From owner-chemistry@ccl.net Wed Sep 26 14:47:00 2018 From: "Grigoriy Zhurko reg_zhurko/a\chemcraftprog.com" To: CCL Subject: CCL:G: Interaction of QC software components via text files Message-Id: <-53476-180926143239-25563-iiUXrOOFmlaeU7rOqXaeCw.:.server.ccl.net> X-Original-From: Grigoriy Zhurko Content-Transfer-Encoding: 7bit Content-Type: text/plain; charset=us-ascii Date: Wed, 26 Sep 2018 21:34:57 +0400 MIME-Version: 1.0 Sent to CCL by: Grigoriy Zhurko [reg_zhurko-x-chemcraftprog.com] I suppose it is well-known that the geometry optimization in Gaussian is sometimes accompanied with serious problems, and the users have to invent their own ways of solving them. Besides that, the energy of the final (optimized) geometry computed with Gaussian is often slightly higher than the energy of the penultimate (previous) step. This is usually not a big problem, but this is slightly annoying (unless you use my program Chemcraft) ). My question is, whether it is possible to implement a third-party algorithm of geometry optimization, which implies invoking the Gaussian by a third-party program. I mean that the third-party program generates a Gaussian input file (.gjf) with single point and gradient computation, then Gaussian processes this job and generates the output file, then the program reads and parses this output file, then it predicts the coordinates of the next step and runs Gaussian again, etc. until the energy minimum is reached. If such an approach is possible, I will probably try to implement the following things: - geometry optimization and frequencies computation with CCSD(T)/CBS, or even a composite method like FPD; - performing several computations with different DFT functionals at same geometry, and passing their results into a neural network to obtain a very accurate energy. So, is this possible to implement such interaction of a third-party program with Gaussian (preferably the Windows version), or other QC codes? Grigoriy Zhuko From owner-chemistry@ccl.net Wed Sep 26 18:22:00 2018 From: "Andrew Rosen rosen ~~ u.northwestern.edu" To: CCL Subject: CCL:G: Interaction of QC software components via text files Message-Id: <-53477-180926153538-3228-qc6mGLtWiVMCN/s3fG5ieA#,#server.ccl.net> X-Original-From: Andrew Rosen Content-Type: multipart/alternative; boundary="0000000000003ff1580576cb5308" Date: Wed, 26 Sep 2018 14:35:21 -0500 MIME-Version: 1.0 Sent to CCL by: Andrew Rosen [rosen^-^u.northwestern.edu] --0000000000003ff1580576cb5308 Content-Type: text/plain; charset="UTF-8" Grigoriy, What you describe can be achieved by using the Atomic Simulation Environment (ASE). The peer-reviewed article for ASE can be found here and main webpage found here . ASE is a Python-based code meant to manipulate, run, and analyse quantum-chemical calculations. ASE has a built-in set of optimization routines (described here ), and the quantum-chemical program of your choice can be used just to calculate the energies and forces required for ASE to update the atomic positions. You then do not have to rely on the optimizers built into Gaussian, and it operates using the exact same workflow you proposed. ASE does support Gaussian as a calculator (as listed here ), but it is currently lacking documentation and would probably require some modifications to suit all your needs. In principle though, this would be an excellent choice, and I would personally recommend giving ASE a try before building something on your own > from scratch. Since ASE is written in Python, it would be fairly trivial to set up your script to run the DFT calculation (using ASE-based optimizers and a Gaussian "calculator") and then feed whatever features you'd like into a Python-based neural network, such as one built with scikit-learn. Andrew On Wed, Sep 26, 2018 at 2:13 PM Grigoriy Zhurko reg_zhurko/ achemcraftprog.com wrote: > > Sent to CCL by: Grigoriy Zhurko [reg_zhurko-x-chemcraftprog.com] > I suppose it is well-known that the geometry optimization in Gaussian is > sometimes accompanied with serious problems, and the users have to invent > their own ways of solving them. Besides that, the energy of the final > (optimized) geometry computed with Gaussian is often slightly higher than > the energy of the penultimate (previous) step. This is usually not a big > problem, but this is slightly annoying (unless you use my program > Chemcraft) ). > My question is, whether it is possible to implement a third-party > algorithm of geometry optimization, which implies invoking the Gaussian by > a third-party program. I mean that the third-party program generates a > Gaussian input file (.gjf) with single point and gradient computation, then > Gaussian processes this job and generates the output file, then the program > reads and parses this output file, then it predicts the coordinates of the > next step and runs Gaussian again, etc. until the energy minimum is reached. > If such an approach is possible, I will probably try to implement the > following things: > - geometry optimization and frequencies computation with CCSD(T)/CBS, or > even a composite method like FPD; > - performing several computations with different DFT functionals at same > geometry, and passing their results into a neural network to obtain a very > accurate energy. > So, is this possible to implement such interaction of a third-party > program with Gaussian (preferably the Windows version), or other QC codes? > Grigoriy Zhuko> > > --0000000000003ff1580576cb5308 Content-Type: text/html; charset="UTF-8" Content-Transfer-Encoding: quoted-printable
Grigoriy,

What you describe can be = achieved by using the Atomic Simulation Environment (ASE). The peer-reviewe= d article for ASE can be found=C2=A0here and main webpage found here. ASE is a Python-based code m= eant to manipulate, run, and analyse quantum-chemical calculations. ASE has= a built-in set of optimization routines (described here), and the quantum-chemical pr= ogram of your choice can be used just to calculate the energies and forces = required for ASE to update the atomic positions. You then do not have to re= ly on the optimizers built into Gaussian, and it operates using the exact s= ame workflow you proposed. ASE does support Gaussian as a calculator (as li= sted here), but it is currently lacking document= ation and would probably require some modifications to suit all your needs.=

In principle though, this would be an excellent c= hoice, and I would personally recommend giving ASE a try before building so= mething on your own from scratch. Since ASE is written in Python, it would = be fairly trivial to set up your script to run the DFT calculation (using A= SE-based optimizers and a Gaussian "calculator") and then feed wh= atever features you'd like into a Python-based neural network, such as = one built with scikit-learn.

Andrew
On Wed, Sep 26, 2018 at 2:13 P= M Grigoriy Zhurko reg_zhurko/achemcra= ftprog.com <owner-chemist= ry#,#ccl.net> wrote:

Sent to CCL by: Grigoriy Zhurko [reg_zhurko-x-chemcraftprog.com= ]
=C2=A0 I suppose it is well-known that the geometry optimization in Gaussia= n is sometimes accompanied with serious problems, and the users have to inv= ent their own ways of solving them. Besides that, the energy of the final (= optimized) geometry computed with Gaussian is often slightly higher than th= e energy of the penultimate (previous) step. This is usually not a big prob= lem, but this is slightly annoying (unless you use my program Chemcraft) ).=
=C2=A0 My question is, whether it is possible to implement a third-party al= gorithm of geometry optimization, which implies invoking the Gaussian by a = third-party program. I mean that the third-party program generates a Gaussi= an input file (.gjf) with single point and gradient computation, then Gauss= ian processes this job and generates the output file, then the program read= s and parses this output file, then it predicts the coordinates of the next= step and runs Gaussian again, etc. until the energy minimum is reached. =C2=A0 If such an approach is possible, I will probably try to implement th= e following things:
=C2=A0- geometry optimization and frequencies computation with CCSD(T)/CBS,= or even a composite method like FPD;
=C2=A0- performing several computations with different DFT functionals at s= ame geometry, and passing their results into a neural network to obtain a v= ery accurate energy.
=C2=A0 So, is this possible to implement such interaction of a third-party = program with Gaussian (preferably the Windows version), or other QC codes?<= br> Grigoriy Zhuko



-=3D This is automatically added to each message by the mailing script =3D-=
E-mail to subscribers: CHEMISTRY#,#ccl.net or use:
=C2=A0 =C2=A0 =C2=A0 http://www.ccl.net/cgi-bin/ccl/s= end_ccl_message

E-mail to administrators: CHEMISTRY-REQUEST#,#ccl.net or use
=C2=A0 =C2=A0 =C2=A0 http://www.ccl.net/cgi-bin/ccl/s= end_ccl_message
=C2=A0 =C2=A0 =C2=A0 http://www.ccl.net/chemistry/sub_un= sub.shtml

Before posting, check wait time at: http://www.ccl.net

Job: http://www.ccl.net/jobs
Conferences: http://server.ccl.net/chemist= ry/announcements/conferences/

Search Messages: http://www.ccl.net/chemistry/sear= chccl/index.shtml
=C2=A0 =C2=A0 =C2=A0 http://www.ccl.net/spammers.txt

RTFI: http://www.ccl.net/chemistry/aboutccl/ins= tructions/


--0000000000003ff1580576cb5308-- From owner-chemistry@ccl.net Wed Sep 26 18:57:00 2018 From: "Christoph Riplinger riplinger,faccts.de" To: CCL Subject: CCL:G: Interaction of QC software components via text files Message-Id: <-53478-180926165216-21341-dPUBvMkrdJeT8HxhKjfdsA/a\server.ccl.net> X-Original-From: Christoph Riplinger Content-Type: multipart/signed; micalg=pgp-sha256; protocol="application/pgp-signature"; boundary="NNFszPWrVCelD7ULR6q1R418MjCwp6VU9" Date: Wed, 26 Sep 2018 22:52:09 +0200 MIME-Version: 1.0 Sent to CCL by: Christoph Riplinger [riplinger---faccts.de] This is an OpenPGP/MIME signed message (RFC 4880 and 3156) --NNFszPWrVCelD7ULR6q1R418MjCwp6VU9 Content-Type: multipart/mixed; boundary="t2v6jM7fkVS637nd53fEJXOllcmqsvNnE"; protected-headers="v1" > From: Christoph Riplinger To: CCL Subscribers Message-ID: <55efe651-9ff8-e02c-80bd-ff5576c01fce _ faccts.de> Subject: Re: CCL:G: Interaction of QC software components via text files References: <53476-180926143239-25563-ElUPykEMPxfmYM7Ji2Eucg _ server.ccl.net> In-Reply-To: <53476-180926143239-25563-ElUPykEMPxfmYM7Ji2Eucg _ server.ccl.net> --t2v6jM7fkVS637nd53fEJXOllcmqsvNnE Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: quoted-printable Content-Language: en-US Dear Grigoriy, the ORCA optimizer is robust and efficient. ORCA's optimizer can be used as an external optimizer and can be used for the described scenario. Besides that, ORCA by itself provides plenty of density functionals and efficient Post-HF methods. Using ORCA's optimizer as external optimizer is well documented in the manual. ORCA is free of charge for academic use (https://orcaforum.cec.mpg.de/). Best regards, Christoph On 09/26/2018 07:34 PM, Grigoriy Zhurko reg_zhurko/achemcraftprog.com wro= te: > Sent to CCL by: Grigoriy Zhurko [reg_zhurko-x-chemcraftprog.com] > I suppose it is well-known that the geometry optimization in Gaussian= is sometimes accompanied with serious problems, and the users have to in= vent their own ways of solving them. Besides that, the energy of the fina= l (optimized) geometry computed with Gaussian is often slightly higher th= an the energy of the penultimate (previous) step. This is usually not a b= ig problem, but this is slightly annoying (unless you use my program Chem= craft) ). > My question is, whether it is possible to implement a third-party alg= orithm of geometry optimization, which implies invoking the Gaussian by a= third-party program. I mean that the third-party program generates a Gau= ssian input file (.gjf) with single point and gradient computation, then = Gaussian processes this job and generates the output file, then the progr= am reads and parses this output file, then it predicts the coordinates of= the next step and runs Gaussian again, etc. until the energy minimum is = reached. > If such an approach is possible, I will probably try to implement the= following things: > - geometry optimization and frequencies computation with CCSD(T)/CBS, = or even a composite method like FPD; > - performing several computations with different DFT functionals at sa= me geometry, and passing their results into a neural network to obtain a = very accurate energy. > So, is this possible to implement such interaction of a third-party p= rogram with Gaussian (preferably the Windows version), or other QC codes?= > Grigoriy Zhuko > > > > -=3D This is automatically added to each message by the mailing script = =3D- > To recover the email address of the author of the message, please chang= e> > Subscribe/Unsubscribe:=20> > Job: http://www.ccl.net/jobs=20=> > --=20 FAccTs GmbH Rolandstrasse 67, 50677 K=C3=B6ln Amtsgericht K=C3=B6ln HRB 88406 Gesch=C3=A4ftsf=C3=BChrer: Dr. Christoph Riplinger https://www.faccts.de --t2v6jM7fkVS637nd53fEJXOllcmqsvNnE-- --NNFszPWrVCelD7ULR6q1R418MjCwp6VU9 Content-Type: application/pgp-signature; name="signature.asc" Content-Description: OpenPGP digital signature Content-Disposition: attachment; filename="signature.asc" -----BEGIN PGP SIGNATURE----- Version: GnuPG v2 iQIcBAEBCAAGBQJbq/F5AAoJEOicIQ7ISfcOu4oP/i9GXFhx2VLs4fIp1z+ZyxRp 2Ya4gZixXFvg7FqX/aiQtPd9Gx4Ox3Tys3G9pZ6+N2st9FTbrcZ3Ku5gP0BOMe1C q25ukhKfy6iCdFOYbrfYb+dcDVLX1iPr193jA1TFnU2ZBMYKaOcX74FeX04WslNF 1YvN2FSVmMCYafREUicUM9pqq8k48hAdH0NwfFPNjLDXfxsa1Irr1MXEcV7Epm3z Rq2QfbjI8kxcAHTfII6Hf90DfAQvZR2M/9H6Xz8Eb3RvguS1pUNA0pKOf2YqGpFA MiRh3Dr5VmoKNaykQ5B3Kic3yeir9NiLX+yw8CeRWjnF5fTuOykiXMsV5hAoHbQi 3vmEy4/c9SIG0ntrJJXR6szGyfLsZvGoU0sNJVELaSeanaj2phhW3vGBXqhJ3bL7 7WJIfshRlNOcgFk1zFGYarZXxVe5yNLO/ZBrKfu3tq3fMhlLZx9HAOVs2TSfZz+W Pb5ZvITnaQc/JsFfdBWPutds2IBx8HqX/qkoNq4dXSskRDyFqkUBT0EdcECBoqZl sn8PfqS4Th4VGPBYG6Icrvi3LLHo+MFTby0mW/r6355sdAiPM4zoUkvuixYoEObd P7y6mdwwGrQol9eLtb/mFm7APomBH3F0xI4tf2FiObwYQqeomX2WRJwoFsGbFrNO VacbWXLRv+u005ivvVcu =2WOO -----END PGP SIGNATURE----- --NNFszPWrVCelD7ULR6q1R418MjCwp6VU9--