Installation Issues

Solutions for common installation and environment setup problems

Table of Contents

• Overview
• Conda Environment Issues
• Python Package Issues
• GPU/CUDA Issues
• Dependency Conflicts
• Path and Import Issues
• Workspace Setup Issues
• See Also

Overview

Installation issues typically fall into these categories:

Category	Symptoms	Common Causes
Conda	Environment creation fails	Dependency conflicts, channel issues
Packages	Import errors	Missing packages, version mismatches
GPU	CUDA errors, slow performance	Driver issues, PyTorch configuration
Paths	Module not found	PYTHONPATH, working directory
Workspace	Permission errors	Directory permissions, missing folders

Conda Environment Issues

Environment Creation Fails

Symptom:

ResolvePackageNotFound: 
  - package_name=version

Solutions:

1. Update conda:

   conda update -n base conda
   

2. Clear cache and retry:

   conda clean --all
   conda env create -f environment.yml
   

3. Use mamba for faster resolution:

   conda install -n base mamba
   mamba env create -f environment.yml
   

4. Create minimal environment and install manually:

   conda create -n OHMind python=3.10
   conda activate OHMind
   pip install -r requirements.txt  # If available
   

Environment Activation Fails

Symptom:

EnvironmentNameNotFound: Could not find conda environment: OHMind

Solutions:

1. List available environments:

   conda env list
   

2. Check if environment was created:

   conda info --envs
   

3. Recreate if missing:

   conda env create -f environment.yml
   

Slow Environment Creation

Symptom: Environment creation takes very long (>30 minutes)

Solutions:

1. Use mamba:

   conda install -n base mamba
   mamba env create -f environment.yml
   

2. Use libmamba solver:

   conda install -n base conda-libmamba-solver
   conda config --set solver libmamba
   conda env create -f environment.yml
   

Python Package Issues

Import Errors

Symptom:

ModuleNotFoundError: No module named 'langchain'

Solutions:

1. Verify environment is activated:

   conda activate OHMind
   which python  # Should point to OHMind env
   

2. Install missing package:

   pip install langchain langchain-core langgraph
   

3. Check package is installed:

   pip show langchain
   

RDKit Import Fails

Symptom:

ImportError: cannot import name 'Chem' from 'rdkit'

Solutions:

1. Install RDKit via conda:

   conda install -c conda-forge rdkit
   

2. Verify installation:

   python -c "from rdkit import Chem; print('RDKit OK')"
   

PyTorch Import Fails

Symptom:

ImportError: libcudart.so.XX.X: cannot open shared object file

Solutions:

1. Install CPU-only PyTorch:

   pip install torch --index-url https://download.pytorch.org/whl/cpu
   

2. Install CUDA-compatible PyTorch:

   # For CUDA 11.8
   pip install torch --index-url https://download.pytorch.org/whl/cu118
      
   # For CUDA 12.1
   pip install torch --index-url https://download.pytorch.org/whl/cu121
   

Version Mismatch Errors

Symptom:

TypeError: __init__() got an unexpected keyword argument 'xxx'

Solutions:

1. Check package versions:

   pip show langchain langchain-core langgraph
   

2. Install specific versions:

   pip install langchain==0.2.0 langchain-core==0.2.0 langgraph==0.1.0
   

3. Upgrade all related packages:

   pip install --upgrade langchain langchain-core langgraph langchain-openai
   

GPU/CUDA Issues

CUDA Not Available

Symptom:

>>> import torch
>>> torch.cuda.is_available()
False

Solutions:

1. Check NVIDIA driver:

   nvidia-smi
   

2. Check CUDA version:

   nvcc --version
   

3. Install matching PyTorch:

   # Check your CUDA version first
   nvidia-smi | grep "CUDA Version"
      
   # Install matching PyTorch
   pip install torch --index-url https://download.pytorch.org/whl/cu118
   

Out of Memory (OOM)

Symptom:

RuntimeError: CUDA out of memory

Solutions:

1. Reduce batch size in VAE operations

2. Clear GPU cache:

   import torch
   torch.cuda.empty_cache()
   

3. Use CPU for inference:

   device = "cpu"  # Instead of "cuda"
   

Wrong GPU Selected

Symptom: Using wrong GPU or GPU 0 when another is preferred

Solutions:

1. Set CUDA_VISIBLE_DEVICES:

   export CUDA_VISIBLE_DEVICES=1  # Use GPU 1
   

2. Set in Python:

   import os
   os.environ["CUDA_VISIBLE_DEVICES"] = "1"
   

Dependency Conflicts

Conflicting Package Versions

Symptom:

ERROR: pip's dependency resolver does not currently take into account all the packages that are installed.

Solutions:

1. Create fresh environment:

   conda deactivate
   conda env remove -n OHMind
   conda env create -f environment.yml
   

2. Use pip-tools for resolution:

   pip install pip-tools
   pip-compile requirements.in
   pip-sync requirements.txt
   

Pydantic Version Issues

Symptom:

pydantic.errors.PydanticUserError: `xxx` is not a valid Pydantic field type

Solutions:

1. Check Pydantic version:

   pip show pydantic
   

2. Install compatible version:

   pip install "pydantic>=2.0,<3.0"
   

LangChain Compatibility

Symptom: Various LangChain-related errors after upgrade

Solutions:

1. Pin compatible versions:

   pip install langchain==0.2.16 langchain-core==0.2.38 langgraph==0.2.28
   

2. Check migration guide:
   • LangChain has breaking changes between major versions
   • Consult the LangChain migration documentation

Path and Import Issues

Module Not Found

Symptom:

ModuleNotFoundError: No module named 'OHMind'

Solutions:

1. Set PYTHONPATH:

   export PYTHONPATH="/path/to/OHMind:$PYTHONPATH"
   

2. Add to .env file:

   PYTHONPATH=/path/to/OHMind
   

3. Install in development mode:

   cd /path/to/OHMind
   pip install -e .
   

Wrong Python Interpreter

Symptom: Commands work in terminal but not in scripts

Solutions:

1. Check which Python:

   which python
   # Should show: /path/to/anaconda3/envs/OHMind/bin/python
   

2. Use full path in scripts:

   /path/to/anaconda3/envs/OHMind/bin/python script.py
   

3. Activate environment in script:

   #!/bin/bash
   source /path/to/anaconda3/etc/profile.d/conda.sh
   conda activate OHMind
   python script.py
   

Import Order Issues

Symptom: Import works sometimes but not always

Solutions:

1. Check for circular imports

2. Ensure proper package structure:

   OHMind/
   ├── __init__.py
   ├── OHVAE/
   │   └── __init__.py
   └── OHPSO/
       └── __init__.py
   

Workspace Setup Issues

Permission Denied

Symptom:

PermissionError: [Errno 13] Permission denied: '/OHMind_workspace/HEM'

Solutions:

1. Create workspace with correct permissions:

   mkdir -p "$OHMind_workspace"
   chmod 755 "$OHMind_workspace"
   

2. Create subdirectories:

   mkdir -p "$OHMind_workspace"/{HEM,QM,MD,Multiwfn}
   chmod 755 "$OHMind_workspace"/{HEM,QM,MD,Multiwfn}
   

3. Check ownership:

   ls -la "$OHMind_workspace"
   # Change owner if needed
   sudo chown -R $USER:$USER "$OHMind_workspace"
   

Directory Not Found

Symptom:

FileNotFoundError: [Errno 2] No such file or directory: '/OHMind_workspace/HEM'

Solutions:

1. Set environment variable:

   export OHMind_workspace="/path/to/workspace"
   

2. Create directories:

   mkdir -p "$OHMind_workspace"/{HEM,QM,MD,Multiwfn}
   

3. Add to .env:

   OHMind_workspace=/path/to/workspace
   HEM_SAVE_PATH=${OHMind_workspace}/HEM
   QM_WORK_DIR=${OHMind_workspace}/QM
   MD_WORK_DIR=${OHMind_workspace}/MD
   MULTIWFN_WORK_DIR=${OHMind_workspace}/Multiwfn
   

Disk Space Issues

Symptom:

OSError: [Errno 28] No space left on device

Solutions:

1. Check disk space:

   df -h "$OHMind_workspace"
   

2. Clean old results:

   # Remove old QM temp files
   find "$QM_WORK_DIR" -name "temp_*" -mtime +7 -exec rm -rf {} \;
      
   # Remove old MD trajectories
   find "$MD_WORK_DIR" -name "*.xtc" -mtime +30 -exec rm {} \;
   

3. Move workspace to larger disk:

   export OHMind_workspace="/mnt/large_disk/OHMind_workspace"
   

Quick Verification Script

#!/bin/bash
# verify_installation.sh

echo "=== OHMind Installation Verification ==="

# Check conda environment
echo -n "Conda environment: "
if [ "$CONDA_DEFAULT_ENV" = "OHMind" ]; then
    echo "✅ OHMind"
else
    echo "❌ Not in OHMind environment (current: $CONDA_DEFAULT_ENV)"
fi

# Check Python version
echo -n "Python version: "
python_version=$(python --version 2>&1)
if [[ "$python_version" == *"3.10"* ]] || [[ "$python_version" == *"3.11"* ]]; then
    echo "✅ $python_version"
else
    echo "⚠️  $python_version (recommended: 3.10+)"
fi

# Check key packages
echo "Key packages:"
for pkg in langchain langgraph rdkit torch pydantic fastapi; do
    if python -c "import $pkg" 2>/dev/null; then
        version=$(pip show $pkg 2>/dev/null | grep Version | cut -d' ' -f2)
        echo "  ✅ $pkg ($version)"
    else
        echo "  ❌ $pkg (not installed)"
    fi
done

# Check PYTHONPATH
echo -n "PYTHONPATH includes OHMind: "
if python -c "import OHMind" 2>/dev/null; then
    echo "✅"
else
    echo "❌ (set PYTHONPATH or install OHMind)"
fi

# Check workspace
echo -n "Workspace: "
if [ -d "$OHMind_workspace" ] && [ -w "$OHMind_workspace" ]; then
    echo "✅ $OHMind_workspace"
else
    echo "❌ Not configured or not writable"
fi

echo "=== Verification Complete ==="

See Also

• Troubleshooting Overview - Main troubleshooting guide
• MCP Issues - Server connection problems
• Quick Start Guide - Initial setup
• Installation Guide - Detailed installation

---

*Last updated: 2025-12-23

OHMind v0.1.0*