Super-resolution of magnetic systems using deep learning.

We construct a deep neural network to enhance the resolution of spin structure images formed by spontaneous symmetry breaking in the magnetic systems. Through the deep neural network, an image is expanded to a super-resolution image and reduced to the original image size to be fitted with the input feed image. The network does not require ground truth images in the training process. Therefore, it can be applied when low-resolution images are provided as training datasets, while high-resolution images are not obtainable due to the intrinsic limitation of microscope techniques. To show the usefulness of the network, we train the network with two types of simulated magnetic structure images; one is from self-organized maze patterns made of chiral magnetic structures, and the other is from magnetic domains separated by walls that are topological defects of the system. The network successfully generates high-resolution images highly correlated with the exact solutions in both cases. To investigate the effectiveness and the differences between datasets, we study the network's noise tolerance and compare the networks' reliabilities. The network is applied with experimental data obtained by magneto-optical Kerr effect microscopy and spin-polarized low-energy electron microscopy.

Topological magnetic structure generation using VAE-GAN hybrid model and discriminator-driven latent sampling.

Recently, deep generative models using machine intelligence are widely utilized to investigate scientific systems by generating scientific data. In this study, we experiment with a hybrid model of a variational autoencoder (VAE) and a generative adversarial network (GAN) to generate a variety of plausible two-dimensional magnetic topological structure data. Due to the topological properties in the system, numerous and diverse metastable magnetic structures exist, and energy and topological barriers separate them. Thus, generating a variety of plausible spin structures avoiding those barrier states is a challenging problem. The VAE-GAN hybrid model can present an effective approach to this problem because it brings the advantages of both VAE's diversity and GAN's fidelity. It allows one to perform various applications including searching a desired sample from a variety of valid samples. Additionally, we perform a discriminator-driven latent sampling (DDLS) using our hybrid model to improve the quality of generated samples. We confirm that DDLS generates various plausible data with large coverage, following the topological rules of the target system.

Optimization of physical quantities in the autoencoder latent space.

We propose a strategy for optimizing physical quantities based on exploring in the latent space of a variational autoencoder (VAE). We train a VAE model using various spin configurations formed on a two-dimensional chiral magnetic system. Three optimization algorithms are used to explore the latent space of the trained VAE. The first algorithm, the single-code modification algorithm, is designed for improving the local energetic stability of spin configurations to generate physically plausible spin states. The other two algorithms, the genetic algorithm and the stochastic algorithm, aim to optimize the global physical quantities, such as topological index, magnetization, energy, and directional correlation. The advantage of our method is that various optimization algorithms can be applied in the latent space containing the abstracted representation constructed by the trained VAE model. Our method based on latent space exploration is utilized for efficient physical quantity optimization.

Searching for the ground state of complex spin-ice systems using deep learning techniques.

Searching for the ground state of a given system is one of the most fundamental and classical questions in scientific research fields. However, when the system is complex and large, it often becomes an intractable problem; there is essentially no possibility of finding a global energy minimum state with reasonable computational resources. Recently, a novel method based on deep learning techniques was devised as an innovative optimization method to estimate the ground state. We apply this method to one of the most complicated spin-ice systems, aperiodic Penrose P3 patterns. From the results, we discover new configurations of topologically induced emergent frustrated spins, different from those previously known. Additionally, a candidate of the ground state for a still unexplored type of Penrose P3 spin-ice system is first proposed through this study. We anticipate that the capabilities of the deep learning techniques will not only improve our understanding on the physical properties of artificial spin-ice systems, but also bring about significant advances in a wide range of scientific research fields requiring computational approaches for optimization.

Estimating the effective fields of spin configurations using a deep learning technique.

The properties of complicated magnetic domain structures induced by various spin-spin interactions in magnetic systems have been extensively investigated in recent years. To understand the statistical and dynamic properties of complex magnetic structures, it is crucial to obtain information on the effective field distribution over the structure, which is not directly provided by magnetization. In this study, we use a deep learning technique to estimate the effective fields of spin configurations. We construct a deep neural network and train it with spin configuration datasets generated by Monte Carlo simulation. We show that the trained network can successfully estimate the magnetic effective field even though we do not offer explicit Hamiltonian parameter values. The estimated effective field information is highly applicable; it is utilized to reduce noise, correct defects in the magnetization data, generate spin configurations, estimate external field responses, and interpret experimental images.

Magnetic Hamiltonian parameter estimation using deep learning techniques.

Understanding spin textures in magnetic systems is extremely important to the spintronics and it is vital to extrapolate the magnetic Hamiltonian parameters through the experimentally determined spin. It can provide a better complementary link between theories and experimental results. We demonstrate deep learning can quantify the magnetic Hamiltonian from magnetic domain images. To train the deep neural network, we generated domain configurations with Monte Carlo method. The errors from the estimations was analyzed with statistical methods and confirmed the network was successfully trained to relate the Hamiltonian parameters with magnetic structure characteristics. The network was applied to estimate experimentally observed domain images. The results are consistent with the reported results, which verifies the effectiveness of our methods. On the basis of our study, we anticipate that the deep learning techniques make a bridge to connect the experimental and theoretical approaches not only in magnetism but also throughout any scientific research.

An innovative magnetic state generator using machine learning techniques.

We propose a new efficient algorithm to simulate magnetic structures numerically. It contains a generative model using a complex-valued neural network to generate k-space information. The output information is hermitized and transformed into real-space spin configurations through an inverse fast Fourier transform. The Adam version of stochastic gradient descent is used to minimize the magnetic energy, which is the cost of our algorithm. The algorithm provides the proper ground spin configurations with outstanding performance. In model cases, the algorithm was successfully applied to solve the spin configurations of magnetic chiral structures. The results also showed that a magnetic long-range order could be obtained regardless of the total simulation system size.

Cholesteric pitch divergence near smectic phase transitions.

The critical behavior of the pitch divergence of cholesteric liquid crystals in the vicinity to smectic-A∗ (SmA∗) and smectic-C∗ (SmC∗) phases is studied experimentally and compared with conflicting theoretical interpretations. Members of two homologous series were studied with varying polymorphism from N∗-SmC∗ to N∗-SmA∗. A modified functionality of the temperature dependence of the pitch is introduced to determine the critical exponent, and it is shown that the latter is independent of sample geometry. In contrast to several earlier investigations aiming to determine the critical exponent, which were inconclusive, the results of our critical exponents for the pitch divergence provide evidence for the model by Chen and Lubensky which predicts a critical exponent of ν=1/2 for the N∗-SmA∗ and ν=1 for the N∗-SmC∗ transition. This specifically implies that fluctuations cannot be neglected in the consideration of the nature of the phase transition.

A case of successful embolotherapy for gastric ulcer bleeding from the intercostal artery after oesophagectomy and gastric reconstruction.

We report a successful treatment with coil embolisation of an intercostal artery for ulcer bleeding in a gastric tube in a 70-year-old man who underwent a total oesophagectomy and gastric tube reconstruction for oesophageal cancer. This case teaches us to search aberrant feeding vessels when active bleeding is suspected in reconstructed gastric tube in the patient with oesophagectomy and oesophagogastrostomy.

Thermostable chitosanase from Bacillus sp. strain CK4: its purification, characterization, and reaction patterns.

A thermostable chitosanase, purified 156-fold to homogeneity in an overall yield of 12.4%, has a molecular weight of about 29,000 +/- 2,000, and is composed of monomer. The enzyme degraded soluble chitosan, colloidal chitosan, and glycol chitosan, but did not degrade chitin or other beta-linked polymers. The enzyme activity was increased about 2.5-fold by the addition of 10 mM Co2+ and 1.4-fold by Mn2+. However, Cu2+ ion strongly inhibited the enzyme. Optimum temperature and pH were 60 degrees C and 6.5, respectively. The enzyme was stable after heat treatment at 80 degrees C for 30 min or 70 degrees C for 60 min and fairly stable in protein denaturants as well. Chitosan was hydrolyzed to (GlcN)4 as a major product, by incubation with the purified enzyme. The effects of ammonium sulfate and organic solvents on the action pattern of the thermostable chitosanase were investigated. The amounts of (GlcN)3-(GlcN)6 were increased about 30% (w/w) in DAC 99 soluble chitosan containing 10% ammonium sulfate, and (GlcN)1 was not produced. The monophasic reaction system consisted of DAC 72 soluble chitosan in 10% EtOH also showed no formation of (GlcN)1, however, the yield of (GlcN)3 approximately (GlcN)6 was lower than DAC 99 soluble chitosan-10% ammonium sulfate. The optimal concentration of ammonium sulfate to be added was 20%. At this concentration, the amount of hexamer was increased by over 12% compared to the water-salt free system.

Alternative embryo transfer on day 3 or day 5 for reducing the risk of multiple gestations.

PURPOSE: This study was carried out to reduce the possibility of high-order multiple gestations and the failure of embryo transfer by determining their replacement date based on the number and quality of 2-day embryos. METHODS: All zygotes were cocultured with cumulus cells in 10 microliters of YS medium containing 10% human follicular fluid (hFF) for 48 or 96 hr. In period I, all embryos were transferred on day 3 (1032 cycles). In period II, the embryos were transferred on either day 3 or day 5 by determining their replacement date based on the number and quality of 2-day embryos: there were 2701 patients in whom embryos were replaced on day 3 (in the case that the number of zygotes was less than eight and the number of good-quality embryos was less than three) and 1952 patients less than 40 years old in whom embryos were replaced on day 5 (in the case that the number of zygotes was eight or more and/or the number of good-quality embryos was three or more). On the other hand, patients who were 40 years old or more were alloted to day 3 transfer cycles, regardless of the number and quality of the 2-day embryos, due to the possibility of their not producing blastocyst-stage embryos in vitro. RESULTS: The number of embryos transferred in period II was 2.9 +/- 0.6, while that in period I was 3.7 +/- 0.5. The multiple pregnancy rate was significantly decreased in period II (30.7%) compared to that (49.6%) in period I, while the pregnancy and implantation rates in period II (36.1 and 16.4%, respectively) were not lower than those (34.9 and 16.1%, respectively) in period I. The rate of triplet or more gestations was significantly minimized in period II (2.3%) compared to that in period I (26.5%). CONCLUSIONS: We propose that determination of the date on which embryos should be transferred based on the number and quality of embryos on day 2 may help to maintain an acceptable pregnancy rate, while minimizing embryo transfer failure and high-order multiple gestations.

Thermostable chitosanase from Bacillus sp. Strain CK4: cloning and expression of the gene and characterization of the enzyme.

A thermostable chitosanase gene from the environmental isolate Bacillus sp. strain CK4, which was identified on the basis of phylogenetic analysis of the 16S rRNA gene sequence and phenotypic analysis, was cloned, and its complete DNA sequence was determined. The thermostable chitosanase gene was composed of an 822-bp open reading frame which encodes a protein of 242 amino acids and a signal peptide corresponding to a 30-kDa enzyme. The deduced amino acid sequence of the chitosanase from Bacillus sp. strain CK4 exhibits 76.6, 15.3, and 14.2% similarities to those from Bacillus subtilis, Bacillus ehemensis, and Bacillus circulans, respectively. C-terminal homology analysis shows that Bacillus sp. strain CK4 belongs to cluster III with B. subtilis. The gene was similar in size to that of the mesophile B. subtilis but showed a higher preference for codons ending in G or C. The enzyme contains 2 additional cysteine residues at positions 49 and 211. The recombinant chitosanase has been purified to homogeneity by using only two steps with column chromatography. The half-life of the enzyme was 90 min at 80 degrees C, which indicates its usefulness for industrial applications. The enzyme had a useful reactivity and a high specific activity for producing functional oligosaccharides as well, with trimers through hexamers as the major products.

Analysis of essential leucine residue for catalytic activity of novel thermostable chitosanase by site-directed mutagenesis.

Bacterial chitosanases share weak amino acid sequence similarities at certain regions of each enzyme. These regions have been assumed to be important for catalytic activities of the enzyme. To verify this assumption, the functional importance of the conserved region in a novel thermostable chitosanase (TCH-2) from Bacillus coagulans CK108 was investigated. Each of the conserved amino acid residues (Leu64, Glu80, Glu94, Asp98, and Gly108) was changed to aspartate and glutamine or asparagine and glutamate by site-directed mutagenesis, respectively. Kinetic parameters for colloidal chitosan hydrolysis were determined with wild-type and 10 mutant chitosanases. The Leu64 --> Arg and Leu64 --> Gln mutations were essentially inactive and kinetic parameters such as Vmax and kcat were approximately 1/10(7) of those of the wild-type enzyme. The Asp98 --> Asn mutation did not affect the Km value significantly, but decreased kcat to 15% of that of wild-type chitosanase. On the other hand, the Asp98 --> Glu mutation affected neither Km nor kcat. The observation that approximately 15% of activity remained after the substitution of Asp98 by Asn indicated that the carboxyl side chain of Asp98 is not absolutely required for catalytic activity. These results indicate that the Leu64 residue is directly involved in the catalytic activity of TCH-2.

Molecular and enzymatic characterization of a maltogenic amylase that hydrolyzes and transglycosylates acarbose.

A gene encoding a maltogenic amylase of Bacillus stearothermophilus ET1 was cloned and expressed in Escherichia coli. DNA sequence analysis indicated that the gene could encode a 69,627-Da protein containing 590 amino acids. The predicted amino acid sequence of the enzyme shared 47-70% identity with the sequences of maltogenic amylase from Bacillus licheniformis, neopullulanase from B. stearothermophilus, and cyclodextrin hydrolase (CDase) 1-5 from an alkalophilic Bacillus 1-5 strain. In addition to starch, pullulan and cyclodextrin, B. stearothermophilus could hydrolyze isopanose, but not panose, to glucose and maltose. Maltogenic amylase hydrolyzed acarbose, a competitive inhibitor of amylases, to glucose and a trisaccharide. When acarbose was incubated with 10% glucose, isoacarbose, containing an alpha-1,6-glucosidic linkage was produced as an acceptor reaction product. B. stearothermophilus maltogenic amylase shared four highly similar regions of amino acids with several amylolytic enzymes. The beta-cyclodextrin-hydrolyzing activity of maltogenic amylase was enhanced to a level equivalent to the activity of CDase when its amino acid sequence between the third and the fourth conserved regions was made more hydrophobic by site-directed mutagenesis. Enhanced transglycosylation activity was observed in most of the mutants. This result suggested that the members of a subfamily of amylolytic enzymes, including maltogenic amylase and CDase, could share similar substrate specificities, enzymatic mechanisms and structure/function relationships.

Pregnancies resulting from in vitro matured oocytes collected from women with regular menstrual cycle.

PURPOSE: To demonstrate that human immature oocytes retrieved from women with regular menstrual cycles can undergo maturation and fertilization, and that the resulting embryos can establish pregnancies. METHODS: Immature oocytes (n = 568) were retrieved from women with regular menstrual cycle. The intact immature oocytes (n = 506) were allowed to mature in YS medium supplemented with 70% human follicular fluid (hFF); the matured oocytes were fertilized with husband sperm. Two pronuclei oocytes were cocultured with cumulus cells in YS medium supplemented with 10% hFF until 2 or 3 days after insemination. The cleaved embryos were transferred in uteri. RESULTS: Follicles were aspirated on Day 9.2 +/- 5.3 of 63 natural cycles from 51 patients (mean age = 34.8 +/- 4.0 years). The average number of retrieved immature oocytes was 9.0. The maturation rate was 74.3% (376/506). The two PN and cleavage rates were 72.6% (273/376) and 89.0% (243/273), respectively. Embryo transfer was achieved in 51 cycles and clinical pregnancy rate was 17.6% (9/51). CONCLUSIONS: The results suggest that in vitro matured oocytes can undergo fertilization and the resulting embryos may successfully lead to pregnancies. However, further research is needed to improve IVM technique to achieve success rate comparable to gonadotrophin stimulated cycles.

Identification of essential amino acid residues for catalytic activity and thermostability of novel chitosanase by site-directed mutagenesis.

The functional importance of a conserved region in a novel chitosanase from Bacillus sp. CK4 was investigated. Each of the three carboxylic amino acid residues (Glu-50, Glu-62, and Asp-66) was changed to Asp and Gln or Asn and Glu by site-directed mutagenesis, respectively. The Asp-66-->Asn and Asp-66-->Glu mutation remarkably decreased kinetic parameters such as Vmax and kcat to approximately 1/1,000 those of the wild-type enzyme, indicating that the Asp-66 residue was essential for catalysis. The thermostable chitosanase contains three Cys residues at positions 49, 72, and 211. The Cys-49-->Ser/Tyr and Cys-72-->Ser/Tyr mutant enzymes were as stable to thermal inactivation and denaturating agents as the wild-type enzyme. However, the half-life of the Cys-211-->Ser/Tyr mutant enzyme was less than 10 min at 80 degrees C, while that of the wild-type enzyme was about 90 min. Moreover, the residual activity of Cys-211-->Ser/Tyr enzyme was substantially decreased by 8 M urea; and it lost all catalytic activity in 40% ethanol. These results show that the substitution of Cys with any amino acid residues at position 211 seems to affect the conformational stability of the chitosanase.