Merge pull request #11 from codegithubka/kimon

Kimon

Author: codegithubka

.gitignore

@@ -9,4 +9,5 @@ results/
 # Jupyter Notebook checkpoints
 .ipynb_checkpoints/
 # Data files
-data/
+data/
+.pytest_cache/

benchmarks/benchmark.txt

@@ -0,0 +1,110 @@
+======================================================================
+   PREDATOR-PREY SIMULATION - OPTIMIZATION BENCHMARKS
+======================================================================
+   Time: 2026-01-22 15:11:24
+   Numba: Available
+   Scipy: Available
+======================================================================
+
+======================================================================
+  PP KERNEL BENCHMARK
+======================================================================
+
+  Grid size: 50x50
+--------------------------------------------------
+  Numba (random)                          8.08 ±  2.38 ms
+  Numba (directed)                       18.19 ±  4.53 ms
+    → Directed hunting overhead: +125.2%
+  Python baseline                       474.44 ± 39.77 ms  ( 58.7x vs Numba)
+
+  Grid size: 100x100
+--------------------------------------------------
+  Numba (random)                         25.59 ±  0.23 ms
+  Numba (directed)                       17.68 ±  0.72 ms
+    → Directed hunting overhead: -30.9%
+
+  Grid size: 150x150
+--------------------------------------------------
+  Numba (random)                         93.28 ± 29.97 ms
+  Numba (directed)                       40.18 ±  4.25 ms
+    → Directed hunting overhead: -56.9%
+
+======================================================================
+  PCF COMPUTATION BENCHMARK
+======================================================================
+
+  Grid: 50x50, Prey: 746
+--------------------------------------------------
+  Cell-list PCF (Numba)                   9.99 ±  6.74 ms
+  Brute-force PCF (Python)              835.34 ±  4.51 ms  ( 83.7x vs Cell-list)
+
+  Grid: 75x75, Prey: 1696
+--------------------------------------------------
+  Cell-list PCF (Numba)                   7.56 ±  0.73 ms
+  Brute-force PCF (Python)             4253.93 ±  8.26 ms  (562.5x vs Cell-list)
+
+  Grid: 100x100, Prey: 2946
+--------------------------------------------------
+  Cell-list PCF (Numba)                   7.49 ±  0.35 ms
+
+======================================================================
+  CLUSTER MEASUREMENT BENCHMARK
+======================================================================
+
+  Grid: 50x50, Prey: 746
+--------------------------------------------------
+  Numba flood-fill                        0.06 ±  0.00 ms
+  Scipy label                             0.93 ±  0.07 ms  ( 14.9x vs Numba)
+
+  Grid: 100x100, Prey: 2946
+--------------------------------------------------
+  Numba flood-fill                        0.25 ±  0.00 ms
+  Scipy label                             6.05 ±  0.06 ms  ( 24.6x vs Numba)
+
+  Grid: 150x150, Prey: 6700
+--------------------------------------------------
+  Numba flood-fill                        0.55 ±  0.01 ms
+  Scipy label                            23.30 ±  0.07 ms  ( 42.6x vs Numba)
+
+======================================================================
+  FULL SIMULATION BENCHMARK
+======================================================================
+
+  Grid: 50x50, Steps: 200
+--------------------------------------------------
+  Full sim (random)                      42.15 ±  0.84 ms
+  Full sim (directed)                    19.25 ±  0.14 ms
+    → Throughput: 4744 steps/sec
+
+  Grid: 100x100, Steps: 200
+--------------------------------------------------
+  Full sim (random)                     164.50 ±  4.51 ms
+  Full sim (directed)                    69.01 ±  0.93 ms
+    → Throughput: 1216 steps/sec
+
+======================================================================
+  SCALING ANALYSIS
+======================================================================
+
+  Size     Kernel (ms)     PCF (ms)        Total (ms)     
+-------------------------------------------------------
+  30       1.07            1.21            2.29           
+  64       5.09            4.91            10.00          
+  98       11.62           9.12            20.73          
+  132      21.23           13.70           34.93          
+  166      33.57           20.54           54.11          
+  200      49.26           27.42           76.68          
+
+======================================================================
+  BENCHMARK SUMMARY
+======================================================================
+
+  KEY FINDINGS:
+--------------------------------------------------
+  • Numba kernel speedup:       58.7x (vs Python)
+  • Cell-list PCF speedup:      83.7x (vs brute-force)
+  • Numba cluster speedup:      24.6x (vs scipy)
+  • Directed hunting overhead: -30.9%
+  • Simulation throughput:      1216 steps/sec (100x100)
+
+======================================================================

benchmarks/benchmark_plots.png

@@ -0,0 +1,21 @@
+ 1424296 function calls (1403459 primitive calls) in 1.094 seconds
+
+   Ordered by: internal time
+   List reduced from 3270 to 15 due to restriction <15>
+
+   ncalls  tottime  percall  cumtime  percall filename:lineno(function)
+      500    0.293    0.001    0.298    0.001 /Users/kimonanagnostopoulos/CSS_Project/scripts/numba_optimized.py:309(update)
+     7030    0.077    0.000    0.082    0.000 /Library/Frameworks/Python.framework/Versions/3.12/lib/python3.12/inspect.py:2729(__init__)
+  835/832    0.062    0.000    0.067    0.000 /Users/kimonanagnostopoulos/CSS_Project/.venv/lib/python3.12/site-packages/llvmlite/binding/ffi.py:210(__call__)
+      257    0.049    0.000    0.049    0.000 {method 'read' of '_io.BufferedReader' objects}
+     7547    0.045    0.000    0.095    0.000 /Library/Frameworks/Python.framework/Versions/3.12/lib/python3.12/inspect.py:879(cleandoc)
+       13    0.043    0.003    0.043    0.003 {built-in method _imp.create_dynamic}
+      251    0.028    0.000    0.028    0.000 {built-in method marshal.loads}
+204142/203012    0.020    0.000    0.020    0.000 {built-in method builtins.len}
+     1069    0.015    0.000    0.015    0.000 /Users/kimonanagnostopoulos/CSS_Project/.venv/lib/python3.12/site-packages/scipy/_lib/_docscrape.py:86(is_unindented)
+      456    0.014    0.000    0.014    0.000 {built-in method builtins.dir}
+      227    0.013    0.000    0.047    0.000 /Users/kimonanagnostopoulos/CSS_Project/.venv/lib/python3.12/site-packages/matplotlib/artist.py:1433(get_aliases)
+    10242    0.013    0.000    0.013    0.000 {method 'search' of 're.Pattern' objects}
+   104235    0.013    0.000    0.013    0.000 {method 'startswith' of 'str' objects}
+    92646    0.011    0.000    0.011    0.000 {method 'lstrip' of 'str' objects}
+     1813    0.011    0.000    0.011    0.000 {method 'reduce' of 'numpy.ufunc' objects}

benchmarks/simulation_profile.txt

@@ -1,12 +1,12 @@
 ### Snellius Usage Breakdown
 
 ```
-ssh <your_username>@snellius.surf.nl
+ssh kanagnostopoul@snellius.surf.nl
 
 # On a separate terminal run the following
 
 # Upload the entire project directory (including your models/ folder)
-scp -r ~/Documents/CSS_Project <your_username>@snellius.surf.nl:~/
+scp -r ~/CSS_Project kanagnostopoul@snellius.surf.nl:~/
 
 # On the Snellius terminal
 
@@ -39,9 +39,13 @@ scancel <JOBID>
 
 tail -f logs_<JOBID>.err
 
+# Watch task completetion
+
+watch -n 10 "ls -1 results_JOBID  | wc -l"
+
 # Fetching the results once the job is done
 
-scp -r <your_username>@snellius.surf.nl:~/results_18514601 ~/Downloads/
+scp -r kanagnostopoul@snellius.surf.nl:~/results_18532145 ~/Downloads/
 ```
 
 The jobscript template can be found in ```run_analysis.sh``` (default rome paritition).

docs/HPC_GUIDE.md

@@ -22,4 +22,175 @@ These phases should be completed sequentially, deepening our understanding at ea
 - Create autocorrelation plot of mean population count, following perturbations around the critical point
 	- Look for critical slowing down: perturbations to states closer to the critical point should more slowly return to the steady state
 ### Phase 4: model extensions
-- Investigate whether hydra effect and SOC still occur with diffusion and directed movement
+- Investigate whether hydra effect and SOC still occur with diffusion and directed movement
+
+---
+
+
+## What we are currently collecting:
+
+### 2D Parameter Sweep
+
+We map the full phase space to find:
+- Hydra regions
+- Critical points
+- Coexistence boundaries
+- Evolutionary advantage zones
+
+For now at least we sweep:
+
+```
+prey_birth in [0.10, 0.35]  
+prey_death in [0.001, 0.10]
+```
+
+Metrics Collected (so far):
+
+1. Population Dynamics
+
+```
+
+prey_mean: time-averaged prey pop
+prey_std: variability in prey
+
+# same for predator as above
+
+prey_survived: did prey persist
+pred_survived: did pred perist
+
+```
+
+2. Cluster structure
+
+```
+
+prey_n_clusters: total number of prey clusters
+pred_n_clusters: total number of pred clusters
+prey_tau: power law exp
+prey_s_c: cutoff cluster sizes
+pred_tau: pred cluster exp
+pred_s_c: pred cutoff
+
+```
+
+3. Order Parameters
+
+```
+prey_largest_fraction_mean
+prey_largest_fraction_std
+pred_largest_fraction_mean
+prey_percolation_prob: fraction of samples with spanning cluster
+pred_percolation_prob: predator percolation prob
+
+```
+
+
+4. Spatial Correlations
+
+```
+
+pcf_distances: distance bins in lattice units
+pcf_prey_prey_mean: prey-prey correlation function
+pcf_pred_pred_mean
+pcf_prey_pred_mean
+segregation_index: measure spatial mixing
+prey_clustering_index: short range prey clustering
+pred_clustering_index
+
+```
+
+5. Evolutionary dynamics
+
+```
+evolved_prey_death_mean: time avg evolved mortality rate
+evolved_prey_death_std
+evolved_prey_death_final
+evolve_sd: mutation strength used
+
+```
+
+---
+
+### Finite-size scaling
+
+We choose a fixed point identified in the main simulation run ```(target_prey_birth, target_prey_death)``` ideally near hydra boundary.
+
+
+For selected grid sizes (TBD) we run independent reps for each size.
+
+
+Metrics:
+
+```
+grid_size
+prey_mean, prey_std
+prey_survived: bool
+prey_largest_fraction: order parameter
+prey_percolation_prob
+prey_tau: grid size dependent exponent
+prey_tau_se: SE on tau
+prey_s_c: cutoff scales
+```
+
+---
+
+### Evo Sensitivity
+
+How does mutation strength affect evolutionary advantage in Hydra regions, speed of adaptation and final evolved mortality rates.
+
+Again. choose fixed point identified from main analysis.
+
+Metrics Dict:
+
+```
+prey_mean: in cell units as the below metrics as well
+prey_std
+pred_mean
+pred_std
+prey_survived: bool
+
++ same cluster metrics and spatial correlation metrics
+
+
+evolved_prey_death_mean: avg mortality across all prey
+evolved_prey_death_std
+evolved_prey_death_final
+evolve_sd
+```
+
+
+## Additions Required:
+
+1. Temporal dynamics for time series analysis. Needed to add critical slowing down effect near phase transitions.
+
+```
+result["prey_timeseries"] = prey_pops[::10]  # Subsample every 10 steps
+result["pred_timeseries"] = pred_pops[::10]
+
+```
+
+```
+def run_perturbation_experiment(...):
+    # Save full time series only for these special runs
+```
+
+2. Snapshots of spatial configurations. This is a costly operation so we need to figure out how and when to do it in the sim.
+
+3. Saving final grid states?
+
+```
+result["final_grid"] = model.grid.copy()
+```
+
+
+
+
+
+
+
+
+
+
+
+
+