A sufficiency certificate from every fibre conditional law

On this page
  1. Goal and scope
  2. Givens and symbols
  3. Rules and sources
  4. Main derivation
  5. Substeps and lemmas
  6. Reverse audits
  7. Error controls and negative controls
  8. Result and limitations
  9. Verification and source binding

← Guide index · Original MKO

derivation-mko-finite-sufficient-statistic-01 · ocme-new142-reader-adoption-20261021

Goal and scope

Audit invariance across theta on every positive-mass fibre, retaining undefined null conditions.

No factorization theorem or estimator-quality inference.

Givens and symbols

  • work
  • G1 — Theta={0,1}, Omega={a,b,c,d,z}; P0=(1/8,3/8,1/4,1/4,0), P1=(1/16,3/16,3/8,3/8,0). S(a)=S(b)=0, S(c)=S(d)=1, S(z)=2.
  • S — total statistic
  • t — fibre label

Rules and sources

  • mko_definition · cited — A statistic is a total map on the full sample carrier and its fibers form a partition. For every (theta,t) first compute P_theta(T=t). Define and calculate the exact fiber conditional law only on a positive-mass cell, then compare all valid theta laws at each t. Leave zero-mass conditions undefined, encoded as null in these outputs. Separate explicit nonsufficiency witnesses, the identity statistic and the constant-statistic boundary. No factorization theorem or estimator-quality conclusion is introduced. · mko-finite-sufficient-statistic
  • work · prerequisite — Finite sufficiency: on each fibre compare full conditional laws only where conditioning has positive mass, requiring theta invariance; null quotients remain undefined.

Main derivation

S01 · implies

Before

P_0,P_1

After

1/8+3/8+1/4+1/4=1,\quad1/16+3/16+3/8+3/8=1

Conditions: none

Rule: work · Uses: G1, work

Leaf basis: work

Both complete laws are normalized and nonnegative.

S02 · implies

Before

S^{-1}(0),S^{-1}(1),S^{-1}(2)

After

\{a,b\},\{c,d\},\{z\}

Conditions: none

Rule: work · Uses: G1, S01, work

Leaf basis: work

S is total and its three nonempty fibres form a complete partition.

S03 · implies

Before

P_\theta(S=t),\quad t=(0,1,2)

After

P_0^S=(1/2,1/2,0),\quad P_1^S=(1/4,3/4,0)

Conditions: none

Rule: work · Uses: G1, S02, work

Leaf basis: work

Compute every fibre mass before deciding where conditioning is defined.

S04 · implies

Before

t=0:\quad P_0(a\mid S=0)=\frac{1/8}{1/2},\quad P_1(a\mid S=0)=\frac{1/16}{1/4}

After

P_0(\cdot\mid S=0)=P_1(\cdot\mid S=0)=(1/4,3/4,0,0,0)

Conditions: none

Rule: work · Uses: G1, S03, work

Leaf basis: work

The conditional masses of a,b agree across theta; the other three outcomes have mass zero.

S05 · implies

Before

t=1:\quad\frac{1/4}{1/2}=\frac{3/8}{3/4}=1/2

After

P_0(\cdot\mid S=1)=P_1(\cdot\mid S=1)=(0,0,1/2,1/2,0)

Conditions: none

Rule: work · Uses: G1, S04, work

Leaf basis: work

The second positive-mass fibre also has a fully invariant conditional law.

S06 · implies

Before

P_0(S=2)=P_1(S=2)=0

After

P_\theta(\cdot\mid S=2)\text{ undefined for both }\theta

Conditions: none

Rule: work · Uses: G1, S05, work

Leaf basis: work

Do not invent a0/0 distribution; this fibre has no valid theta conditions to compare.

S07 · implies

Before

\text{all positive fibre conditionals invariant across valid }\theta

After

S\text{ is sufficient for this finite model}

Conditions: none

Rule: work · Uses: G1, S06, work

Leaf basis: work

Both parameters and all three fibres are exhausted, giving sufficiency by the finite conditional-law definition.

S08 · implies

Before

R(x)\equiv0

After

P_0(a\mid R=0)=1/8\ne1/16=P_1(a\mid R=0)

Conditions: none

Rule: work · Uses: G1, S07, work

Leaf basis: work

The constant statistic R has the single fibre Omega, with probability 1 under each theta. The conditional distribution of the sample outcome given R=0 is P_theta itself and varies with theta, so R is not sufficient for this finite model.

Substeps and lemmas

No substeps in this guide.

Reverse audits

  • S08 · premise_audit · G1, S07, work — The identity statistic instead has point-mass conditionals on valid fibres and is sufficient; the boundaries differ.
  • S07 · premise_audit · G1, S06, work — No factorization theorem replaced the table audit.
  • S06 · premise_audit · G1, S05, work — Undefined is not a zero-vector probability law.
  • S05 · premise_audit · G1, S04, work — Both theta denominators on this cell are positive.
  • S04 · premise_audit · G1, S03, work — List the full five-point conditional law, not only a.
  • S03 · premise_audit · G1, S02, work — Sufficiency does not require the marginal law of S to be invariant.
  • S02 · premise_audit · G1, S01, work — Do not omit null fibre{z}.
  • S01 · premise_audit · G1, work — z has zero mass in both laws but remains in the carrier.

Error controls and negative controls

Mass first

Compute each(theta,t) fibre mass before division.

Complete conditionals

List all five sample probabilities on each valid cell.

  • Sufficiency requires an invariant law of S. → reject — S is sufficient here although P0(S=0)=1/2 and P1(S=0)=1/4.
  • Use(0,0,0,0,0) as the law at S=2. → reject — It sums to 0 and conditioning has zero denominator, so it is undefined.
  • Constant R is sufficient here. → reject — Its conditional probability of a varies as1/8 or1/16.

Result and limitations

S is sufficient and constant R is not; null-fibre conditional laws remain undefined.

    Verification and source binding

    3 declared checks for this guide; only the listed expressions are covered.

    • source commit: b1b2f1dcf8c50d67a93fe52d95f8eef7b3abba9f
    • adopted source SHA-256: 7b86518f187ccbc75abb75a795f774026b2bfa13c74cea0f969179bcaa062304
    • original source SHA-256: 99f4f978f27d4111a01876f24ceb88ab0e95d184de6346874988ea73556fac8f
    • Lean: NOT_RUN · independent AI: SCOPED_LEAD_AND_BILINGUAL_CONTEXT_REVIEW

    Same-source full JSON (formulas, ASTs, controls and revision log)