algo

algo

factory algo

Description:

  • Specifies the macroscopic algorithmic ersatz on VSA symbol.

    • These functions are accessible via the macrovsa:: prefix.

Members

(static) sigma_0 :double

Description:

  • The standard-deviation of the additive noise generated by an approximate operation.

    • Its value order of magnitude is O(1/d) where d is the VSA space dimension, default value is d=10000, and is calibrated on the (Schlegel et al 2020) numerical results, here set to 1/d/64.

The standard-deviation of the additive noise generated by an approximate operation.

  • Its value order of magnitude is O(1/d) where d is the VSA space dimension, default value is d=10000, and is calibrated on the (Schlegel et al 2020) numerical results, here set to 1/d/64.
Type:
  • double

Methods

(static) reduce(symbol) → {Symbol}

Description:

  • Reduces a symbol with respect to binding/unbinding operations.

Parameters:
Name Type Description
symbol Symbol

The symbol to reduce.
Returns:

A reference to the reduced symbol. Available until program end.

Type
Symbol

(static) sim(s1, s2) → {Belief}

Description:

  • Returns the similarity between two symbols.

    • For convinience the symbol can be given by its name, considering that tau = 1, sigma = 0.
Parameters:
Name Type Description
s1 Symbol | String

A reference to the first symbol.
s2 Symbol | String

A reference to the second symbol.
Returns:

The similarity or 0 if out of bounds, as belief, thus with an error estmimation.

Type
Belief

(static) msim({Symbol, s2) → {double}

Description:

  • Returns the mesoscopic similarity between two symbol vectors.

Parameters:
Name Type Description
{Symbol

s1 A reference to the first symbol.
s2 Symbol

A reference to the second symbol.
Returns:

The similarity, using the dotprod.

Type
double

(static) conj() → {Belief}

Description:

  • Calculates the conjuction of belief values (and operator).

    • Here the max(0,min(...)) operator is used as T-Norm.
Parameters:
Name Type Attributes Description
degree... Belief <repeatable>

Degree of belief values, at least 2 and up to 8 arguments, in this implementation.
Returns:

The computed value.

Type
Belief