|
Loading ...
|
|
|
|
pages views since 05/19/2016 : 118748
· Members : 7
· News : 733
· Downloads : 0
· Links : 0
|
|
|
|
Secrets of Neutron Stars Unlocked by "Mirror" Nuclei
|
|
|
Posted by Okachinepa on 09/30/2024 @
Courtesy of SynEvol
Credit: ESO/L.Calcada
The size of an atomic nucleus can be altered by adding or subtracting neutrons. Isotope shifts are the resultant minute variations in the energy levels of the atom's electrons. Researchers can determine an isotope's nucleus radius by precisely measuring these energy fluctuations.
In this study, the nuclear radii of the stable silicon isotopes silicon-28, silicon-29, and silicon-30 were measured with the aid of a laser. The unstable silicon-32 nucleus, which consists of 14 protons and 18 neutrons, was also measured in terms of radius. The researchers put constraints on variables that aid in describing the physics of astrophysical phenomena like neutron stars using the difference in radius between the silicon-32 nucleus and its mirror nucleus, argon-32, which contains 18 protons and 14 neutrons. The findings represent a significant advancement in nuclear theory, the study of nuclei and their constituent parts.
Courtesy of SynEvol
Credit: ESO/L.Calcada
Scientists continue to encounter persistent difficulties in their comprehension of nuclei, even with advancements in nuclear theory. For example, scientists have not made the connection between the strong nuclear force hypothesis and the description of nuclear scale. Furthermore, it's unclear if nuclear theories describing limited atomic nuclei can accurately characterize nuclear matter. The protons and neutrons that make up this unique kind of matter interact with one another. Extremely cold matter, like neutron stars, is considered nuclear matter. These unanswered concerns are addressed in part by precise measurements of charge radii, or the radius of atomic nuclei.
At the BEam COoler and LAser spectroscopy facility (BECOLA) at the Facility for Rare Isotope Beams (FRIB) at Michigan State University, researchers measured the nuclear radius of several silicon isotopes using laser spectroscopy measurements of atomic isotope shifts. The unstable silicon-32, which has 14 protons and 18 neutrons, as well as the stable silicon isotopes silicon-28, silicon-29, and silicon-30, were measured.
The outcomes offer a crucial reference point for the advancement of nuclear theory. The charge radii difference between the silicon-32 nucleus and its mirror nucleus argon-32, which has 18 protons and 14 neutrons, was utilized to limit parameters needed to characterize the properties of dense neutron matter within neutron stars. The reported results agree with the restrictions from gravitational wave observations and other complementing observables.
|
|
|
Brain-Inspired Materials Will Transform Computer Technology
|
|
|
Posted by Okachinepa on 09/30/2024 @
Courtesy of SynEvol
Credit: Texas A&M University
In order to create materials for more effective computing, a group of scientists from Texas A&M University, Sandia National Laboratories – Livermore, and Stanford University are studying the human brain. The newly discovered class of materials is unique in that it spontaneously propagates an electrical signal along a transmission line, simulating the behavior of an axon. These discoveries might have a significant impact on artificial intelligence and computers in the future.
Any electrical signal flowing in a metallic conductor loses amplitude due to the metal’s intrinsic resistance. Approximately thirty miles of tiny copper wires can be found inside modern graphics processing units and central processor units (CPUs) to move electrical impulses around the chip. The pulse integrity must be preserved by amplifiers since these losses mount up quickly. The interconnect-dense chips of today perform differently due to these design limitations.
The researchers looked to axons for insight on how to overcome this constraint. In vertebrates, axons are the parts of nerve cells, or neurons, that have the ability to carry electrical impulses away from the nerve cell body.
According to lead author Dr. Tim Brown, a post-doctoral fellow at Sandia National Lab and a former PhD student in materials science and engineering at Texas A&M, "we want to transmit a data signal from one place to another, more distant location" frequently. For instance, it might be necessary to send an electrical pulse from a CPU chip's edge to some transistors that are close to its center. Resistance at ambient temperature constantly diminishes sent signals, even for the best conducting metals. Therefore, we usually cut into the transmission line and enhance the signal, which consumes energy, time, and space. Biology operates differently: certain signals in the brain are also conveyed over centimeters, but they do so via axons composed of organic matter that is far more resistant, and they do so without ever stopping and strengthening the signals."
Axons are the communication highway, according to Dr. Patrick Shamberger, an associate professor in Texas A&M's Department of Materials Science and Engineering. They transfer signals from one nearby neuron to another. The axons function as fiber optic cables, carrying signals from one neuron to its neighbor, while the neurons are in charge of processing signals.
The materials found in this study are primed, just like the axon model, which enables them to spontaneously amplify a voltage pulse as it travels down the axon. The electronic phase transition in lanthanum cobalt oxide, which increases the material's electrical conductivity when heated, was exploited by the researchers. A positive feedback loop is created when this feature interacts with the little amounts of heat produced as a signal travels through the material.
As a result, a variety of odd phenomena such as the amplification of tiny disturbances, negative electrical resistances, and abnormally high phase shifts in ac signals are found that are not seen in common passive electrical components such as resistors, capacitors, and inductors.
Shamberger claims that these materials are special because they exist in a "Goldilocks state" that is semi-stable. Electrical pulses do not break down or show signs of thermal runaway. If the material is kept at a steady current, however, it will naturally oscillate. The scientists discovered that they could use this phenomenon to produce spikes and increase the strength of a signal that is transmitted across a transmission line.
"Essentially, we take advantage of the material's inherent instabilities, which reinforce an electrical pulse as it travels through the transmission line. Despite our co-author Dr. Stan Williams' theoretical predictions, this is the first evidence of the behavior's existence.
The future of computers, which is driving an increase in energy use, may depend on these discoveries. By 2030, data centers are expected to consume 8% of the nation's electricity, and artificial intelligence may significantly raise that need. Over time, this is a step toward comprehending dynamic materials and applying biological inspiration to encourage computing that is more efficient.
|
|
|
MIT's Quantum Locks Strengthen Cloud AI Security
|
|
|
Posted by Okachinepa on 09/30/2024 @
Courtesy of SynEvol
Credit:MIT
Numerous industries, including financial forecasts and healthcare diagnostics, have found use for deep learning models. Owing to their heavy processing load, these models rely on reliable cloud servers.
But this heavy reliance on cloud computing comes with a lot of security hazards. This is particularly important in delicate industries like healthcare, where hospitals may be discouraged from using AI tools due to privacy concerns around patient data.
In order to address this urgent problem, researchers at MIT have created a security protocol that makes use of light's quantum characteristics to ensure that data is sent securely to and from cloud servers during deep learning calculations.
The protocol takes advantage of the basic ideas of quantum physics to encode data into the laser light used in fiber optic communications systems. This prevents hackers from secretly copying or intercepting the data.
Courtesy of SynEvol
Credit: MIT
Furthermore, the method ensures confidentiality without sacrificing the deep-learning models' accuracy. The researcher conducted tests to show that their procedure could guarantee strong security protections and retain 96 percent accuracy.
"GPT-4 and other deep learning models offer never-before-seen capabilities, but they also demand a lot of processing power. “Our protocol allows users to take advantage of these potent models without jeopardizing the confidentiality of their data or the proprietary nature of the models themselves,” claims Kfir Sulimany, the main author of a paper on this security protocol and an MIT postdoc at the Research Laboratory for Electronics (RLE).
Prahlad Iyengar, a graduate student studying electrical engineering and computer science (EECS), Ryan Hamerly, a former postdoc at NTT Research, Inc., and senior author Dirk Englund, a professor of EECS and principal investigator of the Quantum Photonics and Artificial Intelligence Group and of RLE, join Sulimany on the paper. Recent presentations of the study were made at the Annual Conference on Quantum Cryptography.
The researchers concentrated on a cloud-based compute situation where two parties are involved: a central server that manages a deep learning model and a client that possesses sensitive data, such as medical photos.
Without disclosing any personal information about the patient, the customer wants to utilize the deep-learning model to predict things like if a patient has cancer based on medical pictures.
To make a prediction in this case, sensitive data must be sent. Nonetheless, patient data security must be maintained throughout the procedure.
Furthermore, the server is unwilling to divulge any information on the unique model that OpenAI and other companies have spent years and millions of dollars developing.
Vadlamani continues, "Both parties have something they want to hide."
A malicious party could quickly replicate data transferred from the server or client while using digital computing.
Comparatively speaking, quantum information is not completely replicable. The no-cloning principle is a feature that the researchers take advantage of in their security approach.
The server uses laser light to encode the weights of a deep neural network into an optical field for the researchers' protocol.
Layers of connected nodes, or neurons, that process data are arranged in layers within a neural network, a deep learning model. The parts of the model that do the mathematical operations on each input, layer by layer, are called weights. Up until the last layer produces a forecast, the output of one layer is transferred into the subsequent layer.
The server transmits the network’s weights to the client, which implements operations to get a result based on their private data. The information is kept hidden from the server.
In addition, due of the quantum nature of light, the security protocol forbids the client from replicating the weights and only permits the client to measure a single result.
The protocol is made to cancel out the first layer as soon as the client feeds the first result into the subsequent layer, preventing the client from learning any more about the model.
The client just measures the light required to operate the deep neural network and feed the output into the following layer, as opposed to measuring all of the incoming light from the server. The leftover light is then sent back to the server by the client for security checks, according to Sulimany.
When assessing the model's output, the client inevitably introduces small mistakes because of the no-cloning theorem. The server can measure these mistakes to find out if any information was spilled when it receives the residual light from the client. Crucially, it has been demonstrated that the remaining light conceals the client data.
Optical fibers are usually used in modern telecommunications equipment because they can sustain large bandwidth across long distances. Without the need for additional gear, the researchers can encode data into light for their security protocol because this equipment already has optical lasers.
Upon testing, the researchers discovered that their method could ensure both server and client security while allowing the deep neural network to reach 96 percent accuracy.
Less than 10% of what an adversary would need to retrieve any secret information is contained in the tiny bit of model information that is disclosed when the client executes operations. Conversely, a malevolent server could only acquire roughly 1% of the data required to pilfer the client's information.
Building on years of quantum cryptography work that had also been demonstrated on that testbed, Englund says, "a few years ago, when we developed our demonstration of distributed machine learning inference between MIT's main campus and MIT Lincoln Laboratory, it dawned on me that we could do something entirely new to provide physical-layer security." Nevertheless, a number of complex theoretical obstacles needed to be resolved before it would be possible to implement distributed machine learning with privacy guarantees. This was not conceivable prior to Kfir joining our team, since Kfir was able to establish the unified framework supporting this work by having a unique understanding of both the theory and experimental components.
In the future, the researchers hope to investigate how this protocol may be used in conjunction with federated learning, a method in which several parties train a central deep-learning model using their data. Moreover, it might be applied to quantum operations as opposed to the classical operations they researched for this work, which might offer benefits in terms of precision and security.
In particular, deep learning and quantum key distribution are two domains that are not typically combined, yet this work does so in a creative and captivating way. It provides an extra degree of security to the former and makes what seems like a realistic implementation possible by utilizing techniques from the latter. Regarding maintaining privacy in distributed architectures, this may be of interest. The practical implementation of the protocol and its behavior under suboptimal experimental conditions are exciting to watch, says Eleni Diamanti, a CNRS research director at the Sorbonne University in Paris who was not involved in this work.
|
|
|
Orbitronics: The Future's Energy-Efficient Technology
|
|
|
Posted by Okachinepa on 09/30/2024 @
Courtesy of SynEvol
Credit: Paul Scherrer Institute
The next generation of environmentally friendly technologies may process information using different electron characteristics than electronics, which rely on electron charge for data transport. Spintronics has been the leading candidate for an alternative kind of "tronics" until recently. In this case, information is sent via the electron's spin.
In a developing subject known as orbitronics, scientists are also investigating the potential applications of the orbital angular momentum (OAM) of electrons circling their atomic nucleus. This area has a lot of potential for memory devices, especially since it may be possible to create substantial magnetization with comparatively tiny charge currents, which would result in devices that require less energy. The key topic at hand is which materials are best suited to produce flows of OAMs, which are necessary for orbitronics.
An international research team now demonstrated that chiral topological semi-metals, a new class of materials discovered at PSI in 2019, have properties that make them a highly viable option for generating currents of OAMs. The team was led by scientists from Paul Scherrer Institute PSI and Max Planck Institutes in Halle and Dresden in Germany.
In the hunt for suitable materials for orbitronics, steps ahead have already been made utilizing conventional materials such as titanium. Nevertheless, chiral topological semi-metals have emerged as a compelling candidate since their discovery five years ago. These materials' helical atomic structure lends them a natural "handedness" akin to the double helix of DNA and may equip them with OAM flow-facilitating textures or patterns.
Michael Schüler, an assistant professor of physics at the University of Fribourg and group leader in the PSI Center for Scientific Computing, Theory and Data, says, "This offers a significant advantage to other materials because you don't need to apply external stimuli to get OAM textures—they're an intrinsic property of the material." Schüler co-led the recent study. "This could simplify the process of producing steady and effective OAM currents without the need for particular circumstances."
Researchers have been particularly interested in OAM monopoles, a specific OAM texture that has been postulated in chiral topological semi-metals. OAM extends outward from a central point at these monopoles, resembling the spikes of a frightened hedgehog curled into a ball.
OAM is isotropic, meaning that it is uniform in all directions, which is why these monopoles are so alluring. According to Schüler, this is a very helpful feature since it allows for the generation of OAM flows in any direction.
OAM monopoles are attractive for orbitronics, but up until this most recent work, they were only a theoretical possibility.
Circular Dichroism in Angle-Resolved Photoemission Spectroscopy, or CD-ARPES, is a technique that uses circularly polarized X-rays from a synchrotron light source to study them experimentally. However, in the past, a discrepancy between theory and experiment has made it difficult for researchers to understand the data. "The data may have been available to researchers, but it was buried with evidence supporting OAM monopoles," Schüler claims.
Electrons are ejected from a substance in ARPES when light shines on it. The material's electrical structure can be inferred from the angles and energy of these expelled electrons. The incident light in CD-ARPES has a circular polarization.
According to Schüler, "it is a natural assumption that you are measuring something that is directly proportional to the OAMs if you use circularly polarized light." The issue is that, as our research demonstrates, this turns out to be a quite naive assumption. In truth, it’s really more complex.”
In their investigation, Schüler and colleagues tested two types of chiral topological semi-metals at the Swiss Light Source SLS: those constructed of palladium and gallium or platinum and gallium. With a strong desire to uncover the OAM textures concealed in the intricate network of CD-ARPES data, the team applied rigorous theory to question every presumption.
Then, they carried out an odd but essential additional experimental step in which they changed the photon energies. "The facts didn't make sense at first. According to Schüler, the signal appeared to be fluctuating all over the place.
After carefully separating out the various factors that affected OAM calculations from CD-ARPES data, they discovered that, contrary to what was previously thought, the CD-ARPES signal rotated around the monopoles as photon energy varied. They established the existence of OAM monopoles and closed the gap between theory and experiment in this way.
After successfully seeing OAM monopoles, Schüler and associates demonstrated that a crystal with a mirror image chirality could be used to change the polarity of the monopole, or whether the OAM spikes point inside or outward. According to Schüler, "ortronics devices could potentially be created with different directionality, so this is a very useful property."
With theory and experiment finally coming together, the research community at large may now investigate OAM textures in a range of materials and maximize their orbitronics applications.
|
|
|
Breakthrough in Artificial Photosynthesis Converts CO2 Into Ethylene
|
|
|
Posted by Okachinepa on 09/30/2024 @
Courtesy of SynEvol
Credit: Silvia Cardarelli, Electrical and Computer Engineering, University of Michigan
Reusing CO2 to create sustainable energy requires the ability to bind carbon atoms together. Now, scientists at the University of Michigan have created an artificial photosynthesis system that exhibits previously unheard-of levels of performance in binding two of them into hydrocarbons.
This innovative method outperforms previous artificial photosynthesis systems in terms of efficiency, yield, and lifetime when producing ethylene. Since ethylene is a hydrocarbon that is frequently used to make plastics, the system's direct usage would be to collect carbon dioxide that would otherwise be released into the atmosphere so that plastic might be made.
Courtesy of SynEvol
Credit: Silvia Cardarelli, Electrical and Computer Engineering, University of Michigan
Zetian Mi, an electrical and computer engineering professor at the University of Michigan and the study's corresponding author, said, "The performance, or the activity and stability, is about five to six times better than what is typically reported for solar energy or light-driven carbon dioxide reduction to ethylene."
The world's most produced organic chemical is really ethylene. However, it is usually made using gas and oil at high pressures and temperatures, which releases carbon dioxide.
Courtesy of SynEvol
Credit: Silvia Cardarelli, Electrical and Computer Engineering, University of Michigan
Longer chains of carbon and hydrogen atoms are what's being aimed for in order to create transportable liquid fuels. Eliminating all of the oxygen from water, or H2O, the hydrogen source, and CO2, the carbon source, is a portion of the task.
A forest of gallium nitride nanowires, each only 50 nanometers (a few hundred atoms) broad, and the silicon base on which they were grown are the two types of semiconductors that the device uses to absorb light. Copper clusters, each containing roughly thirty atoms, are scattered throughout the nanowires, where the reaction that converts carbon dioxide and water into ethylene occurs.
Courtesy of SynEvol
Credit: Silvia Cardarelli, Electrical and Computer Engineering, University of Michigan
The nanowires are exposed to light equal to that of the sun at noon while submerged in carbon dioxide-enriched water. The light's energy releases electrons that cause the water close to the gallium nitride nanowires' surface to split. In addition to producing hydrogen for the ethylene synthesis, this also produces oxygen, which the gallium nitride absorbs to form gallium nitride oxide.
Copper has the ability to cling to hydrogen and seize the carbon atoms in carbon dioxide, converting them into carbon monoxide. The researchers think that two carbon monoxide molecules link with the hydrogen when it is combined with the hydrogen and receives an energy boost from the light. It is thought that the reaction is finished at the interface where the three hydrogen atoms from splitting water replace the two oxygen atoms at the copper and gallium nitride oxide.
The researchers discovered that 61 percent of the free electrons produced by the light-emitting semiconductors aided in the production of ethylene. Although a separate catalyst based on silver and copper was able to attain an efficiency of approximately 50%, it was only able to operate for a few hours before degrading, and it need to run in a carbon-based fluid. The Michigan team's equipment, on the other hand, operated without slowing down for 116 hours, and they have operated comparable devices for 3,000 hours.
The fact that oxygen enhances the catalyst and permits a self-healing process is partly responsible for this, as gallium nitride and the water splitting process work in concert. In subsequent research, the device's endurance limits will be investigated.
In the end, the apparatus generated ethylene at a rate that was more than four times greater than that of the closest rival systems.
The paper's first author and assistant research scientist in electrical and computer engineering at the University of Michigan, Bingxing Zhang, stated, "In the future, we want to produce some other multicarbon compounds such as propanol with three carbons or liquid products."
Mi's ultimate goal is liquid fuels, which have the potential to make many of the current transportation technologies sustainable.
|
|
Page : [1] 2 3 4 5 »...
|
|
Members
· Admins : 4 · Members : 3 [List]
· Last : MortaBlack
Who is on-line?
· Guests : 2 · Member : 0 · Admin : 0
|
|
|
Latest wars : | | There are yet no matches |
|
|
|
|