{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "54cf7c4b",
   "metadata": {},
   "source": [
    "# NumPy Tutorial 03: Indexing, Masking, and Reshaping"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6fa31f23",
   "metadata": {},
   "source": [
    "## Download Notebook\n",
    "\n",
    "{download}`Download this notebook <03_indexing_slicing_and_reshaping.ipynb>`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "a22d4443",
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e41a799f",
   "metadata": {},
   "source": [
    "## 1. Basic slicing returns views\n",
    "\n",
    "Changes on a slice may affect the original array."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "3e94260f",
   "metadata": {},
   "outputs": [],
   "source": [
    "a = np.arange(10)\n",
    "s = a[2:7]\n",
    "s[:] = -1\n",
    "print('slice:', s)\n",
    "print('original changed:', a)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7e061c54",
   "metadata": {},
   "source": [
    "## 2. Fancy indexing returns copies\n",
    "\n",
    "Integer-array indexing produces a copy, not a view."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "0c888f93",
   "metadata": {},
   "outputs": [],
   "source": [
    "a = np.arange(10)\n",
    "picked = a[[1, 3, 5, 7]]\n",
    "picked[:] = 100\n",
    "print('picked:', picked)\n",
    "print('original unchanged:', a)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9c2ac082",
   "metadata": {},
   "source": [
    "## 3. Boolean masking\n",
    "\n",
    "Boolean masks are expressive for filtering and assignment."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "901b2efe",
   "metadata": {},
   "outputs": [],
   "source": [
    "rng = np.random.default_rng(0)\n",
    "x = rng.normal(size=12)\n",
    "mask = x > 0\n",
    "\n",
    "print('x:', np.round(x, 3))\n",
    "print('positive values:', np.round(x[mask], 3))\n",
    "\n",
    "x2 = x.copy()\n",
    "x2[x2 < 0] = 0\n",
    "print('negative clipped to 0:', np.round(x2, 3))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5bbce50e",
   "metadata": {},
   "source": [
    "## 4. Multi-dimensional indexing\n",
    "\n",
    "Use row/column selectors and `:` slices to target blocks."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "506938f2",
   "metadata": {},
   "outputs": [],
   "source": [
    "m = np.arange(1, 17).reshape(4, 4)\n",
    "print('m:\\n', m)\n",
    "print('element (2,3):', m[2, 3])\n",
    "print('second column:', m[:, 1])\n",
    "print('middle block:\\n', m[1:3, 1:3])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d6a7dcb5",
   "metadata": {},
   "source": [
    "## 5. Reshape, transpose, and axis semantics\n",
    "\n",
    "Axis 0 is rows, axis 1 is columns for 2D arrays."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "afc4f658",
   "metadata": {},
   "outputs": [],
   "source": [
    "x = np.arange(24)\n",
    "m = x.reshape(2, 3, 4)\n",
    "\n",
    "print('shape:', m.shape)\n",
    "print('sum axis=0 shape:', m.sum(axis=0).shape)\n",
    "print('sum axis=1 shape:', m.sum(axis=1).shape)\n",
    "print('sum axis=2 shape:', m.sum(axis=2).shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "97f9da05",
   "metadata": {},
   "source": [
    "## 6. Practice\n",
    "\n",
    "1. Build a 6x6 matrix from 1 to 36.\n",
    "2. Extract all even numbers using a boolean mask.\n",
    "3. Replace diagonal elements with 0."
   ]
  }
 ],
 "metadata": {
  "language_info": {
   "name": "python"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}