ResMonoidDense — dense exponent-vector implementation of the resolution monoid. More...

#include <iostream>
#include <memory>
#include <vector>
#include "schreyer-resolution/res-monomial-types.hpp"
#include "skew.hpp"

Classes
class	ResMonoidDense
	Dense-format ResMonoid implementation: monomials laid out as fixed-width exponent vectors. More...

Detailed Description

ResMonoidDense — dense exponent-vector implementation of the resolution monoid.

Note: AI-generated documentation. Verify against the source before relying on it.

Declares the implementation res-moninfo.hpp currently picks as the production ResMonoid. Each monomial is encoded as [hash, component, weight_1, ..., weight_nweights, exp_0, ..., exp_{nvars-1}] with firstvar = 2 + nweights and a fixed nslots total, so the slot count is the same for every monomial regardless of which variables actually appear — favouring O(1) access and short-walk comparison over compactness. Hashing is the additive Steel trick: from_expvector stores hash(m) = sum_i hashfcn[i] * e_i (with hashfcn a per-variable precomputed random word), so hash(m*n) = hash(m) + hash(n) and mult updates the hash by a single add. The class-level mask is not part of the hash — it is used only by check_monomial to detect overflow by AND-ing it against each slot. Weight slots front-load the comparison with cached degree contributions, so order tests can short-circuit before walking the exponent suffix. mult(a, b, result) is commutative: skew handling lives separately in skew_vars / skew_mult_sign, which take a SkewMultiplication* by parameter — the class does not own or inherit one.

compare_schreyer(m, n, m0, n0, tie1, tie2) is the live order test; the in-file compare_grevlex is currently #if 0-d out, so only the sparse twin offers a live grevlex comparator. Otherwise the two implementations share the same mult / divide / monomial_size / to_expvector / from_expvector / from_varpower_monomial surface — which is what lets res-moninfo.hpp swap implementations with a single typedef edit. The dense layout wins for rings with few variables or dense exponents (cache-friendly inner loops, predictable layout); the sparse twin res-moninfo-sparse.hpp wins on the opposite end. The mutable unsigned long ncalls_* counters at the bottom of the class are the engine's home-grown profiler hooks.

See also: res-moninfo.hpp; res-moninfo-sparse.hpp; res-monomial-types.hpp; res-poly-ring.hpp; skew.hpp

Definition in file res-moninfo-dense.hpp.

Classes

Detailed Description