The dimensions of quantum computer systems is rising shortly. In 2022, IBM took the top spot with its 433-qubit Osprey chip. Yesterday, Atom Computing announced they’ve one-upped IBM with a 1,180-qubit impartial atom quantum pc.
The brand new machine runs on a tiny grid of atoms held in place and manipulated by lasers in a vacuum chamber. The corporate’s first 100-qubit prototype was a 10-by-10 grid of strontium atoms. The brand new system is a 35-by-35 grid of ytterbium atoms (proven above). (The machine has area for 1,225 atoms, however Atom has thus far run assessments with 1,180.)
Quantum computing researchers are engaged on a spread of qubits—the quantum equal of bits represented by transistors in conventional computing—together with tiny superconducting loops of wire (Google and IBM), trapped ions (IonQ), and photons, amongst others. However Atom Computing and different corporations, like QuEra, imagine impartial atoms—that’s, atoms with no electrical cost—have higher potential to scale.
It is because impartial atoms can keep their quantum state longer, and so they’re naturally plentiful and equivalent. Superconducting qubits are extra vulnerable to noise and manufacturing flaws. Impartial atoms will also be packed extra tightly into the identical area as they haven’t any cost that may intrude with neighbors and might be managed wirelessly. And impartial atoms enable for a room-temperature set-up, versus the near-absolute zero temperatures required by different quantum computer systems.
The corporate could also be onto one thing. They’ve now elevated the variety of qubits of their machine by an order of magnitude in simply two years, and imagine they will go additional. In a video explaining the technology, Atom CEO Rob Hays says they see “a path to scale to thousands and thousands of qubits in lower than a cubic centimeter.”
“We expect that the quantity of problem we needed to face to go from 100 to 1,000 might be considerably greater than the quantity of challenges we’re gonna face when going to no matter we need to go to subsequent—10,000, 100,000,” Atom cofounder and CTO Ben Bloom told Ars Technica.
However scale isn’t all the pieces.
Quantum computer systems are extraordinarily finicky. Qubits might be knocked out of quantum states by stray magnetic fields or fuel particles. The extra this occurs, the much less dependable the calculations. Whereas scaling acquired lots of consideration a number of years in the past, the main target has shifted to error-correction in service of scale. Certainly, Atom Computing’s new pc is greater, however not essentially extra highly effective. The entire thing can’t but be used to run a single calculation, for instance, as a result of accumulation of errors because the qubit depend rises.
There was latest motion on this entrance, nonetheless. Earlier this yr, the corporate demonstrated the ability to check for errors mid-calculation and doubtlessly repair these errors with out disturbing the calculation itself. In addition they have to hold errors to a minimal total by growing the constancy of their qubits. Recent papers, every showing encouraging progress in low-error approaches to impartial atom quantum computing, give recent life to the endeavor. Decreasing errors could also be, partially, an engineering drawback that may be solved with higher tools and design.
“The factor that has held again impartial atoms, till these papers have been printed, have simply been all of the classical stuff we use to regulate the impartial atoms,” Bloom mentioned. “And what that has primarily proven is that if you happen to can work on the classical stuff—work with engineering companies, work with laser producers (which is one thing we’re doing)—you possibly can really push down all that noise. And now unexpectedly, you’re left with this extremely, extremely pure quantum system.”
Along with error-correction in impartial atom quantum computer systems, IBM introduced this yr they’ve developed error correction codes for quantum computing that might scale back the variety of crucial qubits wanted by an order of magnitude.
Nonetheless, even with error-correction, large-scale, fault-tolerant quantum computer systems will want lots of of hundreds or thousands and thousands of bodily qubits. And different challenges—equivalent to how lengthy it takes to maneuver and entangle more and more massive numbers of atoms—exist too. Higher understanding and dealing to unravel these challenges is why Atom Computing is chasing scale concurrently error-correction.
Within the meantime, the brand new machine can be utilized on smaller issues. Bloom mentioned if a buyer is eager about operating a 50-qubit algorithm—the corporate is aiming to supply the pc to companions subsequent yr—they’d run it a number of occasions utilizing the entire pc to reach at a dependable reply extra shortly.
In a subject of giants like Google and IBM, it’s spectacular a startup has scaled their machines so shortly. However Atom Computing’s 1,000-qubit mark isn’t prone to stand alone for lengthy. IBM is planning to complete its 1,121-qubit Condor chip later this yr. The corporate can also be pursuing a modular method—not not like the multi-chip processors widespread in laptops and telephones—the place scale is achieved by linking many smaller chips.
We’re nonetheless within the nascent levels of quantum computing. The machines are helpful for analysis and experimentation however not sensible issues. A number of approaches making progress in scale and error correction—two of the sector’s grand challenges—is encouraging. If that momentum continues within the coming years, certainly one of these machines could lastly clear up the primary helpful drawback that no conventional pc ever might.
Picture Credit score: Atom Computing