HW#7 Complex Constraints (Synchronization)

建議工具

使用 Claude Sonnet 4 推理模式(手動切換，免費用戶額定時間內只能使用三次)

使用 ChatGPT 5 推理模式(自動切換)

使用 Gemini 2.5 Pro 免費額度最高 1M tokens (永遠推理模式)

使用 Grok 4 推理模式(自動切換)

---

課堂練習 

Deadline: This Saturday at 23:59

Send all the share links to  me chang212@gmail.com by email with subject HW#7 [your id, your name]

任選兩題

1. 推理模式 special topic Smart Dog Ball Retrieval 

Given resources as below

• Build an animation of A* search tree (artifact) based on A* template
• more complete and more aesthetic

Build A* tree synced with animation 

Examples: 

• Noodle +  Wave
• Noodle + blue pool 

backtrack

Build an animation of A* search tree (artifact)

\

Optimal solution by A*

 

pruned by A* (illegal state)

 

2. 搶救感恩節晚餐大作戰講義 題目 結果視覺化

至少做兩種工程圖

 狀態圖(State Diagram) 

 (interactive timeline) artifact state diagram

 (interactive timeline) artifact map version

state diagram with aligned timeline

看板圖 (Kanban)

 (interactive timeline)

流程圖(Flow chart)

 (interactive timeline) artifact state diagram

flow

flow

dashboard

3. Dinner Operation A* macro thinking

To-do 1:

Build A* search tree

 

To-do 2:

Build Animation Synced with  A*

Advanced visualization with backtrack included and transit animated (Data visualization techniques)

HW#6 Smart AI

任選2題

 

1. Try smart search. Prompt AI models to generate A* search Algorithm

A* search

2. for 推理模式 special topic Smart Dog Ball Retrieval 

A* search tree

Build an animation of A* search tree based on A* template

3. 搶救感恩節晚餐大作戰講義 題目 

Plan a feasible schedule. Then verify the schedule. Then visualize your results. 

繳交範例

ChatGPT validation, A* Macro*, Visualization

Claude validation, reasoning*, Viz

檢查你的感恩節計畫是否正確

Use validator to check up the plan

validator

如果你希望驗證結果可輸出為app

1. 先進入 validator

2. 按下右上方 customize 鍵

3. 接下來的操作步驟

4. 結果輸出為app

可使用狀態圖(State Diagram)，看板圖 (Dashboard), 流程圖(Flow chart) 進行視覺化

 (interactive timeline)

 (interactive timeline)

flow

Hint 

優化解可以達成所有人四點前到家

Ｈint

如果你要使用現成版型(artifact type) 做行程視覺化，步驟如下

0 找到版型程式  （工程圖下方連結）

1 按介面右上方 customize  鍵

2. 輸入你的文字版時間排程
3 得到專業工程圖

 

Make AI smarter (P3)

 

Puzzle 3 

macro thinking in A*

animation

tree search animation shared, animation artifact

animation synced with A* tree (share)

Make AI smarter (P1)

Build Puzzle 1 viz search tree based on A* macro thinking.  

Tree and animation. (artifact)

Decoration within tree

synced with animation

Make AI smarter (Maze, Smart Dog, Dinner Operation)

Maze syncing with search tree. 2

推理模式 special topic Smart Dog Ball Retrieval 

Build an animation of A* search tree (artifact) based on A* template

more complete and more aesthetic

A* tree synced with animation  (noodle, wave)

Noodle + blue pool 

Way to go

Backtrack

Search Tree based on A* as heuristic (guidance)

Dinner Operation A* macro thinking

 

All in one  A*

Solving Electronics

schematic

A* macro thinking framework

A* macro

Framework  (circuit analysis search tree)

Visualization 

2-stage amp

Bode plot visualization

Colab

CE  (Small-signal transistor gain analysis)

standard strategy

solve for gain

do not ignore bias current

calculate miller effect

calculate poles for lower cut off

calculate poles for higher cut off

plot bode plot gain vs. Hz

plot bode plot phase vs. Hz

1. Solve for the gain. 

source

2. Solve for the gain. 

source

schematic

Gemini 2.5 Pro for s-stage amp, Freq Response

Claude Sonnet 4.5 Freq resp (all in one) staged

3. solve for the gain

 (source)

以上參考答案

(1) -4 (2) 8513 (3) -165 答案誤差10%以內都是正常的

4. Design a two stage amplifier using bjt.

the first stage is class A. the second is class AB. freq response cut off at 10Hz and 25 KHz

 

Make AI smarter (Smart Dog Ball Retrieval)

endgame

化繁為簡 思維鏈 (chain of thought)

深謀遠慮 RL(reinforced learning)

A* 

推理模式 special topic Smart Dog Ball Retrieval 

Build an animation of A* search tree (artifact) based on A* template

Search Tree based on A* as heuristic (guidance)

Supplemental

A* as heuristic (guidance)

A* as heuristic (guidance)

 Model Smart Dog Ball Retrieval as a search problem.

We can use A* to get a solution.

今天我們在課堂上提到 深謀遠慮
(Reinforcement learning) 是AI 能夠聰明推理的主因

茫茫大海中，A* 就常常被用來指引 AI 的推理要往哪個方向前進
不至於盲目搜尋（brute force)

A* 演算法不一定總是得出最佳解，但這題A*剛好可以得到最佳解

 

A* modeling, animated

\

Optimal solution by A*

 

pruned by A* (illegal state)