Researchers have simulated 46 qubits on the world’s top supercomputers
Researchers from the Jülich Supercomputing Center (Germany), Wuhan University (China) and Groningen University (Netherlands) announced that they have broken the world record for simulating the number of qubits on a classical supercomputer. With the ability to simulate 46 qubits, this team broke the previous record, which was 45. They were also able to simulate 32 qubits on a laptop with 16 gigabytes of RAM.
Quantum simulation limits on supercomputers tighten
After quantum computer developers claim that their computer is “superior “Quantum” have reached, first we need to know the maximum limit of classical supercomputers. This limit was previously 45 qubits that could be simulated on a supercomputer; But researchers from Wuhan, Jülich and the University of Groningen were able to add a qubit to it and do so with much less memory. Although the difference between 45 and 46 qubits may seem small; But we must point out that the performance required to simulate qubits grows exponentially for each added qubit. Typically, under the same conditions, for each qubit added to the simulation, the required memory must double.
IBM was previously able to simulate 56 qubits in a narrower type of simulation that used shallow quantum circuits. would do; While the new simulation record has been achieved in quantum computing circuits of any kind.
Prof. Michelsen, leader of the Quantum Information Processing Group at Germany’s Jülich Supercomputing Center, explains to Tom’s Hardware the difference between the two:
Our quantum computer simulator can simulate any quantum circuit with 46 qubits. We simulated a universal quantum computer by computing the omnidirectional vector, i.e. the entire range (power 2 to 46) of the quantum state vector.
- Is the world we live in Is it a computer simulated world?
- IBM using quantum computers Simulates chemical reactions
IBM calculates quantum state amplitudes only for measured outputs and provides experimental evidence that the difficulty of calculating amplitudes The quantum state for measured results may not be as fast as expected (in practice many amplitudes may be zero). Experimental evidence is derived from the simulation of two random quantum global circuits, one with a depth of 27 for 49 qubits and one with a depth of 23 for 56 qubits. Divided as follows: gates that operate on qubits in any given layer without any overlap. He also added: IBM’s limitations in the form of the number of qubits or circuit depth are not yet clear.
How is the simulation achieved?
Both Jülich supercomputers (JUQUEEN) as well as the world’s fastest supercomputer, China’s Sunway TaihuLight.
Professor Michaelsen noted that there are only a few supercomputers that have ever performed a simulation with 45 qubits. ; Because few of them have the required memory, computing nodes and fast enough network connections to perform the simulation. He added:
Software is also very important; Because it must be able to run the simulation effectively.
According to Michaelsen, his team is developing software that works almost perfectly on millions of computing nodes without any performance degradation. He was the first to simulate 42 qubits in 2010 on the previous generation of Yulich supercomputers and was able to improve his record in 2012 with a 43 qubit simulation.
In addition to using a more powerful supercomputer for simulation 46 qubits, the researchers were also able to use only 2 bytes of information to represent the qubit state, down from the 16 bytes previously required. This is what now prevents 32 qubits from being simulated on a laptop with only 16 gigabytes of memory.
By comparison, Rigetti (one of the quantum computing startups) was initially able to simulate only 30 qubits; But now it can simulate 36 qubits. Microsoft initially offered 30 qubit simulations; But now it can provide 40 qubit simulations. Jülich’s latest discovery may force these companies to increase the number of simulated qubits even higher. Jülich’s researchers also believe they have the ability to simulate the relatively large number of qubits that are currently running; This means that once developers excel in quantum, they can test it with algorithms running on quantum computers and prove that they can be more practical.