-
Wisdom of the Silicon Crowd: LLM Ensemble Prediction Capabilities Rival Human Crowd Accuracy
Authors:
Philipp Schoenegger,
Indre Tuminauskaite,
Peter S. Park,
Rafael Valdece Sousa Bastos,
Philip E. Tetlock
Abstract:
Human forecasting accuracy in practice relies on the 'wisdom of the crowd' effect, in which predictions about future events are significantly improved by aggregating across a crowd of individual forecasters. Past work on the forecasting ability of large language models (LLMs) suggests that frontier LLMs, as individual forecasters, underperform compared to the gold standard of a human-crowd forecas…
▽ More
Human forecasting accuracy in practice relies on the 'wisdom of the crowd' effect, in which predictions about future events are significantly improved by aggregating across a crowd of individual forecasters. Past work on the forecasting ability of large language models (LLMs) suggests that frontier LLMs, as individual forecasters, underperform compared to the gold standard of a human-crowd forecasting-tournament aggregate. In Study 1, we expand this research by using an LLM ensemble approach consisting of a crowd of 12 LLMs. We compare the aggregated LLM predictions on 31 binary questions to those of a crowd of 925 human forecasters from a three-month forecasting tournament. Our preregistered main analysis shows that the LLM crowd outperforms a simple no-information benchmark, and is not statistically different from the human crowd. We also observe a set of human-like biases in machine responses, such as an acquiescence effect and a tendency to favour round numbers. In Study 2, we test whether LLM predictions (of GPT-4 and Claude 2) can be improved by drawing on human cognitive output. We find that both models' forecasting accuracy benefits from exposure to the median human prediction as information, improving accuracy by between 17% and 28%, though this leads to less accurate predictions than simply averaging human and machine forecasts. Our results suggest that LLMs can achieve forecasting accuracy rivaling that of the human crowd: via the simple, practically applicable method of forecast aggregation.
△ Less
Submitted 17 June, 2024; v1 submitted 29 February, 2024;
originally announced February 2024.
-
AI-Augmented Predictions: LLM Assistants Improve Human Forecasting Accuracy
Authors:
Philipp Schoenegger,
Peter S. Park,
Ezra Karger,
Philip E. Tetlock
Abstract:
Large language models (LLMs) show impressive capabilities, matching and sometimes exceeding human performance in many domains. This study explores the potential of LLMs to augment judgement in forecasting tasks. We evaluated the impact on forecasting accuracy of two GPT-4-Turbo assistants: one designed to provide high-quality advice ('superforecasting'), and the other designed to be overconfident…
▽ More
Large language models (LLMs) show impressive capabilities, matching and sometimes exceeding human performance in many domains. This study explores the potential of LLMs to augment judgement in forecasting tasks. We evaluated the impact on forecasting accuracy of two GPT-4-Turbo assistants: one designed to provide high-quality advice ('superforecasting'), and the other designed to be overconfident and base-rate-neglecting. Participants (N = 991) had the option to consult their assigned LLM assistant throughout the study, in contrast to a control group that used a less advanced model (DaVinci-003) without direct forecasting support. Our preregistered analyses reveal that LLM augmentation significantly enhances forecasting accuracy by 23% across both types of assistants, compared to the control group. This improvement occurs despite the superforecasting assistant's higher accuracy in predictions, indicating the augmentation's benefit is not solely due to model prediction accuracy. Exploratory analyses showed a pronounced effect in one forecasting item, without which we find that the superforecasting assistant increased accuracy by 43%, compared with 28% for the biased assistant. We further examine whether LLM augmentation disproportionately benefits less skilled forecasters, degrades the wisdom-of-the-crowd by reducing prediction diversity, or varies in effectiveness with question difficulty. Our findings do not consistently support these hypotheses. Our results suggest that access to an LLM assistant, even a biased one, can be a helpful decision aid in cognitively demanding tasks where the answer is not known at the time of interaction.
△ Less
Submitted 12 February, 2024;
originally announced February 2024.
-
Large Language Model Prediction Capabilities: Evidence from a Real-World Forecasting Tournament
Authors:
Philipp Schoenegger,
Peter S. Park
Abstract:
Accurately predicting the future would be an important milestone in the capabilities of artificial intelligence. However, research on the ability of large language models to provide probabilistic predictions about future events remains nascent. To empirically test this ability, we enrolled OpenAI's state-of-the-art large language model, GPT-4, in a three-month forecasting tournament hosted on the…
▽ More
Accurately predicting the future would be an important milestone in the capabilities of artificial intelligence. However, research on the ability of large language models to provide probabilistic predictions about future events remains nascent. To empirically test this ability, we enrolled OpenAI's state-of-the-art large language model, GPT-4, in a three-month forecasting tournament hosted on the Metaculus platform. The tournament, running from July to October 2023, attracted 843 participants and covered diverse topics including Big Tech, U.S. politics, viral outbreaks, and the Ukraine conflict. Focusing on binary forecasts, we show that GPT-4's probabilistic forecasts are significantly less accurate than the median human-crowd forecasts. We find that GPT-4's forecasts did not significantly differ from the no-information forecasting strategy of assigning a 50% probability to every question. We explore a potential explanation, that GPT-4 might be predisposed to predict probabilities close to the midpoint of the scale, but our data do not support this hypothesis. Overall, we find that GPT-4 significantly underperforms in real-world predictive tasks compared to median human-crowd forecasts. A potential explanation for this underperformance is that in real-world forecasting tournaments, the true answers are genuinely unknown at the time of prediction; unlike in other benchmark tasks like professional exams or time series forecasting, where strong performance may at least partly be due to the answers being memorized from the training data. This makes real-world forecasting tournaments an ideal environment for testing the generalized reasoning and prediction capabilities of artificial intelligence going forward.
△ Less
Submitted 17 October, 2023;
originally announced October 2023.
-
Divide-and-Conquer Dynamics in AI-Driven Disempowerment
Authors:
Peter S. Park,
Max Tegmark
Abstract:
AI companies are attempting to create AI systems that outperform humans at most economically valuable work. Current AI models are already automating away the livelihoods of some artists, actors, and writers. But there is infighting between those who prioritize current harms and future harms. We construct a game-theoretic model of conflict to study the causes and consequences of this disunity. Our…
▽ More
AI companies are attempting to create AI systems that outperform humans at most economically valuable work. Current AI models are already automating away the livelihoods of some artists, actors, and writers. But there is infighting between those who prioritize current harms and future harms. We construct a game-theoretic model of conflict to study the causes and consequences of this disunity. Our model also helps explain why throughout history, stakeholders sharing a common threat have found it advantageous to unite against it, and why the common threat has in turn found it advantageous to divide and conquer.
Under realistic parameter assumptions, our model makes several predictions that find preliminary corroboration in the historical-empirical record. First, current victims of AI-driven disempowerment need the future victims to realize that their interests are also under serious and imminent threat, so that future victims are incentivized to support current victims in solidarity. Second, the movement against AI-driven disempowerment can become more united, and thereby more likely to prevail, if members believe that their efforts will be successful as opposed to futile. Finally, the movement can better unite and prevail if its members are less myopic. Myopic members prioritize their future well-being less than their present well-being, and are thus disinclined to solidarily support current victims today at personal cost, even if this is necessary to counter the shared threat of AI-driven disempowerment.
△ Less
Submitted 18 December, 2023; v1 submitted 9 October, 2023;
originally announced October 2023.
-
AI Deception: A Survey of Examples, Risks, and Potential Solutions
Authors:
Peter S. Park,
Simon Goldstein,
Aidan O'Gara,
Michael Chen,
Dan Hendrycks
Abstract:
This paper argues that a range of current AI systems have learned how to deceive humans. We define deception as the systematic inducement of false beliefs in the pursuit of some outcome other than the truth. We first survey empirical examples of AI deception, discussing both special-use AI systems (including Meta's CICERO) built for specific competitive situations, and general-purpose AI systems (…
▽ More
This paper argues that a range of current AI systems have learned how to deceive humans. We define deception as the systematic inducement of false beliefs in the pursuit of some outcome other than the truth. We first survey empirical examples of AI deception, discussing both special-use AI systems (including Meta's CICERO) built for specific competitive situations, and general-purpose AI systems (such as large language models). Next, we detail several risks from AI deception, such as fraud, election tampering, and losing control of AI systems. Finally, we outline several potential solutions to the problems posed by AI deception: first, regulatory frameworks should subject AI systems that are capable of deception to robust risk-assessment requirements; second, policymakers should implement bot-or-not laws; and finally, policymakers should prioritize the funding of relevant research, including tools to detect AI deception and to make AI systems less deceptive. Policymakers, researchers, and the broader public should work proactively to prevent AI deception from destabilizing the shared foundations of our society.
△ Less
Submitted 28 August, 2023;
originally announced August 2023.
-
Devising and Detecting Phishing: Large Language Models vs. Smaller Human Models
Authors:
Fredrik Heiding,
Bruce Schneier,
Arun Vishwanath,
Jeremy Bernstein,
Peter S. Park
Abstract:
AI programs, built using large language models, make it possible to automatically create phishing emails based on a few data points about a user. They stand in contrast to traditional phishing emails that hackers manually design using general rules gleaned from experience. The V-Triad is an advanced set of rules for manually designing phishing emails to exploit our cognitive heuristics and biases.…
▽ More
AI programs, built using large language models, make it possible to automatically create phishing emails based on a few data points about a user. They stand in contrast to traditional phishing emails that hackers manually design using general rules gleaned from experience. The V-Triad is an advanced set of rules for manually designing phishing emails to exploit our cognitive heuristics and biases. In this study, we compare the performance of phishing emails created automatically by GPT-4 and manually using the V-Triad. We also combine GPT-4 with the V-Triad to assess their combined potential. A fourth group, exposed to generic phishing emails, was our control group. We utilized a factorial approach, sending emails to 112 randomly selected participants recruited for the study. The control group emails received a click-through rate between 19-28%, the GPT-generated emails 30-44%, emails generated by the V-Triad 69-79%, and emails generated by GPT and the V-Triad 43-81%. Each participant was asked to explain why they pressed or did not press a link in the email. These answers often contradict each other, highlighting the need for personalized content. The cues that make one person avoid phishing emails make another person fall for them. Next, we used four popular large language models (GPT, Claude, PaLM, and LLaMA) to detect the intention of phishing emails and compare the results to human detection. The language models demonstrated a strong ability to detect malicious intent, even in non-obvious phishing emails. They sometimes surpassed human detection, although often being slightly less accurate than humans. Finally, we make an analysis of the economic aspects of AI-enabled phishing attacks, showing how large language models can increase the incentives of phishing and spear phishing by reducing their costs.
△ Less
Submitted 30 November, 2023; v1 submitted 23 August, 2023;
originally announced August 2023.
-
Diminished Diversity-of-Thought in a Standard Large Language Model
Authors:
Peter S. Park,
Philipp Schoenegger,
Chongyang Zhu
Abstract:
We test whether Large Language Models (LLMs) can be used to simulate human participants in social-science studies. To do this, we run replications of 14 studies from the Many Labs 2 replication project with OpenAI's text-davinci-003 model, colloquially known as GPT3.5. Based on our pre-registered analyses, we find that among the eight studies we could analyse, our GPT sample replicated 37.5% of th…
▽ More
We test whether Large Language Models (LLMs) can be used to simulate human participants in social-science studies. To do this, we run replications of 14 studies from the Many Labs 2 replication project with OpenAI's text-davinci-003 model, colloquially known as GPT3.5. Based on our pre-registered analyses, we find that among the eight studies we could analyse, our GPT sample replicated 37.5% of the original results and 37.5% of the Many Labs 2 results. However, we were unable to analyse the remaining six studies due to an unexpected phenomenon we call the "correct answer" effect. Different runs of GPT3.5 answered nuanced questions probing political orientation, economic preference, judgement, and moral philosophy with zero or near-zero variation in responses: with the supposedly "correct answer." In one exploratory follow-up study, we found that a "correct answer" was robust to changing the demographic details that precede the prompt. In another, we found that most but not all "correct answers" were robust to changing the order of answer choices. One of our most striking findings occurred in our replication of the Moral Foundations Theory survey results, where we found GPT3.5 identifying as a political conservative in 99.6% of the cases, and as a liberal in 99.3% of the cases in the reverse-order condition. However, both self-reported 'GPT conservatives' and 'GPT liberals' showed right-leaning moral foundations. Our results cast doubts on the validity of using LLMs as a general replacement for human participants in the social sciences. Our results also raise concerns that a hypothetical AI-led future may be subject to a diminished diversity-of-thought.
△ Less
Submitted 13 September, 2023; v1 submitted 13 February, 2023;
originally announced February 2023.
-
Probability laws for the distribution of geometric lengths when sampling by a random walk in a Fuchsian fundamental group
Authors:
Peter S. Park
Abstract:
Let $S=Γ\backslash \mathbb{H}$ be a hyperbolic surface of finite topological type, such that the Fuchsian group $Γ\le \operatorname{PSL}_2(\mathbb{R})$ is non-elementary, and consider any generating set $\mathfrak S$ of $Γ$. When sampling by an $n$-step random walk in $π_1(S) \cong Γ$ with each step given by an element in $\mathfrak S$, the subset of this sampled set comprised of hyperbolic elemen…
▽ More
Let $S=Γ\backslash \mathbb{H}$ be a hyperbolic surface of finite topological type, such that the Fuchsian group $Γ\le \operatorname{PSL}_2(\mathbb{R})$ is non-elementary, and consider any generating set $\mathfrak S$ of $Γ$. When sampling by an $n$-step random walk in $π_1(S) \cong Γ$ with each step given by an element in $\mathfrak S$, the subset of this sampled set comprised of hyperbolic elements approaches full measure as $n\to \infty$, and for this subset, the distribution of geometric lengths obeys a Law of Large Numbers, Central Limit Theorem, Large Deviations Principle, and Local Limit Theorem. We give a proof of this known theorem using Gromov's theorem on translation lengths of Gromov-hyperbolic groups.
△ Less
Submitted 8 February, 2019; v1 submitted 10 July, 2018;
originally announced July 2018.
-
Conjugacy growth of commutators
Authors:
Peter S. Park
Abstract:
For the free group $F_r$ on $r>1$ generators (respectively, the free product $G_1 * G_2$ of two nontrivial finite groups $G_1$ and $G_2$), we obtain the asymptotic for the number of conjugacy classes of commutators in $F_r$ (respectively, $G_1 * G_2$) with a given word length in a fixed set of free generators (respectively, the set of generators given by the nontrivial elements of $G_1$ and $G_2$)…
▽ More
For the free group $F_r$ on $r>1$ generators (respectively, the free product $G_1 * G_2$ of two nontrivial finite groups $G_1$ and $G_2$), we obtain the asymptotic for the number of conjugacy classes of commutators in $F_r$ (respectively, $G_1 * G_2$) with a given word length in a fixed set of free generators (respectively, the set of generators given by the nontrivial elements of $G_1$ and $G_2$). Our result is proven by using the classification of commutators in free groups and in free products by Wicks, and builds on the works of Rivin and Sharp, who asymptotically counted the conjugacy classes of commutator-subgroup elements in $F_r$ with a given word length.
△ Less
Submitted 10 February, 2019; v1 submitted 26 February, 2018;
originally announced February 2018.
-
The "Riemann Hypothesis" is True for Period Polynomials of Almost All Newforms
Authors:
Yang P. Liu,
Peter S. Park,
Zhuo Qun Song
Abstract:
The period polynomial $r_f(z)$ for a weight $k \geq 3$ newform $f \in S_k(Γ_0(N),χ)$ is the generating function for special values of $L(s,f)$. The functional equation for $L(s, f)$ induces a functional equation on $r_f(z)$. **, Ma, Ono, and Soundararajan proved that for all newforms $f$ of even weight $k \ge 4$ and trivial nebetypus, the "Riemann Hypothesis" holds for $r_f(z)$: that is, all root…
▽ More
The period polynomial $r_f(z)$ for a weight $k \geq 3$ newform $f \in S_k(Γ_0(N),χ)$ is the generating function for special values of $L(s,f)$. The functional equation for $L(s, f)$ induces a functional equation on $r_f(z)$. **, Ma, Ono, and Soundararajan proved that for all newforms $f$ of even weight $k \ge 4$ and trivial nebetypus, the "Riemann Hypothesis" holds for $r_f(z)$: that is, all roots of $r_f(z)$ lie on the circle of symmetry $|z| =1/\sqrt{N}$. We generalize their methods to prove that this phenomenon holds for all but possibly finitely many newforms $f$ of weight $k \ge 3$ with any nebentypus. We also show that the roots of $r_f(z)$ are equidistributed if $N$ or $k$ is sufficiently large.
△ Less
Submitted 14 August, 2016; v1 submitted 15 July, 2016;
originally announced July 2016.
-
Bounded Gaps Between Products of Distinct Primes
Authors:
Yang P. Liu,
Peter S. Park,
Zhuo Qun Song
Abstract:
Let $r \ge 2$ be an integer. We adapt the Maynard-Tao sieve to produce the asymptotically best-known bounded gaps between products of $r$ distinct primes. Our result applies to positive-density subsets of the primes that satisfy certain equidistribution conditions. This improves on the work of Thorne and Sono.
Let $r \ge 2$ be an integer. We adapt the Maynard-Tao sieve to produce the asymptotically best-known bounded gaps between products of $r$ distinct primes. Our result applies to positive-density subsets of the primes that satisfy certain equidistribution conditions. This improves on the work of Thorne and Sono.
△ Less
Submitted 1 June, 2017; v1 submitted 13 July, 2016;
originally announced July 2016.
-
The van der Waerden complex
Authors:
Richard Ehrenborg,
Likith Govindaiah,
Peter S. Park,
Margaret Readdy
Abstract:
We introduce the van der Waerden complex ${\rm vdW}(n,k)$ defined as the simplicial complex whose facets correspond to arithmetic progressions of length $k$ in the vertex set $\{1, 2, \ldots, n\}$. We show the van der Waerden complex ${\rm vdW}(n,k)$ is homotopy equivalent to a $CW$-complex whose cells asymptotically have dimension at most $\log k / \log \log k$. Furthermore, we give bounds on…
▽ More
We introduce the van der Waerden complex ${\rm vdW}(n,k)$ defined as the simplicial complex whose facets correspond to arithmetic progressions of length $k$ in the vertex set $\{1, 2, \ldots, n\}$. We show the van der Waerden complex ${\rm vdW}(n,k)$ is homotopy equivalent to a $CW$-complex whose cells asymptotically have dimension at most $\log k / \log \log k$. Furthermore, we give bounds on $n$ and $k$ which imply that the van der Waerden complex is contractible.
△ Less
Submitted 20 October, 2016; v1 submitted 2 May, 2016;
originally announced May 2016.
-
Density-Dependent Electron Transport and Precise Modeling of GaN HEMTs
Authors:
Sanyam Bajaj,
Omor F. Shoron,
Pil Sung Park,
Sriram Krishnamoorthy,
Fatih Akyol,
Ting-Hsiang Hung,
Shahed Reza,
Eduardo M. Chumbes,
Jacob Khurgin,
Siddharth Rajan
Abstract:
We report on the direct measurement of two-dimensional sheet charge density dependence of electron transport in AlGaN/GaN high electron mobility transistors. Pulsed IV measurements established increasing electron velocities with decreasing sheet charge densities, resulting in saturation velocity of 1.9 x 10^7 cm/s at a low sheet charge density of 7.8 x 10^11 cm-2. A new optical phonon emission-bas…
▽ More
We report on the direct measurement of two-dimensional sheet charge density dependence of electron transport in AlGaN/GaN high electron mobility transistors. Pulsed IV measurements established increasing electron velocities with decreasing sheet charge densities, resulting in saturation velocity of 1.9 x 10^7 cm/s at a low sheet charge density of 7.8 x 10^11 cm-2. A new optical phonon emission-based electron velocity model for GaN is also presented. It accommodates stimulated LO phonon emission which clamps the electron velocity with strong electron-phonon interaction and long LO phonon lifetime in GaN. A comparison with the measured density-dependent saturation velocity shows that it captures the dependence rather well. Finally, the experimental result is applied in TCAD-based device simulator to predict DC and small signal characteristics of a reported GaN HEMT. Good agreement between the simulated and reported experimental results validated the measurement presented in this report and established accurate modeling of GaN HEMTs.
△ Less
Submitted 27 August, 2015;
originally announced August 2015.
-
Elliptic Curve Variants of the Least Quadratic Nonresidue Problem and Linnik's Theorem
Authors:
Evan Chen,
Peter S. Park,
Ashvin Swaminathan
Abstract:
Let $E_1$ and $E_2$ be $\overline{\mathbb{Q}}$-nonisogenous, semistable elliptic curves over $\mathbb{Q}$, having respective conductors $N_{E_1}$ and $N_{E_2}$ and both without complex multiplication. For each prime $p$, denote by $a_{E_i}(p) := p+1-\#E_i(\mathbb{F}_p)$ the trace of Frobenius. Under the assumption of the Generalized Riemann Hypothesis (GRH) for the convolved symmetric power $L$-fu…
▽ More
Let $E_1$ and $E_2$ be $\overline{\mathbb{Q}}$-nonisogenous, semistable elliptic curves over $\mathbb{Q}$, having respective conductors $N_{E_1}$ and $N_{E_2}$ and both without complex multiplication. For each prime $p$, denote by $a_{E_i}(p) := p+1-\#E_i(\mathbb{F}_p)$ the trace of Frobenius. Under the assumption of the Generalized Riemann Hypothesis (GRH) for the convolved symmetric power $L$-functions $L(s, \mathrm{Sym}^i E_1\otimes\mathrm{Sym}^j E_2)$ where $i,j\in\{0,1,2\}$, we prove an explicit result that can be stated succinctly as follows: there exists a prime $p\nmid N_{E_1}N_{E_2}$ such that $a_{E_1}(p)a_{E_2}(p)<0$ and
\[ p < \big( (32+o(1))\cdot \log N_{E_1} N_{E_2}\big)^2. \] This improves and makes explicit a result of Bucur and Kedlaya.
Now, if $I\subset[-1,1]$ is a subinterval with Sato-Tate measure $μ$ and if the symmetric power $L$-functions $L(s, \mathrm{Sym}^k E_1)$ are functorial and satisfy GRH for all $k \le 8/μ$, we employ similar techniques to prove an explicit result that can be stated succinctly as follows: there exists a prime $p\nmid N_{E_1}$ such that $a_{E_1}(p)/(2\sqrt{p})\in I$ and
\[ p < \left((21+o(1)) \cdot μ^{-2}\log (N_{E_1}/μ)\right)^2. \]
△ Less
Submitted 26 January, 2016; v1 submitted 25 July, 2015;
originally announced July 2015.
-
On Logarithmically Benford Sequences
Authors:
Evan Chen,
Peter S. Park,
Ashvin Swaminathan
Abstract:
Let $\mathcal{I} \subset \mathbb{N}$ be an infinite subset, and let $\{a_i\}_{i \in \mathcal{I}}$ be a sequence of nonzero real numbers indexed by $\mathcal{I}$ such that there exist positive constants $m, C_1$ for which $|a_i| \leq C_1 \cdot i^m$ for all $i \in \mathcal{I}$. Furthermore, let $c_i \in [-1,1]$ be defined by $c_i = \frac{a_i}{C_1 \cdot i^m}$ for each $i \in \mathcal{I}$, and suppose…
▽ More
Let $\mathcal{I} \subset \mathbb{N}$ be an infinite subset, and let $\{a_i\}_{i \in \mathcal{I}}$ be a sequence of nonzero real numbers indexed by $\mathcal{I}$ such that there exist positive constants $m, C_1$ for which $|a_i| \leq C_1 \cdot i^m$ for all $i \in \mathcal{I}$. Furthermore, let $c_i \in [-1,1]$ be defined by $c_i = \frac{a_i}{C_1 \cdot i^m}$ for each $i \in \mathcal{I}$, and suppose the $c_i$'s are equidistributed in $[-1,1]$ with respect to a continuous, symmetric probability measure $μ$. In this paper, we show that if $\mathcal{I} \subset \mathbb{N}$ is not too sparse, then the sequence $\{a_i\}_{i \in \mathcal{I}}$ fails to obey Benford's Law with respect to arithmetic density in any sufficiently large base, and in fact in any base when $μ([0,t])$ is a strictly convex function of $t \in (0,1)$. Nonetheless, we also provide conditions on the density of $\mathcal{I} \subset \mathbb{N}$ under which the sequence $\{a_i\}_{i \in \mathcal{I}}$ satisfies Benford's Law with respect to logarithmic density in every base.
As an application, we apply our general result to study Benford's Law-type behavior in the leading digits of Frobenius traces of newforms of positive, even weight. Our methods of proof build on the work of Jameson, Thorner, and Ye, who studied the particular case of newforms without complex multiplication.
△ Less
Submitted 25 December, 2015; v1 submitted 9 July, 2015;
originally announced July 2015.
-
Linnik's Theorem for Sato-Tate Laws on Elliptic Curves with Complex Multiplication
Authors:
Evan Chen,
Peter S. Park,
Ashvin Swaminathan
Abstract:
Let $E/\mathbb{Q}$ be an elliptic curve with complex multiplication (CM), and for each prime $p$ of good reduction, let $a_E(p) = p + 1 - \#E(\mathbb{F}_p)$ denote the trace of Frobenius. By the Hasse bound, $a_E(p) = 2\sqrt{p} \cos θ_p$ for a unique $θ_p \in [0, π]$. In this paper, we prove that the least prime $p$ such that $θ_p \in [α, β] \subset [0, π]$ satisfies \[ p \ll \left(\frac{N_E}{β- α…
▽ More
Let $E/\mathbb{Q}$ be an elliptic curve with complex multiplication (CM), and for each prime $p$ of good reduction, let $a_E(p) = p + 1 - \#E(\mathbb{F}_p)$ denote the trace of Frobenius. By the Hasse bound, $a_E(p) = 2\sqrt{p} \cos θ_p$ for a unique $θ_p \in [0, π]$. In this paper, we prove that the least prime $p$ such that $θ_p \in [α, β] \subset [0, π]$ satisfies \[ p \ll \left(\frac{N_E}{β- α}\right)^A, \] where $N_E$ is the conductor of $E$ and the implied constant and exponent $A > 2$ are absolute and effectively computable. Our result is an analogue for CM elliptic curves of Linnik's Theorem for arithmetic progressions, which states that the least prime $p \equiv a \pmod q$ for $(a,q)=1$ satisfies $p \ll q^L$ for an absolute constant $L > 0$.
△ Less
Submitted 14 October, 2015; v1 submitted 30 June, 2015;
originally announced June 2015.
-
Unipolar Vertical Transport in GaN/AlGaN/GaN Heterostructures
Authors:
D. N. Nath,
P. S. Park,
Z. C. Yang,
S. Rajan
Abstract:
In this letter, we report on unipolar vertical transport characteristics in c-plane GaN/AlGaN/GaN heterostructures. Vertical current in heterostructures with random alloy barriers was found to be independent of dislocation density and heterostructure barrier height, and significantly higher than theoretical estimates. Percolation-based transport due to random alloy fluctuations in the ternary AlGa…
▽ More
In this letter, we report on unipolar vertical transport characteristics in c-plane GaN/AlGaN/GaN heterostructures. Vertical current in heterostructures with random alloy barriers was found to be independent of dislocation density and heterostructure barrier height, and significantly higher than theoretical estimates. Percolation-based transport due to random alloy fluctuations in the ternary AlGaN is suggested as the dominant transport mechanism, and confirmed through experiments showing that non-random or digital AlGaN alloys and polarization-engineered binary GaN barriers can eliminate percolation transport and reduce leakage significantly. The understanding of vertical transport and methods for effective control proposed here will greatly impact III-nitride unipolar vertical devices.
△ Less
Submitted 16 February, 2013;
originally announced February 2013.
-
Large Area Single Crystal (0001) Oriented MoS2 Thin Films
Authors:
Masihhur R. Laskar,
Lu Ma,
ShanthaKumar K,
Pil Sung Park,
Sriram Krishnamoorthy,
Digbijoy N. Nath,
Wu Lu,
Yiying Wu,
Siddharth Rajan
Abstract:
Layered metal dichalcogenide materials are a family of semiconductors with a wide range of energy band gaps and properties, and potential to open up new areas of physics and technology applications. However, obtaining high crystal quality thin films over a large area remains a challenge. Here we show that chemical vapor deposition (CVD) can be used to achieve large area electronic grade single cry…
▽ More
Layered metal dichalcogenide materials are a family of semiconductors with a wide range of energy band gaps and properties, and potential to open up new areas of physics and technology applications. However, obtaining high crystal quality thin films over a large area remains a challenge. Here we show that chemical vapor deposition (CVD) can be used to achieve large area electronic grade single crystal Molybdenum Disulfide (MoS2) thin films with the highest mobility reported in CVD grown films so far. Growth temperature and choice of substrate were found to critically impact the quality of film grown, and high temperature growth on (0001) orientated sapphire yielded highly oriented single crystal MoS2 films for the first time. Films grown under optimal conditions were found to be of high structural quality from high-resolution X-ray diffraction, transmission electron microscopy, and Raman measurements, approaching the quality of reference geological MoS2. Photoluminescence and electrical measurements confirmed the growth of optically active MoS2 with a low background carrier concentration, and high mobility. The CVD method reported here for the growth of high quality MoS2 thin films paves the way towards growth of a variety of layered 2D chalcogenide semiconductors and their heterostructures.
△ Less
Submitted 25 February, 2013; v1 submitted 13 February, 2013;
originally announced February 2013.
-
Ohmic Contact Formation Between Metal and AlGaN/GaN Heterostructure via Graphene Insertion
Authors:
Pil Sung Park,
Kongara M. Reddy,
Digbijoy N. Nath,
Zhichao Yang,
Nitin P. Padture,
Siddharth Rajan
Abstract:
A simple method for the creation of Ohmic contact to 2-D electron gas (2DEG) in AlGaN/GaN high electron-mobility transistors (HEMTs) using Cr/Graphene layer is demonstrated. A weak temperature dependence of this Ohmic contact observed in the range 77 to 300 K precludes thermionic emission or trap-assisted hop** as possible carrier-transport mechanisms. It is suggested that the Cr/Graphene combin…
▽ More
A simple method for the creation of Ohmic contact to 2-D electron gas (2DEG) in AlGaN/GaN high electron-mobility transistors (HEMTs) using Cr/Graphene layer is demonstrated. A weak temperature dependence of this Ohmic contact observed in the range 77 to 300 K precludes thermionic emission or trap-assisted hop** as possible carrier-transport mechanisms. It is suggested that the Cr/Graphene combination acts akin to a doped n-type semiconductor in contact with AlGaN/GaN heterostructure, and promotes carrier transport along percolating Al-lean paths through the AlGaN layer. This new use of graphene offers a simple and reliable method for making Ohmic contacts to AlGaN/GaN heterostructures, circumventing complex additional processing steps involving high temperatures. These results could have important implications for the fabrication and manufacturing of AlGaN/GaN-based microelectronic and optoelectronic devices/sensors of the future.
△ Less
Submitted 14 February, 2013; v1 submitted 9 January, 2013;
originally announced January 2013.
-
Low Resistance GaN/InGaN/GaN Tunnel Junctions
Authors:
Sriram Krishnamoorthy,
Fatih Akyol,
Pil Sung Park,
Siddharth Rajan
Abstract:
Enhanced interband tunnel injection of holes into a PN junction is demonstrated using P-GaN/InGaN/N-GaN tunnel junctions with a specific resistivity of 1.2 X 10-4 Ω cm2. The design methodology and low-temperature characteristic of these tunnel junctions is discussed, and insertion into a PN junction device is described. Applications of tunnel junctions in III-nitride optoelectronics devices are ex…
▽ More
Enhanced interband tunnel injection of holes into a PN junction is demonstrated using P-GaN/InGaN/N-GaN tunnel junctions with a specific resistivity of 1.2 X 10-4 Ω cm2. The design methodology and low-temperature characteristic of these tunnel junctions is discussed, and insertion into a PN junction device is described. Applications of tunnel junctions in III-nitride optoelectronics devices are explained using energy band diagrams. The lower band gap and polarization fields reduce tunneling barrier, eliminating the need for ohmic contacts to p-type GaN. This demonstration of efficient tunnel injection of carriers in III-Nitrides can lead to a replacement of existing resistive p-type contact material in light emitters with tunneling contact layers, requiring very little metal footprint on the surface, resulting in enhanced light extraction from top emitting emitters.
△ Less
Submitted 21 January, 2013; v1 submitted 20 November, 2012;
originally announced November 2012.
-
GdN Nanoisland-Based GaN Tunnel Junctions
Authors:
Sriram Krishnamoorthy,
Thomas Kent,
**g Yang,
Pil Sung Park,
Roberto Myers,
Siddharth Rajan
Abstract:
We show that GdN nanoislands can enhance inter-band tunneling in GaN PN junctions by several orders of magnitude, enabling low optical absorption low-resistance tunnel junctions (specific resistivity 1.3 X 10-3 Ω-cm2) for various optoelectronic applications. We exploit the ability to overgrow high quality GaN over GdN nanoislands to create new nanoscale heterostructure designs that are not feasibl…
▽ More
We show that GdN nanoislands can enhance inter-band tunneling in GaN PN junctions by several orders of magnitude, enabling low optical absorption low-resistance tunnel junctions (specific resistivity 1.3 X 10-3 Ω-cm2) for various optoelectronic applications. We exploit the ability to overgrow high quality GaN over GdN nanoislands to create new nanoscale heterostructure designs that are not feasible in planar epitaxy. GdN nanoisland assisted inter-band tunneling was found to enhance tunneling in both of the polar orientations of GaN. Tunnel injection of holes was confirmed by low temperature operation of GaN p-n junction with a tunneling contact layer, showing strong electroluminescence down to 20K. The availability of tunnel junctions with negligible absorption could not only improve the efficiency of existing optoelectronic devices significantly, but also enable new electronic and optical devices based on wide band gap materials.
△ Less
Submitted 19 May, 2013; v1 submitted 17 June, 2012;
originally announced June 2012.
-
Comment on "Fock-Darwin States of Dirac Electrons in Graphene-Based Artificial Atoms"
Authors:
P. S. Park,
S. C. Kim,
S. -R. Eric Yang
Abstract:
Chen, Apalkov, and Chakraborty (Phys. Rev. Lett. 98, 186803 (2007)) have computed Fock-Darwin levels of a graphene dot by including only basis states with energies larger than or equal to zero. We show that their results violate the Hellman-Feynman theorem. A correct treatment must include both positive and negative energy basis states. Additional basis states lead to new energy levels in the opti…
▽ More
Chen, Apalkov, and Chakraborty (Phys. Rev. Lett. 98, 186803 (2007)) have computed Fock-Darwin levels of a graphene dot by including only basis states with energies larger than or equal to zero. We show that their results violate the Hellman-Feynman theorem. A correct treatment must include both positive and negative energy basis states. Additional basis states lead to new energy levels in the optical spectrum and anticrossings between optical transition lines.
△ Less
Submitted 14 March, 2012; v1 submitted 17 January, 2012;
originally announced January 2012.
-
Demonstration of Forward Inter-band Tunneling in GaN by Polarization Engineering
Authors:
Sriram Krishnamoorthy,
Pil Sung Park,
Siddharth Rajan
Abstract:
We report on the design, fabrication, and characterization of GaN interband tunnel junction showing forward tunneling characteristics. We have achieved very high forward tunneling currents (153 mA/cm2 at 10 mV, and 17.7 A/cm2 peak current) in polarization-engineered GaN/InGaN/GaN heterojunction diodes grown by plasma assisted molecular beam epitaxy. We also report the observation of repeatable neg…
▽ More
We report on the design, fabrication, and characterization of GaN interband tunnel junction showing forward tunneling characteristics. We have achieved very high forward tunneling currents (153 mA/cm2 at 10 mV, and 17.7 A/cm2 peak current) in polarization-engineered GaN/InGaN/GaN heterojunction diodes grown by plasma assisted molecular beam epitaxy. We also report the observation of repeatable negative differential resistance in interband III-Nitride tunnel junctions, with peak-valley current ratio (PVCR) of 4 at room temperature. The forward current density achieved in this work meets the typical current drive requirements of a multi-junction solar cell.
△ Less
Submitted 19 August, 2011;
originally announced August 2011.
-
Scaling properties of induced density of chiral and non-chiral Dirac fermions in magnetic fields
Authors:
P. S. Park,
S. C. Kim,
S. -R. Eric Yang
Abstract:
We find that a repulsive potential of graphene in the presence of a magnetic field has bound states that are peaked inside the barrier with tails extending over \ell(N + 1), where \ell and N are the magnetic length and Landau level(LL) index. We have investigated how these bound states affect scaling properties of the induced density of filled Landau levels of massless Dirac fermions. For chiral f…
▽ More
We find that a repulsive potential of graphene in the presence of a magnetic field has bound states that are peaked inside the barrier with tails extending over \ell(N + 1), where \ell and N are the magnetic length and Landau level(LL) index. We have investigated how these bound states affect scaling properties of the induced density of filled Landau levels of massless Dirac fermions. For chiral fermions we find, in strong coupling regime, that the density inside the repulsive potential can be greater than the value in the absence of the potential while in the weak coupling regime we find negative induced density. Similar results hold also for non-chiral fermions. As one moves from weak to strong coupling regimes the effective coupling constant between the potential and electrons becomes more repulsive, and then it changes sign and becomes attractive. Different power-laws of induced density are found for chiral and non-chiral fermions.
△ Less
Submitted 24 August, 2011; v1 submitted 17 June, 2011;
originally announced June 2011.
-
Polarization-engineered GaN/InGaN/GaN tunnel diodes
Authors:
Sriram Krishnamoorthy,
Digbijoy N. Nath,
Fatih Akyol,
Pil Sung Park,
Michele Esposto,
Siddharth Rajan
Abstract:
We report on the design and demonstration of polarization-engineered GaN/InGaN/GaN tunnel junction diodes with high current density and low tunneling turn-on voltage. Wentzel-Kramers-Brillouin (WKB) calculations were used to model and design tunnel junctions with narrow bandgap InGaN-based barrier layers. N-polar p-GaN/In0.33Ga0.67N/n-GaN heterostructure tunnel diodes were grown using molecular be…
▽ More
We report on the design and demonstration of polarization-engineered GaN/InGaN/GaN tunnel junction diodes with high current density and low tunneling turn-on voltage. Wentzel-Kramers-Brillouin (WKB) calculations were used to model and design tunnel junctions with narrow bandgap InGaN-based barrier layers. N-polar p-GaN/In0.33Ga0.67N/n-GaN heterostructure tunnel diodes were grown using molecular beam epitaxy. Efficient zero bias tunneling turn-on with a high current density of 118 A/cm2 at a reverse bias of 1V, reaching a maximum current density up to 9.2 kA/cm2 were obtained. These results represent the highest current density reported in III-nitride tunnel junctions, and demonstrate the potential of III-nitride tunnel devices for a broad range of optoelectronic and electronic applications.
△ Less
Submitted 24 August, 2010;
originally announced August 2010.
-
Electronic properties of a graphene antidot in magnetic fields
Authors:
P. S. Park,
S. C. Kim,
S. -R. Eric Yang
Abstract:
We report on several unusual properties of a graphene antidot created by a piecewise constant potential in a magnetic field. We find that the total probability of finding the electron in the barrier can be nearly one while it is almost zero outside the barrier. In addition, for each electron state of a graphene antidot there is a dot state with exactly the same wavefunction but with a different en…
▽ More
We report on several unusual properties of a graphene antidot created by a piecewise constant potential in a magnetic field. We find that the total probability of finding the electron in the barrier can be nearly one while it is almost zero outside the barrier. In addition, for each electron state of a graphene antidot there is a dot state with exactly the same wavefunction but with a different energy. This symmetry is a consequence of Klein tunneling of Dirac electrons. Moreover, in zigzag nanoribbons we find strong coupling between some antidot states and zigzag edge states. Experimental tests of these effects are proposed.
△ Less
Submitted 7 September, 2010; v1 submitted 16 June, 2010;
originally announced June 2010.
-
States near Dirac points of rectangular graphene dot in a magnetic field
Authors:
S. C. Kim,
P. S. Park,
S. -R. Eric Yang
Abstract:
In neutral graphene dots the Fermi level coincides with the Dirac points. We have investigated in the presence of a magnetic field several unusual properties of single electron states near the Fermi level of such a rectangular-shaped graphene dot with two zigzag and two armchair edges. We find that a quasi-degenerate level forms near zero energy and the number of states in this level can be tune…
▽ More
In neutral graphene dots the Fermi level coincides with the Dirac points. We have investigated in the presence of a magnetic field several unusual properties of single electron states near the Fermi level of such a rectangular-shaped graphene dot with two zigzag and two armchair edges. We find that a quasi-degenerate level forms near zero energy and the number of states in this level can be tuned by the magnetic field. The wavefunctions of states in this level are all peaked on the zigzag edges with or without some weight inside the dot. Some of these states are magnetic field-independent surface states while the others are field-dependent. We have found a scaling result from which the number of magnetic field-dependent states of large dots can be inferred from those of smaller dots.
△ Less
Submitted 8 January, 2010;
originally announced January 2010.
-
Effect of many-body quantum fluctuations on matrix Berry phases of a two-dimensional n-type semiconductor quantum dot
Authors:
S. C. Kim,
Y. J. Kim,
P. S. Park,
N. Y. Hwang,
S. -R. Eric Yang
Abstract:
In the presence of spin-orbit coupling and inversion symmetry of the lateral confinement potential a single electron does not exhibit matrix Berry phases in quasi-two-dimensional semiconductor quantum dots. In such a system we investigate whether many-body correlation effects can lead to finite matrix Berry phases. We find that the transformation properties of many-electron wavefunctions under t…
▽ More
In the presence of spin-orbit coupling and inversion symmetry of the lateral confinement potential a single electron does not exhibit matrix Berry phases in quasi-two-dimensional semiconductor quantum dots. In such a system we investigate whether many-body correlation effects can lead to finite matrix Berry phases. We find that the transformation properties of many-electron wavefunctions under two-dimensional inversion operation do not allow finite matrix Berry phases. This effect is exact and is independent of the form of electron-electron interactions. On the other hand, quasi-two-dimensional semiconductor quantum dots with lateral confinement potential without inversion symmetry can have finite matrix Berry phases. We find that many-body quantum fluctuations can change matrix Berry phases significantly in such systems.
△ Less
Submitted 1 October, 2008;
originally announced October 2008.
-
Pum** in quantum dots and non-Abelian matrix Berry phases
Authors:
N. Y. Hwang,
S. C. Kim,
P. S. Park,
S. -R. Eric Yang
Abstract:
We have investigated pum** in quantum dots from the perspective of non-Abelian (matrix) Berry phases by solving the time dependent Schr{ö}dinger equation exactly for adiabatic changes. Our results demonstrate that a pumped charge is related to the presence of a finite matrix Berry phase. When consecutive adiabatic cycles are performed the pumped charge of each cycle is different from the previ…
▽ More
We have investigated pum** in quantum dots from the perspective of non-Abelian (matrix) Berry phases by solving the time dependent Schr{ö}dinger equation exactly for adiabatic changes. Our results demonstrate that a pumped charge is related to the presence of a finite matrix Berry phase. When consecutive adiabatic cycles are performed the pumped charge of each cycle is different from the previous ones.
△ Less
Submitted 14 February, 2008; v1 submitted 7 June, 2007;
originally announced June 2007.
