save work

Manuscript/QUESTDB.bib

@@ -1,13 +1,26 @@
 %% This BibTeX bibliography file was created using BibDesk.
 %% http://bibdesk.sourceforge.net/
 
-%% Created for Pierre-Francois Loos at 2020-12-26 14:08:49 +0100 
+%% Created for Pierre-Francois Loos at 2021-01-01 21:49:37 +0100 
 
 
 %% Saved with string encoding Unicode (UTF-8) 
 
 
 
+@article{Mardirossian_2017,
+	author = {Narbe Mardirossian and Martin Head-Gordon},
+	date-added = {2021-01-01 21:49:17 +0100},
+	date-modified = {2021-01-01 21:49:36 +0100},
+	doi = {10.1080/00268976.2017.1333644},
+	journal = {Mol. Phys.},
+	number = {19},
+	pages = {2315-2372},
+	title = {Thirty years of density functional theory in computational chemistry: an overview and extensive assessment of 200 density functionals},
+	volume = {115},
+	year = {2017},
+	Bdsk-Url-1 = {https://doi.org/10.1080/00268976.2017.1333644}}
+
 @book{Robb_2018,
 	author = {Robb, Michael A},
 	date-added = {2020-11-27 22:23:19 +0100},

Manuscript/QUEST_WIREs.tex

@@ -130,8 +130,8 @@ of Hobza and collaborators which provides benchmark interaction energies for wea
 systems \cite{vanSetten_2013,Bruneval_2016,Caruso_2016,Govoni_2018}. The extrapolated ab initio thermochemistry (HEAT) set designed to achieve high accuracy for enthalpies of formation
 of atoms and small molecules (without experimental data) is yet another successful example of benchmark set \cite{Tajti_2004,Bomble_2006,Harding_2008}. More recently, let us mention the benchmark datasets
 of the \textit{Simons Collaboration on the Many-Electron Problem} providing, for example, highly-accurate ground-state energies for
-hydrogen chains \cite{Motta_2017} as well as transition metal atoms and their ions and monoxides \cite{Williams_2020}. Let us also mention the set of Zhao and Truhlar for small transition metal complexes
-employed to compare the accuracy of density-functional methods \cite{ParrBook} for $3d$ transition-metal chemistry \cite{Zhao_2006}, and finally the popular GMTKN24 \cite{Goerigk_2010},
+hydrogen chains \cite{Motta_2017} as well as transition metal atoms and their ions and monoxides \cite{Williams_2020}. 
+Let us also mention the set of Zhao and Truhlar for small transition metal complexes employed to compare the accuracy of density-functional methods \cite{ParrBook} for $3d$ transition-metal chemistry \cite{Zhao_2006}, \alert{the MGCDB84 molecular database of Mardirossian and Head-Gordon that they use to benchmark a total of 200 density functionals and design (using combinatorial approach) the $\omega$B97M-V functional \cite{Mardirossian_2017},} and finally the popular GMTKN24 \cite{Goerigk_2010},
 GMTKN30 \cite{Goerigk_2011a,Goerigk_2011b} and GMTKN55 \cite{Goerigk_2017} databases for general main group thermochemistry, kinetics, and non-covalent interactions developed by Goerigk, Grimme and
 their coworkers.
 
@@ -240,6 +240,7 @@ This has the advantage to produce a smoother and faster convergence of the SCI e
 The CIPSI energy $E_\text{CIPSI}$ is defined as the sum of the variational energy $E_\text{var}$ (computed via diagonalization of the CI matrix in the reference space) and a PT2 correction $E_\text{PT2}$ which estimates the contribution of the determinants not included in the CI space \cite{Garniron_2017b}.
 By linearly extrapolating this second-order correction to zero, one can efficiently estimate the FCI limit for the total energies.
 These extrapolated total energies (simply labeled as $E_\text{FCI}$ in the remainder of the paper) are then used to compute vertical excitation energies.
+
 Depending on the set, we estimated the extrapolation error via different techniques.
 For example, in Ref.~\cite{Loos_2020b}, we estimated the extrapolation error by the difference between the transition energies obtained with the largest SCI wave function and the FCI extrapolated value.
 This definitely cannot be viewed as a true error bar, but it provides an idea of the quality of the FCI extrapolation and estimate.
@@ -407,6 +408,7 @@ states a very good agreement between the CC3 and CCSDT values, indicating that t
 The estimated values of the excitation energies obtained via a three-point linear extrapolation considering the three largest CIPSI wave functions are also gathered in Table \ref{tab:cycles}.
 In this case, the error bar is estimated via the extrapolation distance, \ie, the difference in excitation energies obtained with the three-point linear extrapolation and the largest CIPSI wave function.
 This strategy has been considered in some of our previous works \cite{Loos_2020b,Loos_2020c,Loos_2020e}.
+
 The deviation from the CCSDT excitation energies for the same set of excitations are depicted in Fig.~\ref{fig:errors}, where the red dots correspond to the excitation energies and error bars estimated via the present method, and the blue dots correspond to the excitation energies obtained via a three-point linear fit and error bars estimated via the extrapolation distance.
 These results contain a good balance between well-behaved and ill-behaved cases.
 For example, cyclopentadiene and furan correspond to well-behaved scenarios where the two flavors of extrapolations yield nearly identical estimates and the error bars associated with these two methods nicely overlap.
@@ -458,9 +460,9 @@ Triazine			&	$^1A_1''(n \ra \pis)$			&	4.85	&	4.84	&	4.77(13)&	5.12(51)	\\
 \hline  % Please only put a hline at the end of the table
 \end{tabular}
 \begin{tablenotes}
-\item $^a$ Excitation energies and error bars estimated via the novel statistical method based on Gaussian random variables (see Sec.~\ref{sec:error}).
+\item $^a$ Excitation energies and error bars \alert{(in eV)} estimated via the novel statistical method based on Gaussian random variables (see Sec.~\ref{sec:error}).
 The error bars reported in parenthesis correspond to one standard deviation.
-\item $^b$ Excitation energies obtained via a three-point linear fit using the three largest CIPSI variational wave functions, and error bars estimated via the extrapolation distance, \ie, the difference in excitation energies obtained with the three-point linear extrapolation and the largest CIPSI wave function.
+\item $^b$ Excitation energies obtained via a three-point linear fit using the three largest CIPSI variational wave functions, and error bars \alert{(in eV)} estimated via the extrapolation distance, \ie, the difference in excitation energies obtained with the three-point linear extrapolation and the largest CIPSI wave function.
 \end{tablenotes}
 \end{threeparttable}
 \end{table}
@@ -1255,10 +1257,12 @@ Concerning the second-order methods (which have the indisputable advantage to be
 A very similar ranking is obtained when one looks at the MSEs.
 It is noteworthy that the performances of EOM-MP2 and CCSD are getting notably worse when the system size increases, while CIS(D) and STEOM-CCSD have a very stable behavior with respect to system size.
 Indeed, the EOM-MP2 MAE attains 0.42 eV for molecules containing between 7 and 10 non-hydrogen atoms, whereas the CCSD tendency to overshoot the transition energies yield a MSE of 0.22 eV for the same set (a rather large error).
+
 For CCSD, this conclusion fits benchmark studies published by other groups \cite{Schreiber_2008,Caricato_2010,Watson_2013,Kannar_2014,Kannar_2017,Dutta_2018}.
 For example, K\'ann\'ar and Szalay obtained a MAE of 0.18 eV on Thiel's set for the states exhibiting a dominant single excitation character.
 The CCSD degradation with system size might partially explain the similar (though less pronounced) trend obtained for CCSDR(3).
 Regarding the apparently better performances of STEOM-CCSD as compared to CCSD, we recall that several challenging states have been naturally removed from the STEOM-CCSD statistics because the active character percentage was lower than $98\%$ (see above).
+
 In contrast to EOM-MP2 and CCSD, the overall accuracy of CC2 and ADC(2) does significantly improve for larger molecules, the performances of the two methods being, as expected, similar \cite{Harbach_2014}.
 Let us note that these two methods show similar accuracies for singlet and triplet transitions, but are significantly less accurate for Rydberg transitions, as already pointed out previously \cite{Kannar_2017}.
 Therefore, both CC2 and ADC(2) offer an appealing cost-to-accuracy ratio for large compounds, which explains their popularity in realistic chemical scenarios \cite{Hattig_2005c,Goerigk_2010a,Send_2011a,Winter_2013,Jacquemin_2015b,Oruganti_2016}.

Response_Letter/Response_Letter.tex

@@ -139,8 +139,8 @@ We look forward to hearing from you.
 	{Finally, on a technical note I suspect that the authors may need to test some more (lower quality) methods to get sufficient statistical accuracy for dieting. 
 	The distributions in Figure 5 suggest quite a lot of correlation between the lower accuracy methods tested so far (e.g. CIS(D), CC2 and ADC(2) have very similar error distributions).}
 	\\
-	\alert{Yes, we are aware of this. We would like to thank the reviewer for pointing this out.
-	We have added a sentence at the end of the manuscript to mention it.}	
+	\alert{Yes, we are aware of this. We would like to thank the reviewer for pointing this out.}
+%	We have added a sentence at the end of the manuscript to mention it.}	
 \end{itemize}
  
 \end{letter}