Nontrivial vector bundles with trivial Chern classes

This paper constructs smooth affine algebras of dimension $p+2$ over an algebraically closed field of characteristic zero that admit nontrivial projective modules of rank $p$ with trivial total Chern classes, thereby demonstrating the existence of nontrivial vector bundles whose topological invariants vanish.

The Gist

Imagine you are an architect trying to build a structure. In the world of mathematics, specifically a field called Algebraic Geometry, these "structures" are shapes defined by equations, and the "materials" you build them with are called vector bundles (or projective modules).

Usually, if a structure looks "flat" or "empty" in a specific mathematical sense (meaning its Chern classes are zero), you expect it to be a simple, boring, free-standing structure (a "free module"). It's like seeing a flat, empty field and assuming there are no hidden trees or hills underneath.

The Big Question:
Can you build a structure that looks completely flat and empty from a distance (trivial Chern classes), but is actually twisted, knotted, and complex underneath (non-trivial in the K-theory group)?

The Answer:
Yes. And this paper, written by Satya Mandal, shows you exactly how to build such a "mathematical magic trick."

Here is the breakdown of the paper using simple analogies:

1. The Setup: The "Seed" and the "Garden"

The author starts with a specific recipe (a polynomial equation) involving a prime number $p$. Think of this as planting a seed.

• The Seed: A specific equation like $X^p + a = 0$.
• The Garden: The author grows a mathematical landscape (an "affine scheme") based on this seed.
• The Twist: In previous work (by N. Mohan Kumar), mathematicians found "weird" structures in this garden, but they were hard to understand and required very heavy, complex calculus (Chow groups) to prove they existed.

2. The Problem: The "Invisible" Knot

The author wants to find a structure that is stably free but not free.

• Free: A structure that is just a stack of identical, straight sheets of paper. Easy to understand.
• Stably Free: A structure that, if you add a few extra sheets of paper to it, becomes a stack of straight sheets.
• Not Free: But on its own, it's actually a tangled knot.

The problem is: How do you prove it's a knot if all your measuring tools (Chern classes) say it's flat?

3. The Solution: The "Dimensional Elevator"

The author uses a clever trick involving dimensions.

• Step 1: Build a Taller Tower.
  The author constructs a new algebra (a new ring of numbers) called $B$. This is like building a taller tower than the original garden. The original garden had a certain number of dimensions; the new tower has two more dimensions than the rank of the bundle they are studying.

  • Analogy: Imagine you have a 2D drawing of a knot. It looks flat. But if you lift that drawing into 3D space, you can see the twist. The author adds "extra dimensions" to the mathematical space to make the knot visible to their specific tools.

• Step 2: The "Splitting" Trick.
  The author uses a recent theorem (from a paper called [ABH26]) which acts like a magic splitter.

  • The theorem says: "If you have a bundle in a space that is just right (specifically, if the dimension of the space is 2 higher than the rank of the bundle), and the top 'twist' measurement is zero, then the bundle must split."
  • Analogy: Imagine a long, twisted rope. If the rope is long enough compared to its thickness, and the very end of the rope is untwisted, the theorem says you can cut off a straight piece of rope, and the rest is still a bundle.
  • The author cuts off a "free" piece (a straight sheet of paper, $B$) from their complex bundle.

• Step 3: The Result.
  What is left after cutting off the straight piece?

  • A bundle $Q$ that is smaller (rank $p$) but lives in a larger space (dimension $p+2$).
  • The Magic: This remaining bundle $Q$ has zero Chern classes (it looks perfectly flat and empty).
  • The Reality: Despite looking empty, it is not a free bundle. It is a "stably free non-free" module. It is a knot that looks like a flat sheet.

4. Why This Matters

Before this paper, finding these "invisible knots" was like finding a needle in a haystack using a very expensive, complicated metal detector (heavy calculus).

• The Old Way: You had to calculate the entire landscape to prove the knot existed.
• The New Way: The author builds a specific "trap" (the algebra $B$) where the knot must exist if you follow the recipe. They prove that even though the "twist meters" (Chern classes) all read zero, the knot is still there.

Summary in One Sentence

Satya Mandal has built a mathematical "optical illusion": a complex, knotted structure that looks perfectly flat and simple to all standard measuring tools, proving that in the world of algebra, things can be much more twisted than they appear.

The Takeaway:
Just because a mathematical object has "zero curvature" (trivial Chern classes) doesn't mean it's simple. Sometimes, you just need to look at it from a slightly higher dimension to see the hidden complexity.

Technical Summary

Here is a detailed technical summary of the paper "Nontrivial vector bundles with trivial Chern classes" by Satya Mandal.

1. Problem Statement

The paper addresses a fundamental question in algebraic K-theory and algebraic geometry: Can one construct non-trivial projective modules (vector bundles) over smooth affine algebras that possess trivial total Chern classes?

• Context: In the study of vector bundles, Chern classes are primary invariants used to distinguish non-trivial bundles from trivial ones. If all Chern classes vanish, a bundle is often suspected to be trivial (or at least stably free).
• The Gap: While examples of stably free non-free modules exist (notably by N. Mohan Kumar), they typically rely on specific rank conditions and complex calculations involving Chow groups. The author seeks to construct examples where the total Chern class is identically 1 ($C(Q) = 1$), yet the class $[Q]$ in the Grothendieck group $K_0(B)$ is non-trivial (i.e., $Q$ is not stably free).
• Goal: To construct a smooth affine algebra $B$ over an algebraically closed field $F_0$ (char 0) and a projective $B$-module $Q$ such that:
  1. $\text{rank}(Q) = \dim(B) - 2$.
  2. The total Chern class $C(Q) = 1$.
  3. $Q$ is not stably free (i.e., $[Q] \neq [B^{\text{rank}(Q)}]$ in $K_0(B)$).

2. Methodology

The construction relies on a recursive geometric setup and a "descent" technique from a larger field to a subring.

A. The Seed Polynomial and Recursive Construction

The author adapts the construction from N. Mohan Kumar [MK85].

1. Seed Polynomial: Fix a prime $p \geq 2$ and a field $F_0$. Let $F = F_0(a)$ where $a$ is a variable. The seed polynomial is chosen as $f(X) = X^p + a$.
2. Recursive Polynomials: A sequence of homogeneous polynomials $F_n(X_0, \dots, X_n)$ is defined inductively:
   • $F_1(X_0, X_1) = X_1^p f(X_0/X_1) = X_0^p + aX_1^p$.
   • $F_n(X_0, \dots, X_n) = F_{n-1}(X_0, \dots, X_{n-1})^p + a^{1+p+\dots} X_n^{p^n}$ (simplified structure).
   • These polynomials are irreducible.
3. The Scheme: Define the open affine subset $X_n = \mathbb{P}^n_F \setminus V(F_n) = \text{Spec}(A_n)$, where $A_n = F[X_0, \dots, X_n]_{(F_n)}$.

B. K-Theory and Chow Groups Analysis

The author analyzes the Grothendieck group $K_0(X_n)$ and its filtration by codimension of support.

• Non-vanishing: Using the localization sequence and properties of Chow groups (specifically $CH_n(X_n) \cong \mathbb{Z}/p\mathbb{Z}$), it is proven that the top filtration step $F^n K_0(X_n)$ is non-zero.
• Construction of Non-Trivial Classes: Elements in $F^n K_0(X_n)$ correspond to differences $[P] - [A_n^n]$ where $P$ is a projective module of rank $n$. Crucially, for these specific elements, the Chern classes $C_k(P)$ vanish for all $1 \leq k \leq n$.

C. Descent to a Subring (The "Common Denominator" Trick)

The algebra $A_n$ is defined over $F = F_0(a)$. To get an algebra over $F_0[a]$, the author performs a descent:

1. Localization: Any projective module $P$ over $A_n$ and any exact sequence representing a class in $K_0$ can be "lifted" to a subring $B \subseteq A_n$ by clearing denominators.
2. The Ring $B$: There exists an element $t \in F_0[a]$ such that $B = F_0[a]_t [X_0, \dots, X_n]_{(F_n)}$.
3. The Module $Q$: A projective $B$-module $Q$ is constructed such that $Q \otimes_B A_n \cong P$.
4. Chern Class Vanishing: By carefully choosing the denominator $t$ (potentially multiplying it further), the author ensures that the Chern classes of $Q$ vanish up to the dimension of $B$.

D. Dimension Reduction via Splitting Theorem

The initial construction yields a module of rank $n = \dim(B) - 1$. To achieve the target rank of $\dim(B) - 2$, the author applies a splitting theorem from a recent work [ABH26]:

• Theorem: For a projective module $P$ over a smooth affine algebra of dimension $d$, if $\text{rank}(P) = d-1$ and the top Chern class $C_{d-1}(P) = 0$, then $P \cong Q \oplus B$ for some projective module $Q$.
• Application: Since the constructed module $\tilde{P}$ has rank $n = \dim(B)-1$ and $C_n(\tilde{P}) = 0$, it splits as $\tilde{P} \cong Q \oplus B$. The resulting module $Q$ has rank $n-1 = \dim(B) - 2$.

3. Key Contributions and Results

1. Existence of Non-Trivial Bundles with Trivial Chern Classes:
   The paper constructs a smooth affine algebra $B$ of dimension $p+2$ (where $p$ is a prime) and a projective module $Q$ of rank $p$ such that:

   • $C(Q) = 1$ (Total Chern class is trivial).
   • $[Q] \neq [B^p]$ in $K_0(B)$ (The module is not stably free).
   • $Q$ is stably free if and only if the underlying class in $K_0(X_n)$ is zero.

2. Dimensional Shift:
   The author successfully shifts the dimension of the counter-example. While Mohan Kumar's examples were of dimension $p+1$ with rank $p$, this construction yields examples of dimension $p+2$ with rank $p$ (i.e., rank = dimension - 2).

3. Refinement of Invariants:
   The paper demonstrates that Chern classes are insufficient to detect the non-freeness of projective modules in this specific "critical rank" setting. The non-triviality is detected purely via the K-theory class $[Q]$, not via characteristic classes.

4. Technical Lemma on Descent:
   The paper provides a rigorous method for descending projective modules and their Chern class properties from a localized ring $A_n$ (over $F_0(a)$) to a subring $B$ (over $F_0[a]$), ensuring the preservation of the vanishing of Chern classes.

4. Significance

• Counter-Examples to Intuition: The results challenge the intuition that trivial Chern classes imply a bundle is stably free (or free) in the context of affine schemes, particularly when the rank is close to the dimension of the base.
• Advancement of Affine Geometry: It provides a new class of "oddities" in affine schemes, distinct from the singular schemes or stably free modules found in previous literature (like the $XY-ZW=0$ or $\sum X_i Y_i = 1$ examples).
• Methodological Bridge: The work bridges the gap between the heavy calculus of Chow groups used in [MK85] and modern splitting theorems in K-theory ([ABH26]), offering a streamlined path to constructing these non-trivial bundles.
• Implications for the Bass-Quillen Conjecture: While not directly solving the conjecture, these examples highlight the complexity of the structure of projective modules over affine algebras, specifically in the range where the rank is less than the dimension of the ring.

Conclusion

Satya Mandal successfully constructs a family of smooth affine algebras $B$ of dimension $p+2$ and projective modules $Q$ of rank $p$ that are not stably free despite having trivial total Chern classes. This is achieved by lifting non-trivial K-theory classes from a specific geometric construction (based on Mohan Kumar's work) to a subring and applying a splitting theorem to reduce the rank by one, thereby creating a robust counter-example to the idea that Chern classes fully classify projective modules in this dimension/rank regime.