boltzmann9 0.1.7__py3-none-any.whl → 0.1.9__py3-none-any.whl

boltzmann9/project.py

@@ -11,7 +11,7 @@ from typing import Any, Dict
 
 def _get_templates_dir() -> Path:
     """Get the path to the templates directory."""
-    return Path(__file__).parent.parent / "templates"
+    return Path(__file__).parent / "templates"
 
 
 def _load_config_template(data_path: str) -> str:

boltzmann9/templates/__init__.py

@@ -0,0 +1 @@
+# Template files for project creation

boltzmann9/templates/config.py

@@ -0,0 +1,65 @@
+"""RBM experiment configuration."""
+
+config = {
+    "device": "auto",  # "auto" | "cpu" | "cuda:0" | "mps"
+
+    "data": {
+        "csv_path": "data.csv",  # Path to training data CSV
+        "drop_cols": ['x'],
+    },
+
+    "model": {
+        "bm_type": "rbm",   # Currently only RBM is supported
+        "visible_blocks": {"v1": 4, "v2": 4},
+        "hidden_blocks": {"h1": 5, "h":10, "h2": 5},
+        "cross_block_restrictions": [("v1", "h2"), ("v2", "h1")],
+        "initialization": "random",
+    },
+
+    "preprocess": {
+        "q_low": 0.001,
+        "q_high": 0.999,
+        "add_missing_bit": True,
+        "max_categories": 200,
+        "min_category_freq": 2,
+    },
+
+    "dataloader": {
+        "batch_size": 256,
+        "split": [0.8, 0.1, 0.1],
+        "seed": 42,
+        "shuffle_train": True,
+        "num_workers": 0,
+        "drop_last_train": True,
+        "pin_memory": "auto",  # "auto" | True | False
+    },
+
+    "train": {
+        "epochs": 100,
+        "lr": 1e-1,
+        "k": 10,
+        "kind": "mean-field",
+        "momentum": 0.9,
+        "weight_decay": 1e-4,
+        "clip_value": 0.05,
+        "clip_norm": 5.0,
+        "lr_schedule": {"mode": "cosine", "min_lr": 1e-4},
+        "sparse_hidden": True,
+        "rho": 0.1,
+        "lambda_sparse": 0.01,
+        "early_stopping": True,
+        "es_patience": 8,
+    },
+
+    "eval": {
+        "recon_k": 1,
+    },
+
+    "conditional": {
+        "clamp_idx": [0, 1, 2, 3],
+        "target_idx": [4, 5, 6, 7],
+        "n_samples": 100,
+        "burn_in": 500,
+        "thin": 10,
+    },
+}

boltzmann9/templates/synthetic_generator.py

@@ -0,0 +1,147 @@
+#!/usr/bin/env python
+"""Generate synthetic data for this project.
+
+This script is auto-generated from the BoltzmaNN9 template.
+Modify the GeneratorConfig parameters below to customize data generation.
+"""
+
+import sys
+from pathlib import Path
+
+# Add parent directory to path for imports
+sys.path.insert(0, str(Path(__file__).parent.parent.parent))
+
+from src.boltzmann9.data_generator import SyntheticDataGenerator, GeneratorConfig
+
+
+def visualize_data(full_df, generator, save_path, show_from=0, show_to=500):
+    """Create and save visualization of the generated data."""
+    try:
+        import matplotlib.pyplot as plt
+        import numpy as np
+    except ImportError as e:
+        print(f"[visualize_data] matplotlib/numpy import failed: {e}")
+        return
+
+    cfg = generator.config
+    K = cfg.k_bins
+    n_bits = generator.n_bits
+
+    # Clamp range to valid indices
+    show_from = max(0, show_from)
+    show_to = min(len(full_df), show_to)
+
+    fig = plt.figure(figsize=(15, 10))
+
+    # Plot 1: Continuous vs Discretized R
+    plt.subplot(3, 2, 1)
+    plt.plot(full_df["t"][show_from:show_to], full_df["R_continuous"][show_from:show_to],
+             alpha=0.5, label="Continuous", linewidth=1)
+    plt.plot(full_df["t"][show_from:show_to], full_df["R"][show_from:show_to],
+             alpha=0.8, label="Discretized", linewidth=1, drawstyle="steps-post")
+    plt.xlabel("Time")
+    plt.ylabel("R")
+    plt.legend()
+    plt.title("Continuous vs Discretized R")
+    plt.grid(True, alpha=0.3)
+
+    # Plot 2: Histogram of discretized R
+    plt.subplot(3, 2, 2)
+    plt.hist(full_df["R"], bins=K, edgecolor="black", alpha=0.7)
+    plt.xlabel("R (discretized)")
+    plt.ylabel("Frequency")
+    plt.title(f"Distribution across {K} bins")
+    plt.grid(True, alpha=0.3)
+
+    # Plot 3: Bin indices over time
+    plt.subplot(3, 2, 3)
+    plt.plot(full_df["t"][show_from:show_to], full_df["R_bin_index"][show_from:show_to],
+             drawstyle="steps-post")
+    plt.xlabel("Time")
+    plt.ylabel("Bin Index")
+    plt.title("Bin Index over Time")
+    plt.yticks(range(K))
+    plt.grid(True, alpha=0.3)
+
+    # Plot 4: Binary bits over time (heatmap style)
+    plt.subplot(3, 2, 4)
+    binary_matrix = np.array([full_df[f"R_bit_{i}"][show_from:show_to].values for i in range(n_bits)])
+    im = plt.imshow(binary_matrix, aspect="auto", cmap="binary", interpolation="nearest")
+    plt.xlabel("Time")
+    plt.ylabel("Bit Position")
+    plt.title(f"Binary Representation over Time ({n_bits} bits)")
+    plt.yticks(range(n_bits), [f"Bit {i}" for i in range(n_bits)])
+    plt.colorbar(im, label="Bit Value")
+
+    # Plot 5: Bin index distribution
+    plt.subplot(3, 2, 5)
+    bin_counts = full_df["R_bin_index"].value_counts().sort_index()
+    plt.bar(bin_counts.index, bin_counts.values, alpha=0.7, edgecolor="black")
+    plt.xlabel("Bin Index")
+    plt.ylabel("Frequency")
+    plt.title("Bin Index Distribution")
+    plt.xticks(range(K))
+    plt.grid(True, alpha=0.3)
+
+    # Plot 6: Decision variable vs bin index
+    plt.subplot(3, 2, 6)
+    df_valid = full_df[full_df["x"].notna()]
+    decision_by_bin = df_valid.groupby("R_bin_index")["x"].mean()
+    plt.bar(decision_by_bin.index, decision_by_bin.values, alpha=0.7, edgecolor="black")
+    plt.xlabel("Bin Index")
+    plt.ylabel("Mean Decision (x)")
+    plt.title("Average Decision by Bin")
+    plt.xticks(range(K))
+    plt.ylim([0, 1])
+    plt.grid(True, alpha=0.3)
+
+    plt.tight_layout()
+
+    fig.savefig(save_path, dpi=150, bbox_inches="tight")
+    plt.close(fig)
+    print(f"Visualization saved to: {save_path}")
+
+
+def main():
+    """Generate synthetic data and save to data.csv."""
+    # ============================================================
+    # MODIFY THESE PARAMETERS TO CUSTOMIZE DATA GENERATION
+    # ============================================================
+    gen_config = GeneratorConfig(
+        n_samples=5000,
+        dt=0.1,
+        r_min=-2.0,
+        r_max=2.0,
+        k_bins=16,
+        spring_k=5.0,
+        sigma=1.0,
+        eq_interval=100,
+        m0=0.25,
+        sigma_eq=0.0,
+        lookahead=10,
+    )
+    seed = 42
+    show_from = 0
+    show_to = 100
+    # ============================================================
+
+    generator = SyntheticDataGenerator(gen_config)
+    generator.print_info()
+
+    full_df, simplified_df = generator.generate(seed=seed)
+
+    # Save to data folder
+    output_path = Path(__file__).parent / "data.csv"
+    simplified_df.to_csv(output_path, index=False)
+
+    print(f"\nSaved training data to: {output_path}")
+    print(f"  Columns: {list(simplified_df.columns)}")
+    print(f"  Rows: {len(simplified_df)}")
+
+    # Generate visualization
+    plot_path = Path(__file__).parent / "data_visualization.png"
+    visualize_data(full_df, generator, plot_path, show_from=show_from, show_to=show_to)
+
+
+if __name__ == "__main__":
+    main()

{boltzmann9-0.1.7.dist-info → boltzmann9-0.1.9.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: boltzmann9
-Version: 0.1.7
+Version: 0.1.9
 Summary: Restricted Boltzmann Machine implementation in PyTorch
 License: MIT
 Requires-Python: >=3.10

{boltzmann9-0.1.7.dist-info → boltzmann9-0.1.9.dist-info}/RECORD

@@ -7,13 +7,16 @@ boltzmann9/data_generator.py,sha256=aXm8b6O4qoXpDa8TCOdDxF_lDaMtcLNR0P06u2cXb1g,
 boltzmann9/model.py,sha256=npBDd_pow4f9gcsLIpWMmJLBB6sGljtaOluvXcFlLP0,31538
 boltzmann9/pipeline.py,sha256=C3L7tHVakrmvdnW9KYMXhXvscwG7TP8xmVHbunx1E8w,5965
 boltzmann9/preprocessor.py,sha256=USxpMCnC80wp4zO72lPxmFeVdfB4I5NE4P2YlojCVPQ,21758
-boltzmann9/project.py,sha256=Iap-nxhhAZDw2Y4yQg53gFQ0CQPCk2zfmZZC0bTPmxQ,5333
+boltzmann9/project.py,sha256=p24xKk6pj1YxgGCfuJy2zEA8dSdG8zAGEGc6_wdxMl0,5326
 boltzmann9/run_utils.py,sha256=1eRDEkN8j1mG5g9iXArWRA1yQq6pw2agVUGV4Yhy0jk,7823
 boltzmann9/tester.py,sha256=yMYhYKWWiG-LhJQvKv7QmMCYPQLbmtMxFVrJ9iHZVs8,5659
 boltzmann9/utils.py,sha256=8ftNcbV4dz52-t0Ewu4E7CmkkVzPDa6kd8ZnxaBvhMM,1337
 boltzmann9/visualization.py,sha256=iPSytX7GMfmbRbR43xOmybLZBLmy58D4Rp8wasISMcs,3407
-boltzmann9-0.1.7.dist-info/METADATA,sha256=4i9cMFOYamgkAeYOCgNl-2P000hYO_FLa38nPihsSNU,3097
-boltzmann9-0.1.7.dist-info/WHEEL,sha256=wUyA8OaulRlbfwMtmQsvNngGrxQHAvkKcvRmdizlJi0,92
-boltzmann9-0.1.7.dist-info/entry_points.txt,sha256=Y2wY1lqICCjTxuaPgme2QxkoAknrASKmhLljcb5FEyw,51
-boltzmann9-0.1.7.dist-info/top_level.txt,sha256=AE0_urGKOFSgKxXf7WQfjGDROWYppyI4tw1VIg-paQI,11
-boltzmann9-0.1.7.dist-info/RECORD,,
+boltzmann9/templates/__init__.py,sha256=LEL_kwiunM5L5fdMLLJQPucOTDb4mL6uUR0U75eA7Ps,38
+boltzmann9/templates/config.py,sha256=2_GRiJ_92lxSr3ypayAGrFoB58Ekyu264OUmoxytOlM,1568
+boltzmann9/templates/synthetic_generator.py,sha256=cz5PZndsIbCh7Qfl40mXH5_yIBei71MvLLdV1mmbPFc,4858
+boltzmann9-0.1.9.dist-info/METADATA,sha256=hrK5IhuTInrxJJwO1R5CYHJJcGSi16Lj2GZCr37kgs0,3097
+boltzmann9-0.1.9.dist-info/WHEEL,sha256=wUyA8OaulRlbfwMtmQsvNngGrxQHAvkKcvRmdizlJi0,92
+boltzmann9-0.1.9.dist-info/entry_points.txt,sha256=Y2wY1lqICCjTxuaPgme2QxkoAknrASKmhLljcb5FEyw,51
+boltzmann9-0.1.9.dist-info/top_level.txt,sha256=AE0_urGKOFSgKxXf7WQfjGDROWYppyI4tw1VIg-paQI,11
+boltzmann9-0.1.9.dist-info/RECORD,,