nimo 2.1.2__tar.gz → 2.1.4__tar.gz

{nimo-2.1.2 → nimo-2.1.4}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: nimo
-Version: 2.1.2
+Version: 2.1.4
 Summary: NIMO package
 Author: NIMO developers
 License-Expression: MIT
@@ -11,6 +11,7 @@ Requires-Dist: physbo>=3.1.0
 Requires-Dist: scikit-learn
 Requires-Dist: scipy
 Requires-Dist: pyDOE3
+Requires-Dist: anthropic
 
 # nimo
 
@@ -33,6 +34,7 @@ NIMO is a Python library to realize a closed loop of robotic experiments and art
 - scikit-learn
 - scipy
 - pyDOE3
+- anthropic
 
 # Install
 

{nimo-2.1.2 → nimo-2.1.4}/README.md

@@ -19,6 +19,7 @@ NIMO is a Python library to realize a closed loop of robotic experiments and art
 - scikit-learn
 - scipy
 - pyDOE3
+- anthropic
 
 # Install
 

{nimo-2.1.2 → nimo-2.1.4}/nimo/__init__.py

@@ -23,6 +23,7 @@ from .ai_tools import ai_tool_es
 from .ai_tools import ai_tool_combi
 from .ai_tools import ai_tool_nts
 from .ai_tools import ai_tool_doe
+from .ai_tools import ai_tool_llmep
 
 from .input_tools import preparation_input_standard
 from .input_tools import preparation_input_naree

nimo-2.1.4/nimo/ai_tools/ai_tool_llmep.py

@@ -0,0 +1,346 @@
+import numpy as np
+import csv
+import json
+import re
+import anthropic
+
+
+class LLMEP():
+    """Class of LLMEP
+
+    This class selects the next experimental candidates using a Large Language Model (LLM).
+    It is a standalone class with no dependency on ai_tool_re.
+
+    The user provides a Markdown prompt file that describes the experimental context.
+    The candidates CSV and any observed data are injected automatically into the prompt.
+
+    """
+
+    def __init__(self, input_file, output_file, num_objectives, num_proposals,
+                 prompt_file, system_prompt_file, llm_model,
+                 num_runs, log_file, api_key, max_tokens):
+        """Constructor
+
+        Args:
+            input_file (str): the file for candidates (CSV with descriptor columns + objective columns)
+            output_file (str): the file for proposals output
+            num_objectives (int): the number of objective columns (rightmost columns in CSV)
+            num_proposals (int): the number of proposals to select
+            prompt_file (str): path to the Markdown file describing the experimental context
+            system_prompt_file (str): path to a file containing the system prompt.
+            llm_model (str): Anthropic model ID to use
+            num_runs (int): number of independent LLM selection runs (for majority vote)
+            log_file (str): if provided, save the full selection log as Markdown
+            api_key (str): Anthropic API key. If None, reads from ANTHROPIC_API_KEY env variable.
+            max_tokens (int): maximum number of tokens in the LLM response (default: 8192).
+
+        """
+        self.input_file         = input_file
+        self.output_file        = output_file
+        self.num_objectives     = num_objectives
+        self.num_proposals      = num_proposals
+        self.prompt_file        = prompt_file
+        self.system_prompt_file = system_prompt_file
+        self.model              = llm_model
+        self.num_runs           = num_runs
+        if num_runs == None: self.num_runs = 10
+        self.log_file           = log_file
+        if log_file == None: self.log_file = "log"
+        self.client             = anthropic.Anthropic(api_key = api_key)
+        self.max_tokens         = max_tokens
+        if max_tokens == None: self.max_tokens = 8192
+
+
+    # ------------------------------------------------------------------
+    # Data loading (ported from RE.load_data)
+    # ------------------------------------------------------------------
+
+    def load_data(self):
+        """Load candidates CSV and split into observed / unobserved rows.
+
+        t_train is returned as a list of raw strings for compatibility with
+        calc_ai's signature, but LLMEP does not use it internally.
+
+        Returns:
+            t_train (list[list[str]]): observed objective values (raw strings)
+            X_all (list[list[str]]): all descriptor rows (raw strings; numeric, categorical, and text all accepted)
+            train_actions (np.ndarray): indices of observed rows
+            test_actions (list[int]): indices of unobserved rows
+        """
+        with open(self.input_file, 'r') as f:
+            reader    = csv.reader(f)
+            _headers  = next(reader)
+            raw_rows  = list(reader)
+
+        X_all         = []
+        t_train       = []
+        train_actions = []
+        test_actions  = []
+
+        for idx, row in enumerate(raw_rows):
+            descriptors = row[:-self.num_objectives]  # kept as strings (numeric, categorical, text all accepted)
+            objectives  = row[-self.num_objectives:]
+            X_all.append(descriptors)
+            # A row is "observed" if all objective cells are non-empty
+            if all(v.strip() != '' for v in objectives):
+                t_train.append(objectives)
+                train_actions.append(idx)
+            else:
+                test_actions.append(idx)
+
+        train_actions = np.array(train_actions)
+
+        return t_train, X_all, train_actions, test_actions
+
+
+    # ------------------------------------------------------------------
+    # Internal helpers
+    # ------------------------------------------------------------------
+
+    def _load_prompt(self):
+        """Load the user-supplied Markdown prompt file."""
+        with open(self.prompt_file, 'r', encoding='utf-8', errors='replace') as f:
+            return f.read()
+
+
+    def _build_candidates_text(self, X_all, train_actions, test_actions):
+        """Build a text table of all candidates for injection into the prompt.
+
+        Descriptor values are kept as raw strings, supporting numeric, categorical, and free-text descriptors.
+        Observed rows include their measured objective value; unobserved rows show 'unknown'.
+
+        Returns:
+            str: CSV-like text block
+        """
+        with open(self.input_file, 'r') as f:
+            reader   = csv.reader(f)
+            headers  = next(reader)
+            raw_rows = list(reader)
+
+        descriptor_headers = headers[:-self.num_objectives]
+        objective_headers  = headers[-self.num_objectives:]
+
+        lines = ["index," + ",".join(descriptor_headers) + "," + ",".join(objective_headers)]
+        for idx, row in enumerate(raw_rows):
+            descriptors    = row[:-self.num_objectives]
+            objectives_str = [v if v.strip() != '' else 'unknown' for v in row[-self.num_objectives:]]
+            lines.append(str(idx) + "," + ",".join(descriptors) + "," + ",".join(objectives_str))
+
+        return "\n".join(lines), objective_headers
+
+
+    def _build_messages(self, user_prompt_md, candidates_text, objective_headers):
+        """Construct the system prompt and user message for the single LLM call."""
+        with open(self.system_prompt_file, 'r', encoding='utf-8') as f:
+            system = f.read()
+
+        obj_cols = ", ".join(f"`{h}`" for h in objective_headers)
+        is_are = 'is' if len(objective_headers) == 1 else 'are'
+
+        user = (
+            f"{user_prompt_md}\n"
+            "\n---\n"
+            "\n## Candidates\n"
+            "\nThe following table lists all candidate experimental conditions.\n"
+            f"- Rows with a value in {obj_cols} are already observed data points.\n"
+            f"- Rows where {obj_cols} {is_are} 'unknown' have not yet been measured.\n"
+            "\n```\n"
+            f"{candidates_text}\n"
+            "```\n"
+            "\n---\n"
+            "\n## Your Task\n"
+            f"\nPerform {self.num_runs} independent selection runs, then aggregate by majority vote.\n"
+            "\n### Step 1 Independent Runs\n"
+            f"\nFor each run (Run 1 through Run {self.num_runs}):\n"
+            "- Choose a **distinct exploration strategy**.\n"
+            f"- Select exactly {self.num_proposals} indices from the 'index' column above.\n"
+            f"- Only select rows where {obj_cols} {is_are} 'unknown'.\n"
+            "- Document the strategy and the selected indices.\n"
+
+            "\n### Step 2 Majority Vote\n"
+            f"\nAfter all {self.num_runs} runs:\n"
+            "- Count how many times each index was selected across all runs.\n"
+            "- Rank indices by vote count (descending). Break ties by preferring lower index values.\n"
+            f"- Select the top {self.num_proposals} indices as the final proposals.\n"
+            "- Document the vote counts for all indices that received at least one vote.\n"
+            "\n### Output Format\n"
+            "\nAfter your reasoning, output **exactly one** JSON object in the following format.\n"
+            "This must be the last thing in your response.\n"
+            "\n```json\n"
+            "{\n"
+            '  "runs": [\n'
+            '    {\n'
+            '      "run": 1,\n'
+            '      "strategy": "<one-line description>",\n'
+            '      "selected_indices": [<index1>, <index2>, ...]\n'
+            '    },\n'
+            '    ...\n'
+            '  ],\n'
+            '  "vote_counts": {\n'
+            '    "<index>": <count>,\n'
+            '    ...\n'
+            '  },\n'
+            '  "final_selected_indices": [<index1>, <index2>, ...],\n'
+            '  "reasons": {\n'
+            '    "<index>": "<one-sentence reason why this composition was selected>",\n'
+            '    ...\n'
+            '  }\n'
+            "}\n"
+            "```\n"
+        )
+        return [{"role": "user", "content": user}], system
+
+
+    def _parse_result(self, response_text):
+        """Extract the result JSON from the LLM response.
+
+        Raises:
+            ValueError: if parsing fails
+        """
+        # Try last ```json ... ``` block first
+        json_blocks = re.findall(r'```json\s*(.*?)\s*```', response_text, re.DOTALL)
+        if json_blocks:
+            try:
+                return json.loads(json_blocks[-1])
+            except json.JSONDecodeError:
+                pass
+
+        # Fallback: find the last {...} containing final_selected_indices
+        for m in reversed(list(re.finditer(r'\{', response_text))):
+            candidate = response_text[m.start():]
+            depth, end = 0, -1
+            for i, ch in enumerate(candidate):
+                if ch == '{':
+                    depth += 1
+                elif ch == '}':
+                    depth -= 1
+                    if depth == 0:
+                        end = i
+                        break
+            if end != -1:
+                try:
+                    data = json.loads(candidate[:end + 1])
+                    if "final_selected_indices" in data:
+                        return data
+                except json.JSONDecodeError:
+                    continue
+
+        raise ValueError(
+            f"Could not parse result JSON from LLM response:\n{response_text[:500]}"
+        )
+
+
+    # ------------------------------------------------------------------
+    # Core selection
+    # ------------------------------------------------------------------
+
+    def calc_ai(self, t_train, X_all, train_actions, test_actions):
+        """Select proposals by delegating all runs + majority vote to the LLM.
+
+        Args:
+            t_train (np.ndarray): observed objective values
+            X_all (np.ndarray): all descriptor rows
+            train_actions (np.ndarray): indices of observed rows
+            test_actions (list[int]): indices of unobserved rows
+
+        Returns:
+            actions (list[int]): selected action indices (length = num_proposals)
+        """
+        user_prompt_md  = self._load_prompt()
+        candidates_text, objective_headers = self._build_candidates_text(X_all, train_actions, test_actions)
+
+        print(f"  [LLMEP] Sending request ({self.num_runs} runs + majority vote) ...")
+
+        messages, system = self._build_messages(user_prompt_md, candidates_text, objective_headers)
+
+        response = self.client.messages.create(
+            model=self.model,
+            max_tokens=self.max_tokens,
+            system=system,
+            messages=messages,
+        )
+
+        response_text = response.content[0].text
+        result        = self._parse_result(response_text)
+
+        # Validate: keep only indices that are actually unobserved
+        test_actions_set = set(test_actions)
+        actions = [int(i) for i in result["final_selected_indices"] if int(i) in test_actions_set]
+        reasons = result.get("reasons", {})
+
+        if len(actions) < self.num_proposals:
+            print(f"  [LLMEP] Warning: only {len(actions)} valid indices returned "
+                  f"(expected {self.num_proposals}). Padding with remaining candidates.")
+            used = set(actions)
+            for a in test_actions:
+                if len(actions) >= self.num_proposals:
+                    break
+                if a not in used:
+                    actions.append(a)
+
+        actions = actions[:self.num_proposals]
+
+        if self.log_file:
+            log_lines = [
+                "# LLMEP Selection Log\n",
+                f"Model    : {self.model}",
+                f"Runs     : {self.num_runs}",
+                f"Proposals: {self.num_proposals}\n",
+                "---\n",
+                "## Full LLM Response\n",
+                response_text,
+                "\n---\n",
+                "## Final Selected Indices (Python-validated)\n",
+            ]
+            for rank, action in enumerate(actions, 1):
+                reason = reasons.get(str(action), "")
+                log_lines.append(f"  {rank}. index={action}, descriptors={X_all[action]}, reason={reason}")
+            with open(self.log_file, 'w', encoding='utf-8') as f:
+                f.write("\n".join(log_lines))
+            print(f"  [LLMEP] Log saved to {self.log_file}")
+
+        print(f"  [LLMEP] Final proposals: {actions}")
+        return actions, reasons
+
+
+    # ------------------------------------------------------------------
+    # Entry point (ported from RE.select)
+    # ------------------------------------------------------------------
+
+    def select(self):
+        """Run the full selection pipeline and write proposals CSV.
+
+        Returns:
+            str: "True" on success
+        """
+        print("Start selection of proposals by LLMEP!")
+
+        t_train, X_all, train_actions, test_actions = self.load_data()
+
+        actions, reasons = self.calc_ai(
+            t_train=t_train, X_all=X_all,
+            train_actions=train_actions, test_actions=test_actions,
+        )
+
+        print('Proposals')
+
+        with open(self.input_file, 'r') as f:
+            indexes = f.readlines()[0].rstrip('\n').split(',')
+
+        indexes       = ["actions"] + indexes[:-self.num_objectives]
+        proposals_all = [indexes]
+
+        for i, action in enumerate(actions):
+            row = [str(action)] + [str(v) for v in X_all[action]]
+            proposals_all.append(row)
+            print("###")
+            print("number =", i + 1)
+            print("actions =", action)
+            print("proposal =", X_all[action])
+            print("###")
+
+        with open(self.output_file, 'w', newline="") as f:
+            csv.writer(f).writerows(proposals_all)
+
+        print("Finish selection of proposals!")
+        return "True"

{nimo-2.1.2 → nimo-2.1.4}/nimo/nimo_modules.py

@@ -21,7 +21,10 @@ class selection():
                  combi_ranges = None, spread_elements = None,
                  sample_mode = None,
                  mode = None, max_iter = None,
-                 output_res = None, training_res = None):
+                 output_res = None, training_res = None,
+                 prompt_file = None, system_prompt_file = None,
+                 llm_model = None, num_runs = None,
+                 log_file = None, api_key = None, max_tokens = None):
 
         """Constructor
         
@@ -49,6 +52,13 @@ class selection():
             max_iter (int): max iteration for DOE
             output_res (str): True or False to output res file
             training_res (str): True or False to training res file
+            prompt_file (str): path to the Markdown file describing the experimental context
+            system_prompt_file (str): path to a file containing the system prompt.
+            llm_model (str): Anthropic model ID to use
+            num_runs (int): number of independent LLM selection runs (for majority vote)
+            log_file (str): if provided, save the full selection log as Markdown
+            api_key (str): Anthropic API key. If None, reads from ANTHROPIC_API_KEY env variable.
+            max_tokens (int): maximum number of tokens in the LLM response (default: 8192).
 
         """
 
@@ -80,6 +90,16 @@ class selection():
         self.output_res = output_res
         self.training_res = training_res
 
+
+        self.prompt_file = prompt_file
+        self.system_prompt_file = system_prompt_file
+        self.llm_model = llm_model
+        self.num_runs = num_runs
+        self.log_file = log_file
+        self.api_key = api_key
+        self.max_tokens = max_tokens
+
+
         res = self.module_selection()
         
 
@@ -153,6 +173,13 @@ class selection():
             res = nimo.ai_tools.ai_tool_doe.DOE(self.input_file, self.output_file, 
             self.num_objectives, self.num_proposals, self.mode, self.max_iter).select()
             return res
+        
+
+        if self.method == "LLMEP":
+            res = nimo.ai_tools.ai_tool_llmep.LLMEP(self.input_file, self.output_file, 
+            self.num_objectives, self.num_proposals,
+            self.prompt_file, self.system_prompt_file, self.llm_model,
+            self.num_runs, self.log_file, self.api_key, self.max_tokens).select()
 
 
 class preparation_input():

{nimo-2.1.2 → nimo-2.1.4}/nimo/visualization/plot_distribution.py

@@ -1,12 +1,13 @@
 import numpy as np
 import csv
+import os
 
 import matplotlib.pyplot as plt
 import time
 from mpl_toolkits.mplot3d import Axes3D
 
 
-def plot(input_file, num_objectives, fig_folder = None, dpi = None):
+def plot(input_file, num_objectives, fig_folder = None, filename = None, dpi = None):
     """Loading candidates
 
     This function do not depend on robot.
@@ -17,18 +18,24 @@ def plot(input_file, num_objectives, fig_folder = None, dpi = None):
 
     """
 
-    if fig_folder == None:
+    if fig_folder is None:
         fig_path = "./fig"
     else:
         fig_path = fig_folder
 
     t_train, X_all, train_actions, test_actions = load_data(input_file, num_objectives)
     
-    if dpi == None:
+    if dpi is None:
         dpi = 72
 
     dt_now = time.localtime()
 
+    if filename is None:
+        name = "distribution_" + time.strftime('%y%m%d%H%M%S', dt_now) + ".png"
+    else:
+        name = filename
+
+
     if num_objectives == 1:
         
         fig = plt.figure()
@@ -36,7 +43,7 @@ def plot(input_file, num_objectives, fig_folder = None, dpi = None):
         plt.hist(t_train)
         plt.xlabel("Objectives")
         plt.ylabel("Counts")
-        plt.savefig(fig_path + "/distribution_" + time.strftime('%y%m%d%H%M%S', dt_now) + ".png", dpi = dpi)
+        plt.savefig(os.path.join(fig_path, name), dpi = dpi)
         plt.clf()
         plt.close() 
 
@@ -55,7 +62,7 @@ def plot(input_file, num_objectives, fig_folder = None, dpi = None):
         plt.scatter(x, y)
         plt.xlabel("Objective 1")
         plt.ylabel("Objective 2")
-        plt.savefig(fig_path + "/distribution_" + time.strftime('%y%m%d%H%M%S', dt_now) + ".png", dpi = dpi)
+        plt.savefig(os.path.join(fig_path, name), dpi = dpi)
         plt.clf()
         plt.close() 
 
@@ -78,7 +85,7 @@ def plot(input_file, num_objectives, fig_folder = None, dpi = None):
         ax.set_xlabel("Objective 1")
         ax.set_ylabel("Objective 2")
         ax.set_zlabel("Objective 3")
-        plt.savefig(fig_path + "/distribution_" + time.strftime('%y%m%d%H%M%S', dt_now) + ".png", dpi = dpi)
+        plt.savefig(os.path.join(fig_path, name), dpi = dpi)
         plt.clf()
         plt.close() 
 

{nimo-2.1.2 → nimo-2.1.4}/nimo/visualization/plot_history.py

@@ -1,9 +1,10 @@
 import matplotlib.pyplot as plt
 import time
 import collections
+import os
 
 
-def cycle(input_file, num_cycles, fig_folder = None, dpi = None):
+def cycle(input_file, num_cycles, fig_folder = None, filename = None, dpi = None):
     """Creating the figure of datapoints depending on the cycles
 
     This function do not depend on robot.
@@ -14,12 +15,12 @@ def cycle(input_file, num_cycles, fig_folder = None, dpi = None):
 
     """
     
-    if fig_folder == None:
+    if fig_folder is None:
         fig_path = "./fig"
     else:
         fig_path = fig_folder
     
-    if dpi == None:
+    if dpi is None:
         dpi = 72
 
     obs_itt = []
@@ -34,18 +35,28 @@ def cycle(input_file, num_cycles, fig_folder = None, dpi = None):
 
     for i in range(len(obs_y[0])):
 
+        if filename is None:
+            name = "history_step_" + time.strftime('%y%m%d%H%M%S', dt_now) + "_" + str(i+1)+ ".png"
+        else:
+            if len(obs_y[0]) == 1: # Use the provided name when the number of objectives is 1
+                name = filename
+            else:
+                root, ext = os.path.splitext(filename)
+                name = f"{root}_{i+1}{ext}"
+
+
         fig = plt.figure()
 
         plt.scatter(obs_itt, [r[i] for r in obs_y], alpha=0.7)
         plt.xlim(0, num_cycles)
         plt.xlabel("Cycle")
         plt.ylabel("Objective"+str(i+1))
-        plt.savefig(fig_path + "/history_step_" + time.strftime('%y%m%d%H%M%S', dt_now) + "_" + str(i+1)+ ".png", dpi = dpi)
+        plt.savefig(os.path.join(fig_path, name), dpi = dpi)
         plt.clf()
         plt.close() 
 
 
-def best(input_file, num_cycles, fig_folder = None):
+def best(input_file, num_cycles, fig_folder = None, filename = None, dpi = None):
     """Creating the figure of best datapoints depending on the cycles
 
     This function do not depend on robot.
@@ -56,11 +67,14 @@ def best(input_file, num_cycles, fig_folder = None):
 
     """
     
-    if fig_folder == None:
+    if fig_folder is None:
         fig_path = "./fig"
     else:
         fig_path = fig_folder
     
+    if dpi is None:
+        dpi = 72
+    
     obs_itt = []
     obs_y = []
 
@@ -105,6 +119,15 @@ def best(input_file, num_cycles, fig_folder = None):
             best_y.append(max_list[target_index[j]])
 
 
+        if filename is None:
+            name = "history_best_" + time.strftime('%y%m%d%H%M%S', dt_now) + "_" + str(i+1)+ ".png"
+        else:
+            if len(obs_y[0]) == 1: # Use the provided name when the number of objectives is 1
+                name = filename
+            else:
+                root, ext = os.path.splitext(filename)
+                name = f"{root}_{i+1}{ext}"
+
         fig = plt.figure()
 
         plt.scatter(best_itt, best_y)
@@ -112,6 +135,6 @@ def best(input_file, num_cycles, fig_folder = None):
         plt.xlim(0, num_cycles)
         plt.xlabel("Cycle")
         plt.ylabel("Best objective"+str(i+1))
-        plt.savefig(fig_path + "/history_best_" + time.strftime('%y%m%d%H%M%S', dt_now) + "_" + str(i+1)+ ".png", dpi = dpi)
+        plt.savefig(os.path.join(fig_path, name), dpi = dpi)
         plt.clf()
         plt.close() 

{nimo-2.1.2 → nimo-2.1.4}/nimo/visualization/plot_phase_diagram.py

@@ -1,5 +1,6 @@
 import numpy as np
 import csv
+import os
 
 import matplotlib.pyplot as plt
 import matplotlib.cm as cm
@@ -10,7 +11,8 @@ import sklearn.semi_supervised
 from sklearn.preprocessing import StandardScaler
 from scipy.spatial import distance
 
-def plot(input_file, fig_folder = None, pdc_estimation = None, num_phases = None, dpi = None):
+def plot(input_file, fig_folder = None, pdc_estimation = None, num_phases = None, 
+         filename_phase = None, filename_diagram = None, dpi = None):
     """Loading candidates
 
     This function do not depend on robot.
@@ -20,7 +22,7 @@ def plot(input_file, fig_folder = None, pdc_estimation = None, num_phases = None
 
     """
     
-    if fig_folder == None:
+    if fig_folder is None:
         fig_path = "./fig"
     else:
         fig_path = fig_folder
@@ -29,10 +31,10 @@ def plot(input_file, fig_folder = None, pdc_estimation = None, num_phases = None
 
     LP_algorithm = pdc_estimation
 
-    if pdc_estimation == None:
+    if pdc_estimation is None:
         LP_algorithm = 'LP'
 
-    if dpi == None:
+    if dpi is None:
         dpi = 72
 
     #Preparation of data    
@@ -93,6 +95,13 @@ def plot(input_file, fig_folder = None, pdc_estimation = None, num_phases = None
 
     dt_now = time.localtime()
 
+    if filename_phase is None:
+        filename_phase = "phase_" + time.strftime('%y%m%d%H%M%S', dt_now) + ".png"
+
+    if filename_diagram is None:
+        filename_diagram = "phase_diagram_" + time.strftime('%y%m%d%H%M%S', dt_now) + ".png"
+
+
     if len(data_list[0]) == 2:
        
         with open(input_file, 'r') as f:
@@ -110,7 +119,7 @@ def plot(input_file, fig_folder = None, pdc_estimation = None, num_phases = None
         plt.xlabel(header[0])
         plt.ylabel(header[1])
         
-        plt.savefig(fig_path + "/phase_" + time.strftime('%y%m%d%H%M%S', dt_now) + ".png", dpi = dpi)
+        plt.savefig(os.path.join(fig_path, filename_phase), dpi = dpi)
         plt.clf()
         plt.close() 
         
@@ -126,7 +135,7 @@ def plot(input_file, fig_folder = None, pdc_estimation = None, num_phases = None
         plt.xlabel(header[0])
         plt.ylabel(header[1])
         
-        plt.savefig(fig_path + "/phase_diagram_" + time.strftime('%y%m%d%H%M%S', dt_now) + ".png", dpi = dpi)
+        plt.savefig(os.path.join(fig_path, filename_diagram), dpi = dpi)
         plt.clf()
         plt.close() 
 
@@ -152,7 +161,7 @@ def plot(input_file, fig_folder = None, pdc_estimation = None, num_phases = None
         ax.set_ylabel(header[1])
         ax.set_zlabel(header[2])
      
-        plt.savefig(fig_path + "/phase_" + time.strftime('%y%m%d%H%M%S', dt_now) + ".png", dpi = dpi)
+        plt.savefig(os.path.join(fig_path, filename_phase), dpi = dpi)
         plt.clf()
         plt.close() 
 
@@ -172,7 +181,7 @@ def plot(input_file, fig_folder = None, pdc_estimation = None, num_phases = None
         ax.set_ylabel(header[1])
         ax.set_zlabel(header[2])
         
-        plt.savefig(fig_path + "/phase_diagram_" + time.strftime('%y%m%d%H%M%S', dt_now) + ".png", dpi = dpi)
+        plt.savefig(os.path.join(fig_path, filename_diagram), dpi = dpi)
         plt.clf()
         plt.close() 
 

{nimo-2.1.2 → nimo-2.1.4}/nimo.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: nimo
-Version: 2.1.2
+Version: 2.1.4
 Summary: NIMO package
 Author: NIMO developers
 License-Expression: MIT
@@ -11,6 +11,7 @@ Requires-Dist: physbo>=3.1.0
 Requires-Dist: scikit-learn
 Requires-Dist: scipy
 Requires-Dist: pyDOE3
+Requires-Dist: anthropic
 
 # nimo
 
@@ -33,6 +34,7 @@ NIMO is a Python library to realize a closed loop of robotic experiments and art
 - scikit-learn
 - scipy
 - pyDOE3
+- anthropic
 
 # Install
 

{nimo-2.1.2 → nimo-2.1.4}/nimo.egg-info/SOURCES.txt

@@ -12,6 +12,7 @@ nimo/ai_tools/ai_tool_bomp.py
 nimo/ai_tools/ai_tool_combi.py
 nimo/ai_tools/ai_tool_doe.py
 nimo/ai_tools/ai_tool_es.py
+nimo/ai_tools/ai_tool_llmep.py
 nimo/ai_tools/ai_tool_nts.py
 nimo/ai_tools/ai_tool_pdc.py
 nimo/ai_tools/ai_tool_physbo.py

{nimo-2.1.2 → nimo-2.1.4}/nimo.egg-info/requires.txt

@@ -4,3 +4,4 @@ physbo>=3.1.0
 scikit-learn
 scipy
 pyDOE3
+anthropic

{nimo-2.1.2 → nimo-2.1.4}/pyproject.toml

@@ -7,7 +7,7 @@ packages = ["nimo", "nimo.ai_tools", "nimo.input_tools", "nimo.output_tools", "n
 
 [project]
 name = "nimo"
-version = "2.1.2"
+version = "2.1.4"
 description = "NIMO package"
 readme = "README.md"
 authors = [
@@ -20,5 +20,6 @@ dependencies = [
     "physbo>=3.1.0",
     "scikit-learn",
     "scipy",
-    "pyDOE3"
+    "pyDOE3",
+    "anthropic"
 ]