arXiv subjects

Neel Patel

Publications and source records attributed to Neel Patel.

33 records · Page 2Linked to original sources

arXiv2022

Delegated Pandora's box

In delegation problems, a principal does not have the resources necessary to complete a particular task, so they delegate the task to an untrusted agent whose interests may differ from their own. Given any family of such problems and space of mechanisms for the principal to choose from, the delegation gap is the worst-case ratio of the principal's optimal utility when they delegate versus their optimal utility when solving the problem on their own. In this work, we consider the delegation gap of the generalized Pandora's box problem, a search problem in which searching for solutions incurs known costs and solutions are restricted by some downward-closed constraint. First, we show that there is a special case when all random variables have binary support for which there exist constant-factor delegation gaps for matroid constraints. However, there is no constant-factor delegation gap for even simple non-binary instances of the problem. Getting around this impossibility, we consider two variants: the free-agent model, in which the agent doesn't pay the cost of probing elements, and discounted-cost approximations, in which we discount all costs and aim for a bicriteria approximation of the discount factor and delegation gap. We show that there are constant-factor delegation gaps in the free-agent model with discounted-cost approximations for certain downward closed constraints and constant discount factors. However, constant delegation gaps can not be achieved under either variant alone. Finally, we consider another variant called the shared-cost model, in which the principal can choose how costs will be shared between them and the agent before delegating the search problem. We show that the shared-cost model exhibits a constant-factor delegation gap for certain downward closed constraints.

cs.GT↗

arXiv2021

Rigidity of Acute Angled Corners for One Phase Muskat Interfaces

We consider the one-phase Muskat problem modeling the dynamics of the free boundary of a single fluid in porous media. We prove local well-posedness for fluid interfaces that are general curves and can have singularities. In particular, the free boundary can have acute angle corners or cusps. Moreover, we show that isolated corners/cusps on the interface must be rigid, meaning the angle of the corner is preserved for a finite time, there is no rotation at the tip, the particle at the tip remains at the tip and the velocity of that particle at the tip points vertically downward.

math.AP↗

arXiv2021

Global Results for the Inhomogeneous Muskat Problem

The inhomogeneous Muskat problem models the dynamics of an interface between two fluids of differing characteristics inside a non-uniform porous medium. We consider the case of a porous media with a permeability jump across a horizontal boundary away from an interface between two fluids of different viscosities and densities. For initial data of explicit medium size, depending on the characteristics of the fluids and porous media, we will prove the global existence and uniqueness of a solution which is instantly analytic and decays in time to the flat interface.

math.AP↗

arXiv2021

The Price is (Probably) Right: Learning Market Equilibria from Samples

Equilibrium computation in markets usually considers settings where player valuation functions are known. We consider the setting where player valuations are unknown; using a PAC learning-theoretic framework, we analyze some classes of common valuation functions, and provide algorithms which output direct PAC equilibrium allocations, not estimates based on attempting to learn valuation functions. Since there exist trivial PAC market outcomes with an unbounded worst-case efficiency loss, we lower-bound the efficiency of our algorithms. While the efficiency loss under general distributions is rather high, we show that in some cases (e.g., unit-demand valuations), it is possible to find a PAC market equilibrium with significantly better utility.

cs.GT↗

arXiv2021

Global Regularity for Gravity Unstable Muskat Bubbles

In this paper, we study the dynamics of fluids in porous media governed by Darcy's law: the Muskat problem. We consider the setting of two immiscible fluids of different densities and viscosities under the influence of gravity in which one fluid is completely surrounded by the other. This setting is gravity unstable because along a portion of the interface, the denser fluid must be above the other. Surprisingly, even without capillarity, the circle-shaped bubble is a steady state solution moving with vertical constant velocity determined by the density jump between the fluids. Taking advantage of our discovery of this steady state, we are able to prove global in time existence and uniqueness of dynamic bubbles of nearly circular shapes under the influence of surface tension. We prove this global existence result for low regularity initial data. Moreover, we prove that these solutions are instantly analytic and decay exponentially fast in time to the circle.

math.AP↗

arXiv2021

Well-posedness for SQG sharp fronts with unbounded curvature

Patch solutions for the surface quasigeostrophic (SQG) equation model sharp temperature fronts in atmospheric and oceanic flows. We establish local well-posedness for SQG sharp fronts of low Sobolev regularity, $H^{2+s}$ for arbitrarily small $s>0$, allowing for fronts with unbounded curvature.

math.AP↗

arXiv2021

High Dimensional Model Explanations: an Axiomatic Approach

Complex black-box machine learning models are regularly used in critical decision-making domains. This has given rise to several calls for algorithmic explainability. Many explanation algorithms proposed in literature assign importance to each feature individually. However, such explanations fail to capture the joint effects of sets of features. Indeed, few works so far formally analyze high-dimensional model explanations. In this paper, we propose a novel high dimension model explanation method that captures the joint effect of feature subsets. We propose a new axiomatization for a generalization of the Banzhaf index; our method can also be thought of as an approximation of a black-box model by a higher-order polynomial. In other words, this work justifies the use of the generalized Banzhaf index as a model explanation by showing that it uniquely satisfies a set of natural desiderata and that it is the optimal local approximation of a black-box model. Our empirical evaluation of our measure highlights how it manages to capture desirable behavior, whereas other measures that do not satisfy our axioms behave in an unpredictable manner.

cs.LG↗

arXiv2020

Model Explanations with Differential Privacy

Black-box machine learning models are used in critical decision-making domains, giving rise to several calls for more algorithmic transparency. The drawback is that model explanations can leak information about the training data and the explanation data used to generate them, thus undermining data privacy. To address this issue, we propose differentially private algorithms to construct feature-based model explanations. We design an adaptive differentially private gradient descent algorithm, that finds the minimal privacy budget required to produce accurate explanations. It reduces the overall privacy loss on explanation data, by adaptively reusing past differentially private explanations. It also amplifies the privacy guarantees with respect to the training data. We evaluate the implications of differentially private models and our privacy mechanisms on the quality of model explanations.

cs.LG↗

arXiv2020

Keeping Your Friends Close: Land Allocation with Friends

We examine the problem of assigning plots of land to prospective buyers who prefer living next to their friends. They care not only about the plot they receive, but also about their neighbors. This externality results in a highly non-trivial problem structure, as both friendship and land value play a role in determining agent behavior. We examine mechanisms that guarantee truthful reporting of both land values and friendships. We propose variants of random serial dictatorship (RSD) that can offer both truthfulness and welfare guarantees. Interestingly, our social welfare guarantees are parameterized by the value of friendship: if these values are low, enforcing truthful behavior results in poor welfare guarantees and imposes significant constraints on agents' choices; if they are high, we achieve good approximation to the optimal social welfare.

cs.GT↗

arXiv2020

Discrete Richman-bidding Scoring Games

We study zero-sum (combinatorial) games, within the framework of so-called Richman auctions (Lazarus et al. 1996) namely, we modify the alternating play scoring ruleset Cumulative Subtraction (CS) (Cohensius et al. 2019), to a discrete bidding scheme (similar to Develin and Payne 2010). Players bid to move and the player with the highest bid wins the move, and hands over the winning bidding amount to the other player. The new game is dubbed Bidding Cumulative Subtraction (BCS). In so-called unitary games, players remove exactly one item out of a single heap of identical items, until the heap is empty, and their actions contribute to a common score, which increases or decreases by one unit depending on whether the maximizing player won the turn or not. We show that there is a unique bidding equilibrium for a much larger class of games, that generalize standard scoring play in the literature. We prove that for all sufficiently large heap sizes, the equilibrium outcomes of unitary BCS are eventually periodic, with period 2, and we show that the periodicity appears at the latest for heaps of sizes quadratic in the total budget.

cs.GT↗

arXiv2019

On the local existence and blow-up for generalized SQG patches

We study patch solutions of a family of transport equations given by a parameter $α$, $0< α<2$, with the cases $α=0$ and $α=1$ corresponding to the Euler and the surface quasi-geostrophic equations respectively. In this paper, using several new cancellations, we provide the following new results. First, we prove local well-posedness for $H^{2}$ patches in the half-space setting for $0<α< 1/3$, allowing self-intersection with the fixed boundary. Furthermore, we are able to extend the range of $α$ for which finite time singularities have been shown in \cite{KYZ} and \cite{KRYZ}. Second, we establish that patches remain regular for $0<α<2$ as long as the arc-chord condition and the regularity of order $C^{1+δ}$ for $δ>α/2$ are time integrable. This finite-time singularity criterion holds for lower regularity than the regularity shown in numerical simulations in \cite{CFMR} and \cite{ScottDritschel} for surface quasi-geostrophic patches, where the curvature of the contour blows up numerically. This is the first proof of a finite-time singularity criterion lower than or equal to the regularity in the numerics. Finally, we also improve results in \cite{G} and in \cite{CCCGW}, giving local-wellposedness for patches in $H^{2}$ for $0<α< 1$ and in $H^3$ for $1<α<2$.

math.AP↗

arXiv2017

On The Muskat Problem With Viscosity Jump: Global In Time Results

The Muskat problem models the filtration of two incompressible immiscible fluids of different characteristics in porous media. In this paper, we consider both the 2D and 3D setting of two fluids of different constant densities and different constant viscosities. In this situation, the related contour equations are non-local, not only in the evolution system, but also in the implicit relation between the amplitude of the vorticity and the free interface. Among other extra difficulties, no maximum principles are available for the amplitude and the slopes of the interface in $L^\infty$. We prove global in time existence results for medium size initial stable data in critical spaces. We also enhance previous methods by showing smoothing (instant analyticity), improving the medium size constant in 3D, together with sharp decay rates of analytic norms. The found technique is twofold, giving ill-posedness in unstable situations for very low regular solutions.

math.AP↗

arXiv2016

Large Time Decay Estimates for the Muskat Equation

We prove time decay of solutions to the Muskat equation in 2D and in 3D. In \cite{JEMS} and \cite{CCGRPS}, the authors introduce the norms $\|f\|_{s}(t)= \int_{\mathbb{R}^{2}} |ξ|^{s}|\hat{f}(ξ)| \ dξ$ in order to prove global existence of solutions to the Muskat problem. In this paper, for the 3D Muskat problem, given initial data $f_{0}\in H^{l}(\mathbb{R}^{2})$ for some $l\geq 3$ such that $\|f_{0}\|_{1} < k_{0}$ for a constant $k_{0} \approx 1/5$, we prove uniform in time bounds of $\|f\|_{s}(t)$ for $-d < s < l-1$ and assuming $\|f_{0}\|_ν < \infty$ we prove time decay estimates of the form $\|f\|_{s}(t) \lesssim (1+t)^{-s+ν}$ for $0 \leq s \leq l-1$ and $-d \leq ν< s$. These large time decay rates are the same as the optimal rate for the linear Muskat equation. We also prove analogous results in 2D.

math.AP↗

arXiv2016

Three New Results on Continuation Criteria for the 3D Relativistic Vlasov-Maxwell System

In this paper, we consider sufficient conditions, called continuation criteria, for global existence and uniqueness of classical solutions to the three-dimensional relativistic Vlasov-Maxwell system. In the compact momentum support setting, we prove that $\|p_{0}^{\frac{18}{5r} - 1+β}f\|_{L^{\infty}_{t}L^{r}_{x}L^{1}_{p}} \lesssim 1$, where $1\leq r \leq 2$ and $β>0$ is arbitrarily small, is a continuation criteria. The previously best known continuation criteria in the compact setting is $\|p_{0}^{\frac{4}{r} - 1+β}f\|_{L^{\infty}_{t}L^{r}_{x}L^{1}_{p}} \lesssim 1$, where $1\leq r < \infty$ and $β>0$ is arbitrarily small, is due to work by Kunze. Thus, our continuation criteria is an improvement in the $1\leq r \leq 2$ range. In addition, we consider also sufficient conditions for a global existence result to the three-dimensional relativistic Vlasov-Maxwell system without compact support in momentum space, building on previous work by Luk-Strain. In their work, it was shown that $\|p_{0}^θf\|_{L^{1}_{x}L^{1}_{p}} \lesssim 1$ is a continuation criteria for the relativistic Vlasov-Maxwell system without compact support in momentum space for $θ> 5$. We improve this result to $θ> 3$. We also build on another result by Luk-Strain, in which the authors proved the existence of a global classical solution in the compact momentum support setting given the condition that there exists a two-dimensional plane on which the momentum support of the particle density remains fixed. We prove well-posedness even if the plane varies continuously in time.

math.AP↗

arXiv2015

On fiber diameters of continuous maps

We present a surprisingly short proof that for any continuous map $f : \mathbb{R}^n \rightarrow \mathbb{R}^m$, if $n>m$, then there exists no bound on the diameter of fibers of $f$. Moreover, we show that when $m=1$, the union of small fibers of $f$ is bounded; when $m>1$, the union of small fibers need not be bounded. Applications to data analysis are considered.

math.MG↗