Merging Devel

.github/workflows/ci_chemistry.yml

@@ -97,6 +97,7 @@ jobs:
         python -m pip install --upgrade pip
         python -m pip install -r requirements.txt
         python -m pip install -e .
+        python -m pip install pyscf
         cd tests
         ls
         pytest test_chemistry.py

README.md

@@ -38,9 +38,7 @@ Currently you need to compile from a separate [fork](https://github.com/kottmanj
 See the github page of this fork for installation instruction.
 Here is a small [tutorial](https://github.com/aspuru-guzik-group/tequila-tutorials/blob/main/ChemistryMadnessInterface.ipynb) that illustrates the usage.
 
-
 # Install from source
-
 **Do not** install like this: (Minecraft lovers excluded)
 <strike>`pip install tequila`</strike>
 
@@ -141,14 +139,14 @@ A Feasible Approach for Automatically Differentiable Unitary Coupled-Cluster on
 Chemical Science, 2021, [doi.org/10.1039/D0SC06627C](https://doi.org/10.1039/D0SC06627C).  
 [arxiv.org/abs/2011.05938](https://arxiv.org/abs/2011.05938)  
 General techniques are implemented in the chemistry modules of tequila.  
-See the [tutorials](https://github.com/aspuru-guzik-group/tequila/tree/master/tutorials) for examples.  
+See the [tutorials](https://github.com/aspuru-guzik-group/tequila-tutorials) for examples.  
 
 J.S. Kottmann, P. Schleich, T. Tamayo-Mendoza, A. Aspuru-Guzik.  
 Reducing Qubit Requirements while Maintaining Numerical Precision for the Variational Quantum Eigensolver: A Basis-Set-Free Approach.  
 J.Phys.Chem.Lett., 2021, [doi.org/10.1021/acs.jpclett.0c03410](https://doi.org/10.1021/acs.jpclett.0c03410).  
 [arxiv.org/abs/2008.02819](https://arxiv.org/abs/2008.02819)  
-[example code](https://github.com/aspuru-guzik-group/tequila/blob/master/tutorials/ChemistryBasisSetFreeVQE.ipynb)  
-[tutorial on the madness interface](https://github.com/aspuru-guzik-group/tequila/blob/master/tutorials/ChemistryMadnessInterface.ipynb)  
+[example code](https://github.com/aspuru-guzik-group/tequila-tutorials/blob/main/ChemistryBasisSetFreeVQE.ipynb)  
+[tutorial on the madness interface](https://github.com/aspuru-guzik-group/tequila-tutorials/blob/main/ChemistryMadnessInterface.ipynb)  
 
 A. Cervera-Lierta, J.S. Kottmann, A. Aspuru-Guzik.  
 The Meta-Variational Quantum Eigensolver.  
@@ -167,10 +165,10 @@ Natural Evolutionary Strategies for Variational Quantum Computation.
 J. S. Kottmann, A. Aspuru-Guzik,  
 Optimized Low-Depth Quantum Circuits for Molecular Electronic Structure using a Separable Pair Approximation,  
 [arxiv.org/abs/2105.03836](https://arxiv.org/abs/2105.03836)  
-example code coming soon  
-implementation currently only on [devel](https://github.com/aspuru-guzik-group/tequila/tree/devel) branch.  
+[example code](https://github.com/aspuru-guzik-group/tequila-tutorials/blob/main/ChemistrySeparablePairAnsatz.ipynb)    
+currently needs the [devel](https://github.com/aspuru-guzik-group/tequila/tree/devel) branch of tequila  
 
-Let us know, if you want your research project to be included in this list!
+Let us know, if you want your research project and/or tutorial to be included in this list!
 
 # Dependencies
 Support for additional optimizers or quantum backends can be activated by intalling them in your environment.

src/tequila/apps/adapt/adapt.py

@@ -5,6 +5,7 @@
 import numpy
 import dataclasses
 import warnings
+from itertools import combinations
 
 
 @dataclasses.dataclass
@@ -139,7 +140,7 @@ def __call__(self, static_variables = None, mp_pool=None, label=None, variables=
                 warnings.warn("variable {} of initial objective not given, setting to 0.0 and activate optimization".format(k), TequilaWarning)
                 variables[k] = 0.0
 
-        energy = 0.0
+        energy = simulate(initial_objective, variables=variables)
         for iter in range(self.parameters.maxiter):
             current_label = (iter,0)
             if label is not None:
@@ -155,7 +156,7 @@ def __call__(self, static_variables = None, mp_pool=None, label=None, variables=
                 print("pool gradient norm is {:+2.8f}, convergence criterion met".format(grad_norm))
                 break
             if numpy.abs(max_grad) < self.parameters.max_gradient_convergence:
-                print("max pool gradient is {:+2.8f}, convergence criterion met".format(grad_norm))
+                print("max pool gradient is {:+2.8f}, convergence criterion |max(grad)|<{} met".format(max_grad, self.parameters.max_gradient_convergence))
                 break
 
             batch_size = self.parameters.batch_size
@@ -291,19 +292,18 @@ def __init__(self, molecule, indices:str):
         self.molecule = molecule
 
         if isinstance(indices, str):
-            if indices.upper() == "UPCCGSD":
-                indices = self.make_indices_doubles(generalized=True)
-                indices += self.make_indices_singles(generalized=True)
-            elif indices.upper() == "UPCCSD":
-                indices = self.make_indices_doubles(generalized=False)
-                indices += self.make_indices_singles(generalized=False)
-            elif indices.upper() == "UPCCD":
-                indices = self.make_indices_doubles(generalized=False)
-            elif indices.upper() == "UPCCGD":
-                indices = self.make_indices_doubles(generalized=True)
-            else:
+            if not "CC" in indices.upper():
                 raise TequilaException("Pool of type {} not yet supported.\nCreate your own by passing the initialized indices".format(indices))
 
+            generalized = True if "G" in indices.upper() else False
+            paired = True if "P" in indices.upper() else False
+            singles = True if "S" in indices.upper() else False
+            doubles = True if "D" in indices.upper() else False
+
+            indices = []
+            if doubles: indices += self.make_indices_doubles(generalized=generalized, paired=paired)
+            if singles: indices += self.make_indices_singles(generalized=generalized)
+
         indices = [tuple(k) for k in indices]
         super().__init__(generators=indices)
 
@@ -323,11 +323,9 @@ def make_indices_singles(self, generalized=False):
                     continue
                 indices.append([(2*p, 2*q)])
                 indices.append([(2*p+1, 2*q+1)])
-        return indices
+        return self.sort_and_filter_unique_indices(indices)
 
     def make_indices_doubles(self, generalized=False, paired=True):
-        # only paired doubles supported currently
-        assert paired
         indices = []
         for p in range(self.molecule.n_electrons//2):
             for q in range(self.molecule.n_electrons//2, self.molecule.n_orbitals):
@@ -343,7 +341,33 @@ def make_indices_doubles(self, generalized=False, paired=True):
                     continue
                 indices.append(idx)
 
-        return indices
+        if not paired:
+            indices += self.make_indices_doubles_all(generalized=generalized)
+
+        return self.sort_and_filter_unique_indices(indices)
+
+    def make_indices_doubles_all(self, generalized=False):
+        singles = self.make_indices_singles(generalized=generalized)
+        unwrapped = [x[0] for x in singles]
+        # now make all combinations of singles
+        indices = [x for x in combinations(unwrapped, 2)]
+        return self.sort_and_filter_unique_indices(indices)
+
+    def sort_and_filter_unique_indices(self, indices):
+        # sort as: [[(a,b),(c,d),(e,f)...],...]with a<c, a<b, c<d
+        sorted_indices = []
+        for idx in indices:
+            idx = tuple([tuple(sorted(pair)) for pair in idx]) # sort internal pairs (a<b, c<d, etc)
+            # avoid having orbitals show up multiple times in excitatin strings
+            idx = tuple([pair for pair in idx if sum([1 for pair2 in idx if pair[0] in pair2 or pair[1] in pair2 ])==1 ])
+            if len(idx) == 0:
+                continue
+            idx = tuple(list(set(idx))) # avoid repetitions (like ((0,2),(0,2)))
+            idx = tuple(sorted(idx, key=lambda x:x[0])) # sort pairs by first entry (a<c)
+            sorted_indices.append(idx)
+        return list(set(sorted_indices))
+
+
 
     def make_unitary(self, k, label):
         return self.molecule.make_excitation_gate(indices=self.generators[k], angle=(self.generators[k], label), assume_real=True)
@@ -386,3 +410,4 @@ def __call__(self, U,  *args, **kwargs):
             S2 = ExpectationValue(H=Qp, U=circuit+Ux.dagger())
             objective += numpy.abs(self.factors[i])*S2
         return objective
+