Merge pull request #40 from ergodicio/multi-species-1

Updates to RTD

configs/1d/defaults.yaml

@@ -125,23 +125,35 @@ other:
     refit_thresh: 0.25
     calc_sigmas: False
 
+plotting:
+    n_sigmas: 3
+    rolling_std_width: 5
+    data_cbar_u: data
+    data_cbar_l: 0
+
+
 data:
     shotnum: 101675
     shotDay: False
     launch_data_visualizer: True
     fit_rng:
         blue_min: 450
         blue_max: 510
-        red_min: 540
-        red_max: 625
+        red_min: 545
+        red_max: 650
+        iaw_min: 525
+        iaw_max: 528
+        iaw_cf_min: 526.4
+        iaw_cf_max: 526.6
     bgscaleE: 1.0
     bgscaleI: 0.1
     bgshotmult: 1
     ion_loss_scale: 1.0
     ele_t0: 0
     ion_t0_shift: 0.0
-    ele_lam_shift: -2.0
-    dpixel: 2
+    ele_lam_shift: 0.0
+    probe_beam: P9
+    dpixel: 3
     lineouts:
         type:
             pixel
@@ -169,7 +181,7 @@ optimizer:
     y_norm: True
     x_norm: False
     grad_method: AD
-    batch_size: 2
+    batch_size: 4
     num_epochs: 200
     learning_rate: 1.0e-4
     parameter_norm: True

configs/1d/inputs.yaml

@@ -10,45 +10,70 @@ parameters:
         lb: 0.01
         ub: 3.75
     amp3:
-        val: 1.
-        active: False
-        lb: 0.
-        ub: 10.
+        val: 1.0
+        active: True
+        lb: 0.01
+        ub: 3.75
     lam:
-        val: 524.0
+        val: 526.5
         active: True
-        lb: 523.
-        ub: 528.
+        lb: 523.0
+        ub: 528.0
     Te:
-        val: .5
+        val: .500
         active: True
-        lb: 0.001
+        lb: 0.01
         ub: 1.5
-        gradient: 0.
+    Te_gradient:
+        val: 0.0
+        active: False
+        lb: 0.
+        ub: 10.
+        num_grad_points: 1
     Ti:
-        val: .2
+        val: 0.05
+        active: True
+        lb: 0.001
+        ub: 1.0
+    Z:
+        val: [1., 6.]
+        active: False
+        lb: 0.5
+        ub: 7.0
+    A:
+        val: [1., 14.]
         active: False
         lb: 0.01
         ub: 3.
     Z:
         val: 8.
         active: False
-        lb: 1.
-        ub: 25.
-    A:
-        val: 40
+    ud:
+        val: 0.0
+        active: True
+        lb: -100.0
+        ub: 100.0
+    Va:
+        val: 0.0
         active: False
+        lb: -20.5
+        ub: 20.5
     ne:
-        val: .2
-        active: True
-        lb: .001
-        ub: 1.
-        gradient: 0.
+        val: 0.35
+        active: False
+        lb: 0.001
+        ub: 1.0
+    ne_gradient:
+        val: 0.
+        active: False
+        lb: 0.
+        ub: 15.
+        num_grad_points: 1
     m:
-        val: 3.
-        active: True
-        lb: 2.
-        ub: 5.
+        val: 2.4
+        active: False
+        lb: 2.0
+        ub: 5.0
     fe:
         val: []
         active: False
@@ -57,23 +82,34 @@ parameters:
             DLM:
         lb: -100.
         ub: -0.5
-    blur:
-        val: []
-        active: False
-        lb: 0.
-        ub: 10.
-    specCurvature:
-        val: []
-        active: False
-        lb: .1
-        ub: 10.
-    fitprops:
-        val: []
-        active: False
+        fe_decrease_strict: False
+        symmetric: False
 
+data:
+    shotnum: 106315
+    lineouts:
+        type:
+            um
+        start: -1000
+        end: 1000
+        skip: 20
+    background:
+        type:
+            Fit
+        slice: 900
 
 other:
+    extraoptions:
+        load_ion_spec: True
+        load_ele_spec: True
+        fit_IAW: True
+        fit_EPWb: False
+        fit_EPWr: False
     refit: False
     refit_thresh: 5.0
     calc_sigmas: True
 
+
+mlflow:
+    experiment: inverse-thomson-scattering
+    run: shot_106315_simul_f14_ion_only

docs/source/FAQ.rst

@@ -0,0 +1,6 @@
+FAQ
+---------------------------------
+
+*Have questions?*
+
+Please reach out and ask, we will add the common ones here.

docs/source/best_practice.rst

@@ -0,0 +1,51 @@
+Best Practices for Fitting
+================================
+
+This page gives input deck snippets which produce the highest quality fits using TSADAR. It is recomended that 
+when fitting new data a small region of the data is fit with a small number of lineouts. This can be accomplished
+by setting the :code:`lineouts: start` and :code:`lineouts: end` to be close or increasing :code:`lineouts: skip`.
+This will allow fast fits that can be used to dial in the starting conditions and the free parameters.
+This can also be used to check and adjust the fitting ranges. Once the best inital conditions have been identified 
+the entire dataset can be fit.
+
+Background and lineout selection
+---------------------------------
+
+There are multiple options for background algorithms and types of fitting. These tend to be the best options for 
+various data types. All of these options are editable in the input deck.
+
+**Best operation for time resolved data:**
+
+.. code-block:: yaml
+
+    background:
+        type: pixel
+        slice: 900
+
+**Best operation for spatially resolved data:**
+
+.. code-block:: yaml
+
+    background:
+        type: fit
+        slice: 900 <or background slice for IAW>
+
+**Best operation for lineouts of angular:**
+
+.. code-block:: yaml
+
+    background:
+	    type: fit
+	    slice: <background shot number>
+
+**Best operation for full angular:**
+
+.. code-block:: yaml
+
+    background:
+	    type: fit
+	    val: <background shot number>
+    lineouts:
+	    type: range
+	    start: 90
+    end: 950

docs/source/contributing.rst

@@ -0,0 +1,6 @@
+Contributing
+---------------------------------
+
+If there are new features you would like to see we encourage you to add to the isssues page on Github 
+or contribute yourself. To contribute, please fork the reposotory or create your own branch. When its time to merge 
+please provide an email so we can give you credit and in case we have any questions.

docs/source/defaults.rst

@@ -0,0 +1,4 @@
+Default options
+---------------------------------
+
+*Coming Soon*

docs/source/examples.rst

@@ -0,0 +1,4 @@
+Examples
+---------------------------------
+
+*Coming Soon*

docs/source/usage.rst → docs/source/getting_started.rst

@@ -1,9 +1,9 @@
-Installation and Usage
-==============================
+Getting Started
+================
 
 *Only MacOS and Linux supported at this time. It may or may not work on Windows but is not yet tested so YMMV*
 
-1. Clone the repo
+1. Clone the repo to the local or remote machine where you will be running analysis
 
 2. Install using below instructions or your own way
 
@@ -18,13 +18,19 @@ Installation and Usage
    source venv/bin/activate        # activate the new environment
    pip install -r requirements.txt # install dependencies
 
-**Conda**
+**Conda CPU**
 
 .. code-block:: bash
 
-   mamba env create -f env.yml
-   mamba activate ts-cpu
+   conda env create -f env.yml
+   conda activate tsadar-cpu
 
+**Conda GPU**
+
+.. code-block:: bash
+
+   conda env create -f env-gpu.yml
+   conda activate tsadar-gpu
 
 **Run command**
 
@@ -41,3 +47,9 @@ And the other just performs a forward pass and gives you the spectra given some
 .. code-block:: bash
 
    python run_tsadar.py --cfg <path>/<to>/<inputs>/<folder> --mode forward
+
+
+The inputs for the code are stored in an input deck. The default location for this input deck and therefore
+the starting path for running jobs is :code:`inverse_thomson_scattering/config/1d`. These inputs should be
+modified to change the specific to fit your analysis needs. More information on the Input deck can be found 
+on the :ref:`inputs` page.

docs/source/index.rst

@@ -7,8 +7,8 @@ Welcome to TSADAR!
 ==================================
 
 TSADAR is a Thomson scattering data analysis software written in Python. It helps determine plasma parameters given
-Thomson scattering measurements by fitting the observed spectra to known parameterized functional forms. More detail on
-the theory is provided elsewhere [1] and will be added to this documentation soon.
+Thomson scattering measurements by fitting the collisionless spectral density function to the observed spectra.
+More detail on the theory is provided elsewhere [1] and the specifics of implementation can be found in :doc:`math`.
 
 The fitting is performed via gradient descent. The gradients are acquired using automatic differentiation and JAX [2].
 More details on the numerics will be added soon.
@@ -18,13 +18,18 @@ More details on the numerics will be added soon.
    :maxdepth: 2
    :caption: Contents:
 
-   usage
-   input
-   fitting
-
+   getting_started
+   best_practice
+   examples
+   inputs
+   defaults
+   FAQ
+   math
+   contributing
+
 *What does TSADAR stand for?*
 
-TBD
+Thomson Scattering with Automatic Differentiation for Analysis and Regression