer a DC motor via sample-efficient reinforcement learning. Using data collected from hardware interactions in the real world, we additionally build a simulator to experiment with a wide range of parameters and learning strategies. With the best parameters found, we learn an effective control policy in one minute and 53 seconds on a simulation and in 10 minutes and 35 seconds on a physical system.

We investigate the power of randomness in two-party communicationcomplexity. In particular, we study the model where theparties can make a constant number of queries to a function that has anefficient one-sided-error randomized protocol. The complexity classesdefined by this model comprise the Randomized Boolean Hierarchy,which is analogous to the Boolean Hierarchy but defined with one-sidederrorrandomness instead of nondeterminism. Our techniques connectthe Nondeterministic and Randomized Boolean Hierarchies, and we providea complete picture of the relationships among complexity classeswithin and across these two hierarchies. In particular, we prove thatthe Randomized Boolean Hierarchy does not collapse, and we prove aquery-to-communication lifting theorem for all levels of the NondeterministicBoolean Hierarchy and use it to resolve an open problem statedin the paper by Halstenberg and Reischuk (CCC 1988) which initiatedthe study of this hierarchy.

An accurate model of building interiors with detailed annotations is critical to protecting the first responders’ safety and building occupants during emergency operations. In collaboration with the City of Memphis, we collected extensive LiDAR and image data for the city’s buildings. We apply machine learning techniques to detect and classify objects of interest for first responders and create a comprehensive 3D indoor space database with annotated safety-related objects. This paper documents the challenges we encountered in data collection and processing, and it presents a complete 3D mapping and labeling system for the environments inside and adjacent to buildings. Moreover, we use a case study to illustrate our process and show preliminary evaluation

Model free techniques have been successful at optimal control of complex systems at an expense of copious amounts of data and computation. However, it is often desired to obtain a control policy in a short period of time with minimal data use and computational burden. To this end, we make use of the NFQ algorithm for steering position control of a golf cart in both a real hardware and a simulated environment that was built from real-world interaction. The controller learns to apply a sequence of voltage signals in the presence of environmental uncertainties and inherent non-linearities that challenge the the control task. We were able to increase the rate of successful control under four minutes in simulation and under 11 minutes in real hardware.

Suppose we have randomized decision trees for an outer function $f$ and an inner function $g$. The natural approach for obtaining a randomized decision tree for the composed function $(f\circ g^n)(x^1,\ldots,x^n)=f(g(x^1),\ldots,g(x^n))$ involves amplifying the success probability of the decision tree for $g$, so that a union bound can be used to bound the error probability over all the coordinates. The amplification introduces a logarithmic factor cost overhead. We study the question: When is this log factor necessary? We show that when the outer function is parity or majority, the log factor can be necessary, even for models that are more powerful than plain randomized decision trees. Our results are related to, but qualitatively strengthen in various ways, known results about decision trees with noisy inputs.

Authors would like to correct the incorrect author references in theonline published article.

Suppose Alice and Bob each start with private randomness and no other input, and they wish to engage in a protocol in which Alice ends up with a set x⊆ [n] and Bob ends up with a set y⊆ [n], such that (x,y) is uniformly distributed over all pairs of disjoint sets. We prove that for some constant β < 1, this requires Ω (n) communication even to get within statistical distance 1− βn of the target distribution. Previously, Ambainis, Schulman, Ta-Shma, Vazirani, and Wigderson (FOCS 1998) proved that Ω (√n) communication is required to get within some constant statistical distance ɛ > 0 of the uniform distribution over all pairs of disjoint sets of size √n.

The complexity class ZPP NP[1] (corresponding to zero-error randomized algorithms with access to one NP oracle query) is known to have a number of curious properties. We further explore this class in the settings of time complexity, query complexity, and communication complexity. • For starters, we provide a new characterization: ZPP NP[1] equals the restriction of BPP NP[1] where the algorithm is only allowed to err when it forgoes the opportunity to make an NP oracle query. • Using the above characterization, we prove a query-to-communication lifting theorem , which translates any ZPP NP[1] decision tree lower bound for a function f into a ZPP NP[1] communication lower bound for a two-party version of f . • As an application, we use the above lifting theorem to prove that the ZPP NP[1] communication lower bound technique introduced by Göös, Pitassi, and Watson (ICALP 2016) is not tight. We also provide a “primal” characterization of this lower bound technique as a complexity class.

The classic TQBF problem is to determine who has a winning strategy in a game played on a given conjunctive normal form formula (CNF), where the two players alternate turns picking truth values for the variables in a given order, and the winner is determined by whether the CNF gets satisfied. We study variants of this game in which the variables may be played in any order, and each turn consists of picking a remaining variable and a truth value for it. For the version where the set of variables is partitioned into two halves and each player may only pick variables from his or her half, we prove that the problem is PSPACE-complete for 5-CNFs and in P for 2-CNFs. Previously, it was known to be PSPACE-complete for unbounded-width CNFs (Schaefer, STOC 1976). For the general unordered version (where each variable can be picked by either player), we also prove that the problem is PSPACE-complete for 5-CNFs and in P for 2-CNFs. Previously, it was known to be PSPACE-complete for 6-CNFs (Ahlroth and Orponen, MFCS 2012) and PSPACE-complete for positive 11-CNFs (Schaefer, STOC 1976).

We study problems in randomized communication complexity when the protocol is only required to attain some small advantage over purely random guessing, i.e., it produces the correct output with probability at least $$\epsilon$$ greater than one over the codomain size of the function. Previously, Braverman and Moitra (in: Proceedings of the 45th symposium on theory of computing (STOC), ACM, pp 161–170, 2013) showed that the set-intersection function requires $$\Theta(\epsilon{n})$$ communication to achieve advantage $$\epsilon$$ . Building on this, we prove the same bound for several variants of set-intersection: (1) the classic “tribes” function obtained by composing with And (provided $$1/\epsilon$$ is at most the width of the And), and (2) the variant where the sets are uniquely intersecting and the goal is to determine partial information about (say, certain bits of the index of) the intersecting coordinate.

We prove that the PNP-type query complexity (alternatively, decision list width) of any Boolean function f is quadratically related to the PNP-type communication complexity of a lifted version of f. As an application, we show that a certain “product” lower bound method of Impagliazzo and Williams (CCC 2010) fails to capture PNP communication complexity up to polynomial factors, which answers a question of Papakonstantinou, Scheder, and Song (CCC 2014).