Welcome to pollywog’s documentation!

Getting Started

Note

Choose your learning path:

User Guide

Reference

Introduction

pollywog is a Python library designed to make it easy to build, manipulate, and visually inspect Leapfrog-style calculation sets—especially when those sets are large, complex, or involve repetitive logic.

When to Use Pollywog

        flowchart TD
   A{How many calculations?} -->|1-5| B[Use Leapfrog UI]
   A -->|10+| C{Repetitive pattern?}
   C -->|Yes| D[✓ Use pollywog]
   C -->|No| E{Complex logic?}
   E -->|Yes| D
   E -->|No| B

   F{Need version control?} -->|Yes| D
   F -->|No| G{Reuse across projects?}
   G -->|Yes| D
   G -->|No| B

   H{ML model integration?} --> D

   style D fill:#7cb342,stroke:#333,stroke-width:2px,color:#fff
   style B fill:#999,stroke:#333,stroke-width:2px,color:#fff

What is Leapfrog?

Leapfrog is industry-leading 3D geological modeling software developed by Seequent. It is widely used in mining and resource estimation for:

  • Drillhole database management and visualization

  • Geological modeling (surfaces, solids, block models)

  • Grade estimation (kriging, IDW, nearest neighbor) with Leapfrog Edge

  • Resource and reserve calculation

Leapfrog uses calculation sets (.lfcalc files) to define formulas and transformations applied to data. These calculations can become complex when dealing with multiple domains, conditional logic, and multi-commodity resources.

Why pollywog?

Problems pollywog solves:

Challenge

Solution

📝 Large calculation sets

Generate hundreds of calculations programmatically instead of point-and-click

🔀 Complex logic

Use Python’s control flow (loops, conditionals) to build calculations

🔁 Repetitive tasks

Automate patterns with helper functions and templates

🔍 Hard to review

Display calculation trees interactively in Jupyter notebooks

🤖 ML integration

Deploy scikit-learn models directly as Leapfrog calculations

📋 Version control

Keep calculation logic in Git-trackable Python scripts

🧪 Testing

Write unit tests for calculation logic before deploying

By using pollywog, you can:

  • ✅ Automate creation and modification of calculation sets

  • ✅ Refactor and query calculations with code

  • ✅ Export directly to Leapfrog .lfcalc format

  • ✅ Visualize and validate logic interactively

  • ✅ Deploy machine learning models in resource models

  • ✅ Maintain calculation logic in version-controlled scripts

Quick Example

Here’s a simple example showing pollywog’s power – creating a complete resource estimation postprocessing workflow in just a few lines:

from pollywog.core import CalcSet, Number
from pollywog.helpers import WeightedAverage, CategoryFromThresholds

# Build a complete calculation set
calcset = CalcSet([
    # 1. Weighted average across geological domains
    WeightedAverage(
        variables=["Au_oxide", "Au_sulfide", "Au_transition"],
        weights=["prop_oxide", "prop_sulfide", "prop_transition"],
        name="Au_composite",
        comment="Domain-weighted gold grade"
    ),

    # 2. Apply dilution and recovery
    Number(name="Au_diluted", expression=["[Au_composite] * 0.95"],
           comment_equation="5% dilution factor"),
    Number(name="Au_recovered", expression=["[Au_diluted] * 0.88"],
           comment_equation="88% metallurgical recovery"),

    # 3. Classify by grade
    CategoryFromThresholds(
        variable="Au_recovered",
        thresholds=[0.3, 1.0, 3.0],
        categories=["waste", "low_grade", "medium_grade", "high_grade"],
        name="ore_class",
        comment="Material classification"
    ),
])

# Export to Leapfrog - done!
calcset.to_lfcalc("resource_model.lfcalc")

Compare this to manually creating 6+ calculations in Leapfrog’s UI! 🎉

Note

This example creates 6 calculations that would take 10-15 minutes to build manually in Leapfrog. With pollywog, it takes seconds and is version-controlled, testable, and reusable!

Key Features

Core Functionality:

  • Read and write Leapfrog .lfcalc files

  • Programmatically create calculations with Number, Category, Variable, and Filter types

  • Build conditional logic with If/Else statements

  • Query and filter calculation sets like pandas DataFrames

Helper Functions:

  • Sum, Product, Average - Basic mathematical operations

  • WeightedAverage - Domain proportion weighting

  • Scale, Normalize - Data transformations

  • CategoryFromThresholds - Threshold-based classification

Machine Learning:

  • Convert scikit-learn decision trees to calculations

  • Convert random forests to calculation ensembles

  • Convert linear models to equations

  • Full support for both regression and classification

Advanced Features:

  • Topological sorting for dependency resolution

  • Dependency analysis and visualization

  • Interactive display in Jupyter notebooks

  • Calculation validation and testing

Where to Start

Indices and tables