Welcome to Assignment 1: Setup and Warm-up

# Welcome to Assignment 1: Setup and Warm-up In this assignment we are going to go through all the setup and background we need for this course! We've divided it into 3 sections as follows: 1. Setting up our python environment. 2. Review math that we will use throughout this course. 3. Get comfortable with Python, which we will use for all assignments and the final project. :::info This homework is due by **Wednesday, January 29th, 2025 at 10:00 PM EST** ::: ## 1. Environment Setup ### Getting started Please click <ins>[here](https://classroom.github.com/a/V22mXH7x)</ins> to get the stencil code. Reference this <ins>[guide](https://hackmd.io/gGOpcqoeTx-BOvLXQWRgQg)</ins> for more information about GitHub and GitHub Classroom. ### Roadmap In order to complete (programming) assignments for this course, you will need a way to code, run, and debug your own Python code. While you are always free to use department machines for this (they have a pre-installed version of the course environment that every assignment has been tested against), you are also free to work on your own machines. Below we give you some information that is helpful for either of these situations. ### A. Configuring Environment #### Developing Locally In order to set up your virtual environment for this couse, we _highly recommend_ (and only formally support) the use of Anaconda to create and manage your Python environment. Once you have cloned the Github Classroom assignment for this assignment, you can do the following to setup your virtual environment: 1. Download the **Anaconda** installer from [here](https://www.anaconda.com/download#downloads), and install it on your computer. We recommend using the Graphical Installer for the correct system (Windows / (Intel) Mac / Mac M1). :::info **Note:** If you have an existing Anaconda or Miniconda installation (such as from CS200), then you don't need to reinstall, and can just use that! You can tell if you have an existing install if the command `conda --version` is recognized. ::: :::warning **Windows**: When installing using the graphical installer, **be sure to check the box which adds `conda` to your `PATH`**. ::: 2. Open a new terminal window and navigate to the root of the cloned assignment in a terminal (such as the one in VSCode) using `cd` and `ls`, and run `./env_setup/Other/conda_create.sh`. This should set up a virtual environment named `csci1470` on your computer. **If you have an Apple M1, this script will be different.** (See below.) :::info You may need to restart your terminal after installing Anaconda in order for this to work. ::: :::warning **Note:** This might be slightly different depending on your platform: - **Apple M1**: We provide a slightly different script which you can call from `"./env_setup/Apple M1/conda_create_m1.sh"` for those who have Apple silicon. - **Windows and Others**: If you are using Windows Powershell, then you can just run `./env_setup/Other/conda_create.sh` (forward slashes), but if you are using *Command Prompt*, then you need to run `.\env_setup\Other\conda_create.sh` (backslashes). **Some users have also experienced problems running the `*.sh` files entirely.** If you are getting a permissions issue try running `chmod a+x` in the command line from inside your repo. If that does not work, you can just open the `conda_create.sh` script in a text editor (such as VSCode), and run each line individually in your terminal. ::: 3. Run `conda activate csci1470`. **You will need to do this in every shell where you want to use the virtual environment**. :::warning If you are the above procedure doesn't work for you (although we **highly recommend trying to troubleshoot that first**), here is another method that does not rely on `conda` commands: :::spoiler 1. **Install Python 3.11** (or Python 3.9, we have not found any differences in functionality between them for our projects). 2. **Install the following packages** in either a virtual environment or your main python environment ipython==8.8.0 matplotlib==3.5.3 numpy==1.23.5 Pillow==9.4.0 scipy==1.9.3 tensorflow==2.11.0 tqdm==4.64.1 - We suggest you create a virtual environment. You can do so by running the following commands: 1. Create a folder where you plan to keep all your homework assignments. 2. In that folder, run `python -m venv cs1470`. This will create a new virtual environment caled `cs1470`. 3. Activate your virtual environment by running `cs1470/Scripts/activate` on Windows machines or `source cs1470/bin/activate` on Mac and Linux machines. Do this before starting any homework asssignment. 4. You can deactivate the virtual environment by typing `deactivate`. - You can install individual packages using `pip` commands (ie `pip install ipython==8.8.0`) - You can install a group of packages by pasting them into a requirements.txt file and running `pip install -r requirements.txt` ::: Once this is complete, you should have a local environment to use for the course! #### Department Machines :::info **Note:** Sometimes even if you set up your local environment correctly, you may experience unexpected bugs and errors that are unique to your local setup. To prevent this from hindering your ablity to complete assignments, we **highly recommend** that you familiarize yourself with the department machines, even if you expect to usually be working locally. ::: Department machines serve as a common, uniform way to work on and debug assignments. There are a variety of ways in which you can use department machines: 1. **In Person.** If you are in the CIT, you can (almost) always head into the Sunlab/[ETC] and work on a department machine. 2. **FastX**. FastX allows you to VNC into a department machine from your own computer, from anywhere! A detailed guide to getting FastX working on your own computer can be found [here](https://cs.brown.edu/about/system/connecting/fastx/). 3. **SSH**. The department machines can also be accessed by SSH (Secure Shell) from anywhere, which should allow you to perform command line activities (cloning repositories, running assignment code). You can check out an SSH guide [here](https://cs.brown.edu/about/system/connecting/ssh/). When using the department machines, you can activate the course virtual environment (which we have already installed) using: ``` source /course/cs1470/cs1470_env/bin/activate ``` Which will activate the course virtual environment. From here, you should be able to clone the repository (see a GitHub guide here for more information on using Git via the command line), and work on your assignment. :::info **Note**: Python files using `tensorflow` may require a little more time on startup to run on department machines (likely because it is pulling files from the department filesystem), but they should all run nonetheless. ::: ### B. Test your environment #### What is an environment? Python packages, or libraries, are external sets of code written by other industry members which might prove really helpful! (Imagine coding how to draw a graph in Python every single time) However, different classes, tasks, and even projects, might require different sets of Python packages. We can manage these as different virtual environments which have different sets of packages installed. #### Conda Specifics If you are using `conda`, you might notice the `(base)` prefix in your terminal. This signifies that you're in the default (hence `(base)`) environment. To access CSCI1470's virtual environment, you can use `conda activate csci1470`. You should now see the `(csci1470)` prefix in your terminal! To return back to the base environment, you can use `conda deactivate`. ## 2. Math Review ### A. Matrix Multiplication 1. Given two column vectors $a \in \mathbb{R}^{m \times 1}, \; b \in \mathbb{R}^{n \times 1}$ the _outer product_ is $$\mathbf{a} \times \mathbf{b} = \begin{bmatrix}a_0 \\ \vdots \\ a_{m-1}\end{bmatrix} \times \begin{bmatrix}b_0 \\ \vdots \\ b_{n-1}\end{bmatrix} = \begin{bmatrix} a_0 b^T\\ \vdots \\ a_{m-1} b^T\\ \end{bmatrix} = \begin{bmatrix} a_0 b_0 & \cdots & a_0 b_{n-1}\\ \vdots & \ddots & \vdots \\ a_{m-1} b_0 & \cdots & a_{m-1} b_{n-1}\\ \end{bmatrix} \in \mathbb{R}^{m\times n} $$ 2. Given two column vectors $\mathbf{a}$ and $\mathbf{b}$ both in $\mathbb{R}^{r\times 1}$, the _inner product_ (or the _dot product_) is defined as: $$ \mathbf{a} \cdot \mathbf{b} = \mathbf{a}^T\mathbf{b} = \begin{bmatrix} a_0\ \cdots\ a_{r-1} \end{bmatrix} \begin{bmatrix}b_0 \\ \vdots \\ b_{r-1}\end{bmatrix} = \sum_{i=0}^{r} a_i b_i $$ where $\mathbf{a}^T$ is the _transpose_ of a vector, which converts between column and row vector alignment. The same idea extends to matrices as well. 3. Given a matrix $\mathbf{M} \in \mathbb{R}^{r\times c}$, and a vector $x\in \mathbb{R}^c$ let $M_i$ be the ith row of the $M$. The matrix product is defined as: $$\mathbf{Mx} \ =\ \mathbf{M}\begin{bmatrix} x_0\\ \vdots \\ x_{c-1}\\ \end{bmatrix} \ =\ \begin{bmatrix} \mathbf{M_0}\\ \vdots \\ \mathbf{M_{r-1}}\\ \end{bmatrix}\mathbf{x} \ =\ \begin{bmatrix} \ \mathbf{M_0 \cdot x}\ \\ \vdots \\ \ \mathbf{M_{r-1} \cdot x}\ \\ \end{bmatrix} $$ Further, given a matrix $N \in \mathbb{R}^{c\times m}$ we define $$ MN = \begin{bmatrix} \mathbf{M_0\cdot N^T_0} \cdots\mathbf{M_0\cdot N^T_c} \\ \vdots \ddots \vdots \\ \mathbf{M_c\cdot N^T_0} \cdots \mathbf{M_c\cdot N^T_m}\end{bmatrix} $$ And we have $MN \in \mathbb{R}^{r\times m}$ 4. $\mathbf{M} \in \mathbb{R}^{r\times c}$ implies that the function $f(x) = \mathbf{Mx}$ can map $\mathbb{R}^{c\times 1} \to \mathbb{R}^{r\times 1}$. 5. $\mathbf{M_1} \in \mathbb{R}^{d\times c}$ and $\mathbf{M_2} \in \mathbb{R}^{r\times d}$ implies $f(x) = \mathbf{M_2M_1x}$ can map $\mathbb{R}^c \to \mathbb{R}^r$. :::info Given this and your own knowledge, try solving these: - __Prove that $(2) + (3)$ implies $(4)$__. In other words, use your understanding of the inner and matrix-vector products to explain why $(4)$ has to be true. - __Prove that $(4)$ implies $(5)$__ ::: ### B. Differentiation Recall that differentiation is finding the rate of change of one variable relative to another variable. Some nice reminders: $$\begin{align} \frac{df(x)}{dx} & \text{ is how $f(x)$ changes with respect to $x$}.\\ \frac{\partial f(x,y)}{\partial x} & \text{ is how $f(x,y)$ changes with respect to $x$ (and ignoring other factors)}.\\ \frac{dz}{dx} &= \frac{dy}{dx} \cdot \frac{dz}{dy} \text{ via chain rule if these factors are easier to compute}. \end{align}$$ Some common derivative patterns include: $$\frac{d}{dx}(2x^3 + 4x + 5) = 6x^2 + 4 $$$$\frac{\partial}{\partial y}(x^2y^3 + xy + 5x^2) = 3x^2y^2 + x % $$$$\frac{d}{dx}(x^3 + 5)^3 = 3(x^3 + 5)^2 \times (3x^2) $$$$\frac{d}{dx}\ln(x) = \frac{1}{x} $$ :::info Given this and your own knowledge: Use (and internalize) the log properties to solve the following: $$\frac{\partial}{\partial y}\ln(x^5/y^2)$$ The properties are as follows: $$\log(x^p) = p\log(x)$$ $$\log(xy) = \log(x) + \log(y)$$ $$\log(x/y) = \log(x) - \log(y)$$ Solve the following partial for a valid $j$ and all valid $i$: $$\frac{\partial}{\partial x_j}\ln\bigg[\sum_i x_iy_i\bigg]$$ Consider using the chain rule. Let $g_1(x) = \sum_i x_iy_i$... ::: ### C. Probability There exist events that are __independent__ of each other, meaning that the probability of each event stays the same regardless of the outcome of other events. For example, consider picking a particular 3-digit number at random: $$P(x = 123) = P(x_0 = 1)P(x_1 = 2)P(x_2 = 3) = (1/10)^3 = 1/1000$$ Alternatively, some events are __dependent__ on other events. For example, consider 3 draws from a set of 1 red, 1 green, and 1 blue ball. $$P(b_0 = R) = 1/3$$ $$P(b_1 = G\ |\ b_0 = R) = 1/2$$ $$P(b_2 = B\ |\ (b_0 = R) \cup (b_1 = G)) = 1/1$$ This starts off the notion of _conditional probability_, where some components are realized conditional to other components. An important formula for conditional probability is Bayes' Theorem: $$P(A|B) = \frac{P(B|A)P(A)}{P(B)}$$ Whenever events happen at random, they happen with some probability. This is governed by some _probability distribution_. For example, $X \sim P(x)$ is a _realization_ (or _variate_, or _random variable_) of the $P(x)$ distribution. Of note: - The distribution may be parameterized by some factors. For example, $X \sim \mathcal{N}(\mu=0, \sigma=1)$ is a distribution _similar to_ (AKA an instance of) the unit normal distribution. - The distribution may depend on something. For example, the variate may depend on the realizations of some other distribution i.e. with $P(X|Z)$. These distributions are equipped with _expectation_ functions $\mathbb{E}$ and $\mathbb{V}$ that reveal their expected behavior (mean and variance, respectively). These also usually suggest the _long-term equilibrium behavior_, or the distribution of realizations after many realizations are drawn and accumulated. - __Discrete Probability Distribution__ governs discrete events $\{e_0, e_1, ...\}$. - If the number of possible events is finite such that $x \in \{e_0, e_1, ..., e_n\}$, there are finite set of associated probabilities $\{P(e_0), P(e_1), ..., P(e_n)\}$. - The list of probabilities must add up to 1. This implies there is a 100\% chance of an event being... one of the possible events. - Probabilities must be non-negative - __Continuous Probability Distribution__ governs continuous values. For example, the unit normal distribution mentioned before. :::info Given this and your own knowledge: - You're trying to train up a cat/dog classifier which outputs prediction between 0 and 1. Given that the input is in fact an image of a cat or dog, the truth is always one of those two. As such, the output is a probability distribution $Y$ with unknown $P(Y = y)$ for all possible $y$ in the domain of $Y$. Your friend knows that their dataset $\mathbb{D} = (\mathbb{X}, \mathbb{Y})$ is balanced between cats and dogs, and so argues that $P(Y=y)$ is equal for all plausible $y$. - If your friend's assumption were correct, what value of $P(Y=y)$ would make this a valid probability distribution for all $y$ in the domain of $Y$? - Is your friend's assumption correct? Why or why not? ::: ## 3. Python Review ### A. Basic Python Review For an overview of python syntax and common python uses we recommend checking out <ins>[this](https://www.w3schools.com/python)</ins> python tutorial for a refresher. ### B. Advanced Python #### I. Dunder Methods ##### Constructors :::info Write up a Python class for a `Square` whose constructor (the `__init__` method) takes in a string `name` and numeric `length` field. You can use this code to verify functionality: ```python square1 = Square("square1", 5) square1.name == "square1" square1.length == 5 ``` ::: ##### Calling We can give a class some special interaction patterns with other Dunder methods. If we give a class, say a `Multiplier`, a `__call__` method, then we can specify what happens when we call an instance of the class! For example: ```python multer = Multiplier() multer(5, 10) ``` This is effectively the same as calling: ```python multer.__call__(5, 10) ``` :::info Write up a Python class `Multiplier` which, when called on two integers, returns the product of the two integers Use this code to check: ```python multer = Multiplier() multer(5, 10) == 50 ``` ::: #### II. Classes and OOP :::warning Throughout this course, we will be working with Python extensively. Though you won't need to be an OOP expert, we do expect some basics which help Deep Learning libraries work in organized and efficient ways. ::: ##### Objects vs Classes OOP (Object Oriented Programming) strongly focuses on objects, which encompass any value, varaible, *etc*. It's really any "tangible" thing. ```python thing1 = "i am an object!" thing2 = 1234567 ... ``` However, we might find it useful to organize these things into classes of things, with properties like instance variables or methods shared across all things that are members of the same class. You might be familiar with Python's `str`, `int`, and `float` classes, for example. When working with Python, you'll almost always be working with objects which are instances of classes. You can check the type of a variable with the built-in `type` method! ##### Inheritance If classes are sets and objects are set elements, then we also need subsets and supersets! In Python, we can make a "child" class inherit all the methods from a "parent" class like so: ```python class ChildClass(ParentClass1): ``` ##### Class-level vs Instance Variables Consider the instance of our `Square` earlier. ```python square1 = Square("square1", 5) square1.name == "square1" square1.length == 5 ``` :::success Notice that the `name` and `length` variables for are instance variables! ::: In contrast with instance variables, class level variables are shared across all instances of the class. Say in the declaration of `Square`, we included this: ```python class Square: shape = "square" def __init__(...): ... ``` Then, if we checked `square1.shape` or `square2.shape`, you'll notice that they both return the string `"square"`. This also applies to checking the class directly: `Square.shape` will also return `"square"`! `shape` is a class level variable because it is shared across the whole class! :::warning Class variables can be redefined from any instance (*e.g.* `square1.shape = "rectangle"`) or directly through the class (*e.g.* `Square.shape = "rectangle"`), but we **strongly** recommend doing it through the class directly. ::: ##### Putting it Together :::info Task 1. __Make a parent class named `Logger`. Above the constructor, include this line:__ ```python logging_tape: LoggingTape | None = None ``` This will be our log of things that happen! The `: LoggingTape | None` indicates that the variable `logging_tape` will either be of type `LoggingTape` (which we'll make in just a second), or `None` ::: #### IV. Context Managers Context managers in Python are a great tool for temporarily defined things. For instance, you have probably seen ```python with open('file.txt', 'r') as f: # do things with the file f #f is now closed, do other things! ``` You'll notice that `f` is only properly defined as being the file opened with read permissions while in the "context" of the `with` statement (within the `with` statement's indent block) This is a context manager! Context managers derive their functionality from special Dunder Methods `__enter__` and `__exit__`. Let's work through an example! Say we want to set the **class variable** `Logger.logging_tape` to be a new `LoggingTape`, but only temporarily. Sounds perfect for a `with` statement, huh? Here's a starter: ```python class LoggingTape: def __init__(self): ... def __enter__(self): ... def __exit__(self, *args): ... def add_to_log(self, new_log): ... def print_logs(self): for log in self.logs: print(log) ``` We might see some code using the `LoggingTape` like ```python= with LoggingTape() as tape: ... ... ``` On line 1, `LoggingTape`'s `__enter__` method is called to enter the `with ` statement. Then, after the indent block (so after line 2 but before line 3), `LoggingTape`'s `__exit__` method is called to exit from the `with` statement. :::info In `LoggingTape`'s constructor, make an empty list called `logs`. We'll store strings as messages of logs of whatever happened Then, in `__enter__`, set `Logger.logging_tape = self` and `return self`. We're setting a class level variable! Next, in `__exit__`, set `Logger.logging_tape = None` In `add_to_log`, append `new_log` to the end of `logs`. ::: Now, check it out this code block: ```python= with LoggingTape() as tape: #runs LoggingTape's __enter__() #Logger.logging_tape is now defined as tape (from line 1)! tape.add_to_log("Hi!") #runs LoggingTape's __exit__() #Now Logger.logging_tape is defined as None ``` This might seem a little trivial now, but what this enables us to do is have any `Logger` class record to `tape` while inside the `with` statement (lines 2-3 in the example)! Say we have a car class: ```python= class Car(Logger): def travel(self, distance): self.logging_tape.add_to_log(f"Traveled Distance {distance}") ``` ```python= car = Car() with LoggingTape() as tape: car.travel(5) tape.print_logs ``` :::success The output will be "Traveled Distance 5". The LoggingTape kept track of the logged item automatically for us. ***I wonder if this will be useful in Homework 2...*** ::: ### C. Libraries for this Course #### I. Numerical Python (NumPy) ##### Making NumPy arrays (also shapes) ```python= import numpy as np #1. Make a NumPy array of zeros with shape (5,10). zeros = ... assert(zeros.shape == (5,10)) assert(np.max(zeros) == np.min(zeros) == 0) # 2. Do it again, but make it full of ones! ones = ... assert(ones.shape == (5,10)) assert(np.max(ones) == np.min(ones) == 1) # 3. Slice the array to get the first row of ones! first_row = ... assert(first_row.shape == (10,)) # 4. Slice the array to get the first *column* of ones! # (Hint: Try passing in `:` as one of the slice indices!) first_col = ... assert(first_col.shape == (5,)) # 5. Create a new dimension on the ones array. # You should end up with shape $(5,10,1)$. (Check out `np.expand_dims`) expanded = ... assert(expanded.shape == (5,10,1)) # 6. Cast a list `[1,2,3]` into a NumPy array. # Then, change the first element to `4`. arr = ... ... np.testing.assert_array_equal(arr, np.array([4,2,3])) # 7. Make a NumPy array of integers $0$ to $9$, # inclusive, in shape $(2,5)$ using `np.arange` and `np.reshape`. incr = ... ... assert(incr.shape == (2,5)) np.testing.assert_array_equal(incr, np.array([[0,1,2,3,4],[5,6,7,8,9]])) # 8. With incr from 7., use `np.vstack` to add # a new row `[10, 11, 12, 13, 14]`. vstacked = ... v_target = [[0, 1, 2, 3, 4], [5, 6, 7, 8, 9], [10, 11, 12, 13, 14]] np.testing.assert_array_equal(vstacked, np.array(v_target)) # 9. With incr from 7., use `np.hstack` to add a new column of 0's. # *Hint: think about the dimensionality of the original matrix. # What dimensions do you need to represent a new column?* hstacked = ... h_target = [[0, 1, 2, 3, 4, 0], [5, 6, 7, 8, 9, 0]] np.testing.assert_array_equal(hstacked, np.array(h_target)) ``` ##### Basic Operations (addition, subtraction, scalars) ```python= import numpy as np # 1. Add two NumPy arrays of ones with shape $(5,10)$. ones_1 = ... ones_2 = ... sum_ones = ... assert(sum_ones.shape == (5,10)) assert(np.max(sum_ones) == np.min(sum_ones) == 2) # 2. Subtract a NumPy arrays of ones with shape $(5,10)$ from another one. # Reuse ones_1 and ones_2 diff_ones = ... assert(diff_ones.shape == (5,10)) assert(np.max(diff_ones) == np.min(diff_ones) == 0) # 3. Multiply a NumPy array of ones with shape $(5,10)$ by the scalar two. scaled = ... assert(scaled.shape == (5,10)) assert(np.max(scaled) == np.min(scaled) == 2) ``` ##### Matrix Operations (matrix product, element-wise product/division, mean, axes) ```python= #1. Use NumPy matrix multiplication (`np.matmul` or using the `@` symbol) # to calculate the inner product of vectors v1, v2 v1 = np.array([1,2,3]) v2 = np.array([3,2,1]) inner_prod = ... assert(inner_prod == 10) # 2. Use NumPy matrix multiplication (`np.matmul` or using the `@` symbol) # to calculate the matrix product of matrices m1 and m2. m1 = np.array([[1, 2, 3],\ [0, 1, 0]]) m2 = np.array([[4, 6],\ [2, 1],\ [0, 5]]) mat_prod = ... m_target = np.array([[8, 23],\ [2, 1]]) np.testing.assert_array_equal(mat_prod, m_target) # 3. Use NumPy element-wise matrix multiplication (using the `*` symbol) # to calculate the element-wise product of matrices m1 (above) and m3). m3 = np.array([[4, 6, 2],\ [1, 0, 5]]) elem_prod = ... e_target = np.array([[4, 12, 6],\ [0, 0, 0]]) np.testing.assert_array_equal(elem_prod, e_target) # 4. Use NumPy element-wise matrix division (using the `/` symbol) # to calculate the element-wise quotient of matrices m1 and m4. m4 = np.array([[4, 6, 2],\ [1, 1, 5]]) quot = ... q_target = np.array([[0.25, 0.33333333, 1.5],\ [0., 1., 0.]]) np.testing.assert_allclose(quot, q_target) # 5. Use NumPy functions to find the average of the entries in matrix m5. # Do it again, but get the average per row # Then, do it per column m5 = np.array([[1,2],\ [0,1]]) avg = ... row_avg = ... col_avg = ... assert(avg == 1) np.testing.assert_allclose(row_avg, [1.5, 0.5]) np.testing.assert_allclose(col_avg, [0.5, 1.5]) ``` ##### Logical Operations (masking, `np.where`, `argmax`) ```python= # 1. Use a masking operation on matrix m1. # We want masked to be a matrix whose entries are `False` where # m1's entries are less than $6$, and `True` otherwise. m1 = np.array([[1, 9, 5],\ [8, 0, 2]]) masked = ... masked_target = np.array([[False, True, False],\ [True, False, False]]) np.testing.assert_array_equal(masked, masked_target) # 2. Use `np.where` on matrix m1 to # keep entries greater than or equal to 6 # and replace any entries less than 6 with 0. replaced = ... replaced_target = np.array([[0, 9, 0],\ [8, 0, 0]]) np.testing.assert_array_equal(replaced, replaced_target) # 3. Use `np.argmax` on matrix m1 to find, per row, # the index of the greatest element. max_inds = ... target_inds = [1,0] np.testing.assert_array_equal(max_inds, target_inds) ``` ##### Broadcasting Broadcasting is a powerful mechanism that allows numpy to work with arrays of different shapes when performing arithmetic operations. Frequently we have a smaller array and a larger array, and we want to use the smaller array multiple times to perform some operation on the larger array. Take a glance through some of these simple examples! ```python import numpy as np # We will add the vector v to each row of the matrix x, # storing the result in the matrix y x = np.array([[1,2,3], [4,5,6], [7,8,9], [10, 11, 12]]) v = np.array([1, 0, 1]) y = x + v # Add v to each row of x using broadcasting print(y) # Prints "[[ 2 2 4] # [ 5 5 7] # [ 8 8 10] # [11 11 13]]" ``` ```python import numpy as np # Let's say we have the following matrix A. A = np.random.random((50, 100, 20)) # We can imagine A as 50 instances of (100,20) matrices. # We have the following matrix B B = np.random.random((20,40)) # We want to multiply each (100,20) instance of A by B, # we can do this because the dimensions match up: 20 = 20 print(A @ B.shape) # output should be of shape (50, 100, 40) # each of the 50 (100,20) is multiplied by the (20,40) matrix to yield (100, 40) ``` :::info Consider the following arrays: ```python A = np.array([[0,1,2],[3,4,5]]) # shape (2,3) B = np.array([[1,1,1]]) # shape (1,3) C = np.array([[-1,-1,-1],[1,1,1]]) # shape (2,3) ``` 1. Create matrix `D` as A - B using broadcasting 2. Create matrix `E` with shape (3,2) by reshaping `C` 3. Create matrix `F` with shape (2,2) by matrix multiplying `D` by `E` You can use the following to confirm your results look as they should! ```python assert(np.all(D == [[-1,0,1],[2,3,4]])) assert(np.all(E == [[-1,-1],[-1,1],[1,1]])) assert(np.all(F == [[2,2],[-1,5]])) ``` ::: #### II. Tensorflow Let's try some of the important examples again, but with Tensorflow. You'll find that for a lot of things, you can just replace `np` with `tf`. However, in some cases, the method might be named something else. Again, you should get used to searching for methods you'd like to use in the [documentation](https://www.tensorflow.org/api_docs). :::success **Hint:** If you know the Numpy method you'd like to use, you can usually get away with googling <numpy method name> in tensorflow. ::: ##### Making Tensors (also shapes) ```python= import tensorflow as tf #1. Make a tf Tensor of zeros with shape (5,10). zeros = ... assert(zeros.shape == (5,10)) assert(tf.reduce_max(zeros) == tf.reduce_min(zeros) == 0) # 2. Slice the array to get the first *column* of ones! # (Hint: Try passing in `:` as one of the slice indices!) first_col = ... assert(first_col.shape == (5,)) # 3. Create a new dimension on the ones array. # You should end up with shape $(5,10,1)$. (Check out `tf.expand_dims`) expanded = ... assert(expanded.shape == (5,10,1)) # 3. Cast a list `[1,2,3]` into a tensor. (Check out tf.convert_to_tensor) arr = ... ... assert(tf.reduce_all(arr == [1,2,3])) # 4. Make a tensor of integers $0$ to $9$, # inclusive, in shape $(2,5)$ using `tf.range` and `tf.reshape`. incr = ... ... assert(incr.shape == (2,5)) assert(tf.reduce_all(incr == [[0,1,2,3,4],[5,6,7,8,9]])) ``` ##### Basic Operations (addition, subtraction, scalars) ```python= import tensorflow as tf # 1. Add two tensors of ones with shape $(5,10)$. ones_1 = ... ones_2 = ... sum_ones = ... assert(sum_ones.shape == (5,10)) assert(tf.reduce_max(sum_ones) == tf.reduce_min(sum_ones) == 2) # 2. Multiply a tensor of ones with shape $(5,10)$ by the scalar two. scaled = ... assert(scaled.shape == (5,10)) assert(tf.reduce_max(sum_ones) == tf.reduce_min(sum_ones) == 2) ``` ##### Matrix Operations (matrix product, element-wise product/division, mean, axes) ```python= # 1. Use Tensorflow matrix multiplication (`tf.matmul` or using the `@` symbol) # to calculate the matrix product of matrices m1 and m2. m1 = tf.convert_to_tensor([[1, 2, 3],\ [0, 1, 0]]) m2 = tf.convert_to_tensor([[4, 6],\ [2, 1],\ [0, 5]]) mat_prod = ... m_target = tf.convert_to_tensor([[8, 23],\ [2, 1]]) tf.debugging.assert_equal(mat_prod, m_target) # 3. Use NumPy element-wise matrix multiplication (using the `*` symbol) # to calculate the element-wise product of matrices m1 (above) and m3). m3 = tf.convert_to_tensor([[4, 6, 2],\ [1, 0, 5]]) elem_prod = ... e_target = tf.convert_to_tensor([[4, 12, 6],\ [0, 0, 0]]) tf.debugging.assert_equal(elem_prod, e_target) # 5. Use Tensorflow functions to find the average of the entries in matrix m5. # Do it again, but get the average per row # Then, do it per column m5 = tf.convert_to_tensor([[1,2],\ [0,1]], dtype=tf.float32) avg = ... row_avg = ... col_avg = ... assert(avg == 1) assert(tf.reduce_all(row_avg == [1.5, 0.5])) assert(tf.reduce_all(col_avg == [0.5, 1.5])) ``` ##### Logical Operations ```python= # 1. Use a masking operation on matrix m1. # We want masked to be a matrix whose entries are `False` where # m1's entries are less than $6$, and `True` otherwise. m1 = tf.convert_to_tensor([[1, 9, 5],\ [8, 0, 2]]) masked = ... masked_target = np.array([[False, True, False],\ [True, False, False]]) tf.debugging.assert_equal(masked, masked_target) # 2. Use `tf.argmax` on matrix m1 to find, per row, # the index of the greatest element. max_inds = ... target_inds = [1,0] assert(tf.reduce_all(max_inds == target_inds)) ``` ##### Broadcasting :::info Consider the following arrays: ```python A = tf.convert_to_tensor([[0,1,2],[3,4,5]]) # shape (2,3) B = tf.convert_to_tensor([[1,1,1]]) # shape (1,3) C = tf.convert_to_tensor([[-1,-1,-1],[1,1,1]]) # shape (2,3) ``` 1. Create matrix `D` as A - B using broadcasting 2. Create matrix `E` with shape (3,2) by reshaping `C` 3. Create matrix `F` with shape (2,2) by matrix multiplying `D` by `E` You can use the following to confirm your results look as they should! ```python assert(tf.reduce_all(D == [[-1,0,1],[2,3,4]])) assert(tf.reduce_all(E == [[-1,-1],[-1,1],[1,1]])) assert(tf.reduce_all(F == [[2,2],[-1,5]])) ``` ::: # Conclusion Wohoo! You just completed your first assignment of CSCI1470! Submit your solutions to gradescope for the autograder to test your numpy/tensorflow sections.

Syntax	Example	Reference
# Header	Header	基本排版
- Unordered List	Unordered List
1. Ordered List	Ordered List
- [ ] Todo List	Todo List
> Blockquote	Blockquote
Bold font	Bold font
Italics font	Italics font
~~Strikethrough~~	~~Strikethrough~~
19^th^	19^th
H~2~O	H₂O
++Inserted text++	Inserted text
==Marked text==	Marked text
[link text](https:// "title")	Link
![image alt](https:// "title")	Image
`Code`	`Code`	在筆記中貼入程式碼
```javascript var i = 0; ```	`var i = 0;`
:smile:		Emoji list
{%youtube youtube_id %}	Externals
$L^aT_eX$	L^aT_eX
:::info This is a alert area. :::	This is a alert area.