tfhe 1.5.5

TFHE-rs is a fully homomorphic encryption (FHE) library that implements Zama's variant of TFHE.
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
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
pub mod noise_simulation;
pub use noise_simulation::traits;

use crate::core_crypto::algorithms::glwe_encryption::decrypt_glwe_ciphertext;
use crate::core_crypto::algorithms::lwe_encryption::{
    allocate_and_encrypt_new_lwe_ciphertext, decrypt_lwe_ciphertext,
};
use crate::core_crypto::algorithms::misc::torus_modular_diff;
use crate::core_crypto::algorithms::test::round_decode;
use crate::core_crypto::commons::dispersion::{DispersionParameter, Variance};
use crate::core_crypto::commons::generators::EncryptionRandomGenerator;
use crate::core_crypto::commons::math::random::{ByteRandomGenerator, Gaussian, Uniform};
use crate::core_crypto::commons::noise_formulas::secure_noise::{
    minimal_lwe_variance_for_132_bits_security_gaussian,
    minimal_lwe_variance_for_132_bits_security_tuniform,
};
use crate::core_crypto::commons::numeric::{CastFrom, UnsignedInteger};
use crate::core_crypto::commons::parameters::{
    CiphertextModulus, DynamicDistribution, LweCiphertextCount, LweDimension, PlaintextCount,
};
use crate::core_crypto::commons::test_tools::{
    arithmetic_mean, equivalent_pfail_gaussian_noise, gaussian_mean_confidence_interval,
    gaussian_variance_confidence_interval, normality_test_f64,
    pfail_clopper_pearson_exact_confidence_interval, variance, NormalityTestResult,
};
use crate::core_crypto::commons::traits::container::Container;
use crate::core_crypto::commons::traits::Encryptable;
use crate::core_crypto::entities::glwe_ciphertext::GlweCiphertext;
use crate::core_crypto::entities::glwe_secret_key::GlweSecretKey;
use crate::core_crypto::entities::lwe_ciphertext::{LweCiphertext, LweCiphertextOwned};
use crate::core_crypto::entities::lwe_secret_key::LweSecretKey;
use crate::core_crypto::entities::{Cleartext, PlaintextList};
use crate::shortint::encoding::ShortintEncoding;
use crate::shortint::parameters::{
    AtomicPatternParameters, CarryModulus, MessageModulus, PBSParameters,
};

#[derive(Clone, Copy, Debug, PartialEq, Eq)]
pub struct PrecisionWithPadding {
    value: u32,
}

pub fn normality_check(
    noise_samples: &[f64],
    check_location: &str,
    alpha: f64,
) -> NormalityTestResult {
    let normality_check =
        normality_test_f64(&noise_samples[..5000.min(noise_samples.len())], alpha);

    if normality_check.null_hypothesis_is_valid {
        println!("Normality check {check_location} is OK\n");
    } else {
        println!("Normality check {check_location} failed\n");
    }

    normality_check
}

pub fn mean_and_variance_check<Scalar: UnsignedInteger>(
    noise_samples: &[f64],
    suffix: &str,
    expected_mean: f64,
    expected_variance: Variance,
    noise_distribution_used_for_encryption: DynamicDistribution<Scalar>,
    decryption_key_lwe_dimension: LweDimension,
    modulus_as_f64: f64,
) -> bool {
    assert!(expected_mean.is_finite(), "Expected mean is infinite");
    assert!(
        expected_variance.0.is_finite(),
        "Expected variance is infinite"
    );
    assert!(expected_variance.0 >= 0.0, "Expected positive variance");

    let measured_mean = arithmetic_mean(noise_samples);
    let measured_variance = variance(noise_samples);

    let mean_ci = gaussian_mean_confidence_interval(
        noise_samples.len() as f64,
        measured_mean,
        measured_variance.get_standard_dev(),
        0.99,
    );

    let variance_ci =
        gaussian_variance_confidence_interval(noise_samples.len() as f64, measured_variance, 0.99);

    println!("measured_variance_{suffix}={measured_variance:?}");
    println!("expected_variance_{suffix}={expected_variance:?}");
    println!("variance_lower_bound={:?}", variance_ci.lower_bound());
    println!("variance_upper_bound={:?}", variance_ci.upper_bound());
    println!("measured_mean_{suffix}={measured_mean:?}");
    println!("expected_mean_{suffix}={expected_mean:?}");
    println!("mean_{suffix}_lower_bound={:?}", mean_ci.lower_bound());
    println!("mean_{suffix}_upper_bound={:?}", mean_ci.upper_bound());

    // Expected mean is 0
    let mean_is_in_interval = mean_ci.mean_is_in_interval(expected_mean);

    if mean_is_in_interval {
        println!(
            "PASS: measured_mean_{suffix} confidence interval \
            contains the expected mean"
        );
    } else {
        println!(
            "FAIL: measured_mean_{suffix} confidence interval \
            does not contain the expected mean"
        );
    }

    // We want to be smaller but secure or in the interval
    let variance_is_ok = if measured_variance <= expected_variance {
        let noise_for_security = match noise_distribution_used_for_encryption {
            DynamicDistribution::Gaussian(_) => {
                minimal_lwe_variance_for_132_bits_security_gaussian(
                    decryption_key_lwe_dimension,
                    modulus_as_f64,
                )
            }
            DynamicDistribution::TUniform(_) => {
                minimal_lwe_variance_for_132_bits_security_tuniform(
                    decryption_key_lwe_dimension,
                    modulus_as_f64,
                )
            }
        };

        let variance_is_secure = measured_variance >= noise_for_security;

        if variance_is_secure {
            println!("PASS: measured_variance_{suffix} is smaller than expected variance.");

            if !variance_ci.variance_is_in_interval(expected_variance) {
                println!(
                    "\n==========\n\
                    Warning: noise formula might be over estimating the noise.\n\
                    ==========\n"
                );
            }
        } else {
            println!("FAIL: measured_variance_{suffix} is NOT secure.")
        }

        variance_is_secure
    } else {
        let interval_ok = variance_ci.variance_is_in_interval(expected_variance);

        if interval_ok {
            println!(
                "PASS: measured_variance_{suffix} confidence interval \
                contains the expected variance"
            );
        } else {
            println!(
                "FAIL: measured_variance_{suffix} confidence interval \
                does not contain the expected variance"
            );
        }

        interval_ok
    };

    mean_is_in_interval && variance_is_ok
}

pub fn encrypt_new_noiseless_lwe<
    Scalar: UnsignedInteger + Encryptable<Uniform, Gaussian<f64>> + CastFrom<u64>,
    InputKeyCont: Container<Element = Scalar>,
    Gen: ByteRandomGenerator,
>(
    lwe_secret_key: &LweSecretKey<InputKeyCont>,
    ciphertext_modulus: CiphertextModulus<Scalar>,
    msg: Scalar,
    encoding: &ShortintEncoding<Scalar>,
    encryption_random_generation: &mut EncryptionRandomGenerator<Gen>,
) -> LweCiphertextOwned<Scalar> {
    let noiseless_distribution = Gaussian::from_dispersion_parameter(Variance(0.0), 0.0);

    let plaintext = encoding.encode(Cleartext(msg));

    allocate_and_encrypt_new_lwe_ciphertext(
        lwe_secret_key,
        plaintext,
        noiseless_distribution,
        ciphertext_modulus,
        encryption_random_generation,
    )
}

#[derive(Clone, Copy)]
pub struct PfailAndPrecision {
    pfail: f64,
    precision_with_padding: PrecisionWithPadding,
}

impl PfailAndPrecision {
    pub fn new(pfail: f64, msg_mod: MessageModulus, carry_mod: CarryModulus) -> Self {
        assert!(msg_mod.0.is_power_of_two());
        assert!(carry_mod.0.is_power_of_two());

        let precision_with_padding = precision_with_padding(msg_mod, carry_mod);

        Self {
            pfail,
            precision_with_padding,
        }
    }

    pub fn new_from_ap_params(ap_params: &AtomicPatternParameters) -> Self {
        Self::new(
            2.0f64.powf(ap_params.log2_p_fail()),
            ap_params.message_modulus(),
            ap_params.carry_modulus(),
        )
    }

    pub fn pfail(&self) -> f64 {
        self.pfail
    }

    pub fn precision_with_padding(&self) -> PrecisionWithPadding {
        self.precision_with_padding
    }
}

#[derive(Clone, Copy)]
pub struct PfailTestMeta {
    original_pfail_and_precision: PfailAndPrecision,
    new_pfail_and_precision: PfailAndPrecision,
    expected_fails: u32,
    total_runs_for_expected_fails: u32,
}

impl PfailTestMeta {
    pub fn new_with_desired_expected_fails(
        original_pfail_and_precision: PfailAndPrecision,
        new_pfail_and_precision: PfailAndPrecision,
        expected_fails: u32,
    ) -> Self {
        let expected_fails_f64: f64 = expected_fails.into();
        let total_runs_for_expected_fails =
            (expected_fails_f64 / new_pfail_and_precision.pfail).round() as u32;

        println!("expected_fails: {expected_fails}");
        println!("total_runs_for_expected_fails: {total_runs_for_expected_fails}");

        Self {
            original_pfail_and_precision,
            new_pfail_and_precision,
            expected_fails,
            total_runs_for_expected_fails,
        }
    }

    pub fn new_with_total_runs(
        original_pfail_and_precision: PfailAndPrecision,
        new_pfail_and_precision: PfailAndPrecision,
        total_runs_for_expected_fails: u32,
    ) -> Self {
        let total_runs_f64: f64 = total_runs_for_expected_fails.into();
        let expected_fails = (total_runs_f64 * new_pfail_and_precision.pfail).round() as u32;

        println!("expected_fails: {expected_fails}");
        println!("total_runs_for_expected_fails: {total_runs_for_expected_fails}");

        Self {
            original_pfail_and_precision,
            new_pfail_and_precision,
            expected_fails,
            total_runs_for_expected_fails,
        }
    }

    pub fn original_pfail_and_precision(&self) -> PfailAndPrecision {
        self.original_pfail_and_precision
    }
    pub fn new_pfail_and_precision(&self) -> PfailAndPrecision {
        self.new_pfail_and_precision
    }
    pub fn expected_fails(&self) -> u32 {
        self.expected_fails
    }
    pub fn total_runs_for_expected_fails(&self) -> u32 {
        self.total_runs_for_expected_fails
    }
}

#[derive(Clone, Copy)]
pub struct PfailTestResult {
    pub measured_fails: f64,
}

pub fn pfail_check(pfail_test_meta: &PfailTestMeta, pfail_test_result: PfailTestResult) {
    let measured_fails = pfail_test_result.measured_fails;
    let total_runs_for_expected_fails = pfail_test_meta.total_runs_for_expected_fails;
    let expected_fails = pfail_test_meta.expected_fails();

    let new_pfail_and_precision = pfail_test_meta.new_pfail_and_precision();
    let expected_pfail = new_pfail_and_precision.pfail();
    let new_precision_with_padding = pfail_test_meta
        .new_pfail_and_precision
        .precision_with_padding();

    let original_pfail_and_precision = pfail_test_meta.original_pfail_and_precision();
    let original_pfail = original_pfail_and_precision.pfail();
    let original_precision_with_padding = original_pfail_and_precision.precision_with_padding();

    let measured_pfail = measured_fails / (total_runs_for_expected_fails as f64);

    println!("measured_fails={measured_fails}");
    println!("expected_fails={expected_fails}");
    println!("measured_pfail={measured_pfail}");
    println!("expected_pfail={expected_pfail}");

    let equivalent_measured_pfail = equivalent_pfail_gaussian_noise(
        new_precision_with_padding.value,
        measured_pfail,
        original_precision_with_padding.value,
    );

    println!("equivalent_measured_pfail={equivalent_measured_pfail}");
    println!("original_expected_pfail  ={original_pfail}");
    println!(
        "equivalent_measured_pfail_log2={}",
        equivalent_measured_pfail.log2()
    );
    println!("original_expected_pfail_log2  ={}", original_pfail.log2());

    if measured_fails > 0.0 {
        let pfail_confidence_interval = pfail_clopper_pearson_exact_confidence_interval(
            total_runs_for_expected_fails as f64,
            measured_fails,
            0.99,
        );

        let pfail_lower_bound = pfail_confidence_interval.lower_bound();
        let pfail_upper_bound = pfail_confidence_interval.upper_bound();
        println!("pfail_lower_bound={pfail_lower_bound}");
        println!("pfail_upper_bound={pfail_upper_bound}");

        let equivalent_pfail_lower_bound = equivalent_pfail_gaussian_noise(
            new_precision_with_padding.value,
            pfail_lower_bound,
            original_precision_with_padding.value,
        );
        let equivalent_pfail_upper_bound = equivalent_pfail_gaussian_noise(
            new_precision_with_padding.value,
            pfail_upper_bound,
            original_precision_with_padding.value,
        );

        println!("equivalent_pfail_lower_bound={equivalent_pfail_lower_bound}");
        println!("equivalent_pfail_upper_bound={equivalent_pfail_upper_bound}");
        println!(
            "equivalent_pfail_lower_bound_log2={}",
            equivalent_pfail_lower_bound.log2()
        );
        println!(
            "equivalent_pfail_upper_bound_log2={}",
            equivalent_pfail_upper_bound.log2()
        );

        if measured_pfail <= expected_pfail {
            if !pfail_confidence_interval.mean_is_in_interval(expected_pfail) {
                println!(
                    "\n==========\n\
                    WARNING: measured pfail is smaller than expected pfail \
                    and out of the confidence interval\n\
                    the optimizer might be pessimistic when generating parameters.\n\
                    ==========\n"
                );
            }
        } else {
            assert!(pfail_confidence_interval.mean_is_in_interval(expected_pfail));
        }
    } else {
        println!(
            "\n==========\n\
            WARNING: measured pfail is 0, it is either a bug or \
            it is way smaller than the expected pfail\n\
            the optimizer might be pessimistic when generating parameters, \
            or some hypothesis does not hold.\n\
            ==========\n"
        );
    }
}

#[derive(Clone, Copy, Debug)]
pub struct NoiseSample {
    pub value: f64,
}

#[derive(Clone, Copy, Debug)]
pub enum DecryptionAndNoiseResult {
    DecryptionSucceeded { noise: NoiseSample },
    DecryptionFailed,
}

impl DecryptionAndNoiseResult {
    pub fn new_from_lwe<Scalar: UnsignedInteger + CastFrom<u64>, CtCont, KeyCont>(
        ct: &LweCiphertext<CtCont>,
        secret_key: &LweSecretKey<KeyCont>,
        expected_msg: Scalar,
        encoding: &ShortintEncoding<Scalar>,
    ) -> Self
    where
        CtCont: Container<Element = Scalar>,
        KeyCont: Container<Element = Scalar>,
    {
        let decrypted_plaintext = decrypt_lwe_ciphertext(secret_key, ct).0;

        let delta = encoding.delta();
        let cleartext_modulus_with_padding = encoding.full_cleartext_space();

        // We apply the modulus on the cleartext + the padding bit
        let decoded_msg = round_decode(decrypted_plaintext, delta) % cleartext_modulus_with_padding;

        let expected_plaintext = expected_msg * delta;

        // decrypted_plaintext = expected_plaintext + error
        // The order below computes:
        // decrypted_plaintext - expected_plaintext in a modular way, which is what we want
        // It only changes the average value sign, so that it is more intuitive when comparing to
        // theory
        let noise = torus_modular_diff(
            decrypted_plaintext,
            expected_plaintext,
            ct.ciphertext_modulus(),
        );

        if decoded_msg == expected_msg {
            Self::DecryptionSucceeded {
                noise: NoiseSample { value: noise },
            }
        } else {
            Self::DecryptionFailed
        }
    }

    pub fn new_from_glwe<Scalar: UnsignedInteger + CastFrom<u64>, CtCont, KeyCont>(
        ct: &GlweCiphertext<CtCont>,
        secret_key: &GlweSecretKey<KeyCont>,
        lwe_per_glwe: LweCiphertextCount,
        expected_msg: Scalar,
        encoding: &ShortintEncoding<Scalar>,
    ) -> Vec<Self>
    where
        CtCont: Container<Element = Scalar>,
        KeyCont: Container<Element = Scalar>,
    {
        let mut decrypted =
            PlaintextList::new(Scalar::ZERO, PlaintextCount(ct.polynomial_size().0));

        let delta = encoding.delta();
        let cleartext_modulus_with_padding = encoding.full_cleartext_space();

        decrypt_glwe_ciphertext(secret_key, ct, &mut decrypted);

        let expected_plaintext = expected_msg * delta;

        decrypted
            .as_ref()
            .iter()
            .take(lwe_per_glwe.0)
            .map(|&decrypted_plaintext| {
                // We apply the modulus on the cleartext + the padding bit
                let decoded_msg =
                    round_decode(decrypted_plaintext, delta) % cleartext_modulus_with_padding;

                let noise = torus_modular_diff(
                    expected_plaintext,
                    decrypted_plaintext,
                    ct.ciphertext_modulus(),
                );

                if decoded_msg == expected_msg {
                    Self::DecryptionSucceeded {
                        noise: NoiseSample { value: noise },
                    }
                } else {
                    Self::DecryptionFailed
                }
            })
            .collect()
    }

    pub fn get_noise_if_decryption_was_correct(&self) -> Option<NoiseSample> {
        match self {
            Self::DecryptionSucceeded { noise } => Some(*noise),
            Self::DecryptionFailed => None,
        }
    }

    /// If decryption failed (in context of pfail evaluation) returns 1.0 else 0.0, to easily sum
    /// failures to evaluate pfail
    pub fn failure_as_f64(&self) -> f64 {
        match self {
            Self::DecryptionSucceeded { .. } => 0.0,
            Self::DecryptionFailed => 1.0,
        }
    }
}

pub fn update_ap_params_msg_and_carry_moduli(
    ap_params: &mut AtomicPatternParameters,
    new_message_modulus: MessageModulus,
    new_carry_modulus: CarryModulus,
) {
    match ap_params {
        AtomicPatternParameters::Standard(pbsparameters) => match pbsparameters {
            PBSParameters::PBS(classic_pbsparameters) => {
                classic_pbsparameters.message_modulus = new_message_modulus;
                classic_pbsparameters.carry_modulus = new_carry_modulus;
            }
            PBSParameters::MultiBitPBS(multi_bit_pbsparameters) => {
                multi_bit_pbsparameters.message_modulus = new_message_modulus;
                multi_bit_pbsparameters.carry_modulus = new_carry_modulus;
            }
        },
        AtomicPatternParameters::KeySwitch32(key_switch32_pbsparameters) => {
            key_switch32_pbsparameters.message_modulus = new_message_modulus;
            key_switch32_pbsparameters.carry_modulus = new_carry_modulus;
        }
    }
}

pub fn update_ap_params_for_pfail(
    ap_params: &mut AtomicPatternParameters,
    new_message_modulus: MessageModulus,
    new_carry_modulus: CarryModulus,
) -> (PfailAndPrecision, PfailAndPrecision) {
    let orig_pfail_and_precision = PfailAndPrecision::new_from_ap_params(&*ap_params);

    println!("original_pfail: {}", orig_pfail_and_precision.pfail());
    println!(
        "original_pfail_log2: {}",
        orig_pfail_and_precision.pfail().log2()
    );

    update_ap_params_msg_and_carry_moduli(ap_params, new_message_modulus, new_carry_modulus);

    let new_expected_pfail = equivalent_pfail_gaussian_noise(
        orig_pfail_and_precision.precision_with_padding().value,
        orig_pfail_and_precision.pfail(),
        precision_with_padding(ap_params.message_modulus(), ap_params.carry_modulus()).value,
    );
    let new_expected_log2_pfail = new_expected_pfail.log2();

    match ap_params {
        AtomicPatternParameters::Standard(pbsparameters) => match pbsparameters {
            PBSParameters::PBS(classic_pbsparameters) => {
                classic_pbsparameters.log2_p_fail = new_expected_log2_pfail;
            }
            PBSParameters::MultiBitPBS(multi_bit_pbsparameters) => {
                multi_bit_pbsparameters.log2_p_fail = new_expected_log2_pfail;
            }
        },
        AtomicPatternParameters::KeySwitch32(key_switch32_pbsparameters) => {
            key_switch32_pbsparameters.log2_p_fail = new_expected_log2_pfail;
        }
    }

    let new_expected_pfail = PfailAndPrecision::new_from_ap_params(&*ap_params);

    println!("new_expected_pfail: {}", new_expected_pfail.pfail());
    println!(
        "new_expected_pfail_log2: {}",
        new_expected_pfail.pfail().log2()
    );

    (orig_pfail_and_precision, new_expected_pfail)
}

pub fn precision_with_padding(
    msg_mod: MessageModulus,
    carry_mod: CarryModulus,
) -> PrecisionWithPadding {
    let cleartext_modulus = msg_mod.0 * carry_mod.0;
    assert!(cleartext_modulus.is_power_of_two());
    PrecisionWithPadding {
        value: cleartext_modulus.ilog2() + 1,
    }
}

pub fn expected_pfail_for_precision(
    precision_with_padding: PrecisionWithPadding,
    variance: Variance,
) -> f64 {
    // The additional 1 is to guarantee proper decryption
    let precision_for_proper_decryption: i32 =
        (precision_with_padding.value + 1).try_into().unwrap();
    let correctness_threshold = 2.0f64.powi(-precision_for_proper_decryption);

    let measured_std_dev = variance.get_standard_dev().0;
    let measured_std_score = correctness_threshold / measured_std_dev;

    statrs::function::erf::erfc(measured_std_score / core::f64::consts::SQRT_2)
}

#[test]
fn test_expected_pfail_for_ci_run_filter() {
    // Practical check on a compression-like scenario, of interest because pfail is known to be very
    // low
    let precision_with_padding = precision_with_padding(MessageModulus(1 << 2), CarryModulus(1));
    let theoretical_variance = Variance(1.0216297411906617e-5);

    assert_eq!(
        expected_pfail_for_precision(precision_with_padding, theoretical_variance).log2(),
        -280.4295428516361
    );
}