The Truth in the Noise
Cheol-jun stared blankly at the paper by the cosmic evolution research team at the National Astronomical Observatory of Japan (NAOJ), which he had read over and over for the past few days.
Then, he recalled what Mr. Cha Seon-hyeok had argued.
"The Japanese team is insisting that those planets will collide and destroy themselves soon. But my intuition tells me there's definitely dark matter holding that orbit tight so it doesn't collapse. The problem is, we can't prove it properly. We found a pattern, but we're stuck on the analysis..."
Cheol-jun narrowed his eyes.
First of all, Japan had clean evidence extracted by a supercomputer based on high-resolution spectroscopic data from the Subaru Telescope.
Cheol-jun had already examined all the Level 2 data released alongside the paper—that is, the refined data.
It was data that the Japanese team's AI algorithm had neatly organized into coordinates A, B, and C after completely wiping out the noise.
In that data, there was absolutely no room for refutation.
That was why Mr. Cha Seon-hyeok was focusing on the patterns found in the SDSS raw data.
However, the public SDSS raw data contained far too much noise.
Once the computer filtered out all the noise, the remaining amount of information was severely lacking.
After listening to a brief explanation, Cheol-jun could understand what kind of pattern Mr. Cha Seon-hyeok had seen to intuitively deduce the presence of dark matter.
Putting aside any Korea-Japan rivalry, Cheol-jun thought the process of making people understand what Mr. Cha Seon-hyeok and he had seen would be quite fun.
It felt like a form of communication, a way of building a broad connection, and he liked that feeling.
So, he was happily pondering over it.
Cheol-jun organized the thoughts that had come to him while watching Si-woo's Baduk game.
Averages, optimization, and the things discarded because of them.
'The noise that the computer discarded during optimization?'
There may still be truth within the things discarded by an optimized system.
The logic he had suddenly derived from Uncle Gödel's Incompleteness Theorems kept swirling in his mind.
Cheol-jun's round eyes sparkled.
'The original raw spectroscopic data files from the Subaru Telescope.'
Level 0 raw data. A massive, tens-to-hundreds-of-terabytes-large junkyard containing everything exactly as captured by the telescope's lens: dust, satellites, stellar flares, and lens vibrations.
*Crunch.*
Cheol-jun stared blankly at his computer tower while chewing on a puffed rice snack.
'I can't do this alone at home.'
He turned his head and looked at the bookshelf.
Out of the countless books lined up, he selected the ones containing the concepts absolutely necessary to properly organize what was currently in his head.
'Topology, differential geometry, functional analysis, Hamiltonian mechanics equations, canonical equations... And using all of these, what I ultimately need to reach are the non-degeneracy condition of the KAM theorem, the perturbation size condition, and the Diophantine condition.'
Cheol-jun, who had piled a stack of thick books on his desk, beamed.
'Wow! This is going to be so much fun!'
Cha Seon-hyeok stood blankly in front of his monitor with his arms crossed.
A similar graph had been displayed on the screen for the past three weeks.
Plots extracted from the SDSS raw data.
No matter how they preprocessed it, the result was always the same.
If they stripped away the noise, the signal became too weak; if they kept the noise, the pattern was buried.
It felt like the answer lay somewhere in the middle, but they didn't know where that middle ground was.
"Did anything come up this time?"
Do Seon-ju asked cautiously.
"It's the same."
"..."
One of the graduate students glared at the monitor before leaning his head back against the chair with a thud.
"Professor, we've changed the algorithm parameters twelve times already."
"I know."
"I feel like the thirteenth time will yield the same result."
"...I know."
Cha Seon-hyeok couldn't take his eyes off the graph.
He had a distinct feeling that something was there.
It was an intuition he had held since his time in Leiden.
The fact that such minute deviations appeared so consistently in a multi-body orbit meant there was a high chance it wasn't just noise.
But he couldn't find a way to capture that feeling with precise numerical values.
"...Professor."
Do Seon-ju crept closer.
"Have you thought about the Subaru raw data?"
Cha Seon-hyeok kept his mouth shut and knit his brows.
"It's been twelve months since the observation, so it should be uploaded to SMOKA by now."
"..."
"Maybe if we look at what the Japanese team filtered out ourselves..."
"Seon-ju."
"Yes?"
"How many terabytes is that?"
"Well, maybe around fifty to a hundred terabytes?"
Cha Seon-hyeok shook his head.
"Who on earth is going to look through all of that again?"
"Well, we could at least do some sampling..."
"We have to know which part to sample to actually do it. How are we supposed to dig through a hundred terabytes when we don't even know what is where? Besides, the Japanese team must have used a far more sophisticated algorithm than ours when they preprocessed it. We're going to recover what their supercomputer filtered out? How?"
Do Seon-ju pouted, and Cha Seon-hyeok looked back at the graph.
There were only about two months left until the summer conference.
That was the deadline set by the department head.
If nothing came out by then, they had agreed to cleanly wrap it up.
'Sigh, what a shame...'
Honestly, in their current state, he wouldn't have confidence even if they had two years instead of two months.
Cha Seon-hyeok let out a quiet sigh and sat down in his chair.
Then, he grabbed the mouse.
There was nothing else to do but adjust the parameters for the thirteenth time, just as they had been doing...
Just then, his phone vibrated.
'Huh?'
As soon as he checked the caller's name, a warm, automatic smile spread across his face.
He quickly picked up the phone.
"Oh, Chul-jun! Hey there! Huh? Today? Sure, sure. Yep."
Cha Seon-hyeok clapped his hands together.
"Our lucky charm is coming. Let's air the place out and run the vacuum. And light that scented candle over there!"
"Yes, sir!"
Vitality suddenly bloomed in the lab, where a heavy, gloomy atmosphere had been lingering.
"Where's Cheol-jun?"
"As soon as he got out of the car, he ran straight to the astronomy department."
At Hyeong-man's words, Kim Seong-cheol, the head of the piano department, shook his head.
"Does he not play the piano anymore?"
"No, he plays occasionally. Just not as much as before."
"Ah... He has quite the musical talent, too. What a waste. Such a waste."
"Haha. He'll figure it out. It's not like that talent is going anywhere."
"So, does he say astronomy is fun?"
"Yeah. He loses himself when he's reading books. These days, even when I tell him to go to sleep, he sometimes pretends not to hear and keeps fidgeting around. Before, he was like clockwork when I told him it was bedtime... *Tsk*."
Kim Seong-cheol chuckled.
It had only been a few months, but Hyeong-man sounded exactly like a doting father.
"Should we grab dinner together later?"
"Hmm, I'm not sure. Cheol-jun said he might finish a bit late today. That's why I tagged along."
"Late?"
"Yeah. He said he has quite a lot to explain. And if the researchers there agree with him, there are a lot of things he needs to try out. Apparently, he can't do it with the computer at home. I have no idea what he's talking about. Do you know anything about astronomy?"
"Haha. I don't really understand natural science stuff either, even when I hear it. So you're going to wait for him?"
"Yeah. I was thinking of ordering them some food around dinner time. The people in that lab have their own work to do, but they still look after Cheol-jun. They adore him so much. I'm really grateful they take care of him like that. Seeing those young folks researching so passionately is also pretty cool."
Kim Seong-cheol suddenly turned his head and looked out the window.
"If we're talking about being cool, you guys were much cooler."
"Hmm?"
"Is it easy to hide in the shadows and take on dangerous work for the country when no one even knows or appreciates it? Thanks to that, a guy like me got his life saved."
"We did it because the hazard pay was high. The kickbacks from the field were pretty sweet, too."
"Anyway, what's impressive is impressive."
"Ha, you think I didn't have any desire for honor or status? Look at me. I couldn't stand the overseas operations anymore, so I came back, and as soon as I got passed over for the deputy director position, I walked right out. You need to at least be a deputy director to show your face in public and act important."
"..."
"In that sense, Jaegang was a guy I truly respected. He had saved up quite a bit of kickback money, and on top of that, a wealthy and beautiful woman was waiting for him with single-minded devotion. Yet, he stubbornly insisted on finishing up the operation before coming back. If he had just returned to Korea, he would have had it made... *Sigh*... but look how it turned out."
"..."
"Leaving that poor kid all alone... *Tsk*. Bastard."
"Sigh..."
The two of them knit their brows and fell silent for a moment.
After a short while, Kim Seong-cheol scrunched up the bridge of his nose and leaned against the window frame.
"But seriously, what on earth is he explaining that takes so long? I'll never understand the natural sciences."
"Haha..."
"..."
Cha Seon-hyeok blinked.
The other researchers were also staring blankly at the electronic whiteboard.
The electronic whiteboard, installed low to match Cheol-jun's height, was packed tight with mathematical formulas.
Standing in front of it, Cheol-jun narrowed his eyes with a serious expression, glared at the formulas for a moment, and then spoke.
"There's a lot of data on minute gravitational deviations that the computer cut out, classifying them as stellar jitter or observational noise. But I don't think all of it is useless."
Cheol-jun pointed to a formula with the tip of his pen.
It was a Hamiltonian mechanics equation that converted the discarded data into the Z-axis, twisting the orbit into three dimensions.
It was perturbation theory, something they hadn't seen in a very long time, as everyone these days had grown accustomed to comfortably leaving numerical integration to supercomputers.
"You all know that the universe isn't coplanar. So why don't we calculate using mutual inclination?"
Cha Seon-hyeok knit his brows slightly.
"...It's not that we don't do it because we don't know about it."
Cha Seon-hyeok moistened his dry lips with a bitter expression.
"When you run an N-body problem on a supercomputer, if you insert the Z-axis—the mutual inclination variable—the number of possibilities explodes to infinity. Because of chaos theory, tiny errors snowball, causing the computer to spit out calculation errors or conclude that the system will immediately collapse. That's why the entire academic community made an implicit compromise."
"A compromise?"
"Yeah. If we don't have precise inclination data, we just assume it's a coplanar system with a zero-degree inclination, like our solar system, to make the computation easier."
Cha Seon-hyeok pointed to the red noise data displayed on the monitor next to Cheol-jun, which the Japanese team had discarded.
"On top of that, when the ever-so-smart artificial intelligence preprocessed the data, it kindly ironed out those minute three-dimensional vertical vibrations, labeling them as statistically insignificant with a 97.9% probability of being stellar flares. Scholars are only human; they had no reason to doubt the clean, two-dimensional data the computer handed them. And when they ran it with that flat data, they got a provocative, sexy result perfect for Nature—that the compact multi-planetary system SBR-714 would meet its doom in a catastrophic collision within a million years."
Cha Seon-hyeok's gaze locked onto the beautiful geometric formula written right in the center of the whiteboard, proving that the perturbation does not diverge.
His pupils trembled slightly.
The **KAM Theorem** (Kolmogorov-Arnold-Moser Theorem).
The only theorem in human history that proved absolute stability in a chaos-filled **Many-Body Problem** using pure mathematics alone.
A formula that had been neglected in coding due to frequent errors ever since humans began relying entirely on computers was now written clearly on the electronic whiteboard.
To look at the many-body problem through the lens of pure mathematical astronomy—of design and proof—which had stalled ever since the field began relying on supercomputer numerical integration.
And by an elementary schooler who had only just been introduced to astronomy?
It was certainly an extraordinary approach.
...That much was certain.
But.
"Doubting the statistical errors of AI, picking up data the computer threw into the trash... and trying to solve three-dimensional variables that even supercomputers choked on, using complex analytical mechanics by hand... There isn't a single crazy astronomer who does that these days."
Cha Seon-hyeok caught his breath and turned his head to look at the small child.
"...Except for you, that is."
The small child tilted his head.
"Am I doing this by myself?"
"Huh?"
Cheol-jun blinked his round eyes and spoke again.
"We're all going to do it together, aren't we?"
"...What?"
Cha Seon-hyeok swallowed hard, and the researchers who had been watching blankly as if it were none of their business flinched.
Looking at them, Cheol-jun smiled brightly.
"It's going to be fun, right?"
Because things that are fun are always better when done together.