gebpy 1.1.3__py3-none-any.whl

gebpy/core/rocks/sedimentary.py

@@ -0,0 +1,385 @@
+#!/usr/bin/env python
+# -*-coding: utf-8 -*-
+
+#-----------------------------------------------
+
+# Name:		sedimentary.py
+# Author:	Maximilian A. Beeskow
+# Version:	1.0
+# Date:		15.12.2025
+
+#-----------------------------------------------
+
+"""
+Module: sedimentary.py
+This module controls the generation of the synthetic data of the oxides minerals.
+"""
+
+# PACKAGES
+import yaml
+import numpy as np
+import pandas as pd
+from pathlib import Path
+
+# MODULES
+from ..minerals.synthesis import MineralDataGeneration
+from ..rocks.common import RockGeneration
+
+# Code
+BASE_PATH = Path(__file__).resolve().parents[2]
+DATA_PATH = BASE_PATH / "data_rocks"
+
+class SedimentaryRocks:
+    _yaml_cache = {}
+    _mineralogy_cache = {}
+    _mineral_groups_cache = {}
+    _rocks = {"Sandstone", "Limestone", "Dolostone", "Marl"}
+
+    def __init__(self, name, random_seed, rounding: int = 3) -> None:
+        self.name = name
+        self.random_seed = random_seed
+        self.rng = np.random.default_rng(random_seed)
+        self.current_seed = int(np.round(self.rng.uniform(0, 1000), 0))
+        self.rounding = rounding
+        self.data_path = DATA_PATH
+        self.rock_gen = RockGeneration()
+        self.conversion_factors = self.rock_gen._determine_oxide_conversion_factors()
+        self.cache = {}
+
+    def _load_yaml(self, rock_name: str) -> dict:
+        # 1) Cache-Hit
+        if rock_name in SedimentaryRocks._yaml_cache:
+            return SedimentaryRocks._yaml_cache[rock_name]
+
+        # 2) Laden von Disk
+        yaml_file = self.data_path/f"{rock_name}.yaml"
+        if not yaml_file.exists():
+            raise FileNotFoundError(f"No YAML file found for {rock_name}.")
+
+        with open(yaml_file, "r") as f:
+            data = yaml.safe_load(f)
+
+        if "mineralogy" in data and rock_name not in SedimentaryRocks._mineralogy_cache:
+            self._compile_mineralogy(rock_name, data["mineralogy"])
+
+        if "mineral_groups" in data:
+            self._compile_mineral_groups(rock_name, data["mineral_groups"])
+
+        # 3) Cache schreiben
+        SedimentaryRocks._yaml_cache[rock_name] = data
+
+        return data
+
+    def _compile_mineralogy(self, rock_name: str, mineralogy_dict: dict):
+        """
+        Extracts and compiles all chemistry formulas from the YAML file.
+        Stores the compiled ASTs in the global formula cache.
+        """
+        if rock_name not in SedimentaryRocks._mineralogy_cache:
+            SedimentaryRocks._mineralogy_cache[rock_name] = {}
+
+        for element, entry in mineralogy_dict.items():
+            mineral = element
+            interval = list(entry.values())
+            lower_limit = interval[0]
+            upper_limit = interval[1]
+            compiled = [lower_limit, upper_limit]
+            SedimentaryRocks._mineralogy_cache[rock_name][mineral] = compiled
+
+    def _compile_mineral_groups(self, rock_name: str, group_dict: dict):
+        if rock_name not in SedimentaryRocks._mineral_groups_cache:
+            SedimentaryRocks._mineral_groups_cache[rock_name] = {}
+
+        for group, entry in group_dict.items():
+            minerals = entry["minerals"]
+            min_val = entry["min"]
+            max_val = entry["max"]
+
+            SedimentaryRocks._mineral_groups_cache[rock_name][group] = {
+                "minerals": minerals,
+                "min": min_val,
+                "max": max_val
+            }
+
+    def _sample_mineralogy(self, rock_name: str, number: int):
+        has_groups = rock_name in SedimentaryRocks._mineral_groups_cache
+
+        if not has_groups:
+            # ---- Modus A: flach ----
+            mineral_limits = SedimentaryRocks._mineralogy_cache[rock_name]
+            mins = [v[0] for v in mineral_limits.values()]
+            maxs = [v[1] for v in mineral_limits.values()]
+            minerals = list(mineral_limits.keys())
+
+            comp = self._sample_bounded_simplex_batch(mins, maxs, number)
+            return minerals, comp
+
+        # ---- Modus B: gruppiert ----
+        groups = SedimentaryRocks._mineral_groups_cache[rock_name]
+        mineral_limits = SedimentaryRocks._mineralogy_cache[rock_name]
+
+        # 1️⃣ Gruppen + freie Minerale
+        group_names = list(groups.keys())
+        group_mins = [groups[g]["min"] for g in group_names]
+        group_maxs = [groups[g]["max"] for g in group_names]
+
+        free_minerals = [
+            m for m in mineral_limits
+            if not any(m in groups[g]["minerals"] for g in groups)
+        ]
+
+        free_mins = [mineral_limits[m][0] for m in free_minerals]
+        free_maxs = [mineral_limits[m][1] for m in free_minerals]
+
+        labels = group_names + free_minerals
+        mins = group_mins + free_mins
+        maxs = group_maxs + free_maxs
+
+        # 2️⃣ Top-Level-Sampling
+        top_comp = self._sample_bounded_simplex_batch(mins, maxs, number)
+
+        # 3️⃣ Gruppen intern auflösen
+        mineral_list = []
+        mineral_comp = []
+
+        for i, label in enumerate(labels):
+            frac = top_comp[:, i]
+
+            if label in groups:
+                minerals = groups[label]["minerals"]
+                n = len(minerals)
+                split = self.rng.dirichlet(np.ones(n), size=number)
+
+                for j, m in enumerate(minerals):
+                    mineral_list.append(m)
+                    mineral_comp.append(frac * split[:, j])
+            else:
+                mineral_list.append(label)
+                mineral_comp.append(frac)
+
+        comp = np.vstack(mineral_comp).T
+        return mineral_list, comp
+
+    def _sample_bounded_simplex_batch(self, min_vals, max_vals, number):
+        min_vals = np.asarray(min_vals, dtype=float)
+        max_vals = np.asarray(max_vals, dtype=float)
+        n = len(min_vals)
+
+        # --- Konsistenz ---
+        if min_vals.sum() > 1:
+            raise ValueError("Sum of minimum fractions > 1")
+
+        if max_vals.sum() < 1:
+            raise ValueError("Sum of maximum fractions < 1")
+
+        span = max_vals - min_vals
+        samples = np.zeros((number, n))
+
+        for i in range(number):
+            remaining = 1.0 - min_vals.sum()
+            order = np.arange(n)
+
+            self.rng.shuffle(order)
+            x = np.zeros(n)
+
+            for j in order[:-1]:
+                upper = min(span[j], remaining)
+                val = self.rng.uniform(0, upper)
+                x[j] = val
+                remaining -= val
+
+            x[order[-1]] = remaining
+            samples[i] = min_vals + x
+
+        return samples
+
+    def _collect_mineral_data(self, list_minerals, number):
+        _mineral_data = []
+        for index, mineral in enumerate(list_minerals):
+            data_init = MineralDataGeneration(mineral, number)
+            data_mineral = data_init.generate_data()
+            is_fixed = data_mineral.shape[0] == 1
+            if is_fixed and number > 1:
+                data_mineral = pd.concat([data_mineral]*number, ignore_index=True)
+            elif not is_fixed and data_mineral.shape[0] != number:
+                raise ValueError(
+                    f"Mineral '{mineral}' returned {data_mineral.shape[0]} rows, "
+                    f"but expected {number}.")
+            _mineral_data.append(data_mineral)
+
+        return _mineral_data
+
+    def _extract_mineral_property_data(self, list_minerals, data_mineral, property):
+        arrays = [data_mineral[i][property].to_numpy() for i in range(len(list_minerals))]
+        return np.vstack(arrays)
+
+    def _update_bulk_density_data(self, _bulk_data, density_fluid, number):
+        porosity = _bulk_data["porosity"]
+        _bulk_data["rho_s"] = _bulk_data["rho"]
+        _bulk_data["rho"] = (1 - porosity)*_bulk_data["rho"] + porosity*density_fluid
+        _bulk_data["rho_f"] = np.ones(number)*density_fluid
+
+        return _bulk_data
+
+    def _update_seismic_velocities(self, _bulk_data):
+        val_G = _bulk_data["G"]*10**9
+        val_rho = _bulk_data["rho"]
+        val_K = _bulk_data["K"]*10**9
+        _bulk_data["vS"] = (val_G/val_rho)**0.5
+        _bulk_data["vP"] = ((val_K + 4/3*val_G)/val_rho)**0.5
+        _bulk_data["vP/vS"] = _bulk_data["vP"]/_bulk_data["vS"]
+
+        return _bulk_data
+
+    def _update_elastic_parameter_data(self, _bulk_data):
+        val_G = _bulk_data["G"]*10**9
+        val_K = _bulk_data["K"]*10**9
+        _bulk_data["E"] = ((9*val_K*val_G)/(3*val_K + val_G))*1e-9
+        _bulk_data["poisson"] = (3*val_K - 2*val_G)/(2*(3*val_K + val_G))
+        _bulk_data["lame"] = (val_K - (2*val_G)/3)*1e-9
+
+        return _bulk_data
+
+    def _extract_element_data(self, data_minerals, list_elements):
+        seen = set(list_elements)
+        ordered = list(list_elements)
+
+        for dataset in data_minerals:
+            for key in dataset.keys():
+                if key.startswith("chemistry."):
+                    element = key[len("chemistry."):]
+                    if element not in seen:
+                        seen.add(element)
+                        ordered.append(element)
+
+        return ordered
+
+    def _extract_oxide_data(self, data_minerals, list_oxides):
+        seen = set(list_oxides)
+        ordered = list(list_oxides)
+
+        for dataset in data_minerals:
+            for key in dataset.keys():
+                if key.startswith("compounds."):
+                    oxide = key[len("compounds."):]
+                    if oxide not in seen:
+                        seen.add(oxide)
+                        ordered.append(oxide)
+
+            # Spezialfall Pyrit
+            if dataset["mineral"][0] == "Py":
+                for oxide in ("Fe2O3", "SO3"):
+                    if oxide not in seen:
+                        seen.add(oxide)
+                        ordered.append(oxide)
+                if "FeS2" in seen:
+                    seen.remove("FeS2")
+                    ordered = [o for o in ordered if o != "FeS2"]
+
+        return ordered
+
+    def _update_chemistry_data(self, _bulk_data, data_minerals, data_composition, element, number):
+        n_minerals = len(data_minerals)
+        helper = np.zeros((n_minerals, number))
+        key_element = "chemistry." + element
+
+        for i, dataset in enumerate(data_minerals):
+            if key_element in dataset:
+                helper[i, :] = dataset[key_element].to_numpy()
+
+        bulk_values = np.sum(data_composition * helper.T, axis=1)
+        _bulk_data["w." + element] = bulk_values
+
+        return _bulk_data
+
+    def _update_oxide_data(self, bulk_data, list_oxides):
+        for oxide in list_oxides:
+            cation, anion = self.rock_gen._get_elements_of_compound(compound=oxide)
+            if anion == "O":
+                key_cation = "w." + cation
+                values = self.conversion_factors[oxide]["factor"]*bulk_data[key_cation]
+                key_oxide = "w." + oxide
+                bulk_data[key_oxide] = values
+            else:
+                print("There is a non-oxide compound part of the list.")
+
+        return bulk_data
+
+    def _assign_mineral_amounts(self, bulk_data, data_amounts):
+        for mineral, values in data_amounts.items():
+            key_mineral = "phi." + mineral
+            bulk_data[key_mineral] = values
+
+        return bulk_data
+
+    def generate_dataset(self, number: int = 1, fluid: str = "water", density_fluid=None) -> None:
+        if density_fluid is None:
+            if fluid == "water":
+                density_fluid = 1000
+            elif fluid == "oil":
+                density_fluid = 800
+            elif fluid == "natural gas":
+                density_fluid = 750
+
+        siliciclastics = {"Sandstone"}
+        data_yaml = self._load_yaml(rock_name=self.name)
+        min_porosity = data_yaml["physical_properties"]["porosity"]["min"]
+        max_porosity = data_yaml["physical_properties"]["porosity"]["max"]
+        porosity = self.rng.uniform(min_porosity, max_porosity, number)
+        mineral_limits = SedimentaryRocks._mineralogy_cache[self.name]
+        _limits = {"lower": [], "upper": []}
+
+        for mineral, values in mineral_limits.items():
+            _limits["lower"].append(values[0])
+            _limits["upper"].append(values[1])
+
+        list_minerals = list(SedimentaryRocks._mineralogy_cache[self.name].keys())
+        _properties = ["rho", "vP", "vS", "K", "G", "GR", "PE"]
+        _bulk_data = {}
+        # Collect mineralogical composition data
+        n_minerals = len(list_minerals)
+        _helper_composition = np.zeros((number, n_minerals))
+        _helper_mineral_amounts = {mineral: np.zeros(number) for mineral in list_minerals}
+
+        _helper_composition = self._sample_bounded_simplex_batch(
+            min_vals=_limits["lower"], max_vals=_limits["upper"], number=number)
+        _helper_mineral_amounts = {mineral: _helper_composition[:, j] for j, mineral in enumerate(list_minerals)}
+
+        # Collect mineral data
+        _helper_elements = []
+        _helper_oxides = []
+        _mineral_data = self._collect_mineral_data(list_minerals=list_minerals, number=number)
+        _helper_elements = self._extract_element_data(data_minerals=_mineral_data, list_elements=_helper_elements)
+        _helper_oxides = self._extract_oxide_data(data_minerals=_mineral_data, list_oxides=_helper_oxides)
+        _helper_bulk_data = {}
+        for property in ["rho", "K", "G", "GR", "PE"]:
+            _helper_property = self._extract_mineral_property_data(
+                list_minerals=list_minerals, data_mineral=_mineral_data, property=property)
+            _helper_bulk_data[property] = np.sum(_helper_composition*_helper_property.T, axis=1)
+        # Assign porosity data
+        _helper_bulk_data["porosity"] = porosity
+        # Update bulk density data
+        _helper_bulk_data = self._update_bulk_density_data(
+            _bulk_data=_helper_bulk_data, density_fluid=density_fluid, number=number)
+        # Update bulk seismic velocity data
+        _helper_bulk_data = self._update_seismic_velocities(_bulk_data=_helper_bulk_data)
+        # Update elastic parameter data
+        _helper_bulk_data = self._update_elastic_parameter_data(_bulk_data=_helper_bulk_data)
+        # Update chemistry data
+        for element in _helper_elements:
+            _helper_bulk_data = self._update_chemistry_data(
+                _bulk_data=_helper_bulk_data, data_minerals=_mineral_data, data_composition=_helper_composition,
+                element=element, number=number)
+        # Update oxide data
+        _helper_bulk_data = self._update_oxide_data(bulk_data=_helper_bulk_data, list_oxides=_helper_oxides)
+        # Update rock composition data
+        _helper_bulk_data = self._assign_mineral_amounts(
+            bulk_data=_helper_bulk_data, data_amounts=_helper_mineral_amounts)
+        # Conversion to a pandas dataframe object
+        _bulk_data = pd.DataFrame(_helper_bulk_data)
+
+        return _bulk_data
+
+# TEST
+if __name__ == "__main__":
+    DEFAULT_DATA = SedimentaryRocks(name="Sandstone", random_seed=42).generate_dataset(number=10)

gebpy/data_minerals/albite.yaml

@@ -0,0 +1,59 @@
+mineral: Albite
+group: Tectosilicates
+formula: NaAlSi3O8
+last updated: 10.12.2025
+
+metadata:
+  key: "Ab"
+  crystal_system: "triclinic"
+
+chemistry:
+  Na: 1
+  Al: 1
+  Si: 3
+  O: 8
+
+oxides:
+  Na2O: 1
+  Al2O3: 1
+  SiO2: 1
+
+physical_properties:
+  K:
+    value: 103.45e9
+    ref: "unknown"
+  G:
+    value: 69.0e9
+    ref: "unknown"
+
+cell_data:
+  space_group:
+    value: "C1"
+    ref: "www.handbookofmineralogy.org"
+  a:
+    value: 8.137
+    ref: "www.handbookofmineralogy.org"
+  b:
+    value: 12.785
+    ref: "www.handbookofmineralogy.org"
+  c:
+    value: 7.1583
+    ref: "www.handbookofmineralogy.org"
+  alpha:
+    value: 94.26
+    ref: "www.handbookofmineralogy.org"
+  beta:
+    value: 116.60
+    ref: "www.handbookofmineralogy.org"
+  gamma:
+    value: 87.71
+    ref: "www.handbookofmineralogy.org"
+  Z:
+    value: 4
+    ref: "www.handbookofmineralogy.org"
+
+sources:
+  - author: "John W. Anthony, Richard A. Bideaux, Kenneth W. Bladh, and Monte C. Nichols"
+    title: "Handbook of Mineralogy"
+    publisher: "Mineralogical Society of America, Chantilly, VA"
+    url: "http://www.handbookofmineralogy.org/"

gebpy/data_minerals/anatase.yaml

@@ -0,0 +1,43 @@
+mineral: Anatase
+group: Oxides
+formula: TiO2
+last updated: 10.12.2025
+
+metadata:
+  key: "Anat"
+  crystal_system: "tetragonal"
+
+chemistry:
+  Ti: 1
+  O: 2
+
+oxides:
+  TiO2: 1
+
+physical_properties:
+  K:
+    value: 36.5e9
+    ref: "unknown"
+  G:
+    value: 45.6e9
+    ref: "unknown"
+
+cell_data:
+  space_group:
+    value: "P321"
+    ref: "www.webmineral.com"
+  a:
+    value: 3.793
+    ref: "www.webmineral.com"
+  c:
+    value: 9.52
+    ref: "www.webmineral.com"
+  Z:
+    value: 4
+    ref: "www.webmineral.com"
+
+sources:
+  - author: "John W. Anthony, Richard A. Bideaux, Kenneth W. Bladh, and Monte C. Nichols"
+    title: "Handbook of Mineralogy"
+    publisher: "Mineralogical Society of America, Chantilly, VA"
+    url: "http://www.handbookofmineralogy.org/"

gebpy/data_minerals/ankerite.yaml

@@ -0,0 +1,47 @@
+mineral: Ankerite
+group: Carbonates
+formula: Ca Fe (CO3)2
+last updated: 10.12.2025
+
+metadata:
+  key: "Ank"
+  crystal_system: "trigonal"
+
+chemistry:
+  Ca: 1
+  Fe: 1
+  C: 2
+  O: 6
+
+oxides:
+  CaO: 1
+  FeO: 1
+  CO2: 1
+
+physical_properties:
+  K:
+    value: 73.0e9
+    ref: "unknown"
+  G:
+    value: 32.0e9
+    ref: "unknown"
+
+cell_data:
+  space_group:
+    value: "R3"
+    ref: "www.handbookofmineralogy.org"
+  a:
+    value: 4.83
+    ref: "www.handbookofmineralogy.org"
+  c:
+    value: 16.167
+    ref: "www.handbookofmineralogy.org"
+  Z:
+    value: 3
+    ref: "www.handbookofmineralogy.org"
+
+sources:
+  - author: "unknown"
+    title: "unknown"
+    publisher: "www.handbookofmineralogy.org"
+    url: ["www.handbookofmineralogy.org", "www.webmineral.com"]

gebpy/data_minerals/annite.yaml

@@ -0,0 +1,57 @@
+mineral: Annite
+group: Phyllosilicates
+formula: KFe3AlSi3O10(OH)2
+last updated: 10.12.2025
+
+metadata:
+  key: "Ann"
+  crystal_system: "monoclinic"
+
+chemistry:
+  K: 1
+  Fe: 3
+  Al: 1
+  Si: 3
+  O: 12
+  H: 2
+
+oxides:
+  K2O: 1
+  FeO: 1
+  Al2O3: 1
+  SiO2: 1
+  H2O: 1
+
+physical_properties:
+  K:
+    value: 114.72e9
+    ref: "unknown"
+  G:
+    value: 58.61e9
+    ref: "unknown"
+
+cell_data:
+  space_group:
+    value: "C2/m"
+    ref: "www.handbookofmineralogy.org"
+  a:
+    value: 5.3860
+    ref: "www.handbookofmineralogy.org"
+  b:
+    value: 9.3241
+    ref: "www.handbookofmineralogy.org"
+  c:
+    value: 10.2683
+    ref: "www.handbookofmineralogy.org"
+  beta:
+    value: 100.63
+    ref: "www.handbookofmineralogy.org"
+  Z:
+    value: 2
+    ref: "www.handbookofmineralogy.org"
+
+sources:
+  - author: "John W. Anthony, Richard A. Bideaux, Kenneth W. Bladh, and Monte C. Nichols"
+    title: "Handbook of Mineralogy"
+    publisher: "Mineralogical Society of America, Chantilly, VA"
+    url: "http://www.handbookofmineralogy.org/"

gebpy/data_minerals/anorthite.yaml

@@ -0,0 +1,59 @@
+mineral: Anorthite
+group: Tectosilicates
+formula: CaAl2Si2O8
+last updated: 10.12.2025
+
+metadata:
+  key: "An"
+  crystal_system: "triclinic"
+
+chemistry:
+  Ca: 2
+  Al: 2
+  Si: 2
+  O: 8
+
+oxides:
+  CaO: 1
+  Al2O3: 1
+  SiO2: 1
+
+physical_properties:
+  K:
+    value: 114.72e9
+    ref: "unknown"
+  G:
+    value: 73.72e9
+    ref: "unknown"
+
+cell_data:
+  space_group:
+    value: "P1"
+    ref: "www.handbookofmineralogy.org"
+  a:
+    value: 8.1768
+    ref: "www.handbookofmineralogy.org"
+  b:
+    value: 12.8768
+    ref: "www.handbookofmineralogy.org"
+  c:
+    value: 14.1690
+    ref: "www.handbookofmineralogy.org"
+  alpha:
+    value: 93.17
+    ref: "www.handbookofmineralogy.org"
+  beta:
+    value: 115.85
+    ref: "www.handbookofmineralogy.org"
+  gamma:
+    value: 92.22
+    ref: "www.handbookofmineralogy.org"
+  Z:
+    value: 8
+    ref: "www.handbookofmineralogy.org"
+
+sources:
+  - author: "John W. Anthony, Richard A. Bideaux, Kenneth W. Bladh, and Monte C. Nichols"
+    title: "Handbook of Mineralogy"
+    publisher: "Mineralogical Society of America, Chantilly, VA"
+    url: "http://www.handbookofmineralogy.org/"