Merge pull request #135 from slacgismo/main

Merge time shift bug correction

.gitignore

@@ -297,3 +297,29 @@ bh_unicode_properties.cache
 GitHub.sublime-settings
 
 # End of https://www.gitignore.io/api/data,macos,python,pycharm,database,sublimetext,jupyternotebook
+
+# Created by https://www.toptal.com/developers/gitignore/api/vim
+# Edit at https://www.toptal.com/developers/gitignore?templates=vim
+
+### Vim ###
+# Swap
+[._]*.s[a-v][a-z]
+!*.svg  # comment out if you don't need vector files
+[._]*.sw[a-p]
+[._]s[a-rt-v][a-z]
+[._]ss[a-gi-z]
+[._]sw[a-p]
+
+# Session
+Session.vim
+Sessionx.vim
+
+# Temporary
+.netrwhist
+*~
+# Auto-generated tag files
+tags
+# Persistent undo
+[._]*.un~
+
+# End of https://www.toptal.com/developers/gitignore/api/vim

README.md

@@ -93,9 +93,16 @@ See [notebooks](/notebooks) folder for examples.
 
 ### 3 ways of setting up, either approach works:
 
-#### 1) Recommended: Set up `conda` environment with provided `.yml` file
+#### 1) Recommended: Install with pip
 
-We recommend setting up a fresh Python virtual environment in which to use `solar-data-tools`. We recommend using the [Conda](https://docs.conda.io/projects/conda/en/latest/index.html) package management system, and creating an environment with the environment configuration file named `pvi-user.yml`, provided in the top level of this repository. This will install the `statistical-clear-sky` package as well.
+In a fresh Python virtual environment, simply run
+
+
+```bash
+$ pip install solar-data-tools
+```
+
+#### 2) Creat conda virtual environment from file
 
 Creating the env:
 
@@ -123,32 +130,17 @@ $ conda env update -f pvi-user.yml
 
 Additional documentation on setting up the Conda environment is available [here](https://github.com/slacgismo/pvinsight-onboarding/blob/main/README.md).
 
-
-#### 2) PIP Package
-
-```sh
-$ pip install solar-data-tools
-```
-
-Alternative: Clone repo from GitHub
-
-Mimic the pip package by setting up locally.
-
-```bash
-$ pip install -e path/to/root/folder
-```
-
-#### 3) Anaconda Package
+#### 3) General Anaconda Package
 
 ```sh
-$ conda install -c slacgismo solar-data-tools
+$ conda install slacgismo::solar-data-tools
 ```
 
 ### Solvers
 
-#### QSS & OSQP
+#### QSS & CLARABEL
 
-By default, [QSS](https://github.com/cvxgrp/qss) and OSQP solvers are used for non-convex and convex problems, respectively. Both are supported by [OSD](https://github.com/cvxgrp/signal-decomposition/tree/main), the modeling language used to solve signal decomposition problems in Solar Data Tools, and both are open source. 
+By default, [QSS](https://github.com/cvxgrp/qss) and CLARABEL solvers are used for non-convex and convex problems, respectively. Both are supported by [OSD](https://github.com/cvxgrp/signal-decomposition/tree/main), the modeling language used to solve signal decomposition problems in Solar Data Tools, and both are open source. 
 
 #### MOSEK
 

solardatatools/data_handler.py

@@ -1141,11 +1141,6 @@ def rescale_signal(signal, minimum, maximum):
             use_ixs = self.daily_flags.clear
         else:
             use_ixs = self.daily_flags.no_errors
-        ### NEW 3/20/24: overwrite use of qss on this function. Forcing CLARABEL here uses the old-style
-        # iteratively reweighted model, rather than the nonconvex piecewise constant model. In testing,
-        # the older model (which is not technically covered by the monograph) seems to work better.
-        if solver != "MOSEK":
-            solver = "CLARABEL"
         ########## Updates to timeshift algorithm, 6/2023 ##########
         # If running with any solver other than QSS: solve convex problem
         # If running with QSS without a set w1: run w1 meta-opt with convex problem,

solardatatools/signal_decompositions.py

@@ -93,6 +93,7 @@ def l2_l1d1_l2d2p365(
             yearly_periodic=yearly_periodic,
             return_all=return_all,
             transition_locs=transition_locs,
+            solver=solver,
             verbose=verbose,
         )
     elif yearly_periodic:
@@ -172,6 +173,7 @@ def tl1_l2d2p365(
             w1=w1,
             yearly_periodic=yearly_periodic,
             return_all=return_all,
+            solver=solver,
             verbose=verbose,
         )
     else:
@@ -226,7 +228,8 @@ def l1_l1d1_l2d2p365(
     if solver == "MOSEK":
         # MOSEK weights set in CVXPY function
         res = _cvx_l1_l1d1_l2d2p365(
-            signal=signal, use_ixs=use_ixs, return_all=return_all, verbose=verbose
+            signal=signal, use_ixs=use_ixs, return_all=return_all,
+            solver=solver, verbose=verbose
         )
     else:
         res = _osd_l1_l1d1_l2d2p365(
@@ -264,7 +267,8 @@ def l2_l1d2_constrained(
     """
     if solver == "MOSEK":
         # MOSEK weights set in CVXPY function
-        res = _cvx_l2_l1d2_constrained(signal, return_all=return_all, verbose=verbose)
+        res = _cvx_l2_l1d2_constrained(signal, return_all=return_all,
+                                       solver=solver, verbose=verbose)
     else:
         res = _osd_l2_l1d2_constrained(
             signal, w0=w0, w1=w1, return_all=return_all, solver=solver, verbose=verbose