Error in band_gap_example

[jinwoolee29]

Hello,
I have a question about the band_gap_example.ipynb file in LLMatDesign.

When running band_gap_example.ipynb, an AttributeError occurs related to MDLCalculator.

The MatDeepLearn_dev environment has been fully installed, and the path to the best_checkpoint.pt file in the checkpoint folder's config.yml file has been correctly set. Additionally, the Material Project API key was initially missing, but I have updated it. After running the notebook, the following error message is displayed:

AttributeError: 'MDLCalculator' object has no attribute 'direct_calculate'

Using the dir function to inspect the modules implemented in MDLCalculator, it seems that the issue is caused by the absence of direct_calculate.

I would greatly appreciate it if you could provide guidance on resolving this issue.

Below are the full details of the error message. Thank you.

---

AttributeError Traceback (most recent call last)
Cell In[8], line 16
5 llm = AskLLM("gpt-4o", api_key=api_key, openai_organization=openai_organization)
7 agent = Agent(
8 llm,
9 save_path="outputs/cifs/",
(...)
13 mp_api_key=mp_api_key
14 )
---> 16 out = discover_bandgap(
17 agent,
18 chemical_formula="SrTiO3"
19 )

File C:\MatDeepLearn_dev\LLMatDesign\llmatdesign\core\discover.py:46, in discover_bandgap(agent, chemical_formula, structure, band_gap, target_value, max_iterations)
43 modification_str = get_action(agent.llm, prompt)
44 modification = ast.literal_eval(modification_str)
---> 46 new_structure, new_band_gap = agent.perform_modification(
47 structures_list[-1],
48 modification["Modification"],
49 calculation_type='band_gap'
50 )
52 # get post action reflection
53 reflection_prompt = get_reflection_prompt(
54 structures_list[-1].get_chemical_formula('metal'),
55 new_structure.get_chemical_formula('metal'),
(...)
59 new_band_gap
60 )

File C:\MatDeepLearn_dev\LLMatDesign\llmatdesign\core\agent.py:232, in Agent.perform_modification(self, structure, modification, calculation_type)
229 else:
230 raise ValueError(f"Invalid modification type: {modification_type}")
--> 232 return self.optimize_and_calculate(new_structure, calculation_type=calculation_type)

File C:\MatDeepLearn_dev\LLMatDesign\llmatdesign\core\agent.py:159, in Agent.optimize_and_calculate(self, atoms, calculation_type)
157 return optimized_atoms[0], val
158 elif calculation_type == "band_gap" and self.bandgap_calculator is not None:
--> 159 val = self.bandgap_calculator.direct_calculate(optimized_atoms[0])
160 return optimized_atoms[0], val
161 else:

AttributeError: 'MDLCalculator' object has no attribute 'direct_calculate'