TFHE-rs includes two main types to represent encrypted data:
FheUint: this is the homomorphic equivalent of Rust unsigned integers u8, u16, ...
FheInt: this is the homomorphic equivalent of Rust (signed) integers i8, i16, ...
In the same manner as many programming languages, the number of bits used to represent the data must be chosen when declaring a variable. For instance:
// let clear_a: u64 = 7;
let mut a = FheUint64::try_encrypt(clear_a, &keys)?;
// let clear_b: i8 = 3;
let mut b = FheInt8::try_encrypt(clear_b, &keys)?;
// let clear_c: u128 = 2;
let mut c = FheUint128::try_encrypt(clear_c, &keys)?;
Operation list
The table below contains an overview of the available operations in TFHE-rs. The notation Enc (for Encypted) either refers to FheInt or FheUint, for any size between 1 and 256-bits.
More details, and further examples, are given in the following sections.
Integer
In TFHE-rs, integers are used to encrypt all messages which are larger than 4 bits. All supported operations are listed below.
Arithmetic operations.
Homomorphic integer types support arithmetic operations.
The list of supported operations is:
A simple example of how to use these operations:
use tfhe::prelude::*;use tfhe::{generate_keys, set_server_key, ConfigBuilder, FheInt8, FheUint8};fnmain() ->Result<(), Box<dyn std::error::Error>> {let config =ConfigBuilder::all_disabled().enable_default_integers().build();let (keys, server_keys) =generate_keys(config);set_server_key(server_keys);let clear_a =15_u64;let clear_b =27_u64;let clear_c =43_u64;let clear_d =-87_i64;letmut a =FheUint8::try_encrypt(clear_a, &keys)?;letmut b =FheUint8::try_encrypt(clear_b, &keys)?;letmut c =FheUint8::try_encrypt(clear_c, &keys)?;letmut d =FheInt8::try_encrypt(clear_d, &keys)?; a = a *&b; // Clear equivalent computations: 15 * 27 mod 256 = 149 b =&b +&c; // Clear equivalent computations: 27 + 43 mod 256 = 70 b = b -76u8; // Clear equivalent computations: 70 - 76 mod 256 = 250 d = d -13i8; // Clear equivalent computations: -87 - 13 = 100 in [-128, 128[let dec_a:u8= a.decrypt(&keys);let dec_b:u8= b.decrypt(&keys);let dec_d:i8= d.decrypt(&keys);assert_eq!(dec_a, ((clear_a * clear_b) %256_u64) asu8);assert_eq!(dec_b, (((clear_b + clear_c).wrapping_sub(76_u64)) %256_u64) asu8);assert_eq!(dec_d, (clear_d -13) asi8);Ok(())}
Bitwise operations.
Homomorphic integer types support some bitwise operations.
The list of supported operations is:
A simple example of how to use these operations:
use tfhe::prelude::*;use tfhe::{generate_keys, set_server_key, ConfigBuilder, FheUint8};fnmain() ->Result<(), Box<dyn std::error::Error>> {let config =ConfigBuilder::all_disabled().enable_default_integers().build();let (keys, server_keys) =generate_keys(config);set_server_key(server_keys);let clear_a =164;let clear_b =212;letmut a =FheUint8::try_encrypt(clear_a, &keys)?;letmut b =FheUint8::try_encrypt(clear_b, &keys)?; a = a ^&b; b = b ^&a; a = a ^&b;let dec_a:u8= a.decrypt(&keys);let dec_b:u8= b.decrypt(&keys);// We homomorphically swapped values using bitwise operationsassert_eq!(dec_a, clear_b);assert_eq!(dec_b, clear_a);Ok(())}
Comparisons.
Homomorphic integers support comparison operations.
Due to some Rust limitations, it is not possible to overload the comparison symbols because of the inner definition of the operations. This is because Rust expects to have a Boolean as an output, whereas a ciphertext is returned when using homomorphic types.
You will need to use different methods instead of using symbols for the comparisons. These methods follow the same naming conventions as the two standard Rust traits:
Homomorphic integers support the min/max operations.
A simple example of how to use these operations:
use tfhe::prelude::*;use tfhe::{generate_keys, set_server_key, ConfigBuilder, FheUint8};fnmain() ->Result<(), Box<dyn std::error::Error>> {let config =ConfigBuilder::all_disabled().enable_default_integers().build();let (keys, server_keys) =generate_keys(config);set_server_key(server_keys);let clear_a:u8=164;let clear_b:u8=212;letmut a =FheUint8::try_encrypt(clear_a, &keys)?;letmut b =FheUint8::try_encrypt(clear_b, &keys)?;let min = a.min(&b);let max = a.max(&b);let dec_min :u8= min.decrypt(&keys);let dec_max :u8= max.decrypt(&keys);// We homomorphically swapped values using bitwise operationsassert_eq!(dec_min, u8::min(clear_a, clear_b));assert_eq!(dec_max, u8::max(clear_a, clear_b));Ok(())}
Ternary conditional operator.
The ternary conditional operator allows computing conditional instructions of the form if cond { choice_if } else { choice_else }.
The syntax is encrypted_condition.if_then_else(encrypted_choice_if, encrypted_choice_else). The encrypted_condition should be an encryption of 0 or 1 in order to be valid.
use tfhe::prelude::*;use tfhe::{generate_keys, set_server_key, ConfigBuilder, FheInt32};fnmain() ->Result<(), Box<dyn std::error::Error>> {// Basic configuration to use homomorphic integerslet config =ConfigBuilder::all_disabled().enable_default_integers().build();// Key generationlet (client_key, server_keys) =generate_keys(config);let clear_a =32i32;let clear_b =-45i32;// Encrypting the input data using the (private) client_key// FheInt32: Encrypted equivalent to i32let encrypted_a =FheInt32::try_encrypt(clear_a, &client_key)?;let encrypted_b =FheInt32::try_encrypt(clear_b, &client_key)?;// On the server side:set_server_key(server_keys);// Clear equivalent computations: 32 > -45let encrypted_comp =&encrypted_a.gt(&encrypted_b);let clear_res:i32= encrypted_comp.decrypt(&client_key);assert_eq!(clear_res, (clear_a > clear_b) asi32);// `encrypted_comp` contains the result of the comparison, i.e.,// a boolean value. This acts as a condition on which the// `if_then_else` function can be applied on.// Clear equivalent computations:// if 32 > -45 {result = 32} else {result = -45}let encrypted_res =&encrypted_comp.if_then_else(&encrypted_a, &encrypted_b);let clear_res:i32= encrypted_res.decrypt(&client_key);assert_eq!(clear_res, clear_a);Ok(())}
Casting.
Casting between integer types is possible via the cast_from associated function or the cast_into method.
use tfhe::prelude::*;use tfhe::{generate_keys, set_server_key, ConfigBuilder, FheInt16, FheUint8, FheUint32, FheUint16};fnmain() ->Result<(), Box<dyn std::error::Error>> {let config =ConfigBuilder::all_disabled().enable_default_integers().build();let (client_key, server_key) =generate_keys(config);// Casting requires server_key to set// (encryptions/decryptions do not need server_key to be set)set_server_key(server_key); {let clear =12_837u16;let a =FheUint16::encrypt(clear, &client_key);// Downcastinglet a:FheUint8= a.cast_into();let da:u8= a.decrypt(&client_key);assert_eq!(da, clear asu8);// Upcastinglet a:FheUint32= a.cast_into();let da:u32= a.decrypt(&client_key);assert_eq!(da, (clear asu8) asu32); } {let clear =12_837u16;let a =FheUint16::encrypt(clear, &client_key);// Upcastinglet a =FheUint32::cast_from(a);let da:u32= a.decrypt(&client_key);assert_eq!(da, clear asu32);// Downcastinglet a =FheUint8::cast_from(a);let da:u8= a.decrypt(&client_key);assert_eq!(da, (clear asu32) asu8); } {let clear =12_837i16;let a =FheInt16::encrypt(clear, &client_key);// Casting from FheInt16 to FheUint16let a =FheUint16::cast_from(a);let da:u16= a.decrypt(&client_key);assert_eq!(da, clear asu16); }Ok(())}
ShortInt Operations
Native small homomorphic integer types (e.g., FheUint3 or FheUint4) can easily compute various operations. In general, computing over encrypted data in TFHE-rs is as easy as computing over clear data, since the same operation symbol is used. The addition between two ciphertexts is done using the symbol + between two FheUint values. Many operations can be computed between a clear value (i.e. a scalar) and a ciphertext.
Arithmetic operations.
Small homomorphic integer types support all common arithmetic operations, meaning +, -, x, /, mod.
The division operation presents a subtlety: since data is encrypted, it is possible to compute a division by 0. In this case, the division is tweaked so that dividing by 0 returns the maximum possible value for the message.
The list of supported operations is:
A simple example of how to use these operations:
use tfhe::prelude::*;use tfhe::{generate_keys, set_server_key, ConfigBuilder, FheUint3};fnmain() ->Result<(), Box<dyn std::error::Error>> {let config =ConfigBuilder::all_disabled().enable_default_uint3().build();let (keys, server_keys) =generate_keys(config);set_server_key(server_keys);let clear_a =7;let clear_b =3;let clear_c =2;letmut a =FheUint3::try_encrypt(clear_a, &keys)?;letmut b =FheUint3::try_encrypt(clear_b, &keys)?;letmut c =FheUint3::try_encrypt(clear_c, &keys)?; a = a *&b; // Clear equivalent computations: 7 * 3 mod 8 = 5 b =&b +&c; // Clear equivalent computations: 3 + 2 mod 8 = 5 b = b -5; // Clear equivalent computations: 5 - 5 mod 8 = 0let dec_a = a.decrypt(&keys);let dec_b = b.decrypt(&keys);// We homomorphically swapped values using bitwise operationsassert_eq!(dec_a, (clear_a * clear_b) %8);assert_eq!(dec_b, ((clear_b + clear_c) -5) %8);Ok(())}
Bitwise operations.
Small homomorphic integer types support some bitwise operations.
The list of supported operations is:
A simple example of how to use these operations:
use tfhe::prelude::*;use tfhe::{generate_keys, set_server_key, ConfigBuilder, FheUint3};fnmain() ->Result<(), Box<dyn std::error::Error>> {let config =ConfigBuilder::all_disabled().enable_default_uint3().build();let (keys, server_keys) =generate_keys(config);set_server_key(server_keys);let clear_a =7;let clear_b =3;letmut a =FheUint3::try_encrypt(clear_a, &keys)?;letmut b =FheUint3::try_encrypt(clear_b, &keys)?; a = a ^&b; b = b ^&a; a = a ^&b;let dec_a = a.decrypt(&keys);let dec_b = b.decrypt(&keys);// We homomorphically swapped values using bitwise operationsassert_eq!(dec_a, clear_b);assert_eq!(dec_b, clear_a);Ok(())}
Comparisons.
Small homomorphic integer types support comparison operations.
Due to some Rust limitations, it is not possible to overload the comparison symbols because of the inner definition of the operations. Rust expects to have a Boolean as an output, whereas a ciphertext is returned when using homomorphic types.
You will need to use different methods instead of using symbols for the comparisons. These methods follow the same naming conventions as the two standard Rust traits:
Using the shortint type allows you to evaluate bivariate functions (i.e., functions that take two ciphertexts as input).
A simple code example:
use tfhe::prelude::*;use tfhe::{generate_keys, set_server_key, ConfigBuilder, FheUint2};fnmain() ->Result<(), Box<dyn std::error::Error>> {let config =ConfigBuilder::all_disabled().enable_default_uint2().build();let (keys, server_keys) =generate_keys(config);set_server_key(server_keys);let clear_a =1;let clear_b =3;let a =FheUint2::try_encrypt(clear_a, &keys)?;let b =FheUint2::try_encrypt(clear_b, &keys)?;let c = a.bivariate_function(&b, std::cmp::max);let decrypted = c.decrypt(&keys);assert_eq!(decrypted, std::cmp::max(clear_a, clear_b) asu8);Ok(())}
Boolean Operations
Native homomorphic Booleans support common Boolean operations.
The list of supported operations is:
name
symbol
type
For division by 0, the convention is to return modulus - 1. For instance, for FheUint8, the modulus is 28=256, so a division by 0 will return an encryption of 255. For the remainder operator, the convention is to return the first input without any modification. For instance, if ct1 = FheUint8(63) and ct2 = FheUint8(0) then ct1 % ct2 will return FheUint8(63).
name
symbol
type
name
symbol
type
name
symbol
type
name
symbol
type
In Rust, operations on native types are modular. For example, computations on u8 are carried out modulo 28. A similar idea applies for FheUintX, where operations are done modulo 2X. For FheUint3, operations are done modulo 8=23.