SearchLibrium 0.0.145__tar.gz → 0.0.147__tar.gz

{searchlibrium-0.0.145/src/SearchLibrium.egg-info → searchlibrium-0.0.147}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: SearchLibrium
-Version: 0.0.145
+Version: 0.0.147
 Summary: A Python package for econometric models driven by search
 Author: Mario Fernando Taco Morales, Alexander Paz, Prithvi Beeramole, Robert Burdett
 Author-email: Zeke Ahern <z.ahern@qut.edu.au>

{searchlibrium-0.0.145 → searchlibrium-0.0.147}/pyproject.toml

@@ -82,7 +82,7 @@ Documentation = "https://pypi.org/project/SearchLibrium/"
 realpython = "SearchLibrium.__main__:main"
 
 [tool.bumpver]
-current_version = "0.0.145"
+current_version = "0.0.147"
 version_pattern = "MAJOR.MINOR.PATCH"
 commit_message = "[skip ci] Bump version {old_version} -> {new_version}"
 commit = true

{searchlibrium-0.0.145 → searchlibrium-0.0.147}/src/SearchLibrium/search.py

@@ -87,6 +87,7 @@ except ImportError:
     from .ordered_logit import OrderedLogitLong
     from . import misc
     from .multinomial_nested import NestedLogit, MultiLayerNestedLogit
+    from addicty import Dict
 
 ''' ---------------------------------------------------------- '''
 ''' CONSTANTS                                                  '''
@@ -788,8 +789,9 @@ class Parameters:
         # TODO I Think we could initialise it this way more effictively
         acceptable_keys = [
             'LCR', 'verbose', 'asc_ind', 'nests', 'lambdas', 'varnest',
-            '_jax', 'all_sig', 'de_init', 'de_popsize', 'de_maxiter',            
-            'de_tol', 'de_polish', 'sd_penalty', 'halton_opts', 'latent_class'
+            '_jax', 'all_sig', 'de_init', 'de_popsize', 'de_maxiter',
+            'de_tol', 'de_polish', 'sd_penalty', 'halton_opts', 'latent_class',
+            'num_classes',
         ]
 
         # Assign all kwargs to self, but only if the key is in the acceptable_keys list
@@ -1985,12 +1987,15 @@ class Search():
         if model_n == 'nested_logit':
             all_vars = list(set(asvars+isvars))
             logging.info('prereqs')
-            if 'ps_alt_vars' in self.param.pres_spec_constr:
-                ps_alt_vars = self.param.pres_spec_constr['ps_alt_vars']
+            # pres_spec_constr defaults to None when no constraints are supplied
+            # at all -- 'x in None' raises TypeError, so guard with `or {}`.
+            _psc = self.param.pres_spec_constr or {}
+            if 'ps_alt_vars' in _psc:
+                ps_alt_vars = _psc['ps_alt_vars']
             else:
                 ps_alt_vars = None
-            if 'ps_nest_vars' in self.param.pres_spec_constr:
-                ps_nest_vars = self.param.pres_spec_constr['ps_nest_vars']
+            if 'ps_nest_vars' in _psc:
+                ps_nest_vars = _psc['ps_nest_vars']
             else:
                 ps_nest_vars = None
             state = Dict({'all_vars': all_vars,
@@ -3749,8 +3754,8 @@ class Search():
     ''' and cache the result if it converged.                              '''
     ''' ------------------------------------------------------------------ '''
 
-    def evaluate_solution(self, sol):
-       
+    def evaluate_solution(self, sol, track_best=True):
+
         sig = self.setup_signature(sol)
 
         if sig not in self.explored_specs:
@@ -3763,7 +3768,7 @@ class Search():
                 sol.update_objective(i, score)
             return sol, True
 
-        sol, converged = self._evaluate_solution(sol)
+        sol, converged = self._evaluate_solution(sol, track_best=track_best)
 
         if converged:
             self.evaluated_solutions[sig] = [sol.obj(i) for i in range(self.nb_crit)]
@@ -3780,7 +3785,7 @@ class Search():
     ''' re-estimated. The function returns the estimated solution  '''
     ''' only if it converges.                                      '''
     ''' ---------------------------------------------------------- '''
-    def _evaluate_solution(self, sol):
+    def _evaluate_solution(self, sol, track_best=True):
     # {
         # apply_constraints() was previously only wired into evaluate_lc/
         # evaluate_nested_logit/evaluate_mixed_nested — plain multinomial/mixed
@@ -3836,7 +3841,7 @@ class Search():
             sol['converged'] = True
             self.update_objectives(self.param.criterions, sol)
 
-            if not hasattr(self, 'best_solution') or self.find_best_sol([sol, self.best_solution]) == sol:
+            if track_best and (not hasattr(self, 'best_solution') or self.find_best_sol([sol, self.best_solution]) == sol):
                 self.best_solution = sol
                 if self.last_printed_solution is None or not self.solutions_equal(sol, self.last_printed_solution):
                     self.print_best_solution(sol)
@@ -4404,7 +4409,10 @@ class Search():
         if len(all_vars) == 0:
             raise ValueError('need a variable: todo debug why')
         all_vars = [var for var in self.param.varnames if var in all_vars]
-        nest_vars = [var for var in self.param.varnest if var in all_vars]
+        # varnest (nest-level utility variables) is optional and has no default
+        # in Parameters.__init__ -- most nested_logit specs don't use nest-level
+        # covariates at all, so treat missing/None as "no nest variables".
+        nest_vars = [var for var in (getattr(self.param, 'varnest', None) or []) if var in all_vars]
 
         X, y = self.param.df[all_vars].values, self.param.choices
         X_nest = self.param.df[nest_vars]

{searchlibrium-0.0.145 → searchlibrium-0.0.147}/src/SearchLibrium/siman.py

@@ -516,7 +516,10 @@ class SA(Search):
         for attempt in range(1, N_trials + 1):
             sol = self.generate_solution()
             sol = self.repair_solution_for_clarity(sol)
-            sol, converged = self.evaluate(sol)
+            # track_best=False: this is a disposable calibration trial, not
+            # part of the accepted search trajectory, so it must not update
+            # (or print) the run's global best solution.
+            sol, converged = self.evaluate(sol, track_best=False)
             if converged:
                 base_sol = sol
                 break
@@ -579,7 +582,10 @@ class SA(Search):
                 print(f"Trial {trial}: Could not generate a new proposed solution, skipping temperature calculation.")
                 continue
 
-            proposed_sol, converged = self.evaluate(proposed_sol)
+            # track_best=False: same reason as above — these proposals are
+            # only used to estimate delta_E for the starting temperature and
+            # are discarded immediately after, they never become current_sol.
+            proposed_sol, converged = self.evaluate(proposed_sol, track_best=False)
             if not converged:
                 print(f"Trial {trial}: Proposed solution did not converge, skipping temperature calculation.")
                 continue
@@ -639,10 +645,10 @@ class SA(Search):
     ''' ---------------------------------------------------------- '''
     ''' Function.  Evaluate how good a solution is                 '''
     ''' ---------------------------------------------------------- '''
-    def evaluate(self, sol):
+    def evaluate(self, sol, track_best=True):
     # {
-        sol, converged = self.evaluate_solution(sol)
-        
+        sol, converged = self.evaluate_solution(sol, track_best=track_best)
+
         return sol, converged
     # }
 

{searchlibrium-0.0.145 → searchlibrium-0.0.147}/src/SearchLibrium/threshold.py

@@ -367,7 +367,7 @@ class TA(Search):
         if self.nb_crit == 1:
         # {
             if is_better(sol.obj(0), self.best_sol.obj(0), self.param.sign_crit(0)):
-                self.best_sol.copy_solution(sol)
+                self.best_sol = self.copy_solution(sol)
                 self.no_impr = 0
         # }
         else:

searchlibrium-0.0.147/src/SearchLibrium/version.txt

@@ -0,0 +1 @@
+0.0.147

{searchlibrium-0.0.145 → searchlibrium-0.0.147/src/SearchLibrium.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: SearchLibrium
-Version: 0.0.145
+Version: 0.0.147
 Summary: A Python package for econometric models driven by search
 Author: Mario Fernando Taco Morales, Alexander Paz, Prithvi Beeramole, Robert Burdett
 Author-email: Zeke Ahern <z.ahern@qut.edu.au>

searchlibrium-0.0.145/src/SearchLibrium/version.txt

@@ -1 +0,0 @@
-0.0.145