145: Chapter 145 Proof of Re-equipping

August 12th, 2:40 AM.

Switzerland, ETH Zurich, the fifth floor of G Building at the Department of Mathematics.

The motion-activated lights on the entire floor had long since gone out.

Only through the crack in the door of Professor Heinrich Voss's office at the end of the corridor did a sliver of light still shine.

Voss was sixty-one years old this year.

He wore a dark gray woolen sweater that had been washed somewhat faded, with reading glasses perched on the bridge of his nose, and his silver-white hair appeared somewhat sparse against the backlight of the desk lamp.

If one only looked at his appearance, he was just like a retired grandfather seen everywhere on the streets of Zurich, quiet, rigorous, and even somewhat rigid.

But in this office piled high with manuscripts and preprints, he was an old-school figure in the direction of the entropy method within Additive Combinatorics.

For thirty years, he had always done only one thing.

Amidst the gears of multivariate probability distributions, searching for the boundaries of symmetry, entropy dissipation, and structural compression.

He was not the type of scholar who liked to express opinions in public forums.

He had no Twitter account.

He did not participate in hot debates in academic circles.

In fact, a week ago, he had not even registered a Lean Zulip account.

He was accustomed to letting mathematics stay on paper, blackboards, and the densely packed pencil annotations on the margins of manuscripts.

His initial impression of the name Jiang Lin came from a phone call made by Manners on August 4th.

At that time, Voss was vacationing in Zermatt at the foot of the Alps.

Calling it a vacation, it was actually just changing to a different place to work.

He sat on the wooden cabin's terrace, with a German novel by Thomas Mann spread across his knees, and a cup of black coffee placed beside the table.

In the distance, the snowline on the ridge of the Matterhorn gleamed with a dazzling white light in the afternoon sun.

His wife walked over carrying a plate of baked muffins, and seeing him stare blankly at the snowline, she simply shook her head helplessly.

She was far too familiar with this state.

A mathematician's so-called vacation was often just moving from the blackboard in the office to the dining table at the foot of the mountain.

His body was in the mountains.

His soul was still imprisoned within those abstract symbols.

It was that very afternoon when Manners's call came in.

After the call was connected, there was not even a single conventional greeting.

Manners's voice exuded an unusual sense of urgency.

" Heinrich, you must look at this, right now. "

Voss clicked open the file package sent by Manners.

The first was the PFR/Marton v0.95 circulation draft manuscript that Jiang Lin had just opened to the professional review circle.

The second was a formalized dependency graph description listing the Lean4 coding progress of nodes 31 through 47.

The third was the natural language blueprint of the dual involution wrapper for node 38.

Manners highlighted only a single sentence in the email.

[ " Start reading from node 38. " ]

Thus Voss did not slowly read from the first page.

He cut straight into Section 4.2.

Third-layer loss recovery.

Residual spectrum accounting.

Entropy Involution Lemma.

High-dimensional boundaries in the case of K greater than or equal to 8.

That was the quagmire he was most familiar with in these thirty years.

He sat in that rattan chair on the terrace, and spent a full three days reading only this single spot.

Reading repeatedly.

Calculating repeatedly.

Repeatedly spreading Jiang Lin's natural language proof, Lean formalized blueprint, and dependency graph nodes on the same tabletop, cross-referencing and pushing back through them.

The wind in the mountains blew across his woolen sweater, but he seemed to have lost his perception of temperature.

By the evening of the third day, he closed his computer and gazed at the snowline of the Matterhorn for a long time.

Afterward, he did something he had rarely done in his thirty-year academic career.

He ended his vacation early, drove three hours back to his office in Zurich, and began writing a boundary test note.

It was not a refutation of the paper.

Nor was it the announcement of a loophole.

Even less was it out of an older generation scholar's jealousy toward a young person.

Voss had no such interest.

His eyes, soaked in the entropy method for thirty years, merely saw a visibility risk at a high-dimensional boundary.

In the v0.95 circulation draft manuscript, to handle complex four-variable posterior measures, Jiang Lin adopted a fairly light dual involution encapsulation route in the third-layer loss recovery section.

That route was extremely beautiful.

Like a thin and sharp scalpel, it forcefully compressed the extremely tedious symmetric structure of node 38 into a relatively short formalized interface.

If viewed solely from the global logic of the main proof, it was not wrong.

Nor did Voss see a counterexample that could overturn the main theorem.

But he was far too familiar with the entropy method.

He knew deeply that in the deep waters of mathematics, things that were overly pretty and concise would often hide certain boundary-layer costs too deeply.

It wasn't hidden away completely.

Instead, it was hidden where external reviewers could not see where it was actually being paid.

Complexity does not vanish out of thin air.

If it is flattened here, it must be clearly absorbed somewhere else.

If the author does not explicitly highlight that absorption process, subsequent readers might get stuck in high-dimensional, high-K scenarios.

Not because the bridge collapsed.

But because a certain load-bearing structure of the bridge was wrapped inside concrete.

The author knew it was there.

External reviewers, however, could not see it.

Voss locked himself in his office and spent five days repeatedly calculating on the whiteboard, little by little compressing that vague intuition in his mind into a test sharp enough.

In a finite-dimensional model like F₂¹⁵ where spectrum cluster backflow was already sufficient to appear, taking a specific K = 8 boundary configuration.

If pushing forward along the dual involution encapsulation route of the v0.95 circulation edition, the first two rounds of iteration were entirely normal.

But by the third round of iteration, an extremely inconspicuous boundary-layer backflow would appear in the residual spectrum loss term.

This backflow would not cause the main theorem to collapse.

Nor was it sufficient to show that Jiang Lin's main proof failed.

But it was sufficient to illustrate one problem.

The lightweight path given by the v0.95 circulation edition did not demonstrate the absorption of this boundary sufficiently.

In other words, it allowed the author to walk across himself, but might not let external reviewers follow with peace of mind.

For an ordinary paper, this might merely be insufficient annotation details.

But for a manuscript claiming to pierce through the core difficulties of finite field PFR/Marton and currently being taken over by a formalized blueprint, this was a review issue that had to be handled seriously.

Voss wrote these five days of calculations into an eleven-page English note.

The title was deliberated by him for a very long time, and finally set very cautiously.

" A High-Dimensional Boundary Test for Jiang's Entropic Involution "

A High-Dimensional Boundary Test for Jiang's Entropic Involution.

He did not use a sensationalist title like " Jiang's proof is wrong ".

Nor did he write " fatal gap " in the abstract.

But he wrote very clearly in the main text derivations:

Under this specific high-dimensional boundary test, the lightweight dual involution path of the v0.95 circulation edition had not yet explicitly demonstrated how the third-layer residual spectrum loss completed its absorption.

This was not a negation of the conclusion.

This was review pressure.

August 12th, 2:47 AM.

Voss sat in front of his desk and rubbed his dry eyes.

He did not immediately post the note out.

He first opened his email.

Voss was an old-school person.

If you were to point out a high-dimensional boundary risk in a young author's manuscript in public, you should first let that young author know what you were truly pointing out.

No sneak attacks.

No engaging in public opinion pressure.

No packaging academic tests into traffic weapons.

He began typing on the keyboard.

Dear Lin:

I have prepared a short note discussing a high-dimensional boundary test of the dual involution route in your v0.95 manuscript.

Please allow me to be precise: I am not claiming that the main theorem is false.

The scope of the problem is narrower, but I believe it is serious. In high-dimensional, high-K regions, especially around F₂¹⁵ and K=8, the dual involution route in the current circulation edition does not display the third-layer residual spectrum absorption process to a degree sufficient for external review. I have placed the relevant calculations in the appendix.

My judgment is that this problem should be handled by explicitly writing out a heavier third-layer parametrization, which should at least be presented in the formalization part or the appendix, rather than doing a superficial rewrite of the current wrapper.

Before circulating this note on a larger scale, I am writing to inform you first.

Best regards,

Heinrich

After finishing writing it, he stared at the screen and read it once.

The mouse cursor moved on the screen, and he changed two expressions that might cause ambiguity, deleting an adverb with a slightly heavy tone.

After making sure this text expressed both serious academic pressure and did not lose an elder's politeness, he pressed the send button.

The moment the email was sent, he glanced at the time in the lower right corner of the screen.

Zurich time, 2:47 AM.

Converted to China time, this moment was precisely 8:47 AM.

Jiang Lin was in the hardware assembly workshop of the Low Entropy Workshop.

After the first round of grayscale testing in the Hengtai unmanned closed tunnel, the G-01C No. 2 machine was transported back to the workshop overnight.

They were conducting a return-to-base debriefing.

On the lifting platform in the center of the workshop, the six-legged engineering prototype internally codenamed G-01C lay quietly there.

On the dark gray carbon fiber shell, it was covered with scratches left by blind testing on gravel slopes and low-speed sections of the closed tunnel.

The protective layers of the six multi-degree-of-freedom mechanical feet had been peeled off by Xu Man along the adhesive layer boundaries according to the disassembly procedure, revealing the high-sensitivity sensor array inside that had been replaced once.

Several colored ribbon cables hung outside like exposed nerves.

Xu Man wore protective goggles, held a torque wrench in her hand, and was replacing the foot-end cushioning module for the left front foot.

Chen Yan held a ruggedized industrial tablet, squatted beside, eyes fixed on the data stream jumping on the screen, recording engineering logs in a spreadsheet.

" Boss Jiang, the radial runout data of the left front foot is still a bit drifting. "

Chen Yan reported without even raising his head.

" Yesterday in the second round of the low-speed section of the closed tunnel, it got stuck once by a piece of wet cinder. The reducer might have taken an impact. Should we raise the response threshold a bit? "

Jiang Lin stood aside, looked at the scarred machine, and shook his head.

" Do not. "

" The purpose of us making discrete automata is to let the machine perceive when physical damage occurs that something is amiss, rather than relying on increasing tolerance to pretend it wasn't injured. "

" Leave the true deviation behind. "

" If subsequent models want to process, they process the damage of the Real World, not data that has been artificially smoothed out. "

Chen Yan nodded and typed a remark into the log.

[ Left front foot radial runout abnormal, do not raise threshold, retain true deviation. ]

Just at this moment, the phone Jiang Lin placed on the tool bench beside vibrated once.

The screen lit up.

New email.

The sender suffix was:

hvoss@math.ethz.ch

Jiang Lin walked over and picked up the phone.

Heinrich Voss.

He clicked open the email.

The motor sounds, fan sounds, and slight clicking sounds of the torque wrench in the workshop seemed at this moment to be pushed away by some invisible force.

When he read the third paragraph and saw the phrases F₂¹⁵, K=8, and third-layer residual-spectrum absorption, his gaze paused on the screen for a second.

Not because of panic at being caught in a loophole.

Nor because of vexation over a calculation error.

But because this extreme boundary case pointed out by Voss, he had seen it.

Not in a library of the Real World.

Nor in a study room of Jiangcheng University.

But in the middle period of the Ninth Wasteland.

At that time, he had once used homemade paint mixed with rust and red clay to write a third-layer parameterization scheme on the north wall of the Stone House that was far larger and more complex than the v0.95 circulation edition.

That route was later temporarily named by him as the triple involution route.

In that scheme, the third-layer spectrum cluster mapping was not naturally induced by the first two layers of dual involution like it was now, but was extremely stubbornly and separately introduced once by him for a third-layer spectrum cluster re-calibration.

That route was heavy.

So heavy that if stuffed directly into the main text, the entire paper would immediately gain at least twenty pages of obscure and difficult-to-understand technical derivations.

So heavy that it would turn the already extremely high-density PFR/Marton proof further into a solid steel block that almost no one could read intuitively.

Every variable substitution was like dancing in shackles.

Every lemma required cumbersome conditional prerequisites.

Therefore, when organizing the v0.95 circulation edition, Jiang Lin did not put that route into the main text.

He compressed it into old branches, technical memos, and self-use indexes.

The main text still retained the lightweight dual involution wrapper.

In the global structure of the main proof, doing so was not a problem.

But Voss reminded him of another matter.

Now this proof was no longer just his own proof.

It was being jointly reviewed by Professor Han Yanshan, Ding Jian, Terence Tao, and Voss.

It had to allow others to enter.

And for others, some heavy things could not be hidden forever in the author's own old branches.

Jiang Lin gently exhaled a breath.

" Chen Yan, Xu Man. "

Both people raised their heads simultaneously.

" I am leaving the workshop side first today. The safety boundaries of the Hengtai grayscale test have been locked, and the rest of the return-to-base debriefing will proceed according to the original plan. "

Jiang Lin's tone was very calm.

" If you encounter any blockages on state machine synchronization, send me a WeChat message first, and I will check it later. "

Chen Yan's hand typing on the keyboard stopped midair.

Xu Man also took off her gloves and asked cautiously, " Boss Jiang, is it that Hengtai changed its mind again? "

" No. "

Jiang Lin picked up his black backpack.

"Someone on the mathematics side did a very good boundary test, forcing out a heavy route I had previously pushed into an old branch."

Xu Man was completely lost.

Chen Yan understood even less.

When Jiang Lin walked to the entrance of the workshop, he paused for a moment and turned back to instruct.

"Chen Yan, record today's failure logs in more detail."

"Especially the latency data of the discrete automaton during the fraction of a second when switching gaits. Don't just write down the average."

"Draw separate curves for p95 and p99."

"What we want to see are the edges, not a steady state whitewashed by averages."

Chen Yan immediately straightened up.

"Understood."

Jiang Lin nodded and turned to leave.

After returning home, he pressed the power button of the computer mainframe.

The system booted up.

He opened a hidden folder.

Path: PFR_Marton/Old_Branches/Layer3_HeavyRoute

Inside the folder quietly lay three files.

triple_involution_sketch.tex

sigma_layer3_parameterization_notes.md

Node38_heavy_route_discarded.lean

Looking at the last file name, Jiang Lin was silent for a moment.

Then, he changed the file name.

Node38_heavy_route_boundary_witness.lean

It had once been shelved outside the public reading path.

But now, it was coming back.

Jiang Lin opened his email and clicked the reply button on Voss's email.

...

Dear Heinrich:

Thank you for your note.

I have checked your boundary test. In the following sense, your diagnosis is correct: the double involution route in the v0.95 circulated version indeed does not display that third-layer boundary absorption process to a degree sufficient for external review.

But for the complete proof tree, this is not a new problem.

I have a heavier route here, using triple involution along with a specific σ_layer-3 parameterization. The reason I didn't put it into v0.95 is that it significantly increases the textual expression load. Your note has convinced me that it should be restored to the formalization section and appendix as a boundary witness route.

After the Lean check passes, I will send the branch to you.

...

Click send.

To his surprise, only six minutes later, Voss's reply arrived.

...

Lin——

This is much stronger than the response I originally expected.

If the triple route already exists, please send it to me.

H.

...

Jiang Lin did not reply again; the conversation was suitable to end here.

Next was the battlefield of code and logic.

August 12, 4:00 PM.

10:00 AM Zurich time.

Voss uploaded that eleven-page note to arXiv.

"A High-Dimensional Boundary Test for Jiang's Entropic Involution"

In the abstract, he deliberately retained a qualification.

[This note does not claim a counterexample to Jiang's theorem. It identifies a boundary visibility issue in the circulated version. (This article does not claim to provide a counterexample to Jiang's theorem. What it points out is a boundary visibility issue in the v0.95 circulated route.)]

This sentence was clear enough.

But the public world never completely retains qualifications.

Due to the time difference, the domestic mathematics circle did not notice this article until evening through automated scraping scripts and academic group forwarding.

The first to sniff the smell of gunpowder was a highly influential mathematics popularization official account in China.

The tweet was sent out at 7:00 PM. The title was somewhat restrained, but it already carried obvious provocation.

"Senior ETH Professor Publishes Article Testing Jiang Lin's PFR Manuscript: Insufficient Visibility of v0.95 High-Dimensional Boundary Paths"

This article was like a small fuse, quickly igniting the already tense domestic public opinion field.

At 9:00 PM, several tech self-media outlets followed up and reposted it.

An entry appeared at the bottom of the trending search list.

#JiangLinPFRManuscriptTestedByBoundary#

The readership surged to tens of millions in a short period, and the comment section quickly split into two camps.

Some netizens, swept up by emotion, immediately pointed their fingers at Voss far away in Switzerland.

"Isn't this just nitpicking?"

"Jiang Lin is only eighteen and just won the ICCM Gold Medal, and already some people can't sit still?"

"Are foreigners starting that academic hegemony routine again?"

...

Another group of people had long found the young and famous Jiang Lin unpleasing to the eye, and immediately mocked him in reverse.

"I said it early on: mathematics is not a web novel, how could it be so easy to cross domains and instantly kill."

"A preprint is a preprint; don't hype it to the skies before passing peer review."

"How many days has it been? Professionals have already pointed out problems."

...

Amidst the full screen of mutual abuse, a third voice quickly emerged.

That was graduate students, postdoctoral fellows genuinely lingering at the bottom of the academic circle, and a small minority of rational researchers.

They did not join a side.

Instead, they went straight to Google Scholar to browse Heinrich Voss's resume.

When they saw the several papers on Additive Combinatorics that Voss co-authored with Tom Sanders in 2008, 2011, and 2013, and saw his record of long-term work on entropy methods and spectral cluster structures, this group of people fell silent.

Subsequently, several long posts appeared under the mathematics topic on Zhihu.

The core meaning was surprisingly consistent.

[Don't curse Voss; he is someone who truly understands this industry. If Heinrich Voss publishes an article under his real name pointing out the insufficient visibility of the high-dimensional boundary paths in v0.95, then it is not nitpicking, but the highest-specification boundary test issued by the academic community to Jiang Lin.]

[Pay attention to the wording: Voss did not say the main theorem is wrong. What he said is that in this boundary case, the v0.95 circulated version did not absorb and display the third-layer residual spectrum clearly enough.]

[This is not fan circle infighting. Whether Jiang Lin's piece can continue to move forward depends on whether he can catch this test.]

These professional voices were not sensational enough.

There weren't many people who could listen and take it in.

At the same time.

August 12, 11:00 PM.

Jiang Lin had been sitting at his desk for nearly twelve hours.

During this time, he had only stood up to go to the bathroom twice.

The outside noise, public opinion reversals, and trending search entries seemed to him to happen in another universe.

At this moment, in his universe, there was only mathematics.

Scattered on the desk were more than twenty used sheets of draft paper.

The bottom few sheets were his steps for reviewing Voss's boundary test.

The dozen or so sheets in the middle were him re-sorting out the visibility breakpoints of the double involution paths under the F₂¹⁵, K = 8 configurations.

On the topmost and newest draft paper, he used a heavy stroke to outline the core definition architecture of triple involution.

The formula itself was not difficult to write.

The true difficulty, and the soul of the entire heavy route, was the specific parameterization of the third-layer spectral cluster mapping.

It required precise adjustment of the mapping kernel weights to explicitly guide the amount of information refluxing at the high-dimensional boundaries back into a controllable ledger without destroying the overall measure symmetry.

This matter, he had done in the Wasteland.

Now, he was going to write it for the Real World to see.

1:15 AM.

Jiang Lin swept the draft papers on the desk aside and opened the Lean4 editor.

He began connecting the sealed heavy route into the formalization appendix branch of Node 38.

This was not patching the main proof.

This was building a bridge for the review chain.

Branch name: formalize/layer3-heavy-witness-voss-test

The first machine check ran.

The terminal window scrolled wildly, and then stopped amidst a patch of dazzling red text.

Error at line 47: type mismatch.

Error at line 47: Type mismatch.

The mapping domain type and the pushforward measure type output from the previous layer could not be strictly aligned at the machine level.

Jiang Lin did not have the slightest emotional fluctuation.

Redefine the intermediate state type conversion lemma.

Connect the breakpoint.

The second machine check ran.

The progress bar advanced to line 312 and stopped again.

Universe polymorphism error in subset conditioning.

Universe polymorphism error in subset conditioning.

This was a very underlying logical architecture problem.

Humans say to take a subset.

But the machine asks, which universe hierarchy does this subset exist in?

Time passed minute by minute.

At 3:20 AM, Jiang Lin finished fixing the polymorphism conflict.

His eyes were bloodshot, but his gaze was frighteningly bright.

The third machine check.

The green checkmarks on the screen began to light up one by one.

The compiler smoothly crossed line 300, line 500, line 800.

The code surged through the logical gates like a tide.

4:58 AM.

The terminal issued a crisp prompt tone.

The final line of result popped up on the screen.

Goal accomplished. No errors.

Goal accomplished. No errors.

All passed.

But Jiang Lin was not in a hurry to push.

He pulled out the printout of Voss's boundary test from that pile of draft papers.

Picking up a pen, facing the Lean code structure that had just run successfully, he manually re-verified the triple involution path on paper.

First-round mapping.

Second-round mapping.

Third-round mapping.

This time, rigorous mathematical logic demonstrated its dominance.

As the third iteration progressed, the residual spectrum loss term left no invisible reflux.

Like a subterranean river originally hidden deep in the stratum, its river channel was re-excavated, clearly guiding it back within the polynomial upper bound.

6:07 AM.

The sky over Jiangcheng was already tinged with morning light.

Jiang Lin typed commands in the terminal, pushing this heavy branch to the remote code repository.

Repository name: pfr-f2-formalization-blueprint

Branch: formalize/layer3-heavy-witness-voss-test

Afterward, he opened his email and sent a third email to Voss.

...

Heinrich:

The triple route has been added as a boundary witness.

Branch: formalize/layer3-heavy-witness-voss-test

Now, the boundary test can be explicitly closed. The loss term is absorbed in the third iteration via σ_layer-3 parameterization.

Please check it before I merge it into the formalization workspace.

...

At the same time the send button was pressed, Zurich had just crossed midnight.

The lights in Voss's office were still on. Holding a cup of hot coffee in his hand, he saw Jiang Lin's email.

He clicked on the email, quickly skimmed the text, and then immediately put the coffee cup aside.

He opened the Lean environment on his computer and cloned the branch Jiang Lin had just pushed locally.

As an older-generation mathematician, he was not skilled at writing Lean code.

But he fully possessed the ability to read formalized code and restore it to underlying mathematical derivations.

One line.

Two lines.

One page.

Two pages.

When reading to the middle section of the code, seeing the complex variable σ_layer-3 and the exceptionally exquisite parameterized definition behind it, his hand holding the mouse froze mid-air, remaining motionless for a long time.

This structure—he was too familiar with it.

After a moment of contemplation, he turned his seat and opened the iron filing cabinet behind him.

From a row of folders classified by year, he pulled out an old leather-bound notebook from 2014.

Flipping open the paper pages that had yellowed due to age, his fingers stopped at a sketch tucked inside.

That was eight years ago, in the late autumn of 2014.

Tom Sanders visited ETH for the second time.

That afternoon, it was drizzling in Zurich.

He and Sanders stood in front of a blackboard in the Department of Mathematics lounge, drinking cheap black tea while discussing a difficult problem regarding spectral cluster reconstruction in Additive Combinatorics.

At that time, Sanders casually outlined a route in the bottom right corner of the blackboard.

The parameterized structure of that route was almost identical to the core logic of the Lean code Jiang Lin wrote on the screen now.

But that afternoon, when the discussion came to an end, the two of them looked closely at the blackboard for a long time.

"This is too fancy, Heinrich."

Sanders said at the time, shaking his head.

"It might solve the local visibility problem, but it will make the entire proof framework incomparably cumbersome."

"Indeed."

Voss at the time agreed as well.

"This approach of building a whole set of heavy parameterizations just to illuminate one boundary layer seems too extravagant."

Thus, Voss picked up the eraser and wiped away the formulas on that blackboard with his own hands.

They discarded that route into the corner of memory as untimely redundancy.

And today, eight years later.

A Chinese teenager far away in Asia, only eighteen years old, walking alone in an unprecedented dark tunnel.

When external reviewers asked him to illuminate the high-dimensional boundaries, he took back out from his old branch this framework that his predecessors had erased.

Not only did he take it out.

He also completed and compacted it, and ran it through using modern formalization language.

Sitting in the empty office, Voss's mood was extremely complicated.

In fact, he did not feel the shame and annoyance of a senior being counterattacked by a junior, let alone the loss of having his life's work surpassed.

He only felt a sense of emotion at the incredible continuity of mathematics itself, along with a distinct awe of a sense of destiny.

Indeed, who would have thought that a certain once-broken old problem could cross time and space and be caught by another person.

Voss opened the email reply interface.

This time, his writing was no longer in that restrained official style; between the lines, it was full of the candor of a mathematics apprentice facing the truth.

...

Lin:

This branch is correct.

The triple route explicitly and completely closes this boundary test.

A personal addition: The parameterization of σ_layer-3 is very close to something Sanders and I outlined on a blackboard in Zurich in 2014 and subsequently abandoned. At the time, we thought it was too cumbersome.

You have found the same object again.

More precisely, it is your proof that has made it necessary.

I am sixty-one years old this year, and I have spent thirty years in this direction. Today I understand once again that what we discard is sometimes just waiting for a better theorem.

When you are ready, merge it.

Heinrich

...

When Jiang Lin woke up, washed his face, and sat back down in front of the computer to finish reading this email, it was already 11:00 AM on August 13.

Looking at Voss's reply on the screen, he opened GitHub.

In the private repository pfr-f2-formalization-blueprint, he merged the massive branch formalize/layer3-heavy-witness-voss-test into the formalization workspace.

It was not merged into the text of the seventh edition manuscript.

v7-final remained the fixed baseline.

This heavy route would enter the appendix and the Lean dependency graph as a formalization boundary witness for Node 38.

Over the past five days, formalization nodes 42, 43, and 44 had successively passed review.

Now, with this branch merged, the background continuous integration server began to spin frantically.

Eleven minutes later.

All test cases finished running.

The interface lit up in a patch of green.

The formalization progress changed from 44/47 to 45/47.

In the merge confirmation dialog box, Jiang Lin solemnly typed a passage into the commit message field.

...

Added the third-layer heavy witness route.

The triple involution makes the high-dimensional boundary test proposed by Voss completely visible.

Acknowledgments: Heinrich Voss proposed the boundary test; the Sanders-Voss 2014 Zurich blackboard sketch provided the initial parameterization seed for σ_layer-3.

...

After completing the operation, he switched out of the browser and opened the Lean Zulip discussion area, which was only open to formalization reviewers, a few peers, and maintainers.

In the discussion zone, he posted a short update.

...

The boundary test proposed by Voss has been explicitly closed via the newly added third-layer heavy witness route.

v0.96 review package progress: 44/47 → 45/47.

This is not a modification to the manuscript baseline, but a newly added formalization/appendix route for boundary visibility.

Thanks to Heinrich Voss for pointing out this high-dimensional boundary test, and thanks to the Sanders-Voss 2014 Zurich blackboard sketch for providing the initial seed for σ_layer-3.

I once omitted this route from v0.95 because it was too heavy.

Now it seems that for a proof entering the formalization review stage, this trade-off is no longer appropriate.

...

After this message was sent out, the channel fell silent for about a dozen seconds.

Afterward, replies began to appear one by one.

Those who truly understood saw what happened in the exact same instant.

The most-liked comment came from Terence Tao.

...

This is a great example of how rigorous review should operate.

A boundary problem is accurately isolated.

A heavier route is explicitly presented.

The corresponding contribution is also preserved under the correct names.

This is the way mathematics should work.

...

This was not exaggerated praise, but it carried more weight than exaggeration.

August 13, 1:00 PM.

The online academic storm began to cool down.

The domestic public opinion field, however, had just entered the second stage.

Voss published a very short statement on his ETH personal homepage and simultaneously updated the arXiv note to v2, adding an Author's Note on the first page.

[Jiang Lin has given the heavy boundary witness route of triple involution, which completely responds to the high-dimensional boundary test proposed in this paper.]

[This route has a clear structural connection with a parameterized draft that Tom Sanders and I erased during a Zurich blackboard discussion in 2014.]

[We thought it was too cumbersome back then. Now it seems it was simply waiting for a theorem that needed it.]

This statement was transported to the mathematics circle.

Then it was translated into Chinese, flowed back into the country, and exploded in several mathematics, theoretical computer science, and Lean formalization discussion groups.

Then it accurately pierced through the high-intelligence circles.

# Voss Responds to Jiang Lin #

# The Erased Blackboard #

# Jiang Lin Triple Involution #

Several trending terms spread rapidly in mathematics, scientific research, and technology circles.

Many ordinary netizens could not understand F₂¹⁵, nor could they understand K = 8.

They could not understand the third-layer residual spectrum loss, nor could they understand the σ_layer-3 parameterization.

But they understood a story.

A senior Swiss professor proposed the sharpest boundary test.

Jiang Lin did not issue a statement, did not engage in a war of words, and did not deploy any public relations rhetoric.

He only brought out a route that could be formally reviewed.

Then, that senior professor publicly admitted that this route was related to a blackboard he had wiped away with his own hands eight years ago.

This was more impactful than a simple plot twist.

A young Chinese mathematician wrote a paragraph in a long Weibo post, which was quickly reposted in large numbers.

[The truly worthwhile part of this matter is not whether Jiang Lin was questioned, nor whether Voss admitted his mistake.]

[What it showcases is the rarest and most precious side of the serious mathematical community: a senior points out a boundary, a young author responds with a heavier proof object, and an old concept that was once erased re-enters the records.]

[The outside world thinks this is questioning and counterattacking, but those in the know know that this is the process of the proof being taken over by the community.]

[The most terrifying thing about Jiang Lin is not that he never encounters boundary tests, but that when boundary tests appear, he can immediately pull out more verifiable reserve routes from his own arsenal.]

August 13, 9:00 PM.

In the GitHub repository, Terence Tao submitted a new pull request.

It was a rearrangement suggestion for a conditional lemma in Node 46, along with the corresponding Lean skeleton.

This was the penultimate formalization node.

Of the entire formalization blueprint stretching up to twenty thousand lines, only the last two puzzle pieces remained.

Just as Jiang Lin was about to open the PR page, the phone screen beside his desk suddenly lit up.

It was an instant message sent by Voss.

...

Lin——One more thing.

Sanders called me an hour ago. He saw the update of v0.96.

He wanted to ask you whether you would allow him to write a brief historical note for the 2014 draft and circulate it along with v1.0.

He said that if you agree, he hopes that blackboard can be officially recorded.

——H.

...

Jiang Lin looked at the screen.

Tom Sanders.

Professor in the Department of Mathematics at the University of Cambridge, UK.

Over the past twenty-odd years, he has been one of the most important practical contributors in the field of Additive Combinatorics.

The quasi-polynomial Bogolyubov-Ruzsa bounds proposed by Sanders in 2012 is foundational work that everyone in this direction cannot bypass.

During the long years in the Wasteland World, Jiang Lin had read the photocopy of Sanders's 2012 paper back and forth countless times in the Stone House.

The margins of that paper were filled with derivation annotations he made with pens of different colors.

For Jiang Lin, that was not just a paper.

That was a lighthouse left by predecessors when he was groping in the mathematical Wasteland.

Now, Sanders had seen the 2014 blackboard architecture he restored in the code base.

He wanted to write a historical note for it.

Put that once-erased blackboard back into the records.

Jiang Lin solemnly replied to this.

...

Heinrich:

Please tell Tom, yes.

Also please tell him that this is my honor.

In addition, may I ask if he would allow me to cite that 2014 blackboard in the acknowledgments and appendix notes, while writing both his and your names.

I hope that blackboard truly exists in the records.

Lin

...

After clicking send, he flipped his phone over and placed it facedown on the desktop.

Then he opened Terence Tao's PR and began code review line by line.

Exactly 1:00 AM.

With the final type alignment, Lemma_46 passed merger.

The formalization workspace progress of the main branch changed from 45/47 to 46/47.

Of the PFR formalization review blueprint watched by the entire internet, only the last checkpoint remained.

Node 47.

The ultimate wrap-up of the main theorem.

But Jiang Lin did not press his advantage to write the last piece of code.

He opened the file named v1.0_Epilogue.tex in the local manuscript folder.

At the end of the document, he added a new piece of text.

...

Epilogue, Postscript on August 13:

During the final week of preparation for this manuscript, Heinrich Voss of ETH Zurich pointed out a high-dimensional boundary visibility test in the circulating version of v0.95. This test prompted this paper to restore a heavier triple involution witness route; this route is based on a third-layer parameterization structure, which is related to an unpublished Sanders-Voss blackboard sketch in Zurich in 2014.

The author thanks Sanders and Voss. In different ways, they ensured that the boundary layer of this paper is not only correct for the author, but also visible to the reader.

Mathematics, in the final analysis, is a long dialogue.

This paper is merely a single line of record in this dialogue.

...

Early morning on August 14.

A sudden rain fell in Jiangcheng.

Jiang Lin was awakened by the sound of raindrops hitting his bedroom windowsill.

Unusually, he did not get up immediately, but instead opened his eyes, lay quietly in bed, and listened to the pattering sound of rain outside.

Until the rain stopped, he got up and pushed the window open.

The wind blew into the room, and the air was filled with the mixed scent of rain-wetted camphor tree leaves and soil.

While eating breakfast, Jiang Lin took a sip of hot soy milk, raised his head, and said casually: "Mom, I might have to go to Beijing today."

Zhang Xiufen's hand holding the chopsticks paused for a second, and there was not too much surprise on her face.

"How long are you going for?"

"At least three or four days this time, and I might need to stay a few more days after the opening."

"When are you leaving?"

"This afternoon's high-speed rail."

Zhang Xiufen did not ask much more, just picked up another fried dough stick and put it into her son's bowl.

She did not ask her son why he suddenly wanted to go to Beijing.

Nor did Jiang Lin actively explain too much.

On the surface, this trip to Beijing was for the World Robot Conference.

The G-01C No. 3 exhibition machine would be transported to Beijing by dedicated vehicle on August 16.

The Low Entropy Workshop needed to handle entry, insurance, equipment lists, lithium battery transportation instructions, on-site electricity applications, and desensitized demonstration scripts in advance.

Besides that, if Sanders's original European itinerary could be successfully adjusted, there might also be a brief face-to-face meeting.

After finishing breakfast, Jiang Lin returned to his room and began packing his luggage.

His luggage was always simple.

In a black backpack, he only packed a few things.

A performance-grade laptop.

A fast charger.

An original English edition of Additive Combinatorics with margins filled with annotations.

A mobile hard drive containing all current code and manuscript versions of PFR v1.0.

And a few pieces of clothes to change into.

He placed the zipped backpack on the living room sofa and lowered his head to glance at his wristwatch.

There were still a few hours left before the high-speed rail departed.

Jiang Lin took out his phone from his pocket, opened the Zulip private chat interface, and sent a message to Terence Tao.

...

Terry——I am leaving for Beijing this afternoon to handle the preparations for the World Robot Conference and matters before the circulation of v1.0.

The current progress of v1.0 is 46/47.

We will wrap up Node 47 before August 17.

Then, circulate v1.0.

...

Three minutes later, Terence Tao replied.

Wrap it.

(Wrap it up.)

I'm reading.

(I'm watching.)