Showing 1–50 of 56 results for author: Taki, M

Hierarchical Associative Memory, Parallelized MLP-Mixer, and Symmetry Breaking

Authors: Ryo Karakida, Toshihiro Ota, Masato Taki

Abstract: Transformers have established themselves as the leading neural network model in natural language processing and are increasingly foundational in various domains. In vision, the MLP-Mixer model has demonstrated competitive performance, suggesting that attention mechanisms might not be indispensable. Inspired by this, recent research has explored replacing attention modules with other mechanisms, in… ▽ More Transformers have established themselves as the leading neural network model in natural language processing and are increasingly foundational in various domains. In vision, the MLP-Mixer model has demonstrated competitive performance, suggesting that attention mechanisms might not be indispensable. Inspired by this, recent research has explored replacing attention modules with other mechanisms, including those described by MetaFormers. However, the theoretical framework for these models remains underdeveloped. This paper proposes a novel perspective by integrating Krotov's hierarchical associative memory with MetaFormers, enabling a comprehensive representation of the entire Transformer block, encompassing token-/channel-mixing modules, layer normalization, and skip connections, as a single Hopfield network. This approach yields a parallelized MLP-Mixer derived from a three-layer Hopfield network, which naturally incorporates symmetric token-/channel-mixing modules and layer normalization. Empirical studies reveal that symmetric interaction matrices in the model hinder performance in image recognition tasks. Introducing symmetry-breaking effects transitions the performance of the symmetric parallelized MLP-Mixer to that of the vanilla MLP-Mixer. This indicates that during standard training, weight matrices of the vanilla MLP-Mixer spontaneously acquire a symmetry-breaking configuration, enhancing their effectiveness. These findings offer insights into the intrinsic properties of Transformers and MLP-Mixers and their theoretical underpinnings, providing a robust framework for future model design and optimization. △ Less

Submitted 17 June, 2024; originally announced June 2024.

Comments: 18 pages

CAM Back Again: Large Kernel CNNs from a Weakly Supervised Object Localization Perspective

Authors: Shunsuke Yasuki, Masato Taki

Abstract: Recently, convolutional neural networks (CNNs) with large size kernels have attracted much attention in the computer vision field, following the success of the Vision Transformers. Large kernel CNNs have been reported to perform well in downstream vision tasks as well as in classification performance. The reason for the high-performance of large kernel CNNs in downstream tasks has been attributed… ▽ More Recently, convolutional neural networks (CNNs) with large size kernels have attracted much attention in the computer vision field, following the success of the Vision Transformers. Large kernel CNNs have been reported to perform well in downstream vision tasks as well as in classification performance. The reason for the high-performance of large kernel CNNs in downstream tasks has been attributed to the large effective receptive field (ERF) produced by large size kernels, but this view has not been fully tested. We therefore revisit the performance of large kernel CNNs in downstream task, focusing on the weakly supervised object localization (WSOL) task. WSOL, a difficult downstream task that is not fully supervised, provides a new angle to explore the capabilities of the large kernel CNNs. Our study compares the modern large kernel CNNs ConvNeXt, RepLKNet, and SLaK to test the validity of the naive expectation that ERF size is important for improving downstream task performance. Our analysis of the factors contributing to high performance provides a different perspective, in which the main factor is feature map improvement. Furthermore, we find that modern CNNs are robust to the CAM problems of local regions of objects being activated, which has long been discussed in WSOL. CAM is the most classic WSOL method, but because of the above-mentioned problems, it is often used as a baseline method for comparison. However, experiments on the CUB-200-2011 dataset show that simply combining a large kernel CNN, CAM, and simple data augmentation methods can achieve performance (90.99% MaxBoxAcc) comparable to the latest WSOL method, which is CNN-based and requires special training or complex post-processing. The code is available at https://github.com/snskysk/CAM-Back-Again. △ Less

Submitted 11 March, 2024; originally announced March 2024.

Comments: Accepted by CVPR2024. Code: https://github.com/snskysk/CAM-Back-Again

Hypernuclear event detection in the nuclear emulsion with Monte Carlo simulation and machine learning

Authors: A. Kasagi, W. Dou, V. Drozd, H. Ekawa, S. Escrig, Y. Gao, Y. He, E. Liu, A. Muneem, M. Nakagawa, K. Nakazawa, C. Rappold, N. Saito, T. R. Saito, S. Sugimoto, M. Taki, Y. K. Tanaka, A. Yanai, J. Yoshida, M. Yoshimoto, H. Wang

Abstract: This study developed a novel method for detecting hypernuclear events recorded in nuclear emulsion sheets using machine learning techniques. The artificial neural network-based object detection model was trained on surrogate images created through Monte Carlo simulations and image-style transformations using generative adversarial networks. The performance of the proposed model was evaluated using… ▽ More This study developed a novel method for detecting hypernuclear events recorded in nuclear emulsion sheets using machine learning techniques. The artificial neural network-based object detection model was trained on surrogate images created through Monte Carlo simulations and image-style transformations using generative adversarial networks. The performance of the proposed model was evaluated using $α$-decay events obtained from the J-PARC E07 emulsion data. The model achieved approximately twice the detection efficiency of conventional image processing and reduced the time spent on manual visual inspection by approximately 1/17. The established method was successfully applied to the detection of hypernuclear events. This approach is a state-of-the-art tool for discovering rare events recorded in nuclear emulsion sheets without any real data for training. △ Less

Submitted 1 May, 2023; originally announced May 2023.

Comments: 32 pages, 13 figures

iMixer: hierarchical Hopfield network implies an invertible, implicit and iterative MLP-Mixer

Authors: Toshihiro Ota, Masato Taki

Abstract: In the last few years, the success of Transformers in computer vision has stimulated the discovery of many alternative models that compete with Transformers, such as the MLP-Mixer. Despite their weak inductive bias, these models have achieved performance comparable to well-studied convolutional neural networks. Recent studies on modern Hopfield networks suggest the correspondence between certain e… ▽ More In the last few years, the success of Transformers in computer vision has stimulated the discovery of many alternative models that compete with Transformers, such as the MLP-Mixer. Despite their weak inductive bias, these models have achieved performance comparable to well-studied convolutional neural networks. Recent studies on modern Hopfield networks suggest the correspondence between certain energy-based associative memory models and Transformers or MLP-Mixer, and shed some light on the theoretical background of the Transformer-type architectures design. In this paper, we generalize the correspondence to the recently introduced hierarchical Hopfield network, and find iMixer, a novel generalization of MLP-Mixer model. Unlike ordinary feedforward neural networks, iMixer involves MLP layers that propagate forward from the output side to the input side. We characterize the module as an example of invertible, implicit, and iterative mixing module. We evaluate the model performance with various datasets on image classification tasks, and find that iMixer, despite its unique architecture, exhibits stable learning capabilities and achieves performance comparable to or better than the baseline vanilla MLP-Mixer. The results imply that the correspondence between the Hopfield networks and the Mixer models serves as a principle for understanding a broader class of Transformer-like architecture designs. △ Less

Submitted 1 April, 2024; v1 submitted 25 April, 2023; originally announced April 2023.

Comments: 19 pages. v2: minor improvements

FFT-based Dynamic Token Mixer for Vision

Authors: Yuki Tatsunami, Masato Taki

Abstract: Multi-head-self-attention (MHSA)-equipped models have achieved notable performance in computer vision. Their computational complexity is proportional to quadratic numbers of pixels in input feature maps, resulting in slow processing, especially when dealing with high-resolution images. New types of token-mixer are proposed as an alternative to MHSA to circumvent this problem: an FFT-based token-mi… ▽ More Multi-head-self-attention (MHSA)-equipped models have achieved notable performance in computer vision. Their computational complexity is proportional to quadratic numbers of pixels in input feature maps, resulting in slow processing, especially when dealing with high-resolution images. New types of token-mixer are proposed as an alternative to MHSA to circumvent this problem: an FFT-based token-mixer involves global operations similar to MHSA but with lower computational complexity. However, despite its attractive properties, the FFT-based token-mixer has not been carefully examined in terms of its compatibility with the rapidly evolving MetaFormer architecture. Here, we propose a novel token-mixer called Dynamic Filter and novel image recognition models, DFFormer and CDFFormer, to close the gaps above. The results of image classification and downstream tasks, analysis, and visualization show that our models are helpful. Notably, their throughput and memory efficiency when dealing with high-resolution image recognition is remarkable. Our results indicate that Dynamic Filter is one of the token-mixer options that should be seriously considered. The code is available at https://github.com/okojoalg/dfformer △ Less

Submitted 17 December, 2023; v1 submitted 7 March, 2023; originally announced March 2023.

Comments: The 38th Annual AAAI Conference on Artificial Intelligence (AAAI'24)

Example-Based Explainable AI and its Application for Remote Sensing Image Classification

Authors: Shin-nosuke Ishikawa, Masato Todo, Masato Taki, Yasunobu Uchiyama, Kazunari Matsunaga, Peihsuan Lin, Taiki Ogihara, Masao Yasui

Abstract: We present a method of explainable artificial intelligence (XAI), "What I Know (WIK)", to provide additional information to verify the reliability of a deep learning model by showing an example of an instance in a training dataset that is similar to the input data to be inferred and demonstrate it in a remote sensing image classification task. One of the expected roles of XAI methods is verifying… ▽ More We present a method of explainable artificial intelligence (XAI), "What I Know (WIK)", to provide additional information to verify the reliability of a deep learning model by showing an example of an instance in a training dataset that is similar to the input data to be inferred and demonstrate it in a remote sensing image classification task. One of the expected roles of XAI methods is verifying whether inferences of a trained machine learning model are valid for an application, and it is an important factor that what datasets are used for training the model as well as the model architecture. Our data-centric approach can help determine whether the training dataset is sufficient for each inference by checking the selected example data. If the selected example looks similar to the input data, we can confirm that the model was not trained on a dataset with a feature distribution far from the feature of the input data. With this method, the criteria for selecting an example are not merely data similarity with the input data but also data similarity in the context of the model task. Using a remote sensing image dataset from the Sentinel-2 satellite, the concept was successfully demonstrated with reasonably selected examples. This method can be applied to various machine-learning tasks, including classification and regression. △ Less

Submitted 2 February, 2023; originally announced February 2023.

Comments: 10 pages, 4 figures, accepted for publication in International Journal of Applied Earth Observation and Geoinformation

Breathing of dissipative light bullets of nonlinear polarization mode in Kerr resonators

Authors: S. S. Gopalakrishnan, M. Tlidi, M. Taki, K. Panajotov

Abstract: We demonstrate the existence of breathing dissipative light bullets in a birefringent optical resonator filled with Kerr media. The propagation of light inside the cavity for each polarized component, which is coupled by cross-phase modulation, is described by the coupled Lugiato-Lefever equations. The space-time dynamics of breathing light bullets are described using Stokes parameters and frequen… ▽ More We demonstrate the existence of breathing dissipative light bullets in a birefringent optical resonator filled with Kerr media. The propagation of light inside the cavity for each polarized component, which is coupled by cross-phase modulation, is described by the coupled Lugiato-Lefever equations. The space-time dynamics of breathing light bullets are described using Stokes parameters and frequency spectra. △ Less

Submitted 18 May, 2022; originally announced May 2022.

Comments: 4

Sequencer: Deep LSTM for Image Classification

Authors: Yuki Tatsunami, Masato Taki

Abstract: In recent computer vision research, the advent of the Vision Transformer (ViT) has rapidly revolutionized various architectural design efforts: ViT achieved state-of-the-art image classification performance using self-attention found in natural language processing, and MLP-Mixer achieved competitive performance using simple multi-layer perceptrons. In contrast, several studies have also suggested… ▽ More In recent computer vision research, the advent of the Vision Transformer (ViT) has rapidly revolutionized various architectural design efforts: ViT achieved state-of-the-art image classification performance using self-attention found in natural language processing, and MLP-Mixer achieved competitive performance using simple multi-layer perceptrons. In contrast, several studies have also suggested that carefully redesigned convolutional neural networks (CNNs) can achieve advanced performance comparable to ViT without resorting to these new ideas. Against this background, there is growing interest in what inductive bias is suitable for computer vision. Here we propose Sequencer, a novel and competitive architecture alternative to ViT that provides a new perspective on these issues. Unlike ViTs, Sequencer models long-range dependencies using LSTMs rather than self-attention layers. We also propose a two-dimensional version of Sequencer module, where an LSTM is decomposed into vertical and horizontal LSTMs to enhance performance. Despite its simplicity, several experiments demonstrate that Sequencer performs impressively well: Sequencer2D-L, with 54M parameters, realizes 84.6% top-1 accuracy on only ImageNet-1K. Not only that, we show that it has good transferability and the robust resolution adaptability on double resolution-band. △ Less

Submitted 12 January, 2023; v1 submitted 4 May, 2022; originally announced May 2022.

Comments: Accepted in NeurIPS 2022; camera ready edition

Application of QUBO solver using black-box optimization to structural design for resonance avoidance

Authors: Tadayoshi Matsumori, Masato Taki, Tadashi Kadowaki

Abstract: Quadratic unconstrained binary optimization (QUBO) solvers can be applied to design an optimal structure to avoid resonance. QUBO algorithms that work on a classical or quantum device have succeeded in some industrial applications. However, their applications are still limited due to the difficulty of transforming from the original optimization problem to QUBO. Recently, black-box optimization (BB… ▽ More Quadratic unconstrained binary optimization (QUBO) solvers can be applied to design an optimal structure to avoid resonance. QUBO algorithms that work on a classical or quantum device have succeeded in some industrial applications. However, their applications are still limited due to the difficulty of transforming from the original optimization problem to QUBO. Recently, black-box optimization (BBO) methods have been proposed to tackle this issue using a machine learning technique and a Bayesian treatment for combinatorial optimization. We employed the BBO methods to design a printed circuit board for resonance avoidance. This design problem is formulated to maximize natural frequency and simultaneously minimize the number of mounting points. The natural frequency, which is the bottleneck for the QUBO formulation, is approximated to a quadratic model in the BBO method. We demonstrated that BBO using a factorization machine shows good performance in both the calculation time and the success probability of finding the optimal solution. Our results can open up QUBO solvers' potential for other applications in structural designs. △ Less

Submitted 19 July, 2022; v1 submitted 11 April, 2022; originally announced April 2022.

Comments: This is a preprint of an article published in Scientific Reports. The final authenticated version is available online at: 10.1038/s41598-022-16149-8

Journal ref: Sci Rep 12, 12143 (2022)

Drift instabilities in localised Faraday patterns

Authors: Juan F. Marín, Rafael Riveros Ávila, Saliya Coulibaly, Majid Taki, Mónica A. García-Ñustes

Abstract: Nature is intrinsically heterogeneous, and remarkable phenomena can only be observed in the presence of intrinsically nonlinear heterogeneities. Spontaneous pattern formation in nature has fascinated humankind for centuries, and the understanding of the underlying symmetry-breaking instabilities has been of longstanding scientific interest. In this article, we provide theoretical and experimental… ▽ More Nature is intrinsically heterogeneous, and remarkable phenomena can only be observed in the presence of intrinsically nonlinear heterogeneities. Spontaneous pattern formation in nature has fascinated humankind for centuries, and the understanding of the underlying symmetry-breaking instabilities has been of longstanding scientific interest. In this article, we provide theoretical and experimental evidence that heterogeneities can generate convection (drift instabilities) in the amplitude of localised patterns. We derive a minimal theoretical model describing the growth of localised Faraday patterns under heterogeneous parametric drive, unveiling the presence of symmetry-breaking nonlinear gradients. The model reveals new dynamics in the phase of the underlying patterns, exhibiting convective instabilities when the system crosses a secondary bifurcation point. We discuss the impact of our results in the understanding of convective instabilities induced by heterogeneities in generic nonlinear extended systems far from equilibrium. △ Less

Submitted 7 December, 2021; originally announced December 2021.

Comments: 18 pages, 4 figures. Under review

Counterfactual Explanation of Brain Activity Classifiers using Image-to-Image Transfer by Generative Adversarial Network

Authors: Teppei Matsui, Masato Taki, Trung Quang Pham, Junichi Chikazoe, Koji Jimura

Abstract: Deep neural networks (DNNs) can accurately decode task-related information from brain activations. However, because of the nonlinearity of the DNN, the decisions made by DNNs are hardly interpretable. One of the promising approaches for explaining such a black-box system is counterfactual explanation. In this framework, the behavior of a black-box system is explained by comparing real data and rea… ▽ More Deep neural networks (DNNs) can accurately decode task-related information from brain activations. However, because of the nonlinearity of the DNN, the decisions made by DNNs are hardly interpretable. One of the promising approaches for explaining such a black-box system is counterfactual explanation. In this framework, the behavior of a black-box system is explained by comparing real data and realistic synthetic data that are specifically generated such that the black-box system outputs an unreal outcome. Here we introduce a novel generative DNN (counterfactual activation generator, CAG) that can provide counterfactual explanations for DNN-based classifiers of brain activations. Importantly, CAG can simultaneously handle image transformation among multiple classes associated with different behavioral tasks. Using CAG, we demonstrated counterfactual explanation of DNN-based classifiers that learned to discriminate brain activations of seven behavioral tasks. Furthermore, by iterative applications of CAG, we were able to enhance and extract subtle spatial brain activity patterns that affected the classifier's decisions. Together, these results demonstrate that the counterfactual explanation based on image-to-image transformation would be a promising approach to understand and extend the current application of DNNs in fMRI analyses. △ Less

Submitted 28 October, 2021; originally announced October 2021.

Comments: 28 pages, 6 figures, 3 tables, 2 supplementary figures, 1 supplementary table

RaftMLP: How Much Can Be Done Without Attention and with Less Spatial Locality?

Authors: Yuki Tatsunami, Masato Taki

Abstract: For the past ten years, CNN has reigned supreme in the world of computer vision, but recently, Transformer has been on the rise. However, the quadratic computational cost of self-attention has become a serious problem in practice applications. There has been much research on architectures without CNN and self-attention in this context. In particular, MLP-Mixer is a simple architecture designed usi… ▽ More For the past ten years, CNN has reigned supreme in the world of computer vision, but recently, Transformer has been on the rise. However, the quadratic computational cost of self-attention has become a serious problem in practice applications. There has been much research on architectures without CNN and self-attention in this context. In particular, MLP-Mixer is a simple architecture designed using MLPs and hit an accuracy comparable to the Vision Transformer. However, the only inductive bias in this architecture is the embedding of tokens. This leaves open the possibility of incorporating a non-convolutional (or non-local) inductive bias into the architecture, so we used two simple ideas to incorporate inductive bias into the MLP-Mixer while taking advantage of its ability to capture global correlations. A way is to divide the token-mixing block vertically and horizontally. Another way is to make spatial correlations denser among some channels of token-mixing. With this approach, we were able to improve the accuracy of the MLP-Mixer while reducing its parameters and computational complexity. The small model that is RaftMLP-S is comparable to the state-of-the-art global MLP-based model in terms of parameters and efficiency per calculation. In addition, we tackled the problem of fixed input image resolution for global MLP-based models by utilizing bicubic interpolation. We demonstrated that these models could be applied as the backbone of architectures for downstream tasks such as object detection. However, it did not have significant performance and mentioned the need for MLP-specific architectures for downstream tasks for global MLP-based models. The source code in PyTorch version is available at \url{https://github.com/okojoalg/raft-mlp}. △ Less

Submitted 12 January, 2023; v1 submitted 9 August, 2021; originally announced August 2021.

Comments: ACCV2022 accepted

Optical Crystals and Light-Bullets in Kerr Resonators

Authors: M. Tlidi, S. S. Gopalakrishnan, M. Taki, K. Panajotov

Abstract: Stable light bullets and clusters of them are presented in the monostable regime using the mean-field Lugiato-Lefever equation [Gopalakrishnan, Panajotov, Taki, and Tlidi, Phys. Rev. Lett. 126, 153902 (2021)]. It is shown that three-dimensional (3D) dissipative structures occur in a strongly nonlinear regime where modulational instability is subcritical. We provide a detailed analysis on the forma… ▽ More Stable light bullets and clusters of them are presented in the monostable regime using the mean-field Lugiato-Lefever equation [Gopalakrishnan, Panajotov, Taki, and Tlidi, Phys. Rev. Lett. 126, 153902 (2021)]. It is shown that three-dimensional (3D) dissipative structures occur in a strongly nonlinear regime where modulational instability is subcritical. We provide a detailed analysis on the formation of optical 3D crystals in both the super- and sub-critical modulational instability regimes, and we highlight their link to the formation of light bullets in diffractive and dispersive Kerr resonators. We construct bifurcation diagrams associated with the formation of optical crystals in both monostable and bistable regimes. An analytical study has predicted the predominance of body-centered-cubic (bcc) crystals in the intracavity field over a large variety of other 3D solutions with less symmetry. These results have been obtained using a weakly nonlinear analysis but have never been checked numerically. We show numerically that indeed the most robust structures over other self-organized crystals are the bcc crystals. Finally, we show that light-bullets and clusters of them can occur also in a bistable regime. △ Less

Submitted 30 July, 2021; originally announced July 2021.

Machine learning application to Fermi-LAT data: sharpening all-sky map and emphasizing variable sources

Authors: Shogo Sato, Jun Kataoka, Soichiro Ito, Jun'ichi Kotoku, Masato Taki, Asuka Oyama, Takaya Toyoda, Yuki Nakamura, Marino Yamamoto

Abstract: A novel application of machine-learning (ML) based image processing algorithms is proposed to analyze an all-sky map (ASM) obtained using the Fermi Gamma-ray Space Telescope. An attempt was made to simulate a one-year ASM from a short-exposure ASM generated from one-week observation by applying three ML based image processing algorithms: dictionary learning, U-net, and Noise2Noise. Although the in… ▽ More A novel application of machine-learning (ML) based image processing algorithms is proposed to analyze an all-sky map (ASM) obtained using the Fermi Gamma-ray Space Telescope. An attempt was made to simulate a one-year ASM from a short-exposure ASM generated from one-week observation by applying three ML based image processing algorithms: dictionary learning, U-net, and Noise2Noise. Although the inference based on ML is less clear compared to standard likelihood analysis, the quality of the ASM was generally improved. In particular, the complicated diffuse emission associated with the galactic plane was successfully reproduced only from one-week observation data to mimic a ground truth (GT) generated from a one-year observation. Such ML algorithms can be implemented relatively easily to provide sharper images without various assumptions of emission models. In contrast, large deviations between simulated ML maps and GT map were found, which are attributed to the significant temporal variability of blazar-type active galactic nuclei (AGNs) over a year. Thus, the proposed ML methods are viable not only to improve the image quality of an ASM, but also to detect variable sources, such as AGNs, algorithmically, i.e., without human bias. Moreover, we argue that this approach is widely applicable to ASMs obtained by various other missions; thus, it has the potential to examine giant structures and transient events, both of which are rarely found in pointing observations. △ Less

Submitted 2 April, 2021; originally announced April 2021.

Comments: 12 pages, 11 figures, accepted for publication in ApJ

Dissipative soliton interaction in Kerr resonators with high-order dispersion

Authors: Andrei G. Vladimirov, Mustapha Tlidi, Majid Taki

Abstract: We consider an optical resonator containing a photonic crystal fiber and driven coherently by an injected beam. This device is described by a generalized Lugiato-Lefever equation with fourth order dispersion. We use an asymptotic approach to derive interaction equations governing the slow time evolution of the coordinates of two interacting dissipative solitons. We show that Cherenkov radiation in… ▽ More We consider an optical resonator containing a photonic crystal fiber and driven coherently by an injected beam. This device is described by a generalized Lugiato-Lefever equation with fourth order dispersion. We use an asymptotic approach to derive interaction equations governing the slow time evolution of the coordinates of two interacting dissipative solitons. We show that Cherenkov radiation induced by positive fourth-order dispersion leads to a strong increase of the interaction force between the solitons. As a consequence, large number of equidistant soliton bound states in the phase space of the interaction equations can be stabilized. We show that the presence of even small spectral filtering not only dampens the Cherenkov radiation at the soliton tails and reduces the interaction strength, but can also affect the the bound state stability. △ Less

Submitted 7 February, 2021; originally announced February 2021.

Comments: 7 pages, 7 figures

MSC Class: 78A60; 35Q55; 35Q60

Journal ref: Phys. Rev. A 103, 063505 (2021)

Dissipative light bullets in Kerr cavities: multi-stability, clustering, and rogue waves

Authors: Shyam Sunder Gopalakrishnan, Krassimir Panajotov, Majid Taki, Mustapha Tlidi

Abstract: We report the existence of stable dissipative light bullets in Kerr cavities. These three-dimensional (3D) localized structures consist of either an isolated light bullet (LB), or could occur in clusters forming well-defined 3D patterns. They can be seen as stationary states in the reference frame moving with the group velocity of light within the cavity. The number of LBs and their distribution i… ▽ More We report the existence of stable dissipative light bullets in Kerr cavities. These three-dimensional (3D) localized structures consist of either an isolated light bullet (LB), or could occur in clusters forming well-defined 3D patterns. They can be seen as stationary states in the reference frame moving with the group velocity of light within the cavity. The number of LBs and their distribution in 3D settings are determined by the initial conditions while their maximum peak power remains constant for a fixed value of the system parameters. Their bifurcation diagram allows us to explain this phenomenon as a manifestation of homoclinic snaking for dissipative light bullets. However, when the strength of the injected beam is increased, LBs lose their stability and the cavity field exhibits giant, short living three-dimensional pulses. The statistical characterization of pulse amplitude reveals a long tail probability distribution indicating the occurrence of extreme events; often called rogue waves. △ Less

Submitted 18 November, 2020; originally announced November 2020.

Journal ref: Phys. Rev. Lett. 126, 153902 (2021)

CNN-based event classification for alpha-decay events in nuclear emulsion

Authors: J. Yoshida, H. Ekawa, A. Kasagi, M. Nakagawa, K. Nakazawa, N. Saito, T. R. Saito, M. Taki, M. Yoshimoto

Abstract: We developed an efficient classifier that sorts alpha-decay events from various vertex-like objects in nuclear emulsion using a convolutional neural network (CNN). Alpha-decay events in the emulsion are standard calibration sources for the relation between the track length and kinetic energy in each emulsion sheet. We trained the CNN using 15,885 images of vertex-like objects including 906 alpha-d… ▽ More We developed an efficient classifier that sorts alpha-decay events from various vertex-like objects in nuclear emulsion using a convolutional neural network (CNN). Alpha-decay events in the emulsion are standard calibration sources for the relation between the track length and kinetic energy in each emulsion sheet. We trained the CNN using 15,885 images of vertex-like objects including 906 alpha-decay events and tested it using a dataset of 46,948 images including 255 alpha-decay events. By tuning the hyperparameters of the CNN, the trained models achieved an Average Precision Score of 0.740 +/- 0.009 for the test dataset. For the model obtained, a discrimination threshold of the classification can be arbitrarily adjusted according to the balance between the precision and recall. The precision and recall of the classification using previous method without a CNN were 0.081 +/- 0.006 and 0.788 +/- 0.056, respectively, for the same dataset. By contrast, the developed classifier obtained a precision of 0.547 +/- 0.025 when a similar recall value of 0.788 was set. The developed CNN method reduced the human load for further visual inspection after the classification by approximately 1/7 compared to the estimated load of the former method without a CNN. △ Less

Submitted 12 September, 2020; originally announced September 2020.

Turbulence-induced rogue waves in Kerr resonators

Authors: Saliya Coulibaly, Majid Taki, Abdelkrim Bendahmane, Guy Millot, Bertrand Kibler, Marcel Gabriel Clerc

Abstract: Spontaneous emergence of self-organized patterns and their bifurcations towards a regime of complex dynamics in non-equilibrium dissipative systems is a paradigm of phase transition. Indeed, the behavior of these patterns in the highly nonlinear regime remains less explored, even in recent high-quality-factor resonators such as Kerr-nonlinear optical ones. Here, we investigate theoretically and ex… ▽ More Spontaneous emergence of self-organized patterns and their bifurcations towards a regime of complex dynamics in non-equilibrium dissipative systems is a paradigm of phase transition. Indeed, the behavior of these patterns in the highly nonlinear regime remains less explored, even in recent high-quality-factor resonators such as Kerr-nonlinear optical ones. Here, we investigate theoretically and experimentally the alteration of the resulting Kerr frequency combs from the weakly to the highly nonlinear regime, in the frameworks of spatiotemporal chaos, and dissipative phase transitions. We reveal the existence of a striking and easily accessible scenario of spatiotemporal chaos, free of cavity solitons, in a monostable operating regime, wherein a transition to amplitude turbulence via spatiotemporal intermittency is evidenced. Moreover, statistics of the light bursts in the resulting turbulent regime unveils the existence of rogue waves as extreme events characterized by long-tail statistics. △ Less

Submitted 9 January, 2019; originally announced January 2019.

Journal ref: Phys. Rev. X 9, 011054 (2019)

Deep Residual Networks and Weight Initialization

Authors: Masato Taki

Abstract: Residual Network (ResNet) is the state-of-the-art architecture that realizes successful training of really deep neural network. It is also known that good weight initialization of neural network avoids problem of vanishing/exploding gradients. In this paper, simplified models of ResNets are analyzed. We argue that goodness of ResNet is correlated with the fact that ResNets are relatively insensiti… ▽ More Residual Network (ResNet) is the state-of-the-art architecture that realizes successful training of really deep neural network. It is also known that good weight initialization of neural network avoids problem of vanishing/exploding gradients. In this paper, simplified models of ResNets are analyzed. We argue that goodness of ResNet is correlated with the fact that ResNets are relatively insensitive to choice of initial weights. We also demonstrate how batch normalization improves backpropagation of deep ResNets without tuning initial values of weights. △ Less

Submitted 9 September, 2017; originally announced September 2017.

Comments: 10 pages, 4 figures

Report number: RIKEN-iTHEMS-Report-17

Charged Higgs Signals in $t\,\overline{t}\,H$ Searches

Authors: Daniele S. M. Alves, Sonia El Hedri, Anna Maria Taki, Neal Weiner

Abstract: New scalars from an extended Higgs sector could have weak scale masses and still have escaped detection. In a Type I Two Higgs Doublet Model, for instance, even the charged Higgs can be lighter than the top quark. Because electroweak production of these scalars is modest, the greatest opportunity for their detection might come from rare top decays. For mass hierarchies of the type… ▽ More New scalars from an extended Higgs sector could have weak scale masses and still have escaped detection. In a Type I Two Higgs Doublet Model, for instance, even the charged Higgs can be lighter than the top quark. Because electroweak production of these scalars is modest, the greatest opportunity for their detection might come from rare top decays. For mass hierarchies of the type $m_t>m_{H^\pm}>m_{A^0,\,H^0}$, the natural signal can arise from top quark pair production, followed by the decay chain $t \rightarrow b\,H^+$, $H^+ \rightarrow W^{+(*)} φ^0$, $φ^0\rightarrow b\overline{b},\,τ^+τ^-$, where $φ^0=A^0,\,H^0$. These final states largely overlap with those of the Standard Model $t\,\bar{t}\,H^0_{_\text{SM}}$ process, and therefore can potentially contaminate $t\,\bar{t}\,H^0_{_\text{SM}}$ searches. We demonstrate that existing $t\,\bar{t}\,H^0_{_\text{SM}}$ analyses can already probe light extended Higgs sectors, and we derive new constraints from their results. Furthermore, we note that existing excesses in $t\,\bar{t}\,H^0_{_\text{SM}}$ searches can be naturally explained by the contamination of rare top decays to new light Higgses. We discuss how to distinguish this signal from the Standard Model process. △ Less

Submitted 24 March, 2017; v1 submitted 20 March, 2017; originally announced March 2017.

Comments: Added discussion about CDF ttH measurements; several typos fixed; figures 1-3 adjusted; references added

Journal ref: Phys. Rev. D 96, 075032 (2017)

Localized Faraday patterns under heterogeneous parametric excitation

Authors: Héctor Urra, Juan F. Marín, Milena Páez-Silva, Majid Taki, Saliya Coulibaly, Leonardo Gordillo, Mónica A. García-Ñustes

Abstract: Faraday waves are a classic example of a system in which an extended pattern emerges under spatially uniform forcing. Motivated by systems in which uniform excitation is not plausible, we study both experimentally and theoretically the effect of heterogeneous forcing on Faraday waves. Our experiments show that vibrations restricted to finite regions lead to the formation of localized subharmonic w… ▽ More Faraday waves are a classic example of a system in which an extended pattern emerges under spatially uniform forcing. Motivated by systems in which uniform excitation is not plausible, we study both experimentally and theoretically the effect of heterogeneous forcing on Faraday waves. Our experiments show that vibrations restricted to finite regions lead to the formation of localized subharmonic wave patterns and change the onset of the instability. The prototype model used for the theoretical calculations is the parametrically driven and damped nonlinear Schrödinger equation, which is known to describe well Faraday-instability regimes. For an energy injection with a Gaussian spatial profile, we show that the evolution of the envelope of the wave pattern can be reduced to a Weber-equation eigenvalue problem. Our theoretical results provide very good predictions of our experimental observations provided that the decay length scale of the Gaussian profile is much larger than the pattern wavelength. △ Less

Submitted 14 March, 2019; v1 submitted 8 February, 2017; originally announced February 2017.

Comments: 10 pages, 9 figures, Accepted

Journal ref: Phys. Rev. E 99, 033115 (2019)

Characterization of spatiotemporal chaos in Kerr optical frequency {comb generators} and in fiber cavity

Authors: Z. Liu, M. Ouali, S. Coulibaly, M. G. Clerc, M. Taki, M. Tlidi

Abstract: Complex spatiotemporal dynamics have been a subject of recent experimental investigations in optical frequency comb microresonators and in driven fiber cavities with a Kerr-type media. We show that this complex behavior has a spatiotemporal chaotic nature. We determine numerically the Lyapunov spectra, allowing to characterize different dynamical behavior occurring in these simple devices. The Yor… ▽ More Complex spatiotemporal dynamics have been a subject of recent experimental investigations in optical frequency comb microresonators and in driven fiber cavities with a Kerr-type media. We show that this complex behavior has a spatiotemporal chaotic nature. We determine numerically the Lyapunov spectra, allowing to characterize different dynamical behavior occurring in these simple devices. The Yorke-Kaplan dimension is used as an order parameter to characterize the bifurcation diagram. We identify a wide regime of parameters where the system exhibits a coexistence between the spatiotemporal chaos, the oscillatory localized structure, and the homogeneous steady state. The destabilization of an oscillatory localized state through radiation of counter propagative fronts between the homogeneous and the spatiotemporal chaotic states is analyzed. To characterize better the spatiotemporal chaos, we estimate the front speed as a function of the pump intensity. △ Less

Submitted 21 January, 2017; originally announced January 2017.

Quark confinement due to creation of micro AdS black holes in quarkonium model

Authors: Mehran Taki, Abolfazl Mirjalili

Abstract: We use the solution of the Dirac equation for quarkonium atom in the 4D Anti de sitter (AdS$_{4}$) space to investigate the effect of the large negative cosmological constant on the phenomenon of quark confinement. We do the required calculations in the AdS$_{4}$ space to indicate that large cosmological constant can describe the quark confinement. In fact using the coulomb potential in Dirac equa… ▽ More We use the solution of the Dirac equation for quarkonium atom in the 4D Anti de sitter (AdS$_{4}$) space to investigate the effect of the large negative cosmological constant on the phenomenon of quark confinement. We do the required calculations in the AdS$_{4}$ space to indicate that large cosmological constant can describe the quark confinement. In fact using the coulomb potential in Dirac equation while we employ the AdS metric will additionally lead us to a linear potential in the quark-antiquark interaction which can be considered to explain the quark confinement. This confining term is arising essentially from the geometrical features of the space. On the other hand the origin of the large cosmological constant can be justified by assuming the appearance of micro black holes in the recent hadronic collision process which is now current, for instance, at the LHC project. △ Less

Submitted 15 November, 2016; originally announced November 2016.

Extended and localized Hopf-Turing mixed-mode in non-instantaneous Kerr cavities

Authors: M. Ouali, S. Coulibaly, M. Taki, M. Tlidi

Abstract: We investigate the spatio-temporal dynamics of a ring cavity filled with a non-instantaneous Kerr medium and driven by a coherent injected beam. We show the existence of a stable mixed-mode solution that can be either extended or localized in space. The mixed-mode solutions are obtained in a regime where Turing instability (often called modulational instability) interacts with self-pulsing phenome… ▽ More We investigate the spatio-temporal dynamics of a ring cavity filled with a non-instantaneous Kerr medium and driven by a coherent injected beam. We show the existence of a stable mixed-mode solution that can be either extended or localized in space. The mixed-mode solutions are obtained in a regime where Turing instability (often called modulational instability) interacts with self-pulsing phenomenon (Andronov-Hopf bifurcation). We numerically describe the transition from stationary inhomogeneous solutions to a branch of mixed-mode solutions. We characterize this transition by constructing the bifurcation diagram associated with these solutions. Finally, we show stable localized mixed-mode solutions, which consist of time-periodic oscillations that are localized in space. △ Less

Submitted 13 November, 2016; originally announced November 2016.

Comments: 5 pages, 5 figures

More on 5d descriptions of 6d SCFTs

Authors: Hirotaka Hayashi, Sung-Soo Kim, Kimyeong Lee, Masato Taki, Futoshi Yagi

Abstract: We propose new five-dimensional gauge theory descriptions of six-dimensional $\mathcal{N}=(1,0)$ superconformal field theories arising from type IIA brane configurations including an $ON^0$-plane. The new five-dimensional gauge theories may have $SO$, $Sp$, and $SU$ gauge groups and further broaden the landscape of ultraviolet complete five-dimensional $\mathcal{N}=1$ supersymmetric gauge theories… ▽ More We propose new five-dimensional gauge theory descriptions of six-dimensional $\mathcal{N}=(1,0)$ superconformal field theories arising from type IIA brane configurations including an $ON^0$-plane. The new five-dimensional gauge theories may have $SO$, $Sp$, and $SU$ gauge groups and further broaden the landscape of ultraviolet complete five-dimensional $\mathcal{N}=1$ supersymmetric gauge theories. When we include an $O8^-$-plane in addition to an $ON^0$-plane, T-duality yields two $O7^-$-planes at the intersections of an $ON^0$-plane and two $O5^0$-planes. We propose a novel resolution of the $O7^-$-plane with four D7-branes in such a configuration, which enables us to obtain three different types of five-dimensional gauge theories, depending on whether we resolve either none or one or two $O7^-$-planes. Such different possibilities yield a new five-dimensional duality between a D-type $SU$ quiver and an $SO-Sp$ quiver theories. We also claim that a twisted circle compactification of a six-dimensional superconformal field theory may lead to a five-dimensional gauge theory different from those obtained by a simple circle compactification. △ Less

Submitted 30 October, 2016; v1 submitted 27 December, 2015; originally announced December 2015.

Comments: 41 pages, published version

Report number: IFT-UAM/CSIC-15-134, KIAS-Q15006, RIKEN-STAMP-23

Journal ref: JHEP10(2016)126

Spectral curves of $\mathcal{N}=1$ theories of class $\mathcal{S}_k$

Authors: Ioana Coman, Elli Pomoni, Masato Taki, Futoshi Yagi

Abstract: We study the Coulomb branch of class $\mathcal{S}_k$ $\mathcal{N} = 1$ SCFTs by constructing and analyzing their spectral curves. We study the Coulomb branch of class $\mathcal{S}_k$ $\mathcal{N} = 1$ SCFTs by constructing and analyzing their spectral curves. △ Less

Submitted 18 December, 2015; originally announced December 2015.

Comments: 37 pages, 13 figures

Peregrine comb: multiple compression points for Peregrine rogue waves in periodically modulated nonlinear Schr{ö}dinger equations

Authors: Gaston Thiofack, Saliya Coulibaly, Majid Taki, Stephan De Bievre, Guillaume Dujardin

Abstract: It is shown that sufficiently large periodic modulations in the coefficients of a nonlinear Schr{ö}dinger equation can drastically impact the spatial shape of the Peregrine soliton solutions: they can develop multiple compression points of the same amplitude, rather than only a single one, as in the spatially homogeneous focusing nonlinear Schr{ö}dinger equation. The additional compression points… ▽ More It is shown that sufficiently large periodic modulations in the coefficients of a nonlinear Schr{ö}dinger equation can drastically impact the spatial shape of the Peregrine soliton solutions: they can develop multiple compression points of the same amplitude, rather than only a single one, as in the spatially homogeneous focusing nonlinear Schr{ö}dinger equation. The additional compression points are generated in pairs forming a comb-like structure. The number of additional pairs depends on the amplitude of the modulation but not on its wavelength, which controls their separation distance. The dynamics and characteristics of these generalized Peregrine soliton are analytically described in the case of a completely integrable modulation. A numerical investigation shows that their main properties persist in nonintegrable situations, where no exact analytical expression of the generalized Peregrine soliton is available. Our predictions are in good agreement with numerical findings for an interesting specific case of an experimentally realizable periodically dispersion modulated photonic crystal fiber. Our results therefore pave the way for the experimental control and manipulation of the formation of generalized Peregrine rogue waves in the wide class of physical systems modeled by the nonlinear Schr{ö}dinger equation. △ Less

Submitted 23 September, 2015; originally announced September 2015.

Journal ref: Phys. Rev. A 92, 043837 (2015)

A new 5d description of 6d D-type minimal conformal matter

Authors: Hirotaka Hayashi, Sung-Soo Kim, Kimyeong Lee, Masato Taki, Futoshi Yagi

Abstract: We propose a new 5d description of the circle-compactified 6d $(D_{N+4}, D_{N+4})$ minimal conformal matter theory which can be approached by the 6d $\mathcal{N}=(1,0)$ $Sp(N)$ gauge theory with $N_f=2N+8$ flavors and one tensor multiplet. Compactifying the brane set-up for the 6d theory, we arrive at a 5-brane Tao diagram for 5d $\mathcal{N}=1$ $SU(N+2)$ theory of the vanishing Chern-Simons level… ▽ More We propose a new 5d description of the circle-compactified 6d $(D_{N+4}, D_{N+4})$ minimal conformal matter theory which can be approached by the 6d $\mathcal{N}=(1,0)$ $Sp(N)$ gauge theory with $N_f=2N+8$ flavors and one tensor multiplet. Compactifying the brane set-up for the 6d theory, we arrive at a 5-brane Tao diagram for 5d $\mathcal{N}=1$ $SU(N+2)$ theory of the vanishing Chern-Simons level with $2N+8$ flavors. We conjecture that the 6d theory is recovered as the UV fixed point of this 5d theory. We show that the global symmetry of this 5d theory is $SO(4N+16)$ identical to that of the 6d theory by analyzing the 7-brane monodromy. By using the Tao diagram, we also find the instanton fugacity is exactly given by the circle radius. By decoupling flavors in this 5d theory, one can obtain all the 5d $SU(N+2)$ gauge theories of various Chern-Simons levels and corresponding enhanced global symmetries at the 5d UV fixed point. △ Less

Submitted 20 August, 2015; v1 submitted 17 May, 2015; originally announced May 2015.

Comments: v1: 21 pages, 10 figures. v2: mirror correction and references added. v3: published version

Report number: IFT-UAM/CSIC-15-044, KIAS-P15022, RIKEN-STAMP-9

Journal ref: JHEP 1508:097,2015

Tao Probing the End of the World

Authors: Sung-Soo Kim, Masato Taki, Futoshi Yagi

Abstract: We introduce a new type IIB 5-brane description for the E-string theory which is the world-volume theory on the M5-brane probing the end of the world M9-brane. The E-string in the new realization is depicted as spiral 5-branes web equipped with the cyclic structure which is key to uplifting to six dimensions. Utilizing the topological vertex to the 5-brane web configuration enables us to write dow… ▽ More We introduce a new type IIB 5-brane description for the E-string theory which is the world-volume theory on the M5-brane probing the end of the world M9-brane. The E-string in the new realization is depicted as spiral 5-branes web equipped with the cyclic structure which is key to uplifting to six dimensions. Utilizing the topological vertex to the 5-brane web configuration enables us to write down a combinatorial formula for the generating function of the E-string elliptic genera, namely the full partition function of topological strings on the local 1/2 K3 surface. △ Less

Submitted 2 August, 2015; v1 submitted 14 April, 2015; originally announced April 2015.

Comments: 15 figures, 41 pages, v2: minor revision, v3: published version

Report number: KIAS-P15010, RIKEN-STAMP-3

Journal ref: Prog. Theor. Exp. Phys. 2015, 083B02

Solitons and frequency combs in silica microring resonators: Interplay of the Raman and higher-order dispersion effects

Authors: Carles Milián, Andrey V. Gorbach, Majid Taki, Alexey V. Yulin, Dmitry V. Skryabin

Abstract: The influence of Raman scattering and higher order dispersions on solitons and frequency comb generation in silica microring resonators is investigated. The Raman effect introduces a threshold value in the resonator quality factor above which the frequency locked solitons can not exist and, instead, a rich dynamics characterized by generation of self-frequency shift- ing solitons and dispersive wa… ▽ More The influence of Raman scattering and higher order dispersions on solitons and frequency comb generation in silica microring resonators is investigated. The Raman effect introduces a threshold value in the resonator quality factor above which the frequency locked solitons can not exist and, instead, a rich dynamics characterized by generation of self-frequency shift- ing solitons and dispersive waves is observed. A mechanism of broadening of the Cherenkov radiation through Hopf instability of the frequency locked solitons is also reported. △ Less

Submitted 15 September, 2015; v1 submitted 2 March, 2015; originally announced March 2015.

Comments: 12 pages, 10 figures

Journal ref: Phys. Rev. A 92, 033851 (2015)

Fiber-Base Duality and Global Symmetry Enhancement

Authors: Vladimir Mitev, Elli Pomoni, Masato Taki, Futoshi Yagi

Abstract: We show that the 5D Nekrasov partition functions enjoy the enhanced global symmetry of the UV fixed point. The fiber-base duality is responsible for the global symmetry enhancement. For $SU(2)$ with $N_f\leq 7$ flavors the fiber-base symmetry together with the manifest flavor $SO(2N_f)$ symmetry generate the $E_{N_f+1}$ global symmetry, while in the higher rank case the manifest global symmetry of… ▽ More We show that the 5D Nekrasov partition functions enjoy the enhanced global symmetry of the UV fixed point. The fiber-base duality is responsible for the global symmetry enhancement. For $SU(2)$ with $N_f\leq 7$ flavors the fiber-base symmetry together with the manifest flavor $SO(2N_f)$ symmetry generate the $E_{N_f+1}$ global symmetry, while in the higher rank case the manifest global symmetry of the two dual theories related by the fiber-base duality map generate the symmetry enhancement. The symmetry enhancement at the level of the partition function is manifest once we chose an appropriate reparametrization for the Coulomb moduli. △ Less

Submitted 10 November, 2014; originally announced November 2014.

Comments: 61 pages, 22 figures

Impact of third order dispersion on nonlinear bifurcations in optical resonators

Authors: François Leo, Stéphane Coen, Pascal Kockaert, Philippe Emplit, Marc Haelterman, Arnaud Mussot, Majid Taki

Abstract: It is analytically shown that symmetry breaking, in dissipative systems, affects the nature of the bifurcation at onset of instability resulting in transitions from super to subcritical bifurcations. In the case of a nonlinear fiber cavity, we have derived an amplitude equation to describe the nonlinear dynamics above threshold. An analytical expression of the critical transition curve is obtained… ▽ More It is analytically shown that symmetry breaking, in dissipative systems, affects the nature of the bifurcation at onset of instability resulting in transitions from super to subcritical bifurcations. In the case of a nonlinear fiber cavity, we have derived an amplitude equation to describe the nonlinear dynamics above threshold. An analytical expression of the critical transition curve is obtained and the predictions are in excellent agreement with the numerical solutions of the full dynamical model. △ Less

Submitted 15 May, 2014; originally announced May 2014.

Comments: 5 pages, 2 figures

Journal ref: Phys. Lett. A, 379 (34), pp. 1934-1937 (2015)

On AGT-W Conjecture and q-Deformed W-Algebra

Authors: Masato Taki

Abstract: We propose an extension of the Alday-Gaiotto-Tachikawa-Wyllard conjecture to 5d SU(N) gauge theories. A Nekrasov partition function then coincides with the scalar product of the corresponding Gaiotto-Whittaker vectors of the q-deformed W_N algebra. We propose an extension of the Alday-Gaiotto-Tachikawa-Wyllard conjecture to 5d SU(N) gauge theories. A Nekrasov partition function then coincides with the scalar product of the corresponding Gaiotto-Whittaker vectors of the q-deformed W_N algebra. △ Less

Submitted 27 March, 2014; originally announced March 2014.

Comments: 18 pages

Seiberg Duality, 5d SCFTs and Nekrasov Partition Functions

Authors: Masato Taki

Abstract: It is known that a 4d N = 1 SCFT lives on D3-branes probing a local del Pezzo Calabi-Yau singularity. The Seiberg (or toric) duality of this SCFT arises from the Picard-Lefshetz transformation of the affine E_N 7-brane background that is associated with the Calabi-Yau threefold. In this paper we study the duality of the affine E_N background itself and a 5-brane probing it. We then find that many… ▽ More It is known that a 4d N = 1 SCFT lives on D3-branes probing a local del Pezzo Calabi-Yau singularity. The Seiberg (or toric) duality of this SCFT arises from the Picard-Lefshetz transformation of the affine E_N 7-brane background that is associated with the Calabi-Yau threefold. In this paper we study the duality of the affine E_N background itself and a 5-brane probing it. We then find that many different Type IIB 5-brane webs describe the same SCFT in 5d. We check this duality by comparing the Nekrasov partition functions of these 5-brane web configurations. △ Less

Submitted 28 January, 2014; originally announced January 2014.

Comments: 66 pages, 37 figures

Notes on Enhancement of Flavor Symmetry and 5d Superconformal Index

Authors: Masato Taki

Abstract: The UV fixed point theory of SU(2) gauge theory with N_f = 0,1,...,7 flavors is believed to have the enlarged E_{N_f +1} flavor symmetry. Actually it is not easy to check this conjecture because the UV theory is strongly-coupled, however, computation of certain SUSY protected quantities provides strong evidence for the enhancement of flavor symmetry. We study the superconformal index for SU(2) gau… ▽ More The UV fixed point theory of SU(2) gauge theory with N_f = 0,1,...,7 flavors is believed to have the enlarged E_{N_f +1} flavor symmetry. Actually it is not easy to check this conjecture because the UV theory is strongly-coupled, however, computation of certain SUSY protected quantities provides strong evidence for the enhancement of flavor symmetry. We study the superconformal index for SU(2) gauge theory with N_f = 0, 1 flavors in details, and we give a support for the enhancement by studying combinatorial structure of the superconformal indexes of these theories. We also give a nontrivial evidence that the local F_2 geometry leads to the E_1 superconformal field theory. △ Less

Submitted 28 October, 2013; originally announced October 2013.

Comments: 32 pages, 4 figures

Non-Lagrangian Theories from Brane Junctions

Authors: Ling Bao, Vladimir Mitev, Elli Pomoni, Masato Taki, Futoshi Yagi

Abstract: In this article we use 5-brane junctions to study the 5D T_N SCFTs corresponding to the 5D N=1 uplift of the 4D N=2 strongly coupled gauge theories, which are obtained by compactifying N M5 branes on a sphere with three full punctures. Even though these theories have no Lagrangian description, by using the 5-brane junctions proposed by Benini, Benvenuti and Tachikawa, we are able to derive their S… ▽ More In this article we use 5-brane junctions to study the 5D T_N SCFTs corresponding to the 5D N=1 uplift of the 4D N=2 strongly coupled gauge theories, which are obtained by compactifying N M5 branes on a sphere with three full punctures. Even though these theories have no Lagrangian description, by using the 5-brane junctions proposed by Benini, Benvenuti and Tachikawa, we are able to derive their Seiberg-Witten curves and Nekrasov partition functions. We cross-check our results with the 5D superconformal index proposed by Kim, Kim and Lee. Through the AGTW correspondence, we discuss the relations between 5D superconformal indices and n-point functions of the q-deformed W_N Toda theories. △ Less

Submitted 8 November, 2013; v1 submitted 14 October, 2013; originally announced October 2013.

Comments: 66 pages, 28 figures

Holomorphic Blocks for 3d Non-abelian Partition Functions

Authors: Masato Taki

Abstract: The most recent studies on the supersymmetric localization reveal many non-trivial features of supersymmetric field theories in diverse dimensions, and 3d gauge theory provides a typical example. It was conjectured that the index and the partition function of a 3d N=2 theory are constructed from a single component: the holomorphic block. We prove this conjecture for non-abelian gauge theories by c… ▽ More The most recent studies on the supersymmetric localization reveal many non-trivial features of supersymmetric field theories in diverse dimensions, and 3d gauge theory provides a typical example. It was conjectured that the index and the partition function of a 3d N=2 theory are constructed from a single component: the holomorphic block. We prove this conjecture for non-abelian gauge theories by computing exactly the 3d partition functions and holomorphic blocks. △ Less

Submitted 24 March, 2013; originally announced March 2013.

Comments: 24 pages, 4 figures

W_3 irregular states and isolated N=2 superconformal field theories

Authors: Hiroaki Kanno, Kazunobu Maruyoshi, Shotaro Shiba, Masato Taki

Abstract: We explore the proposal that the six-dimensional (2,0) theory on the Riemann surface with irregular punctures leads to a four-dimensional gauge theory coupled to the isolated N=2 superconformal theories of Argyres-Douglas type, and to two-dimensional conformal field theory with irregular states. Following the approach of Gaiotto-Teschner for the Virasoro case, we construct W_3 irregular states by… ▽ More We explore the proposal that the six-dimensional (2,0) theory on the Riemann surface with irregular punctures leads to a four-dimensional gauge theory coupled to the isolated N=2 superconformal theories of Argyres-Douglas type, and to two-dimensional conformal field theory with irregular states. Following the approach of Gaiotto-Teschner for the Virasoro case, we construct W_3 irregular states by colliding a single SU(3) puncture with several regular punctures of simple type. If n simple punctures are colliding with the SU(3) puncture, the resulting irregular state is a simultaneous eigenvector of the positive modes L_n, ..., L_{2n} and W_{2n}, ..., W_{3n} of the W_3 algebra. We find the corresponding isolated SCFT with an SU(3) flavor symmetry as a nontrivial IR fixed point on the Coulomb branch of the SU(3) linear quiver gauge theories, by confirming that its Seiberg-Witten curve correctly predicts the conditions for the W_3 irregular states. We also compare these SCFT's with the ones obtained from the BPS quiver method. △ Less

Submitted 1 February, 2013; v1 submitted 4 January, 2013; originally announced January 2013.

Comments: 1+53 pages, 1 figure; v2: minor corrections

Report number: KEK-TH-1595, RIKEN-MP-66

Asymmetric Cloaking Theory Based on Finsler Geometry ~ How to design Harry Potter's invisibility cloak with a scientific method ~

Authors: Tomohiro Amemiya, Daisuke Nishiyama, Masato Taki

Abstract: Is it possible to actually make Harry's invisibility cloaks? The most promising approach for realizing such magical cloaking in our real world would be to use transformation optics, where an empty space with a distorted geometry is imitated with a non-distorted space but filled with transformation medium having appropriate permittivity and permeability. An important requirement for true invisibili… ▽ More Is it possible to actually make Harry's invisibility cloaks? The most promising approach for realizing such magical cloaking in our real world would be to use transformation optics, where an empty space with a distorted geometry is imitated with a non-distorted space but filled with transformation medium having appropriate permittivity and permeability. An important requirement for true invisibility cloaks is nonreciprocity; that is, a person in the cloak should not be seen from the outside but should be able to see the outside. This invisibility cloak, or a nonreciprocal shield, cannot be created as far as we stay in conventional transformation optics. Conventional transformation optics is based on Riemann geometry with a metric tensor independent of direction, and therefore cannot be used to design the nonreciprocal shield. To overcome this problem, we propose an improved theory of transformation optics that is based on Finsler geometry, an extended version of Riemann geometry. Our theory shows that nonreciprocal shielding can be realized by covering cloaking space with transformation medium having anisotropic, nonreciprocal permittivity and permeability. This theory includes conventional transformation optics as special cases. We show the method for designing the spatial distribution of the permittivity and permeability required to make the nonreciprocal shield. △ Less

Submitted 31 January, 2013; v1 submitted 13 November, 2012; originally announced November 2012.

Comments: 10 pages, 2 figures

Generalized Whittaker states for instanton counting with fundamental hypermultiplets

Authors: Hiroaki Kanno, Masato Taki

Abstract: M-theoretic construction of N=2 gauge theories implies that the instanton partition function is expressed as the scalar product of coherent states (Whittaker states) in the Verma module of an appropriate two dimensional conformal field theory. We present the characterizing conditions for such states that give the partition function with fundamental hypermultiplets for SU(3) theory and SU(2) theory… ▽ More M-theoretic construction of N=2 gauge theories implies that the instanton partition function is expressed as the scalar product of coherent states (Whittaker states) in the Verma module of an appropriate two dimensional conformal field theory. We present the characterizing conditions for such states that give the partition function with fundamental hypermultiplets for SU(3) theory and SU(2) theory with a surface operator. We find the states are no longer the coherent states in the strict sense but we can characterize them in terms of a few annihilation operators of lower levels combined with the zero mode (Cartan part) of the Virasoro algebra L_0 or the sl(2) current algebra J_0^0. △ Less

Submitted 19 March, 2012; v1 submitted 7 March, 2012; originally announced March 2012.

Comments: 46 pages; minor changes

M5-branes, toric diagrams and gauge theory duality

Authors: Ling Bao, Elli Pomoni, Masato Taki, Futoshi Yagi

Abstract: In this article we explore the duality between the low energy effective theory of five-dimensional N=1 SU(N)^{M-1} and SU(M)^{N-1} linear quiver gauge theories compactified on S^1. The theories we study are the five-dimensional uplifts of four-dimensional superconformal linear quivers. We study this duality by comparing the Seiberg-Witten curves and the Nekrasov partition functions of the two dual… ▽ More In this article we explore the duality between the low energy effective theory of five-dimensional N=1 SU(N)^{M-1} and SU(M)^{N-1} linear quiver gauge theories compactified on S^1. The theories we study are the five-dimensional uplifts of four-dimensional superconformal linear quivers. We study this duality by comparing the Seiberg-Witten curves and the Nekrasov partition functions of the two dual theories. The Seiberg-Witten curves are obtained by minimizing the worldvolume of an M5-brane with nontrivial geometry. Nekrasov partition functions are computed using topological string theory. The result of our study is a map between the gauge theory parameters, i.e., Coulomb moduli, masses and UV coupling constants, of the two dual theories. Apart from the obvious physical interest, this duality also leads to compelling mathematical identities. Through the AGTW conjecture these five-dimentional gauge theories are related to q-deformed Liouville and Toda SCFTs in two-dimensions. The duality we study implies the relations between Liouville and Toda correlation functions through the map we derive. △ Less

Submitted 3 February, 2012; v1 submitted 21 December, 2011; originally announced December 2011.

Comments: 58 pages, 17 figures; v2: minor corrections, references added

Line operators on S^1xR^3 and quantization of the Hitchin moduli space

Authors: Yuto Ito, Takuya Okuda, Masato Taki

Abstract: We perform an exact localization calculation for the expectation values of Wilson-'t Hooft line operators in N=2 gauge theories on S^1xR^3. The expectation values are naturally expressed in terms of the complexified Fenchel-Nielsen coordinates, and form a quantum mechanically deformed algebra of functions on the associated Hitchin moduli space by Moyal multiplication. We propose that these expecta… ▽ More We perform an exact localization calculation for the expectation values of Wilson-'t Hooft line operators in N=2 gauge theories on S^1xR^3. The expectation values are naturally expressed in terms of the complexified Fenchel-Nielsen coordinates, and form a quantum mechanically deformed algebra of functions on the associated Hitchin moduli space by Moyal multiplication. We propose that these expectation values are the Weyl transform of the Verlinde operators, which act on Liouville/Toda conformal blocks as difference operators. We demonstrate our proposal explicitly in SU(N) examples. △ Less

Submitted 29 March, 2016; v1 submitted 17 November, 2011; originally announced November 2011.

Comments: 58 pages; v.2 minor corrections, references added; v.3 corrections corresponding to JHEP erratum

Report number: UT-Komaba/11-12; YITP-11-96

Nonlinear supratransmission in multicomponent systems

Authors: P. Anghel-Vasilescu, J. Dorignac, F. Geniet, J. Leon, M. Taki

Abstract: A method is proposed to solve the challenging problem of determining the supratransmission threshold (onset of instability of harmonic boundary driving inside a band gap) in multicomponent nonintegrable nonlinear systems. It is successfully applied to the degenerate three-wave resonant interaction in a birefringent quadratic medium where the process generates spatial gap solitons. No analytic expr… ▽ More A method is proposed to solve the challenging problem of determining the supratransmission threshold (onset of instability of harmonic boundary driving inside a band gap) in multicomponent nonintegrable nonlinear systems. It is successfully applied to the degenerate three-wave resonant interaction in a birefringent quadratic medium where the process generates spatial gap solitons. No analytic expression is known for this model showing the broad applicability of the method to nonlinear systems. △ Less

Submitted 11 August, 2010; originally announced August 2010.

Comments: 4 pages, 3 figures

Journal ref: Phys. Rev. Lett. 105 (2010) 074101

Surface Operator, Bubbling Calabi-Yau and AGT Relation

Authors: Masato Taki

Abstract: Surface operators in N=2 four-dimensional gauge theories are interesting half-BPS objects. These operators inherit the connection of gauge theory with the Liouville conformal field theory, which was discovered by Alday, Gaiotto and Tachikawa. Moreover it has been proposed that toric branes in the A-model topological strings lead to surface operators via the geometric engineering. We analyze the su… ▽ More Surface operators in N=2 four-dimensional gauge theories are interesting half-BPS objects. These operators inherit the connection of gauge theory with the Liouville conformal field theory, which was discovered by Alday, Gaiotto and Tachikawa. Moreover it has been proposed that toric branes in the A-model topological strings lead to surface operators via the geometric engineering. We analyze the surface operators by making good use of topological string theory. Starting from this point of view, we propose that the wave-function behavior of the topological open string amplitudes geometrically engineers the surface operator partition functions and the Gaiotto curves of corresponding gauge theories. We then study a peculiar feature that the surface operator corresponds to the insertion of the degenerate fields in the conformal field theory side. We show that this aspect can be realized as the geometric transition in topological string theory, and the insertion of a surface operator leads to the bubbling of the toric Calabi-Yau geometry. △ Less

Submitted 15 June, 2011; v1 submitted 15 July, 2010; originally announced July 2010.

Comments: 36 pages, 14 figures. v2: minor changes and typos corrected

Journal ref: JHEP 1107:047,2011

Deformed Prepotential, Quantum Integrable System and Liouville Field Theory

Authors: Kazunobu Maruyoshi, Masato Taki

Abstract: We study the dual descriptions recently discovered for the Seiberg-Witten theory in the presence of surface operators. The Nekrasov partition function for a four-dimensional N=2 gauge theory with a surface operator is believed equal to the wave-function of the corresponding integrable system, or the Hitchin system, and is identified with the conformal block with a degenerate field via the AGT rela… ▽ More We study the dual descriptions recently discovered for the Seiberg-Witten theory in the presence of surface operators. The Nekrasov partition function for a four-dimensional N=2 gauge theory with a surface operator is believed equal to the wave-function of the corresponding integrable system, or the Hitchin system, and is identified with the conformal block with a degenerate field via the AGT relation. We verify the conjecture by showing that the null state condition leads to the Schrodinger equations of the integrable systems. Furthermore, we show that the deformed prepotential emerging from the period integrals of the principal function corresponds to monodromy operation of the conformal block. We also give the instanton partition functions for the asymptotically free SU(2) gauge theories in the presence of the surface operator via the AGT relation. We find that these partition functions involve the counting of two- and four-dimensional instantons. △ Less

Submitted 19 August, 2010; v1 submitted 23 June, 2010; originally announced June 2010.

Comments: 47 pages; v2: references added, typos corrected; v3: a version to appear in Nuclear Physics B

Report number: YITP-10-45

Journal ref: Nucl.Phys.B841:388-425,2010

On AGT Conjecture for Pure Super Yang-Mills and W-algebra

Authors: Masato Taki

Abstract: Recently Alday, Gaiotto and Tachikawa have proposed relation between 2- and 4-dimensional conformal field theories. The relation implies that the Nekrasov partition functions of N=2 superconformal gauge theories are equal to conformal blocks associated with the conformal algebra. Likewise, a counterpart in pure super Yang-Mills theory exists in conformal field theory. We propose a simple relatio… ▽ More Recently Alday, Gaiotto and Tachikawa have proposed relation between 2- and 4-dimensional conformal field theories. The relation implies that the Nekrasov partition functions of N=2 superconformal gauge theories are equal to conformal blocks associated with the conformal algebra. Likewise, a counterpart in pure super Yang-Mills theory exists in conformal field theory. We propose a simple relation between the Shapovalov matrix of the W_3-algebra and the Nekrasov partition function of N=2 SU(3) Yang-Mills theory. △ Less

Submitted 24 December, 2009; originally announced December 2009.

Comments: 22 pages

Journal ref: JHEP 1105:038,2011

Spectral Efficiency Optimization for an Interfering Cognitive Radio with Adaptive Modulation and Coding

Authors: Mehrdad Taki, Farshad Lahouti

Abstract: In this paper, we consider a primary and a cognitive user transmitting over a wireless fading interference channel. The primary user transmits with a constant power and utilizes an adaptive modulation and coding (AMC) scheme satisfying a bit error rate requirement. We propose a link adaptation scheme to maximize the average spectral efficiency of the cognitive radio, while a minimum required spe… ▽ More In this paper, we consider a primary and a cognitive user transmitting over a wireless fading interference channel. The primary user transmits with a constant power and utilizes an adaptive modulation and coding (AMC) scheme satisfying a bit error rate requirement. We propose a link adaptation scheme to maximize the average spectral efficiency of the cognitive radio, while a minimum required spectral efficiency for the primary user is provisioned. The resulting problem is constrained to also satisfy a bit error rate requirement and a power constraint for the cognitive link. The AMC mode selection and power control at the cognitive transmitter is optimized based on the modified signal to noise plus interference ratio feedback of both links. The problem is then cast as a nonlinear discrete optimization problem for which a fast and efficient suboptimum solution is presented. We also present a scheme with rate adaptive and constant power cognitive radio. An important characteristic of the proposed schemes is that no computation or coordination overhead is imposed on the primary radio due to the cognitive radio activity. Numerical results and comparison with the interweave approach to cognitive radio demonstrate the efficiency of the proposed solutions. △ Less

Submitted 7 December, 2009; originally announced December 2009.

Comments: 26 pages, 7 figures submitted to IEEE journal on selected areas in communication

M2-branes Theories without 3+1 Dimensional Parents via Un-Higgsing

Authors: Masato Taki

Abstract: N=2 quiver Chern-Simons theory has lately attracted attention as the world volume theory of multiple M2 branes on a Calabi-Yau 4-fold. We study the connection between the stringy derivation of M2 brane theories and the forward algorithm which gives the toric Calabi-Yau 4-fold as the moduli space of the quiver theory. Then the existence of the 3+1 dimensional parent, which is the consistent 3+1 d… ▽ More N=2 quiver Chern-Simons theory has lately attracted attention as the world volume theory of multiple M2 branes on a Calabi-Yau 4-fold. We study the connection between the stringy derivation of M2 brane theories and the forward algorithm which gives the toric Calabi-Yau 4-fold as the moduli space of the quiver theory. Then the existence of the 3+1 dimensional parent, which is the consistent 3+1 dimensional superconformal theory with the same quiver diagram, is crucial for stringy derivation of M2 brane theories. We also investigate the construction of M2 brane theories that do not have 3+1 dimensional parents. The un-Higgsing procedure plays a key role to construct these M2 brane theories. We find some N=2 quiver Chern-Simons theories which correspond to interesting Calabi-Yau singularities. △ Less

Submitted 2 October, 2009; originally announced October 2009.

Comments: 60 pages, 52 figures

New Seiberg Dualities from N=2 Dualities

Authors: Kazunobu Maruyoshi, Masato Taki, Seiji Terashima, Futoshi Yagi

Abstract: We propose a number of new Seiberg dualities of N=1 quiver gauge theories. The new Seiberg dualities originate in new S-dualities of N=2 superconformal field theories recently proposed by Gaiotto. N=2 S-dual theories deformed by suitable mass terms flow to our N=1 Seiberg dual theories. We show that the number of exactly marginal operators is universal for these Seiberg dual theories and the 't… ▽ More We propose a number of new Seiberg dualities of N=1 quiver gauge theories. The new Seiberg dualities originate in new S-dualities of N=2 superconformal field theories recently proposed by Gaiotto. N=2 S-dual theories deformed by suitable mass terms flow to our N=1 Seiberg dual theories. We show that the number of exactly marginal operators is universal for these Seiberg dual theories and the 't Hooft anomaly matching holds for these theories. These provide strong evidence for the new Seiberg dualities. Furthermore, we study in detail the Klebanov-Witten type theory and its dual as a concrete example. We show that chiral operators and their non-linear relations match between these theories. These arguments also give non-trivial consistency checks for our proposal. △ Less

Submitted 1 September, 2009; v1 submitted 15 July, 2009; originally announced July 2009.

Comments: 31 pages, 7 figures. v2:version to appear in JHEP

Mechanisms of rogue wave formation

Authors: Arnaud Mussot, Alexandre Kudlinski, Mikhail I. Kolobov, Eric Louvergneaux, Marc Douay, Majid Taki

Abstract: Freak waves, or rogue waves, are one of the fascinating manifestations of the strength of nature. These devastating walls of water appear from nowhere, are short-lived and extremely rare. Despite the large amount of research activities on this subject, neither the minimum ingredients required for their generation nor the mechanisms explaining their formation have been given. Today, it is possibl… ▽ More Freak waves, or rogue waves, are one of the fascinating manifestations of the strength of nature. These devastating walls of water appear from nowhere, are short-lived and extremely rare. Despite the large amount of research activities on this subject, neither the minimum ingredients required for their generation nor the mechanisms explaining their formation have been given. Today, it is possible to reproduce such kind of waves in optical fibre systems. In this context, we demonstrate theoretically and numerically that convective instability is the basic ingredient for the formation of rogue waves. This explains why rogues waves are extremely sensitive to noisy environments. △ Less

Submitted 26 March, 2009; originally announced March 2009.

Comments: two movies, 1 text file