S06 The Working Core - The Symmetric Core at Work: the Floor below Addition, Logic and Learning in Its Charts, the Limits of Hardness, and the Navier–Stokes Equations in Speed and Direction

The Symmetric Core is an infinite ladder of commutative, associative operations obtained by transporting addition through successive iterated-logarithmic charts. Room k has the law

x odot_k y

= E_k(ell_k(x)+ell_k(y)),

where ell_k=ln^[k] and E_k=exp^[k]. Every room is a copy of (R,+), adjacent rooms form an ordered field, and the exponential transports each field to the next.

S06 "The Working Core" studies what this architecture does in practice: below addition, in many-valued logic and machine learning, in computational complexity, in fluid mechanics, and in the chart dynamics at the frontier of the shelf.

Below addition

The first room below addition is log-sum-exp:

LSE(x,y) = ln(exp(x)+exp(y)).

More generally,

x odot_(-n) y

= ell_n(E_n(x)+E_n(y)).

Room -n is ordinary addition on the half-line

D_n = (e_(n-1),infinity)

transported back through E_n. Its lower endpoint is the absorber of room n. Only room -1 has an identity, namely -infinity; deeper rooms have finite unattained floors

f_n = ell_n(2*e_(n-1)).

These floors descend asymptotically like the negative Bell tower.

Every descending room lies above max(x,y) and converges to it as the room index tends downward. The excess has a sharp bound controlled by the iterated-exponential height of the larger operand rather than merely by the gap between the operands.

There are two distinct roads from addition to max. Maslov dequantisation gives

T*LSE(a/T,b/T) -> max(a,b).

For every deeper room, the same zero-temperature scaling gives

max(a,b,0),

whereas descending through the rooms at fixed temperature gives max(a,b). The difference between the two iterated limits is exactly a rectifier:

max(a,b,0)-max(a,b)

= ReLU(-max(a,b)).

Thus the rectifier records the failure of temperature scaling and room descent to commute. The HyperCore meets the same boundary through MaxZeration, whose storage mean is max. Together with min, these are the only symmetric means invariant under every increasing change of chart.

Logic on the rooms

Every room has a logic interval between its absorber and identity:

J_n = [e_(n-1),e_n].

The chart ell_(n-1) maps this interval onto [0,1] and turns room n into ordinary multiplication. Consequently, every room restricted to its logic interval is product logic.

The adjacent-room field carries all three fundamental continuous conjunctions:

• product conjunction is room n;
• Lukasiewicz conjunction is room n-1, translated and truncated at the absorber;
• Godel conjunction is min.

Negation and implication are formed from the corresponding room regressors. The Goguen implication is the regressor of room n; the Lukasiewicz implication uses the regressor of room n-1; the Godel implication is the order-theoretic residual of min.

The logic intervals tile the extended line at the Bell tower

..., -infinity, 0, 1, e, e^e, ....

Gluing the room conjunctions with min produces one continuous t-norm across the shelf. It is an ordinal sum of product components, with the regressor serving as implication. This places the room construction inside the classical Mostert-Shields-Ling classification and H<ajek's Basic Logic.

Charts, activations and semiring computation

Many familiar activation and aggregation functions are chart presentations of room operations.

Temperature is a chart parameter of room -1. The soft maximum is log-sum-exp at finite temperature and approaches max as temperature tends to zero. The sigmoid carries room -1 onto Hamacher's sum, while tanh carries addition onto Einstein's velocity addition.

The curvature-merge family

Merge_K(a,b)

= (a+b)/(1-K*a*b)

is the finite projective completion of two adjacent room fields. On the unit interval it reproduces the Hamacher family, with circular, flat and hyperbolic members arising from the sign of K.

Room -2 supplies a threshold-free rectifier,

r(x) = ln(ln(1+exp(exp(x)))).

It is strictly convex, has the floor ln(ln(2)), and approaches the identity doubly exponentially fast.

Dynamic-programming algorithms also admit a room reading. Viterbi decoding, forward recursion and sum-product decoding are one recurrence evaluated in different rooms or semirings. Chart choice changes the aggregation law while preserving the underlying state-transition architecture.

What rooms can make hard

Every room is an abelian group, so an equation inside one room is solved by one regressor:

x ominus_k y

= E_k(ell_k(x)-ell_k(y)).

Search and inversion inside a declared room are therefore explicit. The rapid numerical growth of higher operations does not itself imply hard search.

Lifting a computation to a higher room has an exact representation cost. At room two, one gate can create output with quadratic bit length relative to its input; every fixed higher room remains subexponential in input length. These are growth laws, not general lower bounds on inversion.

Associative power,

Apow(a,b)=exp(ln(a)*ln(b)),

is functionally transcendental over the field operations: no finite algebraic circuit computes it on an open complex set. Once logarithm and exponential are admitted as gates, however, Apow is elementary. The result separates gate vocabularies rather than complexity classes.

Starting from 2 with addition and multiplication, the largest integer obtainable in s straight-line steps is

2^(2^s).

Repeated squaring reaches the bound. Towers of twos are therefore extremely cheap relative to their magnitude.

The Symmetric-Core machine has different power under two cost conventions:

P_SC^(bit) = P,

P_SC^(unit) = P^PosSLP.

At bit cost, the operational separation problem is the ordinary Boolean P versus NP problem. At unit cost, equivalence with P depends on the unresolved status of PosSLP. Branch censuses provide lower bounds only inside declared computation-tree models, where results of Michael Ben-Or and Dima Grigoriev with Nicolai Vorobjov apply. Relativisation, natural proofs and algebrisation remain requirements that any Boolean separation route must overcome.

Fluids in magnitude and direction

For a nonzero vector field, write

u = rho*u_hat,

where rho=|u| and u_hat is the direction on the unit sphere. The room chart separates magnitude from direction.

For both velocity and vorticity, the direction satisfies a weighted harmonic-map heat flow. Diffusion of direction drains the magnitude through the direction energy

|grad(u_hat)|^2.

Stretching, pressure and transport then appear as explicitly identified source terms.

In two dimensions, writing velocity as

u = exp(ln|u|+i*phi)

combines speed and direction into one complex logarithmic variable. Where vorticity vanishes, the resulting equation is the classical log-hodograph formulation associated with Gustav Kirchhoff and Tullio Levi-Civita, and reduces to a complex Cole-Hopf-type equation.

The velocity-observer hierarchy contains a critical L^3 level. In the corresponding balance law, pressure is the only source. The fractional-height ladder of related observers lies below the critical band [1,3/2] and is bounded a priori.

This chart decomposition locates the regularity problem but does not solve it. A complete proof must control the pressure source, direction energy, critical scaling and continuation criteria simultaneously. The volume formulates these requirements and isolates routes that meet the structural checklist without claiming a resolution of the Navier-Stokes problem.

Charts that straighten and irrational flux

The powers of a room are straightened by three classical linearisations at successive dynamical regimes:

• Bottcher coordinates at a superattracting end;
• Koenigs coordinates at a hyperbolic fixed point;
• Abel coordinates for translation along iteration height.

For room powers, these linearisations occupy consecutive levels of the shelf. The room's own chart is the Koenigs coordinate of its native power map, while regular superlogarithms supply Abel charts for exponential iteration.

The branch-flux dial at an irrational angle produces a symbolic sign sequence. The derivative-sign word is a Rote word rather than a Sturmian word. Its factor frequencies are obtained by halving the gaps of the doubled rotation, so at most three frequencies occur by the three-gap theorem.

This provides an exact symbolic description of irrational flux without identifying flux with operational rank, room index or dynamical curvature.

S06 shows the Symmetric Core functioning across several mathematical domains. Below addition, its rooms descend toward max, with the rectifier measuring the difference between temperature scaling and room descent. On their logic intervals, the rooms form product logics and glue into one continuous t-norm. Their charts produce activation functions, fuzzy-operation families and semiring algorithms. Their group structure explains both exact inversion and the limits of operational complexity claims. In fluid mechanics, the same charts separate magnitude from direction and expose harmonic-map direction flows and critical observers. The volume's governing principle is that charts can reveal the native form of a problem and locate its obstruction, but cannot remove the obstruction merely by changing coordinates.