技术笔记

Getting help

Emacs is a sophisticated self-documenting editor. Every facet of Emacs is searchable or describable.

Emacs’s help system is roughly divided into three parts and knowing which one you need and when will save you time.

The Info Manual

M-x info or C-h i to access.

Look up documentation for a command by typing C-h F and at the prompt enter the name of a command.

Apropos

Describe System

Moving by character

Four fundamental movement

Moving by line

Moving by word

Subword and superword movement

Moving by s-expression

Down and up List

Forward and backward list

Other movement commands

Moving by Paragraph

a set of variables that define the beginning and end of a paragraph:

Moving by sentence

You can alter this behavior by customizing (with M-x customize-option) the following variables:

Moving by Defun

Moving by Pages

Scrolling

In Mac iterm2, you should use M-PageUp and M-PageDn to scroll the other window.

Move to begin and end

When you move to the beginning or end of the buffer, Emacs will place the mark – an invisible location marker – where you came from, so you can return to your old position. For instance, if you type M-< to jump to the beginning of the buffer, you can type C-u C-<SPC> to go back.

Numeric arguments

Move down 50 lines

1

M-5 0 C-n

C-u alone has the special meaning of “four times”, moves forward sixteen characters

1

C-u C-u C-f

Take a breath and be relax

My favorite player Kobe has gone. So sad these days. The beginning of 2020 seems so hard. Hold on, man, take a breath and be relax. We will win the battle!

We always want run faster, but not solid on the way. We should slow down, make little things right.

Learning basics

• Math basic is so important, calculus, linear algebra, probability. Figure out every detail.
• Emacs, thinking in emacs way, try to forget vim.
• Dig deep into any concept, try to proof each formula.

Sushi and ramen spirit

• Sushi and ramen is a spirit, they tell us to focus on one thing, do it all life time, and do it better.
• To be a coder, we first need have faith, and then working hard, keep going, focus and do one little thing better.

Let's begin

I'll begin the basics from now on. Study every details I face.

8 and 24 R.I.P &

A new day has come!~

What is reinforcement learning

Reinforcement learning is every where in the world. I learn to write blogs in English. I learn to use emacs to do coding jobs. I do some stock trading.

Reinforcement learning includes:

• Policy: agent's behavior function
• Value function: how good is each state and/or action
• Reward signal: defines the goal of a reinforcement learning problem

Policy

Policy is the agent's behavior, it is a map from state to action

• Deteministic policy:
• Stocastic policy:

Value function

Value funciton is a prediction of future reward, used to evaluate the goodness/badness of states,and therefor to select between actions.

Exploration and Exploitation

To obtain a lot of reward, a reinforcement learning agent must prefer actions that it has tried in the pat and found to be effective in producing reward. But to discover such actions, it has to try actions that it has not selected before. The agent has to exploit what it has already exprienced in order to obtain reward, but it also has to explore in order to make better action selections in the future.

Like variance and bias in machine learning, we always need to make trade-off. So to be or not to be, this is a problem.

I used to use vi as my coding tools. I want expore a new tool called emacs. It may bring more reward in the future, but it seems very hard at the beginning.

Don't repeat yourself

When you are coding, remember that never do same thing twice. For example you want indent some lines.

1
2
3
4
5

auto main() -> int
{
std::cout << "indent me\n";
std::cout << "indent me again\n";
}

Never move to line 3, keypress 4 spaces and move to line 4, keypress 4 space again.

Use vi, there are many ways

• move to line 3, shift v enter vi visual mode, j select line 4, shift > to increase indent
• move to line 3, shift v enter vi visual mode, :normal i and press 4 spaces and enter
• move to line 3, shift v enter vi visual mode, shift > and then move to line 4, press .
• move to line 3, ctrl v block select, and j to select next line, shift i to insert 4 spaces, esc

Also, you can use awk and sed to achieve the same effects.

Actions on the code

Actions on the code, or code transformations, is a serial's operations on the code. Record all operations into tape, called macro, and the macro can be applied to some other places, mapping the code from origin state to desired state.

Operation space

All possible operations compose an Action space. AI should learn to choice some actions, transform the code to a stable state, a state may be pass compile or pass unit test. If AI can do this, we call it auto coding.

Learning from history and some basic rules

Bayesian learning

We init some prior rules to the bot. He will update the prior with posterior.

Reinforcement Learning

Use some reinforcement strategies to perform better actions on the code.

Interactive with human

Consider the indent example above, when I indent line 3, bot help me indent line 4, is it stable? Yes, done!

Hello Yuguo

Bayesian

"Pattern Recognition and Machine Learning" said everything has a Bayesian version.

liyapunov

For origin coding system, given some input , and make the system "lyapunov stability".

Reinforcement Learning

The bot keep mute the code, and at last it pass all tests.

Meta programming

Code and be easily manipulated in multiple levels, source, AST, IR.

2019 bye~

I'am yuguo, happy new year and I'll learn how to auto coding in 2020.