opendp 0.14.2-dev.20260401.2

A library of differential privacy algorithms for the statistical analysis of sensitive private data.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26

# type: ignore
def make_randomized_response_bool(prob: f64, constant_time: bool):
    input_domain = AtomDomain(bool)
    input_metric = DiscreteMetric()
    output_measure = MaxDivergence()
    
    if (prob < 0.5 or prob > 1):  # |\label{line:range}|
        raise Exception("probability must be in [0.5, 1]")

    if prob == 1.0:
        c = float("inf")
    else:
        c = prob.inf_div((1).neg_inf_sub(prob)).inf_ln()
    
    def privacy_map(d_in: u32) -> f64:  # |\label{line:map}|
        if d_in == 0:
            return 0
        else:
            return c

    def function(arg: bool) -> bool:  # |\label{line:fn}|
        sample = not sample_bernoulli_float(prob, constant_time)

        return arg ^ sample
    
    return Measurement(input_domain, function, input_metric, output_measure, privacy_map)