Optical AI chips have struggled to implement a crucial algorithm used to train neural networks–backpropagation. But in a new paper in the journal Science, a team from Stanford University has described the first ever implementation of the training approach on a photonic chip.
A new kind of 3D optical lattice traps atoms using focused laser spots replicated in multiple planes and could eventually serve as a quantum computing platform.
Researchers have produced 3D lattices of trapped atoms for possible quantum computing tasks, but the standard technology doesn’t allow much control over atom spacing. Now a team has created a new type of 3D lattice by combining optical tweezers—points of focused light that trap atoms—with an optical phenomenon known as the Talbot effect [1]. The team’s 3D tweezer lattice has sites for 10,000 atoms, but with some straightforward modifications, the system could reach 100,000 atoms. Such a large atom arrangement could eventually serve as a platform for a quantum computer with error correction.
3D optical lattices have been around for decades. The standard method for creating them involves crossing six laser beams to generate a 3D interference pattern that traps atoms in either the high-or low-intensity spots (see Synopsis: Pinpointing Qubits in a 3D Lattice). These cold-atom systems have been used as precision clocks and as models of condensed-matter systems. However, the spacing between atoms is fixed by the wavelength of the light, which can limit the control researchers have over the atomic behavior.
Twenty years ago, Ferenc Krausz, Theodor Hänsch, and their collaborators used a femtosecond near-infrared (NIR) laser to compel neon atoms to emit pulses of extreme ultraviolet (XUV) light that lasted a few hundred attoseconds. The landmark feat depended on the laser’s strong oscillating electric field, which tore away the atoms’ valence electrons and hurled them back half a cycle later. Now Tobias Heldt of the Max Planck Institute for Nuclear Physics in Germany and his collaborators have developed a new experimental technique that is, in a sense, a mirror image of the 2003 demonstration: they used attosecond XUV pulses to free the valence electrons and to then track their response to femtosecond NIR laser pulses [1].
When a few-cycle femtosecond NIR pulse passes through helium gas, the atoms’ dipole moments fluctuate as the electrons move away and then recollide. Those fluctuations in turn are manifest in the gas’s absorption spectrum. Heldt and his collaborators set out to measure the fluctuations and, from them, infer the electrons’ trajectories.
The attosecond XUV pulse in their experiment did double duty. It ionized the helium atoms to bring the electrons under the influence of the NIR pulse. It also interfered with the fluctuating dipole moments. As a result, the XUV pulse carried away the dipoles’ spectral imprint, which the team measured with a grating spectrometer.
A proposed technique would use light and nanowires to generate electron beams with nearly pure spin polarization.
In a polarized electron beam, the particles’ spins are not randomly oriented but favor a particular direction. The polarization serves as a useful property for studying the magnetism of materials or for probing the spins of atoms or nuclei. But such a beam typically has a low degree of polarization unless it is produced at a synchrotron facility. Theorists have proposed creating these beams using laser light, but so far these approaches have involved extremely intense lasers and have not been expected to produce high polarization. Now Deng Pan of East China Normal University and Hongxing Xu of Wuhan University, China, have proposed a method that reduces the required laser intensity by up to 10 billion times compared with previous laser-based approaches and that should produce a pair of beams that are nearly 100% polarized [1].
In Pan and Xu’s proposal, a wide laser beam broadsides an array of parallel conducting nanowires with 100-nm spacing and excites them to emit electromagnetic waves. An unpolarized electron beam is sent across the array, perpendicular to the wires, about 100 nm away from them. Some electrons absorb or emit photons, causing their spins to align parallel or antiparallel to the local electric field. They also gain or lose a photon’s worth of energy. This interaction with the radiation near the wires generates two new beams with nearly pure spin polarizations and slightly different energies, allowing them to be easily separated. Pan and Xu say that the technique should be implementable with current technology and that it may even lead to new ways of manipulating electrons.
The current work springs from a 2007 study in which Cavalleri and his team reported using terahertz laser pulses to distort a lattice into favoring a particular ground state [2]. The pulses excited specific, quantized vibrations—phonons—that changed the electronic state of a crystal, yielding a transient drop in electrical resistance of 5 orders of magnitude.
In their new experiment, the researchers selected three laser frequencies that were separately coupled to one of several possible lattice distortions in YTiO3. Using a magneto-optical pump-probe setup, they examined how each of the excitations affected the crystal’s structure and its magnetism. Specifically, they observed whether the polarization of the light reflected by the crystal changed when viewed in opposite directions. A clockwise–counterclockwise shift in the polarization of the reflected light would be a sure sign of time-reversal invariance, which happens only in the presence of magnetic order.
They found that ultrafast laser pulses tuned to a phonon frequency of 9 THz caused the YTiO3 crystal to fully magnetize just above zero K. They then showed that this order, instead of vanishing at 27 K, remained stable up to at least 80 K, the highest temperature that they measured. What’s more, the magnetism persisted for many nanoseconds, 6 orders of magnitude longer than the femtoseconds-long laser pulses. The team attribute this long-lasting state to the stability of the structural distortions induced by energy deposited by the laser.
Vertical Takeoff Vehicles, and other tech like electric cars are older than you think This one is from the 50’s.
Here are a series of videos that are mainly played in our gallery alongside exhibits, but we wanted patrons unable to visit the museum on a regular basis the chance to access them at a moment’s notice. Enjoy!
Years after shutting down its robotics division, OpenAI is now back in the game after raising funding for Norwegian robotics company 1X.
Learn about the causes, symptoms, diagnosis & treatment from the MSD Manuals — Medical Consumer Version.
Google is using its enormous Chrome browser testing base to help examine the prospect of continuing the security of the digital age into the uncertainty of the quantum one.
Proteins are made from chains of amino acids that fold into three-dimensional shapes, which in turn dictate protein function. Those shapes evolved over billions of years and are varied and complex, but also limited in number. With a better understanding of how existing proteins fold, researchers have begun to design folding patterns not produced in nature.
But a major challenge, says Kim, has been to imagine folds that are both possible and functional. “It’s been very hard to predict which folds will be real and work in a protein structure,” says Kim, who is also a professor in the departments of molecular genetics and computer science at U of T. “By combining biophysics-based representations of protein structure with diffusion methods from the image generation space, we can begin to address this problem.”
The new system, which the researchers call ProteinSGM, draws from a large set of image-like representations of existing proteins that encode their structure accurately. The researchers feed these images into a generative diffusion model, which gradually adds noise until each image becomes all noise. The model tracks how the images become noisier and then runs the process in reverse, learning how to transform random pixels into clear images that correspond to fully novel proteins.
