like the plugins you can always add your own CMake variables toįor run-time, you may also wish to install the optional runtime add-ons ( e.g., adcc) Who want a minimal build, there’s a -disable-addons, but it is generally notįunctional cmake command, but those are just setting up CMake cache
ForĮxample, at GaTech psi4-path-advisor -intel works. There’s other options that will also pre-configure compilers. It gives you a basic cmake command covering python, sphinx, link-time qcĪddons, and run-time qc addons. Include -c psi4/label/dev to get dependencies toīuild current master, as opposed to latest release.Īctivate environment, conda activate p4dev. Substitute desired python version: 3.6, 3.7, 3.8, 3.9. > `psi4-path-advisor ` -Dany_addl_cmake_vals=ON > cd objdir & make -j`getconf _NPROCESSORS_ONLN` > make install # execute or adapt `cmake ` commands above DepsCache handles python & addons # DepsMKLCache handles math further psi4-path-advisor options handle compilers. gcc Engage conda's psi4-dev-provided gcc/g++/gfortran compilers.
intel-multiarch Engage self-provided icc/icpc/ifort compilers backed by conda's psi4-dev-provided gcc/g++ PLUS compile for multiple architectures (useful for cluster deployments). intel Engage self-provided icc/icpc/ifort compilers backed by conda's psi4-dev-provided gcc/g++. disable-mkl Disengage building against the psi4-dev-provided MKL libraries (`libmkl_rt`). > cd objdir & make -j`getconf _NPROCESSORS_ONLN` > make install -disable-addons Disengage building against the psi4-dev-provided _optional_ link-time Add-Ons like CheMPS2. > conda create -n p4dev python=x.x psi4-dev -c psi4/label/dev -c psi4 > conda activate p4dev > cd psi4-dev -c psi4 > conda activate p4dev > psi4-path-advisor # execute or adapt `cmake ` commands above DepsCache handles python & addons # DepsMKLCache handles math further psi4-path-advisor options handle compilers. v1.2rc1 #or Mac or Windows # substitute x.x by 3.5 | 3.6 | 3.7 for alternate python versions
The P SI4 executable into the main conda environment at Installs Miniconda+Psi4+AddOns into $HOME/psi4conda and Mac, Windows (Ubuntu shell for Windows accepts Linux installers), or native Windows. Sequence of commands to get you to a working P SI4 on Linux,
How to install a Psi4 binary with the Psi4conda installer, command-line ¶ Explicit commands at How to install a Psi4 binary with the Psi4conda installer, command-line. With psi4 -test (green and yellow good red bad). PATH and PSI_SCRATCH that it suggests at the end. Follow the prompts and do make the adjustments to How to install a Psi4 binary with the Psi4conda installer, download site ¶ĭownload one of the nine installers (Linux/Mac/Windows P圓6/37/38/39).īash it.
Users proficient with conda may prefer to consult What do the conda packages psi4 & psi4-dev and the installer psi4conda contain.įor more flexibility and a detailed explanation, go toĭetailed Installation of Miniconda and Detailed Installation of. The P SI4 binary repository is at Anaconda (formerly Binstar).įor commands to get a default installation, go to How to install a Psi4 binary with the Psi4conda installer, command-line Standardizes python distribution so no need to find/install libpython packagesĪdd-ons (plugins, extra features requiring Fortran compiler, etc.) can be made available as conda packagesĭevelop P SI4 through plugins without a pre-existing development environment, see Creating a New Plugin Using a Conda Pre-compiled Binary. Updated nightly so new features accessible
Lightweight software stack (<100 MB w/o P SI4 ~1 GB including P SI4, numpy, and MKL) No root, administrator, or sudo access requiredīuilt with high-performance math libraries Through Anaconda (formerly Continuum Analytics, the company that producesįull-fledged scientific python environment with package manager conda) and, more particularly, Miniconda (a lightweight Pythonĭistribution with same package manager conda). Windows, through the Ubuntu shell) and native Windows architectures
P SI4 is available as a pre-compiled binary for Mac and Linux (and Psi4conda installers are not ready for either platform. As of v1.2rc1, new (conda build 3 updated compilers)Ĭonda packages are available for Linux but not Mac).