floating_point.secure_bit_length module

Module containing a protocol to securely compute the bit length of an MPyC secure integer.

floating_point.secure_bit_length.bit_length(secure_element, *, two_to_pow=False, inv_two_to_pow=False, lower_bound=0, upper_bound=None)[source]

Implementation of Thijs Veugen’s secure bit length protocol.

!! The current implementation only supports non-negative inputs !!

Parameters:

  • secure_element (SecureInteger) – secret shared value of which the bit length needs to be determined. The secret value is assumed to be non-negative.

  • two_to_pow (bool) – If True, also compute and return 2 ** bit_length. This is more efficient than computing the exponentiation afterwards.

  • inv_two_to_pow (bool) – If True, also compute and return 2 ** -bit_length. This is more efficient than computing the exponentiation afterwards.

  • lower_bound (int) – Lower bound on the bit length of the provided value. This parameter allows for a more efficient protocol, but it will also yield an incorrect result if the lower bound is larger than the actual bit length.

  • upper_bound (int | None) – Upper bound on the bit length of the provided value. This parameter allows for a more efficient protocol, but it will also yield an incorrect result if the upper bound is smaller than the actual bit length.

Return type:

SecureInteger | tuple[SecureInteger, SecureInteger] | tuple[SecureInteger, SecureInteger, SecureInteger]

Returns:

A secret-shared value representing the bit length of the input. If two_to_pow=True, the second return value is its power base two. If inv_two_to_pow=True, the final return value is the inverse of its power base two.

floating_point.secure_bit_length.most_significant_bit(secure_element, *, two_to_pow=False, inv_two_to_pow=False, lower_bound=0, upper_bound=None)[source]

Implementation of Thijs Veugen’s secure most significant bit (MSB) protocol.

The function will return the index of the most significant bit. Thus, MSB(1) = 0, MSB(2) = 1, MSB(3) = 1, etc. The edge case MSB(0) returns -1. For consistency, 2 ** -MSB(0) returns 2 and 2 ** MSB(0) returns the multiplicative inverse of 2 in the finite field.

Parameters:

  • secure_element (SecureInteger) – Secret shared value of which the most significant bit needs to be determined.

  • two_to_pow (bool) – If True, also compute and return 2 ** MSB. This is more efficient than computing the exponentiation afterwards.

  • inv_two_to_pow (bool) – If True, also compute and return 2 ** -MSB. This is more efficient than computing the exponentiation afterwards.

  • lower_bound (int) – Lower bound on the most significant bit index of the provided value. This parameter allows for a more efficient protocol, but it will also yield an incorrect result if the lower bound is larger than the actual MSB index.

  • upper_bound (int | None) – Upper bound on the most significant bit index of the provided value. This parameter allows for a more efficient protocol, but it will also yield an incorrect result if the upper bound is smaller than the actual MSB index.

Raises:

ValueError – The provided bounds are unacceptable.

Return type:

SecureInteger | tuple[SecureInteger, SecureInteger] | tuple[SecureInteger, SecureInteger, SecureInteger]

Returns:

The index of the most significant bit of input secure_element. If two_to_pow=True, the second return value is its power base two. If inv_two_to_pow=True, the final return value is the inverse of its power base two.