* pfx readme

Author: Joshuaalbert

dsa2000_cal/README.md

@@ -61,10 +61,10 @@ cd DSA2000-Cal
 
 ```bash
 conda activate dsa2000_cal_py
-pip install dsa2000_call
+pip install dsa2000_cal
 ```
 
-8. Set up PyCharm for development
+8. Set up PyCharm for development (optional but recommended).
 
     1. Make sure you have created a `dsa2000_cal_py` conda env as above, and installed requirements files.
     2. Create a new project in PyCharm in the repo root directory `/home/username/git/DSA2000-Cal`. Use an empty project

dsa2000_cal/notebooks/explore_sky_loss.ipynb

@@ -228,15 +228,15 @@
     "                                  delta)\n",
     "        return accumulate, None\n",
     "\n",
-    "    accumulate = (\n",
+    "    init_accumulate = (\n",
     "        jnp.zeros(l.shape, jnp.float32), jnp.zeros(l.shape, jnp.float32),\n",
-    "        jnp.zeros(1, jnp.float32), jnp.zeros(1, jnp.float32),\n",
-    "        jnp.zeros(1, jnp.float32), jnp.zeros(1, jnp.float32),\n",
-    "        jnp.zeros(1, jnp.float32), jnp.zeros(1, jnp.float32)\n",
+    "        jnp.zeros((), jnp.float32), jnp.zeros((), jnp.float32),\n",
+    "        jnp.zeros((), jnp.float32), jnp.zeros((), jnp.float32),\n",
+    "        jnp.zeros((), jnp.float32), jnp.zeros((), jnp.float32)\n",
     "    )\n",
-    "    accuulate, _ = jax.lax.scan(\n",
+    "    accumulate, _ = jax.lax.scan(\n",
     "        accumulate_over_freq,\n",
-    "        accumulate,\n",
+    "        init_accumulate,\n",
     "        (freqs, jax.random.split(key, len(freqs)))\n",
     "    )\n",
     "\n",
@@ -248,13 +248,13 @@
     "    ) = accumulate\n",
     "\n",
     "    # Compute RMS and image normal stats\n",
-    "    rms_no_noise = jnp.sqrt(jnp.sum((image - zero_point) ** 2))\n",
+    "    rms_no_noise = jnp.sqrt(jnp.mean((image - zero_point) ** 2))\n",
     "    max_no_noise = jnp.max(image)\n",
     "    min_no_noise = jnp.min(image)\n",
     "    mean_no_noise = jnp.mean(image)\n",
     "    std_no_noise = jnp.std(image)\n",
     "\n",
-    "    rms_noise = jnp.sqrt(jnp.sum((image_noise - zero_point) ** 2))\n",
+    "    rms_noise = jnp.sqrt(jnp.mean((image_noise - zero_point) ** 2))\n",
     "    max_noise = jnp.max(image_noise)\n",
     "    min_noise = jnp.min(image_noise)\n",
     "    mean_noise = jnp.mean(image_noise)\n",