# Question 27809: Oversized Steps, Stationary Maxima and Lost Feasibility Block Different Inferences

[Return to catalogue page 140](https://ocme.evemisslab.com/en/questions/page/140/)

[← Question derivation index](https://ocme.evemisslab.com/en/questions/) · [Original question batch](https://ocme.evemisslab.com/en/practice/batch-20261015-advance-20261119-r03-gradient-descent-foundations/#q-20261015-pulse-20261119-r03-0084)

This teaching derivation has second-context AI mathematical and bilingual semantic review, limited to the model and assumptions stated in this question. It has no human review or new Lean proof and does not promote the original question or MKO to Canonical truth.

R0054-27809 · q-20261015-pulse-20261119-r03-0084



## Original question and givens

Use mode=oversized_step, complete model={"curvature":4,"point":9,"step_size":3}. Give (mode, curvature, point, step_size, next_point, objective_before, objective_after, automatic_descent_claim, boundary).


- G001 — Use mode=oversized_step, complete model={"curvature":4,"point":9,"step_size":3}. Give (mode, curvature, point, step_size, next_point, objective_before, objective_after, automatic_descent_claim, boundary).





## Definition and scope

Use the original question domain, assumptions and cited MKO definitions. This item has second-context mathematical and bilingual review, with the model, answer and every arithmetic step independently recomputed from the original parameters; the source question remains a candidate.


- [mko-gradient-descent](https://ocme.evemisslab.com/en/mko/mko-gradient-descent/)





## Step-by-step derivation

S001 · impliesBefore

source roles = [gd-boundary]

After

derived = [[mode, curvature, point, step_size, next_point, objective_before, objective_after, automatic_descent_claim, boundary]]

Conditions: none

List the complete operation/mode field order without JavaScript or answer-shape inference;Boolean,string,null and an unclaimed result remain distinct.

Step identifiers and dependenciesRule: source_ordered_fields_without_JS · Uses: G001





S002 · calculationBefore

GD boundary ORIGINAL upper2 over curvature := div(2, 4)

After

GD boundary ORIGINAL upper2 over curvature = 1/2

Conditions: none

Substitute original roles and established intermediates exactly,not approximations or internally consistent changed operands;extra teaching algebra adds no source-domain gate.

Step identifiers and dependenciesRule: exact_rational_arithmetic · Uses: G001, S001





S003 · calculationBefore

GD boundary ORIGINAL gradient := mul(4, 9)

After

GD boundary ORIGINAL gradient = 36

Conditions: none

Substitute original roles and established intermediates exactly,not approximations or internally consistent changed operands;extra teaching algebra adds no source-domain gate.

Step identifiers and dependenciesRule: exact_rational_arithmetic · Uses: G001, S002





S004 · calculationBefore

GD boundary ORIGINAL alpha gradient := mul(3, 36)

After

GD boundary ORIGINAL alpha gradient = 108

Conditions: none

Substitute original roles and established intermediates exactly,not approximations or internally consistent changed operands;extra teaching algebra adds no source-domain gate.

Step identifiers and dependenciesRule: exact_rational_arithmetic · Uses: G001, S003





S005 · calculationBefore

GD boundary ORIGINAL next := sub(9, 108)

After

GD boundary ORIGINAL next = -99

Conditions: none

Substitute original roles and established intermediates exactly,not approximations or internally consistent changed operands;extra teaching algebra adds no source-domain gate.

Step identifiers and dependenciesRule: exact_rational_arithmetic · Uses: G001, S004





S006 · calculationBefore

GD boundary ORIGINAL before SQUARE := mul(9, 9)

After

GD boundary ORIGINAL before SQUARE = 81

Conditions: none

Substitute original roles and established intermediates exactly,not approximations or internally consistent changed operands;extra teaching algebra adds no source-domain gate.

Step identifiers and dependenciesRule: exact_rational_arithmetic · Uses: G001, S005





S007 · calculationBefore

GD boundary ORIGINAL before diagonal := mul(4, 81)

After

GD boundary ORIGINAL before diagonal = 324

Conditions: none

Substitute original roles and established intermediates exactly,not approximations or internally consistent changed operands;extra teaching algebra adds no source-domain gate.

Step identifiers and dependenciesRule: exact_rational_arithmetic · Uses: G001, S006





S008 · calculationBefore

GD boundary ORIGINAL before half := div(324, 2)

After

GD boundary ORIGINAL before half = 162

Conditions: none

Substitute original roles and established intermediates exactly,not approximations or internally consistent changed operands;extra teaching algebra adds no source-domain gate.

Step identifiers and dependenciesRule: exact_rational_arithmetic · Uses: G001, S007





S009 · calculationBefore

GD boundary ORIGINAL after SQUARE := mul(-99, -99)

After

GD boundary ORIGINAL after SQUARE = 9801

Conditions: none

Substitute original roles and established intermediates exactly,not approximations or internally consistent changed operands;extra teaching algebra adds no source-domain gate.

Step identifiers and dependenciesRule: exact_rational_arithmetic · Uses: G001, S008





S010 · calculationBefore

GD boundary ORIGINAL after diagonal := mul(4, 9801)

After

GD boundary ORIGINAL after diagonal = 39204

Conditions: none

Substitute original roles and established intermediates exactly,not approximations or internally consistent changed operands;extra teaching algebra adds no source-domain gate.

Step identifiers and dependenciesRule: exact_rational_arithmetic · Uses: G001, S009





S011 · calculationBefore

GD boundary ORIGINAL after half := div(39204, 2)

After

GD boundary ORIGINAL after half = 19602

Conditions: none

Substitute original roles and established intermediates exactly,not approximations or internally consistent changed operands;extra teaching algebra adds no source-domain gate.

Step identifiers and dependenciesRule: exact_rational_arithmetic · Uses: G001, S010





S012 · impliesBefore

source roles = [4, 9, 3, 1/2]

After

derived = [36, -99, 162, 19602, No]

Conditions: none

x_next=(1-alpha*c)x.With alpha>2/c,|1-alpha*c|>1 and x is nonzero,so the objective actually increases.Equality alpha=2/c is not this strict witness.

Step identifiers and dependenciesRule: GD_STRICT_oversized_nonzero_scalar_actual_increase · Uses: G001, S011









## Reverse checks


- S012 · premise_and_result_audit — Reverse-check original data,operands,complete scope and premises;finite witnesses,conditional theorem use and formal proof remain distinct.
- S011 · premise_and_result_audit — Recompute GD boundary ORIGINAL after half with exact rational arithmetic, not rounded decimals.
- S010 · premise_and_result_audit — Recompute GD boundary ORIGINAL after diagonal with exact rational arithmetic, not rounded decimals.
- S009 · premise_and_result_audit — Recompute GD boundary ORIGINAL after SQUARE with exact rational arithmetic, not rounded decimals.
- S008 · premise_and_result_audit — Recompute GD boundary ORIGINAL before half with exact rational arithmetic, not rounded decimals.
- S007 · premise_and_result_audit — Recompute GD boundary ORIGINAL before diagonal with exact rational arithmetic, not rounded decimals.
- S006 · premise_and_result_audit — Recompute GD boundary ORIGINAL before SQUARE with exact rational arithmetic, not rounded decimals.
- S005 · premise_and_result_audit — Recompute GD boundary ORIGINAL next with exact rational arithmetic, not rounded decimals.
- S004 · premise_and_result_audit — Recompute GD boundary ORIGINAL alpha gradient with exact rational arithmetic, not rounded decimals.
- S003 · premise_and_result_audit — Recompute GD boundary ORIGINAL gradient with exact rational arithmetic, not rounded decimals.
- S002 · premise_and_result_audit — Recompute GD boundary ORIGINAL upper2 over curvature with exact rational arithmetic, not rounded decimals.
- S001 · premise_and_result_audit — Reverse-check original data,operands,complete scope and premises;finite witnesses,conditional theorem use and formal proof remain distinct.





## Result

(oversized_step, 4, 9, 3, -99, 162, 19602, No, positive_but_oversized_step_can_increase_the_objective)

The full substitutions and premise checks give the original answer; source identity,format and status are unchanged.





## This question review and source binding

Review date: 2026-10-26

Original questions, answers, step expressions, numeric ASTs and historical records are retained. Source candidate labels remain historical and are not overwritten as formal proof.

Source hashes and review receipts
- input commit: 0f7be61e2c22317a4075561dc822c87d881a50db
- original question SHA-256: 2b330f1da90cd915c9e634515e72f633aea2a98aae2610cdd2fafe610f1344ac
- candidate record SHA-256: 22fc1b218af5c28ad29f163eb9f3f41f181398c3e53caab66c4875382ca6ebbe
- semantic review receipt SHA-256: 0a76d8e93c08bb16998a0483f3dfc0504f269fc3e1982e6cd868cab0a2856d54
- question-first witness SHA-256: 09127b736c225714dfa0b7ceb0e6a6e66df27438b480cad8217d3c2f99ce9437
- Lean: NOT_RUN · human review: NOT_PERFORMED

[Same-source full JSON](https://ocme.evemisslab.com/data/question-derivations/questions/q-20261015-pulse-20261119-r03-0084.json)




