用快速哈希优化回测内部索引

crates/fidc-core/src/data.rs

@@ -2,6 +2,7 @@ use std::borrow::Cow;
 use std::collections::{BTreeMap, HashMap, HashSet};
 use std::sync::{Arc, OnceLock};
 
+use ahash::AHashMap;
 use chrono::{NaiveDate, NaiveDateTime};
 use rayon::prelude::*;
 use serde::{Deserialize, Serialize};
@@ -1136,12 +1137,12 @@ pub struct DataSet {
     execution_quotes_by_date: HashMap<NaiveDate, HashMap<String, Vec<IntradayExecutionQuote>>>,
     order_book_depth_index: HashMap<(NaiveDate, String), Vec<IntradayOrderBookDepthLevel>>,
     benchmark_by_date: BTreeMap<NaiveDate, BenchmarkSnapshot>,
-    market_series_by_symbol: Arc<HashMap<String, Arc<SymbolPriceSeries>>>,
-    adjusted_close_series_by_symbol: Arc<HashMap<String, Arc<AdjustedCloseSeries>>>,
+    market_series_by_symbol: Arc<AHashMap<String, Arc<SymbolPriceSeries>>>,
+    adjusted_close_series_by_symbol: Arc<AHashMap<String, Arc<AdjustedCloseSeries>>>,
     market_series_by_symbol_id: Arc<Vec<Option<Arc<SymbolPriceSeries>>>>,
     adjusted_close_series_by_symbol_id: Arc<Vec<Option<Arc<AdjustedCloseSeries>>>>,
     benchmark_series_cache: BenchmarkPriceSeries,
-    symbol_id_by_code: Arc<HashMap<String, u32>>,
+    symbol_id_by_code: Arc<AHashMap<String, u32>>,
     eligible_universe_by_date: Arc<OnceLock<BTreeMap<NaiveDate, Vec<EligibleUniverseSnapshot>>>>,
     benchmark_code: String,
     futures_params_by_symbol: HashMap<String, Vec<FuturesTradingParameter>>,
@@ -1305,27 +1306,32 @@ impl DataSet {
 
         let mut factor_by_date = group_arc_by_date(&factors, |item| item.date);
         sort_arc_groups_by_symbol(&mut factor_by_date, |item| item.symbol.as_str());
-        let mut market_rows_by_symbol = HashMap::<String, Vec<&DailyMarketSnapshot>>::new();
+        let mut market_rows_by_symbol = AHashMap::<String, Vec<&DailyMarketSnapshot>>::new();
         for row in &market {
             market_rows_by_symbol
                 .entry(row.symbol.clone())
                 .or_default()
                 .push(row.as_ref());
         }
+        let market_rows_by_symbol = market_rows_by_symbol.into_iter().collect::<Vec<_>>();
         let market_series_by_symbol = market_rows_by_symbol
             .into_par_iter()
             .map(|(symbol, rows)| {
                 let series = Arc::new(SymbolPriceSeries::new(symbol.clone(), rows));
                 (symbol, series)
             })
-            .collect::<HashMap<_, _>>();
+            .collect::<Vec<_>>()
+            .into_iter()
+            .collect::<AHashMap<_, _>>();
         let adjusted_close_series_by_symbol = market_series_by_symbol
             .par_iter()
             .filter_map(|(symbol, market)| {
                 AdjustedCloseSeries::new(market, &factor_by_date)
                     .map(|series| (symbol.clone(), Arc::new(series)))
             })
-            .collect::<HashMap<_, _>>();
+            .collect::<Vec<_>>()
+            .into_iter()
+            .collect::<AHashMap<_, _>>();
         let factor_texts = factor_texts
             .into_iter()
             .filter_map(|mut item| {
@@ -3298,7 +3304,7 @@ fn build_symbol_id_index(
     market_by_date: &BTreeMap<NaiveDate, Vec<Arc<DailyMarketSnapshot>>>,
     factor_by_date: &BTreeMap<NaiveDate, Vec<Arc<DailyFactorSnapshot>>>,
     candidate_by_date: &BTreeMap<NaiveDate, Vec<Arc<CandidateEligibility>>>,
-) -> HashMap<String, u32> {
+) -> AHashMap<String, u32> {
     let mut symbols = instruments.keys().cloned().collect::<HashSet<_>>();
     for rows in market_by_date.values() {
         for row in rows {
@@ -3337,7 +3343,7 @@ fn build_symbol_id_index(
 
 fn build_group_symbol_ids<T, F>(
     groups: &BTreeMap<NaiveDate, Vec<Arc<T>>>,
-    symbol_id_by_code: &HashMap<String, u32>,
+    symbol_id_by_code: &AHashMap<String, u32>,
     symbol_of: F,
 ) -> BTreeMap<NaiveDate, Vec<u32>>
 where