circkit.arithmetic

class circkit.arithmetic.Table(iterable=(), /)

Bases: tuple

class circkit.arithmetic.ArithmeticCircuit(*args, **kwargs)

Bases: Circuit

Generic class for arithmetic circuits.

class Operations

Bases: Operations

Class gathering Operation`s of :class:.ArithmeticCircuit`.

class ADD(*values, **kvalues)

Bases: Binary

eval(a, b)

Evaluate the operation on given inputs (typically values, not nodes).

class SUB(*values, **kvalues)

Bases: Binary

eval(a, b)

Evaluate the operation on given inputs (typically values, not nodes).

class MUL(*values, **kvalues)

Bases: Binary

eval(a, b)

Evaluate the operation on given inputs (typically values, not nodes).

class DIV(*values, **kvalues)

Bases: Binary

eval(a, b)

Evaluate the operation on given inputs (typically values, not nodes).

class EXP(*values, **kvalues)

Bases: Unary

eval(a)

Evaluate the operation on given inputs (typically values, not nodes).

class INV(*values, **kvalues)

Bases: Unary

eval(a)

Evaluate the operation on given inputs (typically values, not nodes).

class NEG(*values, **kvalues)

Bases: Unary

eval(a)

Evaluate the operation on given inputs (typically values, not nodes).

class LUT(*values, **kvalues)

Bases: Unary

eval(idx)

Evaluate the operation on given inputs (typically values, not nodes).

class RND(*values, **kvalues)

Bases: Nullary

eval()

Evaluate the operation on given inputs (typically values, not nodes).

Node

alias of Node

to_matrix(zero=0, one=1, n_tests=0)

Attempts to express the circuit as an affine map C(x) = A*x + b, where A is a matrix b is a vector, b = 0 when the map is linear

Done by evaluating the circuit at unit-vector inputs and at zero vector. Does not verify that the circuit is actually linear.

Note: current implementation does batch execution using Array inputs

may be doing single calls is more reliable for some circuit variants.

Todo

add option for non-batch execution

Parameters

• zero (const) – constant representing zero

• one (const) – constant representing one

Returns

matrix A Array: vector b

Return type

Matrix

class ADD(*values, **kvalues)

Bases: ADD

class CONST(*values, **kvalues)

Bases: CONST

class DIV(*values, **kvalues)

Bases: DIV

class EXP(*values, **kvalues)

Bases: EXP

class GET(*values, **kvalues)

Bases: GET

class INPUT(*values, **kvalues)

Bases: INPUT

class INV(*values, **kvalues)

Bases: INV

class LUT(*values, **kvalues)

Bases: LUT

class MUL(*values, **kvalues)

Bases: MUL

class NEG(*values, **kvalues)

Bases: NEG

Node_unlinked

alias of Node

class RND(*values, **kvalues)

Bases: RND

class SUB(*values, **kvalues)

Bases: SUB

class circkit.arithmetic.OptArithmeticCircuit(*args, **kwargs)

Bases: ArithmeticCircuit

Optimized arithmetic circuits.

Optimizations include:

• caching operations and nodes;

• precomputing annihilator operations, e.g. a*0 = 0;

• precomputing identity operations, e.g. a*1 = a, a+0 = 0;

• precomputing constant operations;

Node

alias of Node

class ADD(*values, **kvalues)

Bases: ADD

class CONST(*values, **kvalues)

Bases: CONST

class DIV(*values, **kvalues)

Bases: DIV

class EXP(*values, **kvalues)

Bases: EXP

class GET(*values, **kvalues)

Bases: GET

class INPUT(*values, **kvalues)

Bases: INPUT

class INV(*values, **kvalues)

Bases: INV

class LUT(*values, **kvalues)

Bases: LUT

class MUL(*values, **kvalues)

Bases: MUL

class NEG(*values, **kvalues)

Bases: NEG

Node_unlinked

alias of Node

class RND(*values, **kvalues)

Bases: RND

class SUB(*values, **kvalues)

Bases: SUB