]> git.bitcoin.ninja Git - rust-lightning/commitdiff
Scale the success probability of channels without info down by 75%
authorMatt Corallo <git@bluematt.me>
Mon, 24 Apr 2023 22:53:28 +0000 (22:53 +0000)
committerMatt Corallo <git@bluematt.me>
Tue, 19 Sep 2023 21:23:28 +0000 (21:23 +0000)
If we are examining a channel for which we have no information at
all, we traditionally assume the HTLC success probability is
proportional to the channel's capacity. While this may be the case,
it is not the case that a tiny payment over a huge channel is
guaranteed to succeed, as we assume. Rather, the probability of
such success is likely closer to 50% than 100%.

Here we try to capture this by simply scaling the success
probability for channels where we have no information down
linearly. We pick 75% as the upper bound rather arbitrarily - while
50% may be more accurate, its possible it would lead to an
over-reliance on channels which we have paid through in the past,
which aren't necessarily always the best candidates.

Note that we only do this scaling for the historical bucket
tracker, as there we can be confident we've never seen a successful
HTLC completion on the given channel. If we were to apply the same
scaling to the simple liquidity bounds based scoring we'd penalize
channels we've never tried over those we've only ever fails to pay
over, which is obviously not a good outcome.

lightning/src/routing/scoring.rs

index acb29a0d5549bc4fa741b16389abc2fb00210bc9..34a91aa12613c5d58d24f776df068be35b0d4392 100644 (file)
@@ -1007,11 +1007,25 @@ const BASE_AMOUNT_PENALTY_DIVISOR: u64 = 1 << 30;
 /// (recently) seen an HTLC successfully complete over this channel.
 #[inline(always)]
 fn success_probability(
-       amount_msat: u64, min_liquidity_msat: u64, max_liquidity_msat: u64, _capacity_msat: u64,
-       _params: &ProbabilisticScoringFeeParameters, _min_zero_implies_no_successes: bool,
+       amount_msat: u64, min_liquidity_msat: u64, max_liquidity_msat: u64, capacity_msat: u64,
+       _params: &ProbabilisticScoringFeeParameters, min_zero_implies_no_successes: bool,
 ) -> (u64, u64) {
+       debug_assert!(min_liquidity_msat <= amount_msat);
+       debug_assert!(amount_msat < max_liquidity_msat);
+       debug_assert!(max_liquidity_msat <= capacity_msat);
+
        let numerator = max_liquidity_msat - amount_msat;
-       let denominator = (max_liquidity_msat - min_liquidity_msat).saturating_add(1);
+       let mut denominator = (max_liquidity_msat - min_liquidity_msat).saturating_add(1);
+       if min_zero_implies_no_successes && min_liquidity_msat == 0 &&
+               denominator < u64::max_value() / 21
+       {
+               // If we have no knowledge of the channel, scale probability down by ~75%
+               // Note that we prefer to increase the denominator rather than decrease the numerator as
+               // the denominator is more likely to be larger and thus provide greater precision. This is
+               // mostly an overoptimization but makes a large difference in tests.
+               denominator = denominator * 21 / 16
+       }
+
        (numerator, denominator)
 }
 
@@ -2956,47 +2970,47 @@ mod tests {
                        inflight_htlc_msat: 0,
                        effective_capacity: EffectiveCapacity::Total { capacity_msat: 950_000_000, htlc_maximum_msat: 1_000 },
                };
-               assert_eq!(scorer.channel_penalty_msat(42, &source, &target, usage, &params), 4375);
+               assert_eq!(scorer.channel_penalty_msat(42, &source, &target, usage, &params), 6262);
                let usage = ChannelUsage {
                        effective_capacity: EffectiveCapacity::Total { capacity_msat: 1_950_000_000, htlc_maximum_msat: 1_000 }, ..usage
                };
-               assert_eq!(scorer.channel_penalty_msat(42, &source, &target, usage, &params), 2739);
+               assert_eq!(scorer.channel_penalty_msat(42, &source, &target, usage, &params), 4634);
                let usage = ChannelUsage {
                        effective_capacity: EffectiveCapacity::Total { capacity_msat: 2_950_000_000, htlc_maximum_msat: 1_000 }, ..usage
                };
-               assert_eq!(scorer.channel_penalty_msat(42, &source, &target, usage, &params), 2236);
+               assert_eq!(scorer.channel_penalty_msat(42, &source, &target, usage, &params), 4186);
                let usage = ChannelUsage {
                        effective_capacity: EffectiveCapacity::Total { capacity_msat: 3_950_000_000, htlc_maximum_msat: 1_000 }, ..usage
                };
-               assert_eq!(scorer.channel_penalty_msat(42, &source, &target, usage, &params), 1983);
+               assert_eq!(scorer.channel_penalty_msat(42, &source, &target, usage, &params), 3909);
                let usage = ChannelUsage {
                        effective_capacity: EffectiveCapacity::Total { capacity_msat: 4_950_000_000, htlc_maximum_msat: 1_000 }, ..usage
                };
-               assert_eq!(scorer.channel_penalty_msat(42, &source, &target, usage, &params), 1637);
+               assert_eq!(scorer.channel_penalty_msat(42, &source, &target, usage, &params), 3556);
                let usage = ChannelUsage {
                        effective_capacity: EffectiveCapacity::Total { capacity_msat: 5_950_000_000, htlc_maximum_msat: 1_000 }, ..usage
                };
-               assert_eq!(scorer.channel_penalty_msat(42, &source, &target, usage, &params), 1606);
+               assert_eq!(scorer.channel_penalty_msat(42, &source, &target, usage, &params), 3533);
                let usage = ChannelUsage {
                        effective_capacity: EffectiveCapacity::Total { capacity_msat: 6_950_000_000, htlc_maximum_msat: 1_000 }, ..usage
                };
-               assert_eq!(scorer.channel_penalty_msat(42, &source, &target, usage, &params), 1331);
+               assert_eq!(scorer.channel_penalty_msat(42, &source, &target, usage, &params), 3172);
                let usage = ChannelUsage {
                        effective_capacity: EffectiveCapacity::Total { capacity_msat: 7_450_000_000, htlc_maximum_msat: 1_000 }, ..usage
                };
-               assert_eq!(scorer.channel_penalty_msat(42, &source, &target, usage, &params), 1387);
+               assert_eq!(scorer.channel_penalty_msat(42, &source, &target, usage, &params), 3211);
                let usage = ChannelUsage {
                        effective_capacity: EffectiveCapacity::Total { capacity_msat: 7_950_000_000, htlc_maximum_msat: 1_000 }, ..usage
                };
-               assert_eq!(scorer.channel_penalty_msat(42, &source, &target, usage, &params), 1379);
+               assert_eq!(scorer.channel_penalty_msat(42, &source, &target, usage, &params), 3243);
                let usage = ChannelUsage {
                        effective_capacity: EffectiveCapacity::Total { capacity_msat: 8_950_000_000, htlc_maximum_msat: 1_000 }, ..usage
                };
-               assert_eq!(scorer.channel_penalty_msat(42, &source, &target, usage, &params), 1363);
+               assert_eq!(scorer.channel_penalty_msat(42, &source, &target, usage, &params), 3297);
                let usage = ChannelUsage {
                        effective_capacity: EffectiveCapacity::Total { capacity_msat: 9_950_000_000, htlc_maximum_msat: 1_000 }, ..usage
                };
-               assert_eq!(scorer.channel_penalty_msat(42, &source, &target, usage, &params), 1355);
+               assert_eq!(scorer.channel_penalty_msat(42, &source, &target, usage, &params), 3250);
        }
 
        #[test]
@@ -3161,7 +3175,7 @@ mod tests {
                };
 
                // With no historical data the normal liquidity penalty calculation is used.
-               assert_eq!(scorer.channel_penalty_msat(42, &source, &target, usage, &params), 47);
+               assert_eq!(scorer.channel_penalty_msat(42, &source, &target, usage, &params), 168);
                assert_eq!(scorer.historical_estimated_channel_liquidity_probabilities(42, &target),
                        None);
                assert_eq!(scorer.historical_estimated_payment_success_probability(42, &target, 42, &params),
@@ -3169,7 +3183,7 @@ mod tests {
 
                scorer.payment_path_failed(&payment_path_for_amount(1), 42);
                assert_eq!(scorer.channel_penalty_msat(42, &source, &target, usage, &params), 2048);
-               assert_eq!(scorer.channel_penalty_msat(42, &source, &target, usage_1, &params), 128);
+               assert_eq!(scorer.channel_penalty_msat(42, &source, &target, usage_1, &params), 249);
                // The "it failed" increment is 32, where the probability should lie several buckets into
                // the first octile.
                assert_eq!(scorer.historical_estimated_channel_liquidity_probabilities(42, &target),
@@ -3183,7 +3197,7 @@ mod tests {
                // Even after we tell the scorer we definitely have enough available liquidity, it will
                // still remember that there was some failure in the past, and assign a non-0 penalty.
                scorer.payment_path_failed(&payment_path_for_amount(1000), 43);
-               assert_eq!(scorer.channel_penalty_msat(42, &source, &target, usage, &params), 32);
+               assert_eq!(scorer.channel_penalty_msat(42, &source, &target, usage, &params), 105);
                // The first points should be decayed just slightly and the last bucket has a new point.
                assert_eq!(scorer.historical_estimated_channel_liquidity_probabilities(42, &target),
                        Some(([31, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 32, 0, 0, 0, 0, 0],
@@ -3193,17 +3207,17 @@ mod tests {
                // simply check bounds here.
                let five_hundred_prob =
                        scorer.historical_estimated_payment_success_probability(42, &target, 500, &params).unwrap();
-               assert!(five_hundred_prob > 0.66);
-               assert!(five_hundred_prob < 0.68);
+               assert!(five_hundred_prob > 0.59);
+               assert!(five_hundred_prob < 0.60);
                let one_prob =
                        scorer.historical_estimated_payment_success_probability(42, &target, 1, &params).unwrap();
-               assert!(one_prob < 1.0);
-               assert!(one_prob > 0.95);
+               assert!(one_prob < 0.85);
+               assert!(one_prob > 0.84);
 
                // Advance the time forward 16 half-lives (which the docs claim will ensure all data is
                // gone), and check that we're back to where we started.
                SinceEpoch::advance(Duration::from_secs(10 * 16));
-               assert_eq!(scorer.channel_penalty_msat(42, &source, &target, usage, &params), 47);
+               assert_eq!(scorer.channel_penalty_msat(42, &source, &target, usage, &params), 168);
                // Once fully decayed we still have data, but its all-0s. In the future we may remove the
                // data entirely instead.
                assert_eq!(scorer.historical_estimated_channel_liquidity_probabilities(42, &target),
@@ -3367,9 +3381,9 @@ mod tests {
                        inflight_htlc_msat: 0,
                        effective_capacity: EffectiveCapacity::Total { capacity_msat, htlc_maximum_msat: capacity_msat },
                };
-               // With no historical data the normal liquidity penalty calculation is used, which in this
-               // case is diminuitively low.
-               assert_eq!(scorer.channel_penalty_msat(42, &source, &target, usage, &params), 0);
+               // With no historical data the normal liquidity penalty calculation is used, which results
+               // in a success probability of ~75%.
+               assert_eq!(scorer.channel_penalty_msat(42, &source, &target, usage, &params), 1269);
                assert_eq!(scorer.historical_estimated_channel_liquidity_probabilities(42, &target),
                        None);
                assert_eq!(scorer.historical_estimated_payment_success_probability(42, &target, 42, &params),