AdvancedResearch

Avatar Extensions

Avatar Extensions is an abstraction generalization technique that exploits symmetries inside “simpler” theories.

This page is for organizing all important ideas about Avatar Extension in one place.

Reading Sequence of papers

Introduction: The Problem of Over-Constrained Theories

In mathematics, a common problem is that some language is too simple to fit some task. For example, in Group Theory, one if a constructed group does not fit a problem, it kind of falls together like a house of cards, or merely represents a “shadow” of the whole structure.

Avatar Extensions is about extending “simpler” theories abstractly using avatars. This is done by taking some over-constrained language and introducing avatars into it, from within, like manipulating the source code. The idea is to search for these “simpler” theories first and later extend them to fit a particular use case, instead of looking for the most generic theory that fits all use cases. It is sometimes more pragmatic to fix over-constrained theories than to add axioms to under-constrained theories.

Technical comment:
For example, First Order Logic has unconstrained predicates, which requires adding axioms to even talk about uniqueness. In practice, one needs automatic theorem provers to report an error if two symbols are used ambiguously, which technically is not part of logic since good pure logic design should treat two propositions a, b as equal if they are tautologically equal (a == b)^true. In some theories this is merely an assumption, which means that ambiguity of uniqueness properties can not be implemented as pure logical design. This means, while logical languages can be applied to almost every problem, in practice they do not have sufficient semantics to be practical and efficient. This is why Avatar Extensions is a technique to take a “simpler” language, with missing desirable logical properties, and gradually introduce new properties into it until the design problem is solved.

Avatar Extensions deals with this form of abstract generalization in general, by applying specific theories that might be use as guidance to solve problems, such as using 1-avatars to “cover” a language design that includes higher n-avatars.

Avatar Logic

For an experimental implementation, see Avalog.

Avatar Logic replaces predicates with binary relations with some additional axioms (written as an Avatar Table):

p(a, b)         b : p           p(a) = b
p(a, q'(b))     q'(b) : p       p(a) = {q'(_)} ∈ q'(b)

Translated to Zermelo-Fraenkel Set Theory (see paper):

(a, b) ∧ b : p ∧ uniq(b)
---------------------------------------------------------
(p, (a, b)) ∧ ∃! z { (p, (a, z)) } ∧ ∃! r { (r, (a, b)) }

(a, q'(b)) ∧ q'(b) : p
--------------------------------------------------------------------------------
(p, (a, (q, b))) ∧ ∀ x { ∃! z { (p, (a, (z, x))) } } ∧ ∃! r { (r, (a, (q, b))) }

This means that (a, b) contains all information needed to represent a labeled edge.

b : p             `b` has the role `p`
q'(b)             the `q` 1-avatar of `b`
p(a) = q'(b)      the `p` property of `a` is `q'(b)`
p(a) => q'(b)     the `p` property of `a` has `q'(b)`
.q'(b) = b        the inner operator is a `.` in front of an expression
uniq q            make the `q` 1-avatar behave uniquely

Avatar Algebra

Avatar Algebra relates theorems in algebra to theorems about Avatar Extensions.

*     introduction operator
1     the natural 1-avatar
0     the contracting 1-avatar
-     negation is an involution that covers symmetries in products
+     symmetry operator

Theorems:

a * 1 = 1 * a = a                       introduction of new 1-avatar
a * a => error                          no avatar can introduce itself
-(a * b) = (-a) * b = a * (-b)          coverage of products using 1->1 avatar involutions
(a + b) * f(x) = a * f(x) + b * f(x)    superposition of mathematical objects related to 1-avatar symmetry
(a * b) * c = a * (b * c)               associativity is a 6-avatar of a 9-vertex filled avatar graph
a + b = b + a                           symmetry between 1-avatars

Avatar Graphs

Avatar Graphs might be thought of as discrete topological qualities of manifolds.

For visualization, see Avatar-Graph.

The “core self” is an object which has no internal relations for introspection. Avatars are created to model relations.

Information must be passed down from higher avatars in the direction of the core. An Avatar Graph “forgets” which node is the core.

square 5-avatar wagner wagner-mobius counter example to 2^n conjecture associativity

A filled Avatar Graph consists only of core candidates.

A reflection witnessed by a product (x, x) is impossible to model in an Avatar Graph. This means that a “copy” of x must be constructed, which eventually requires copies of 1-avatars. Translated into a theorem about Avatar Extensions, this means 1->1 avatars “covers” products.

Avatar Semantics

The core axiom of Path Semantics can be naively modeled in Avatar Logic as following:

(C, D) :- (A, B), (B, A), (A, C), (B, D).

This propagates products from every identity morphism using outgoing arrows. Adding this axiom to Category Theory collapses a category into a partial groupoid. It is kind of like extending a square matrix. Can also be thought of as a compact representation of an equivalence class.

The analogue for isomorphisms (without identity morphisms) constructs the following way:

step1 step2 step3

Step 1 is the smallest possible example of a Möbius topology.

In Path Semantical Logic, propositions are separated by levels, which gets rid of Möbius topology. Applying the core axiom to a single level would contract all propositions along implications, because of reflexivity x = x is a tautology.

In the language of graphs, it is possible to avoid contraction in a single level since identity morphisms are not necessary.

The diagonals corresponds to highest N-avatars in hypercubes, which can be interpreted as the smallest reduction of an inconsistent theory to a consistent theory.

For example, Girard’s paradox is avoided in type systems by using cumulative universes or types. The highest N-avatars in this context are morphisms between type levels, which are just type memberships. Therefore, Type Theory is essentially a highest N-avatar reduction to make modeling of mathematics consistent.

The diagonals on a cube can be thought of as swap operations, which generates all permutations by selecting a face + rotation in the plane:

swap-cube

This interpretation is the only one that preserves structure.

Group Theory might be thought of as operations on highest N-avatars satisfying the notion of “resources” in linear logic.

Avatar Witness Theory

witness-duality

Witness Duality can be thought of as two different models of equivalence.

Normally, = is used for equality and ~= for existence of isomorphisms, but for Witness Duality this depends on the interpretation. However, the Product Witness can not distinguish between multiple isomorphisms, while the Loop Witness can not express the global constraint of an equivalence relation.

The analogue of the Univalence Axiom says that Witness Duality can only be witnessed with a Loop Witness:

(A = B) ~= (A ~= B)

Symmetry Forcing, a property where proofs of reflexivity requires an isomorphic reconstruction, is related to the Product Witness.

There are two consistent type theories that interprets Path Semantical Logic:

These two theories are cocategory of each other, where morphisms are type memberships.

The first theory satisfies univalence, while the second does not. However, by applying avatar extensions using univalence, one can unify these two theories, by introducing a “special type membership” for which univalence holds, as a kind of proof of transitivity.

An example is Cubical Type Theory, where this special type membership is parameterised by the unit interval. The end points of the unit interval are 0 and 1.

Avatar Univalence might be represented as an L-system:

/ ~~                The core axiom of Path Semantics
~ ~~                Homotopy
/ /~                Inside univalence
~ /~                Outside univalence

/~ => / ~~ ~ ~~     Inserting the core axiom (termination)
/~ => / /~ ~ /~     Inserting inside and outside univalence (higher dimensional continuation)

In the formalisation of Path Semantical Intuitionistic Propositional Logic (PSI) in Avatar Logic:

PSI uses a notion of equality that is non-reflexive, called “quality” (~~). That means, x ~~ x is not a tautology.

Seshatism vs Platonism

Seshatism vs Platonism are related to the core axiom of Path Semantics:

From studying functions of type T -> T as models of equivalence groups (see paper), one can extend Witness Duality with a Moment Witness in the following way:

Notice that “Seshatic” and “Platonic” is used instead of “Seshatism” and “Platonism” since this is the projected theory from quality to equality.

This means that one can view Witness Duality indirectly in terms of Seshatism vs Platonism.

Seshatism vs Platonism are located in the Global Outside of the Logi circle.

Interior vs Exterior

The Interior is constructed using a core with an identity morphism as a “ground state”, getting topological properties via Product Witness (due to Symmetry Forcing).

The Exterior is constructed using a loop onto the identity morphism with a 1-avatar extension as the Loop Witness. This is related to exterior algebra.

Philosophical argument about self-referential Avatar Extensions as the interior:

  1. A theory must have an exterior to be used as “core self”
  2. If the exterior of Avatar Extensions is a dual theory, then they are related through an involution
  3. Involutions can be extended to Avatar Extensions, hence Avatar Extensions being avatar extendable
  4. Avatar Extensions witnesses its own extension, therefore is a Loop Witness
  5. The Exterior Avatar Extensions is a Product Witness, therefore, a candidate for the exterior might be Avatar Graphs

Avatar Modal Subjectivity

Avatar Modal Subjectivity can be thought of as guarded dialetheism as a technique for language boundaries.

The classical notion of subjectivity and objectivity is replaced with a unified concept “modal subjectivity” (e.g. s).

◇s            non-uniform modal subjectivity = classical subjectivity
□s            uniform modal subjectivity = classical objectivity
□¬s => □s     uniform modal non-subjectivity implies uniform modal subjectivity

Uniform modal non-subjectivity □¬s might be thought of as classical absolute objectivity, but this is weakened to uniform modal subjectivity.

Normal axioms:

Modal logic:
¬□¬p <=> ◇p
¬◇¬p <=> □p

Propositional logic:
¬¬p <=> p
(p => q) <=> (¬q => ¬p)

Associativity:
(¬□)(¬p) <=> ¬(□¬p)

Weak Monotonicity

The full experience of an observer in a given moment can never be repeated in practice. To model this phenomena, one can think of an infinite time counter, forever increasing. A time counter is monotonic, hence approximations to subjective experiences are thought of as having a weak monotonicity property.

Relations to Avatar Witness Theory:

For example, witness duality implies that it is impossible to talk about/define objectiveness without self referentiality.

In theories concerning weak monotonicity, the product of uniform and non-uniform subjectivity can be simplified:

□p ∧ ◇p <=> ◇p

Paradoxes and Tautologies

Some kind of paradoxes are related to Möbius topologies in Avatar Graphs. These graphs can be transformed into orthogonal mathematical objects using topological surgery.

For example, a Wagner graph can be transformed in a 3-cube graph.

The topological surgery operation is an involution and relates uniformity of paradoxes with non-uniformity of tautologies.

With self referentiality, it is easier to visualize the involution.

For example:

  1. “this sentence is false” is known as Liar’s paradox
  2. “this sentence is true” is a tautology with two solutions

The second sentence is related to the first one using involution on “this sentence is X”.

The Liar’s paradox offer no potential for non-uniform subjective experience.

Witness duality: In the tautology, an observer can experience one interpretation and later reinterpret it in future experiences, but it is impossible to reinterpret the original experience.

Avatar Schema Theory

Inside vs Outside Theories form an underlying structure that integrates Kent Palmer’s (Ph.D) Schema Theory.

Definition of an Inside Theory:

An Inside Theory is a mathematical language which models external objects as unknowns

Inside Theory is covered by the core axiom of Path Semantics using strictest order.

Definition of an Outside Theory:

An Outside Theory is a mathematical language with at least one symbol which does not refer to its theory

A hypothesis is that Outside Theory is related to weakening the order assumption in the core axiom of Path Semantics.

For philosophical argument of Inside vs Outside, see the blog post Plato’s Cave Found in Mathematics.

Axioms:

consistency ⊼ completeness
verification ⊼ validation
clarity ⊼ coherence
inside ⊼ outside
	
consistency ∨ completeness ∨ clarity => inside
verification ∨ validation => outside
coherence => inside ∨ outside

⊼ is logical NAND and represents Gödelian pairs.

Definition of Formal and Semantics:

formal := consistency ∨ completeness ∨ clarity
semantics := verification ∨ validation

This is sufficient to prove the Gödelian pair:

formal ⊼ semantics

There are 13 satisfied models of the axioms:

- - - - - - - -    4 x 4: Consistency Verification
- - - - - - - -    2 x 2: Completeness Validation
- - - - - - - O    1 x 1: Clarity Coherence
- - - - - - - O
- - - - - - - O       X Y    ¬X Y
- - - - - - - O     
- - - I - I - ⊻       X¬Y    ¬X¬Y
- - I I I I I ⊥

⊻ is just coherence that is both Inside and Outside. ⊥ is absence of all properties which is neither Inside or Outside.

Maximum mathematical languages can have at most 2 properties, given by the grammar:

verification|validation coherence
consistency|completeness coherence|clarity

Avatar Zen Theory

A Zen language is of form mem^N . pow^N where N : nat, mem is set membership and pow is the power set.

An Avatar Zen Ladder is used to study Zen languages formally:

azl(s : set, x : [set]) = ∀ i len(x)-1 { x[i+1] ∈ x[i] } ∧ (x[0] ∈ pow^len(x)(s))

x : [azl(s)] true (x is an Avatar Zen Ladder of s)

Avatar Zen Ladders avoid terminating choices, which means there is some minimum amount of structure preservation.

For example, in language design, a self-hosting compiler is an “ouroborus” Zen level.

For example, in biology, DNA is the sub-language which lets species mutate into new species (“ouroborus” Zen level).

Abstract Corruption

See paper about Abstract Corruption.

A 0-avatar is usually considered “empty inside”. This is often interpreted as a private logical language behind some protective interface. The emptiness inside, presenting total or arbitrary absurdity, makes it possible for the 0-avatar to exert influence on the Outside world.

In a more complex case, a 0-avatar might be given some limited amount of absurdity, but protect the consequences to some extent, e.g. in unsafe blocks of some programming language. The unsafe blocks might have undefined behavior, but the specification or design constraints of the language might limit the consequences of this behavior, or at least set desirable invariant properties.

In system thinking using a security mindset, one can often simplify an adverserial agent as a perfect 0-avatar, which is perfectly evil (from some perspective of another agent) and perfectly capable of lying convincingly to all other agents. This leads to the idea that Abstract Corruption is an theoretical embodiement of corruption in a pure abstract form.

To handle Abstract Corruption pragmatically, one can design systems to swap the node of the 0-avatar in locations of a graph, such that multiple nodes can be potentially 0-avatars. These are called “core candidates”.

This site is open source. Improve this page.