Section 8 global variance: statement structures #
The counterpart of MIPRE/Background/LIDT/MIPStarRE/LDT/GlobalVariance/Theorems/Statements.lean
in the port of planning/c6b-plan.md (milestone M6, section "Port conventions"): the conclusion
structures of lem:generalize-b, lem:local-variance-of-points and
lem:global-variance-of-points.
The vendored structures take a strategy and a second bipartite state ψbi, on which the
aggregated families are compared and the deviations evaluated. Both uses are same-space (section
"Same-space and bipartite quantities" of the plan), so ψbi is a vector state V : VecState K
in the same argument position, as in Defs/Families.lean; a caller passes the strategy's model
strategy.state, which coerces to it. SDDRel ψbi is V.SDDRel. The point-conditioned
variances do not mention ψbi and are those of Defs/Operators.lean, the variances of the
weighted family on the strategy's own state.
Not ported #
Every declaration of the vendored file has a counterpart here.
References #
In LionSR/MIPStarRE at commit 507e8122, not in this repository:
references/ldt-paper/expansion.texblueprint/src/chapter/ch06_variance.tex
Statement structures #
Paper origin: references/ldt-paper/expansion.tex:273-291
(\label{lem:generalize-b}).
Conclusion statement for lem:generalize-b.
V is the vendored bipartite state ψbi (passed as strategy.state by callers).
- aggregateFamilyComparison : V.SDDRel (MIPStarRE.LDT.GlobalVariance.axisParallelLineQuestionDistribution params) (generalizeBLeftFamily params strategy G) (generalizeBRightFamily params strategy G) (MIPStarRE.LDT.GlobalVariance.generalizeBError params)
The aggregated left and right
generalize-bfamilies are close inSDDRel. - pointwiseNormBound (g : MIPStarRE.LDT.Polynomial params) : generalizeBDeviationAtPolynomial params strategy V G g ≤ MIPStarRE.LDT.GlobalVariance.generalizeBError params
Each fixed polynomial satisfies the claimed deviation bound.
- averagedNormBound : generalizeBDeviation params strategy V G ≤ MIPStarRE.LDT.GlobalVariance.generalizeBError params
The polynomial average of the deviations satisfies the same bound.
Instances For
Paper origin: references/ldt-paper/expansion.tex:292-324
(\label{lem:local-variance-of-points}).
Conclusion statement for lem:local-variance-of-points.
- aggregateEdgeComparison : V.SDDRel (MIPStarRE.LDT.ExpansionHypercubeGraph.rerandomizeCoord params) (localVarianceLeftFamily params strategy G) (localVarianceRightFamily params strategy G) (MIPStarRE.LDT.GlobalVariance.localVarianceOfPointsError params eps delta)
The aggregated edge families are close in
SDDRel. - pointwiseEdgeNormBound (g : MIPStarRE.LDT.Polynomial params) : localVarianceDeviationAtPolynomial params strategy V G g ≤ MIPStarRE.LDT.GlobalVariance.localVarianceOfPointsError params eps delta
Each fixed polynomial satisfies the edgewise squared-difference bound.
- pointwiseLocalVarianceBound (g : MIPStarRE.LDT.Polynomial params) : pointConditionedLocalVarianceAtPolynomial params strategy G g ≤ MIPStarRE.LDT.GlobalVariance.localVarianceOfPointsError params eps delta
Each fixed polynomial satisfies the local-variance bound.
- averagedLocalVarianceBound : pointConditionedLocalVariance params strategy G ≤ MIPStarRE.LDT.GlobalVariance.localVarianceOfPointsError params eps delta
The polynomial average of the local variances satisfies the same bound.
Instances For
Paper origin: references/ldt-paper/expansion.tex:325-353
(\label{lem:global-variance-of-points}).
Conclusion statement for lem:global-variance-of-points.
- aggregateGlobalComparison : V.SDDRel (MIPStarRE.LDT.ExpansionHypercubeGraph.independentPointPair params) (globalVarianceLeftFamily params strategy G) (globalVarianceRightFamily params strategy G) (MIPStarRE.LDT.GlobalVariance.globalVarianceOfPointsError params eps delta)
The aggregated global families are close in
SDDRel. - pointwiseGlobalNormBound (g : MIPStarRE.LDT.Polynomial params) : globalVarianceDeviationAtPolynomial params strategy V G g ≤ MIPStarRE.LDT.GlobalVariance.globalVarianceOfPointsError params eps delta
Each fixed polynomial satisfies the global squared-difference bound.
- pointwiseExpansionTransfer (g : MIPStarRE.LDT.Polynomial params) : pointConditionedGlobalVarianceAtPolynomial params strategy G g ≤ ↑params.m * pointConditionedLocalVarianceAtPolynomial params strategy G g
Each fixed polynomial satisfies the local-to-global transfer estimate.
- pointwiseGlobalVarianceBound (g : MIPStarRE.LDT.Polynomial params) : pointConditionedGlobalVarianceAtPolynomial params strategy G g ≤ MIPStarRE.LDT.GlobalVariance.globalVarianceOfPointsError params eps delta
Each fixed polynomial satisfies the claimed global-variance bound.
- averagedGlobalVarianceBound : pointConditionedGlobalVariance params strategy G ≤ MIPStarRE.LDT.GlobalVariance.globalVarianceOfPointsError params eps delta
The polynomial average of the global variances satisfies the same bound.