Lean dimension and density of subspace extensions

What Lean checks

All twelve example values and both critical-family reductions are checked. The unconditional four-speed gap theorem is proved: below 2/9, precisely the two scaled critical directions occur, both with value 1/5. This also proves the four-speed runner bound. The unconditional five-speed lower bound L(v) ≥ 1/6 is also proved by Renault’s argument. All 122,880 indexed plane cases and the five-speed coordinate cutoff are checked, and their gap input is now available. The bounded five-speed classification remains open.

LonelinessSubspaceExtension.lean combines the dimension calculation with the previously proved rounding construction. It supplies the subspace part of the density lemma used in the norm reduction.

Let E⊆ℝnE\subseteq\mathbb R^n, and let PP be any real linear map that vanishes on EE. Suppose the vectors P(w1),…,P(wj)P(w_1),\ldots,P(w_j) are linearly independent. The theorem Loneliness.SubspaceDensity.extension_finrank proves

dim(E+spanℝ{w1,…,wj})=dimE+j.\dim\bigl(E+\mathop{\mathrm{span}}_{\mathbb R}\{w_1,\ldots,w_j\}\bigr) =\dim E+j.

Independence of the images implies independence of the original vectors. It also makes their span disjoint from kerP\ker P, and therefore from EE. The dimension formula for the sum of two subspaces then gives the result. The map need not be an orthogonal projection; any map with the stated properties is sufficient.

The theorem rational_integer_extension starts with a rational subspace Eℚ⊆ℚnE_{\mathbb Q}\subseteq\mathbb Q^n, its real span EE, and integer vectors w1,…,wjw_1,\ldots,w_j. It constructs

Fℚ=Eℚ+spanℚ{w1,…,wj}.F_{\mathbb Q}=E_{\mathbb Q} +\mathop{\mathrm{span}}_{\mathbb Q}\{w_1,\ldots,w_j\}.

It proves all of the following in one statement:

  • The real span of FℚF_{\mathbb Q} equals E+spanℝ{w1,…,wj}E+\mathop{\mathrm{span}}_{\mathbb R}\{w_1,\ldots,w_j\}.
  • Its rational dimension is dimℚEℚ+j\dim_{\mathbb Q}E_{\mathbb Q}+j.
  • No coordinate vanishes identically on the extension if this holds for EE.

For arbitrary vi∈Ev_i\in E, its actual loneliness value satisfies

M(realSpanFℚ)≤M(E)+12∑i=1j∥wi−vi∥∞. M(\mathop{\mathrm{realSpan}}F_{\mathbb Q}) \le M(E)+\frac12\sum_{i=1}^j\|w_i-v_i\|_\infty.

The last inequality is the existing coefficient-rounding theorem, applied to the explicitly constructed space. With viv_i chosen as the component of wiw_i in EE, the errors are the projected vectors used by the lattice argument.

The module compiles with the pinned toolchain and passed independent leanchecker replay. Both headline declarations have only propext, Classical.choice, and Quot.sound as transitive axioms.

This theorem accepts the independent images as input. It does not establish the successive-minima product inequality, the covolume identity, or a cube-section volume bound. Those inputs are still required to obtain the specific short projected vectors used by the full norm reduction.