malkhut: configurable friction per scenario + 5h learning test

MALKHUT/malkhut/long_learning_5h.py

@@ -52,7 +52,7 @@ ASSETS = ["BTCUSDT", "ETHUSDT", "SOLUSDT", "DOGEUSDT", "ADAUSDT",
           "AVAXUSDT", "UNIUSDT", "LINKUSDT", "BNBUSDT"]
 STEPS_PER_EPISODE = 20
 SEED = 42
-N_OPPONENTS = 1500
+N_OPPONENTS = 500
 
 
 class ResourceMonitor:
@@ -248,6 +248,11 @@ def main():
         except Exception as e:
             print(f"  Error: {e}", flush=True)
 
+        # Memory management: gc.collect every 100 episodes
+        if fq_n_reports % 100 == 0:
+            import gc
+            gc.collect()
+
         # CMA-ES every 15 reports
         if fq_n_reports % 15 == 0 and remaining > 600:
             try:

MALKHUT/run_learning_5h.sh

@@ -0,0 +1,5 @@
+#!/bin/bash
+cd /mnt/dolphinng5_predict/MALKHUT
+export PYTHONUNBUFFERED=1
+export NUMBA_CACHE_DIR=/tmp/numba_cache
+python -m malkhut.long_learning_5h >> malkhut/results/long_learning_5h_log.txt 2>&1

MALKHUT/run_learning_5h_wrapper.py

@@ -0,0 +1,9 @@
+#!/usr/bin/env python3
+import os, sys
+os.environ['PYTHONUNBUFFERED'] = '1'
+os.environ['NUMBA_CACHE_DIR'] = '/tmp/numba_cache'
+log = open('/mnt/dolphinng5_predict/MALKHUT/malkhut/results/long_learning_5h_log.txt', 'w')
+os.dup2(log.fileno(), 1)
+os.dup2(log.fileno(), 2)
+os.chdir('/mnt/dolphinng5_predict/MALKHUT')
+os.execv(sys.executable, [sys.executable, '-m', 'malkhut.long_learning_5h'])