Rebirth: Starting from Lighting Up the Tech Tree

Chapter 217 Quantum Chips

On the 180th day of the quantum laboratory's official operation, Shen Yiming placed the eighth version of the test report on Zuo Cheng's desk.

The conclusion section on the first page of the report contained only two words: Failure.

It's not that the solution failed, but that the approach was flawed. The coherence time of the superconducting qubit was stuck at 80 microseconds, and after six rounds of optimization, it was pushed to a maximum of 93. The error correction code overhead ratio was 16 to 1, meaning that protecting each logical qubit required 16 physical qubits. At this ratio, to achieve stable operation with 72 logical qubits, at least 1,152 physical qubits would be needed.

IBM, the world's leading company, has only just broken through the threshold of 1,000 physical qubits. With 402's production capacity, it would take at least three years to reach that scale.

After reading the report, Zuo Cheng asked, "What about the topological route?"

Shen Yiming turned to the next few pages. "Even worse. The Majorana zero-mode signal was completely drowned out by thermal noise. We conducted seven rounds of experiments at -272 degrees Celsius, and the best signal-to-noise ratio was only two decibels, far below the readable threshold."

Zuo Cheng remained silent for a moment. Then he opened the system panel.

The beam of light from the perception of civilization in the quantum direction was still faint, but six months of research had accumulated a vast amount of data. He invoked the scanning function, pulling up all the patents, papers, and experimental data in the field of quantum computing worldwide. Hundreds of laboratories, thousands of papers, and tens of thousands of experimental records unfolded simultaneously in his mind.

The vast majority of these are dead ends. The superconducting route has almost reached the limit of surface code error correction. IBM spent three years reducing the error rate from 1.2% to 0.9%, and pushing it further requires entirely new materials and processes.

But there are two different paths.

The first is a paper from a Finnish university lab, published in Physical Review Letters, with very few citations. They discovered that weak signals from topological qubits can be enhanced using artificial intelligence algorithms, increasing the signal-to-noise ratio from two dB to seven dB in simulations. Reviewers described the paper as theoretically interesting, but virtually impossible to implement in real-world noisy environments.

The second finding is more discreet, hidden in a technical memo within Google's quantum team and never publicly released. They discovered that the readout signal of a quantum state and the multipath fading channel in a communication system are mathematically highly similar, and that an adaptive equalizer could significantly improve the readout error rate. However, Google's AI team and quantum team have never collaborated, and the memo remained unpublished.

Zuo Cheng put the two clues together. Then he knew what to do.

Technological integration.

He activated the capability on the system panel. Three leaves were selected: a deep learning framework for the AI ​​branch, an intelligent adaptive modulation system for the communications engineering branch, and gradient sparsity compression for the Internet of Things branch. The three leaves from different branches began to merge on the main trunk of the technology tree.

Five points were used up, reducing the score from 622 to 617.

The fusion result takes shape in consciousness. Deep learning's adaptive filters can directly act on weak signals of topological qubits, and the modulated multipath equalization algorithm can be reformulated as an error correction algorithm for quantum state readout. Two seemingly unrelated threads become the same thing in the instant of fusion.

Adaptive quantum error correction algorithm.

Zuo Cheng snapped out of his reverie from the system panel and wrote all the technical details into a document. From the algorithm framework to the physical implementation, from noise modeling to signal extraction, each step included a complete derivation and verification scheme. The document was forty-seven pages long.

He sent the document to Shen Yiming. The title was a single line.

Topology route. Signal-to-noise ratio from two decibels to twelve decibels.

Shen Yiming stormed into Zuo Cheng's office that afternoon. He held the forty-seven-page document in his hand, but paused at the door for three seconds before speaking. His expression wasn't one of surprise, but rather one of utter shock, as if his brain hadn't quite caught up in the impact.

"How did you do that?" he asked softly.

Zuo Cheng asked, "Will it be useful?"

"Is it useful?" Shen Yiming turned to page thirty-two of the document, pointing to the derivation of the quantum state equalizer. "This approach uses the same mathematical framework as multipath fading equalization in communication systems, but nobody has thought of it in this direction, at least not in any publicly available literature. How did you connect the dots?"

Zuo Cheng said, "I just happened to have read that Finnish paper."

Shen Yiming stared at him for three seconds, then smiled. It wasn't a happy smile, but the kind of smile that comes from finally accepting defeat and letting go of the lingering resentment. He didn't press the matter.

The next three weeks were the most intense period for the quantum lab. Shen Yiming led a team of five working day and night in shifts to transform the adaptive error correction algorithm from documentation into hardcore code that could run on quantum chips. Yu Ying derived three rounds of mathematical verification, confirming that the algorithm could converge even under the worst noise conditions. Fang Ze's team simultaneously optimized the front-end readout circuit of the quantum chip, reducing the noise floor by three decibels.

Wednesday night of the third week, the seventh round of experiments.

After the Majorana zero-mode signal exits the quantum chip, it first undergoes a 3-decibel noise reduction process in the front-end readout circuit before entering the processing pipeline of the adaptive quantum error correction algorithm. The waveform that was originally submerged in noise on the screen begins to emerge clearly, like stones being revealed one by one at the bottom of a river after sediment has settled.

Signal-to-noise ratio. 12.4 dB.

It was 0.4 decibels higher than Zuo Cheng predicted.

Shen Yiming took a step back from the screen. It wasn't just the readable threshold of the quantum state; this data was approaching the signal-to-noise ratio of a medical-grade brain-computer interface. He removed his hands from the keyboard, as if afraid of accidentally messing up the data.

Yu Ying wrote a number in her notebook: Stability improvement factor: 10.7 times. Previously, a 2-decibel signal could only last a few milliseconds before collapsing; now, 12.4 decibels can be stably maintained for over a minute.

"That's enough," Zuo Cheng said.

"It's enough." Shen Yiming repeated, then added, "It's far more than just enough."

The next step is chip finalization. Fang Ze completed the design of Galaxy-1 within a month. It has 72 topological qubits, a quantum volume of 1,024, and power consumption controlled within 120 watts. All key indicators surpass IBM's Falcon chip released at the same time, with a quantum volume nearly twice as large.

On the day the chip was finalized, Zuo Cheng stood outside the cleanroom of the quantum laboratory, looking through the glass at the tiny silver cube, no bigger than a fingernail. Star River One. The same size as the seed buried beneath the Taklamakan Desert four billion years ago, but with a completely different purpose. The seed was used for planting; Star River One was used for calculation.

On the system panel, the progress bar for quantum technology jumped from zero to forty-three percent. The bark texture of the eighth branch was already faintly visible on the main panel.

He looked at the progress bar and said something to Shen Yiming beside him.

"Not fast enough."

Shen Yiming didn't turn around, his eyes fixed on the silver cube in the cleanroom. He certainly felt it wasn't fast enough. Zuo Cheng felt it wasn't fast enough after every breakthrough. From AI to unmanned systems to space photovoltaics to commercial aerospace to brain-computer interfaces, he had never been satisfied with the speed.

But this time, Shen Yiming felt that he was right.