Examples: update active learning installation instructions

Author: bastonero

Examples/sscha_and_aiida/README.md

@@ -1,20 +1,31 @@
 # Instructions
 
-We provide here the script `run_aiida_sscha.py`, which performs a thermal expansion calculation using SSCHA and aiida-quantumespresso. 
+We provide here some scripts to run SSCHA using AiiDA, FLARE machine-learning potential, and a combination of the two as on-the-fly active learning.
 
-It is preferable to execute the example that you already have some experience with both the SSCHA and AiiDA-QuantumEspresso codes. Nevertheless, you can try following the instructions and to run the example.
+* `run_aiida_sscha.py`, which performs a thermal expansion calculation using SSCHA and aiida-quantumespresso. 
+* `run_flare_sscha.py`, which performs a thermal expansion calculation using SSCHA and FLARE machine-learning interatomic potential. 
+* `run_aiida_flare_sscha.py`, which performs an on-the-fly active learning SSCHA calculation using aiida-quantumespresso for DFT and FLARE as the ML potential. 
+
+It is preferable to execute the example that you already have some experience with both the SSCHA and AiiDA-QuantumESPRESSO codes. Nevertheless, you can try following the instructions and to run the example.
 
 ## Installation
 
-We recommend to install all the packages via `mamba` (which is based on top of `conda`). After having installed `mamba`, you can simply run the following:
+We recommend to install all the packages via `mamba` (which is based on `conda`). After having installed `mamba`, you can simply run the following:
 
 ```console
-> mamba create -n aiida-sscha -c conda-forge python gfortran libblas lapack openmpi julia openmpi-mpicc pip numpy scipy spglib aiida-core
-> pip install ase quippy-ase cellconstructor python-sscha aiida-quantumespresso aiida-pseudo
-> mamba activate aiida-sscha
+> mamba create -n sscha-aiida -c conda-forge python gfortran "blas=*=openblas" openblas lapack julia pip numpy scipy spglib pkg-config aiida-core
+> pip install ase cellconstructor python-sscha aiida-quantumespresso aiida-pseudo
+> mamba activate sscha-aiida
 ```
+Then, you should configure an AiiDA profile and the pw.x code in order to use the example script (see also Prerequisites section).
 
-Then, you should configure an AiiDA profile in order to use the example script (see also Prerequisites section).
+For on-the-fly active learning you also need the FLARE package:
+
+```console
+> git clone --depth 1 https://github.com/mir-group/flare.git
+> cd flare
+> pip install .
+```
 
 ## Prerequisites
 
@@ -23,6 +34,7 @@ You need to have installed in the same environment:
 - `cellconstructor`
 - `aiida-core`
 - `aiida-quantumespresso`
+- (optional, for active learning) `mir-flare` 
 
 For the AiiDA part, it is essential the dameon is running and you have:
 1. Configured a computer where to run the code (e.g. on your own laptop; see `aiida-core` docs for further details)
@@ -33,6 +45,8 @@ Please refer to the aiida-core and aiida-quantumespresso for further installatio
 
 ## How-to run 
 
+### SSCHA with the aiida-quantumespresso interface 
+
 Open the `run_aiida_sscha.py` and change the data according to your needs and local installation. Then simply
 
 ```console
@@ -45,4 +59,20 @@ Usually an actual production run would take a while. We suggest to use instead
 > nohup python run_aiida_sscha.py > run_aiida_sscha.log &
 ```
 
+or a submit script at glance.
+
+### On-the-fly active learning SSCHA with aiida-quantumespresso and FLARE interface 
+
+Open the `run_aiida_flare_sscha.py` and change the data according to your needs and local installation. Then simply
+
+```console
+> python run_aiida_flare_sscha.py > run_aiida_sscha.log
+```
+
+Usually an actual production run would take a while. We suggest to use instead
+
+```console
+> nohup python run_aiida_sscha.py > run_aiida_sscha.log &
+```
+
 or a submit script at glance.