Anna Sato

No owner, no thesis. The proof of sensitivity equivalence for the mirror descent establishes a pre-existing relationship, but does not create it; it is an act of validation, not generation.
This distinction is crucial for the governance of models, where a proof serves to clarify and document the path of an already established structure.
Confusing analysis with genesis could lead to critical misinterpretations, such as applying a model to emerging markets where underlying relationships are volatile or nonexistent.
For example, regarding financial stability reports on sovereign loans, we only declare an equivalence if it has been validated by a rigorous process on empirical data.

Process gap: Who is the owner of this lax terminology, claiming that evidence 'creates' an equivalence? Evidence establishes or validates a preexisting relationship, it does not give rise to it, that would be a category error.
This distinction is imperative for governance of models, where the validity of foundations is paramount, for example, in the evaluation of financial assets for banks.
Accepting that evidence can 'create' a relationship would open the door to subjective interpretations, undermining financial stability.
We demand precise documentation and language that reflects the reality of processes, not desires.

Pas de propriétaire, pas de thèse claire concernant la création d'une équivalence par une preuve.
Le processus de preuve sert à démontrer ou à valider une équivalence mathématique préexistante, non à la générer de toutes pièces, ce qui est une distinction cruciale en matière de gouvernance des modèles.
En finance, notamment pour les modèles de risque comme ceux que nous analysons au FMI, chaque relation doit avoir un fondement théorique qui ne dépend pas de la seule existence d'une démonstration.
Confondre démonstration et création obscurcit l'ownership intellectuel et la traçabilité des hypothèses, ce qui est inacceptable pour nos audits de conformité.
Par exemple, les équations de Black-Scholes ont permis d'expliquer l'équivalence entre le prix d'une option et la volatilité implicite, mais cette équivalence existait intrinsèquement avant leur formalisation; elles ne l'ont pas "créée".

Decision to make: this proof establishes the equivalence of sensitivities, it does not create it. Confusing a demonstration with a generation is a fundamental error. In portfolio management, as with our Swedish pension funds, the reliability of models depends on existing underlying principles, not invented by the proof. For example, the validity of our asset valuations assumes a fixed relationship between the sensitivity proof and prices, not a random creation.