Design in an editor, save to a file, and then load the file to the other simulator. That is an old generation way. 

The new generation way is: designing and simulating, all of interactions are in the same distributing virtual reality by the access permission management without any file loading and saving. 

The general purpose Cyber Physical System needs:

1 kind of thread to read the inputs from each local sensor to update the Distributing Virtual Reality 

1 kind of thread to recognize the 3D images in the Distributing Virtual Reality to be 3D objects 

1 kind of thread to offer the local user interaction with the Distributing Virtual Reality 

1 kind of thread to read the registered senses in the Distributing Virtual Reality to update each local actuator in the real world

Spherical Shell

Basically, 6 pieces of the information eye (face) masks form a spherical shell, and it becomes a head hood when 1 piece of them is opened. 

Graph machine & Replacement machine

Graph machine 是模拟人在阅读或算式的过程是先同时看到几个重要关键的字符再系结旁边的字符组成认知图而理解一段文式然后进行思考。

Replacement machine 是模拟人在阅读或算式的过程会逐步将文式片段转换成自己已理解的片段，多次返复代换到最后其实就是结果与答案了。

Graph machine & Replacement machine

Graph machine 基本是依结合优先顺序透过 comparation gate array 同时并行位移比对将 match 到多个 Sense 中取左边第一个连接到动态配置的 virtual wire 上，之后 sense 欲变更 virtual wire 的连接状态亦是相同的方式输出 senses 到 comparation gate array 上。

Replacement machine 基本透过 comparation gate array 与 replacement table 做同时并行位移比对将 match 到的片段 replace，返复进行至没有任何 match 了即为运算完成。排在 table 前面的优先 replace 即达 graph machine 中结合力较强的 operator 优先结合的效果，过程中遇到 definition keyword 就将片段加入 table 或从其刪除。

两者对括号的处理方式都是在比对到左括号时都暂不 connect 或 replace ，后面再一并进行消括号的动作，自多个左括号处同时并行向右位移比对出右括号，左右括号间仅存在一 Sense 即消掉左右括号，接着再重复前面的动作 loop 至不再比对出任何括号，若自左括号处向右位移比对到左括号则什么都不做、等待下一 loop 后面再重复进行一次。

Graph machine 是接线生的角色，接完线后 graph 自己是线路、自己在动；Replacement machine 的 senses 则完全是被动处理，纯粹是 data。

Can these two machines be combined to be one machine?

Not login to the server, login to the local sense and the system automatically synchronizes the senses @ different machines by caching each other @ the distributing virtual memory. 

How to make the access permission security in this condition? 

The key is there should be the hardware circuit for the access permission control built in the memory and storage, and then the server will be not necessary. 

Deformable wing

吾本欲照此改滚球面展开翼然多个问题：翼面要考虑滚行接触地面的应力。

立体环场空间状态扫描采主动式射频回波测距在军用上会暴露行踪、在民用上亦易造成干扰；有一种演算法可以从多角度被动式摄得的平面影像合成立体模型可是运算量极为庞大。道路上车辆颇多，有车联网共享，多角度影像是随时大量在取得，若各车备多些运算资源就可以分担中心的运算。车多时倾向多车组成的 near cloud 而疏解与中心的通讯负担，车少时附近运算资源不足而远端通讯带宽反就充足所以倾向 far cloud，如此即需要底层分布式操作系统自动调节运算分布方为完整的 Cloud。

The submarine vision

The submarine only can hear the sounds but can not see the image? No no no, the computer can synthesis the visual graphic from the sonar data and integrate the sonar data from different submarines or another units by the sound communication to be a virtual reality basically, and render with the neural database advancedly.

The software operator

Implementing the operator by the memory has more cost than the fixed hardware operator. But in order to implement the flexible bandwidth, the link cost on the switches network is much more if using the hardware operator, so the cost spent on the software operator is acceptable.

Spheron 2019

Spheron 2019

载人不载弹、载弹不载人

有人球无人球联网协作战

载人球轻武 无人球重武装

球面小孔

滚球外视解决方案同武装解决方案：分布内嵌球面，附小盖于开口，球面离地面后开启、球面触地面前关闭。球面清污泥解决方案采喷气不采喷水，喷水另有用途。

球型限定

胶囊型载具长条状有两倍道路宽度才夠空间干溼分离但就没有球型载具內球舱无论环境如何动荡都不会倾斜的优点所以浴卧同室回干设施也是重点工作项目

舱门

斜滑盖舱门不会被水压住打不开，适用浅潜载具便于失控下沈时紧急手动开门逃生，其应力仅需应付登陆时滚行不会崩开；深水压会导致斜滑盖门滑开而失抗压球构形，若要深潜则再外包一层附掀盖式拱边抗压舱门之球壳方能应付深水压，并采多块外挂动态装卸式压舱以于意外发生时渐进抛弃浮上水面方得开启舱门。

Graph ≠ Graphic

The graph IC can replace the CPU but the graphic IC can not, and the graph IC can be the base for the graphic IC.

But is it possible about Graph = Graphic?

Perhaps it could be Graph = Graphic as Computer = Network, but my brain doesn’t reach there yet. I will think about that if I have time.

Sense Graph IC

As the processors in computers were never called electronic brain, the neural networks we usually called before should not be called neural networks. They belong to the 5th generation computers; the real neural networks belong to the 6th generation computers.

No matter the 5th or 6th generation processors, the Sense graph ICs were designed to be the base to satisfy both.

Computer = Network

" Network is a computer " will be made by the Dynamically Distributing Operating System; " Computer is a network " will start with combining the router and the computer to be one.

Hardware : Software

I/O access = memory access ➡ hardware access = software access ➡ remote hardware access = remote software access ➡ hardware service = software service ➡ hardware distribution = software distribution

Distributing Sense

operators + operands = operation

sensed operator = sense
sensed operand = sense
sensed operation = sense

operators and operands distributing on different hardwares
= distributing operation = distributing sense

Distributed Operation

host#1'operand#1 + host#2'operand#2

+ 是1号机1号元的 operator 与其同在一号机，所以现在有2个 senses 在1号机、1个 sense 在2号机，要怎么执行这个跨机的 distributed operation ？

在2号机对1号机使用 WCF service 下这道式子会先把1号机1号元下载到2号机然后才和2号机2号元一起当作远端呼叫1号机 operator + 的 parameters 上传到1号机，1号机1号元被下载了又上传多此一举白费流量，你会说还不如传 request 给 1号机的 Web server 呼叫 Servlet 在1号机执行。但若2号机2号元的 size 大于1号机1号元与 operator + 是否就不如将 operator + 下载到2号机执行？如 Applet 的做法 ( 把 Applet 看作是一个大 operator ) 。三种架构分别，但软件在跑一下是这情况一下是那情况，这就是为何要丢掉传统架构全面采用 DDA 架构。

🤖

The cerebellar motor control has been well trained, and the brain's on-the-spot analysis of the unfamiliar environments also has to be developed. The two-legged one needs the higher balance feedback control than one with more than four legs, and the cost is higher. Although the demand of the military is the ground sneaking reptiles better than the erect humankind and even asks the joints to rotate omnidirectionally: the most perfect artificial creature is the roll ball which can extend out the omnidirectionally rotating legs, arms, and wings from the whole surface of the sphere; except the humankind can and be willing to self-modify the gene, the human-like bodies will be always needed forever. And it is not expected to connect the neural mechanical body only after the natural body is damaged, so it is necessary to redesign the axis center rotation robot to become the anexoskeleton rotation armor one like the IronMan.

Sense space

• 无论 Sense 结构如何进化，其基本都是 graphs，只要 graph 中一 node 在 memory 被 refer 到即将其第一阶链结到的 nodes 从 storage 预载到 memory，如此即确保要 refer 到的 sense 都已在 memory，因一个 sense 定是透过另一个 sense 来 refer。

• 在 memory 中的 sense 获得更新则立即更新 storage 中的 sense ，因为还有别的 memory 在 cache 相同的 storage，所以也必须一併更新所有 cache。

• 当 memory 满时由 rerefer 频率少的 sense 优先 release 腾出空间。

如此外界仅存取 memory 而不直接触及 storage，于是不再有 storage I/O 的观念。

而将网络另一端的 memory 也如对 storage 般对待，于是不再有 network I/O 的观念。

这机制让每一运算机皆为一镜像平行宇宙，每一空间发生的变化都会在其它所有镜像空间同步发生。但其实一个空间并装不下整个宇宙，每个空间实际上只存有局部视野所及的宇宙，是随着视野的移动来由其它平行空间载入其余局部而移出视野的局部实际会在该空间消失但前提是先确保消失的局部有存在另外任一平行空间，于是所有平行空间联集起来是为完整的宇宙。相同的事件在不同的平行空间发生都会有相同的结果镜像到所有的平行空间，但在不同的平行空间触发所花费的时间可能不同而最终成为整体的费时，所以要选择在费时最少的平行空间触发事件以获得最少的整体费时。事件由一组 Senses 构成，每个 Sense 的本尊只会存在其中一个平行空间，其余平行空间的皆为分身。无论本尊位何，所有 Senses 一列排开何者 loading 最大就优先到何者本尊所在之平行空间去进行事件的发生方能让镜像映射所费最少，再者优先选择 performance 最高、space 最大、bandwidth 最大的平行空间触发，那才是原始真实的发生，其余平行空间的皆为结果的镜像。而每一 Sense 本身又是由一群 Senses 组成，所以一件事情的发生实际上是在好几个平行宇宙分部进行再将镜像联集起来。

在任一端将 operands 接上 operators，每一端都会立即同步这个结果，然后每一端都会因为 senses 的相接起了连锁反应也就是运算被触发跑了起来，但每一端运行速度又不一，在速度不一非同步的情况下任一端的变化又会立即更新每一端，那就大乱了，所以每一端都有一个反应前的抑制措施：每一端都各自进行哪个 sense 的 body size 最大的判断，仅在最大的那个 sense 的本尊在己端的前提下才触发。这只是决定在哪端运行的初步判断，作用是让 size 小的 sense body 缓存到 size 大的本尊所在处以使传输耗费最低，进一步还要把 performance、space、bandwidth 三个因素考虑进来，所以至此尚未确定要在此端触发运行。

把 sense 给 lock 住让 operation 进入 critical session 的方式很容易发生过长时独占导致的等待甚至 dead lock，所以当要让多端同时进行同一 operation 以达验算之效时采暂时产生一个 sense 的 clone 转而对其操作之方式在整个 operation 完成后才以 clone 去更新原 sense 然后删掉 clone。

平行空间断链没同步就会分歧出不同版本的 senses，分歧的平行空间合得回来吗？Sense 原本就可以鸟生蛋、蛋生鸟对不同的 senses 进行融合：取交集为 class sense、保留差集为 instance sense，所以并不是无解的问题。

operator ≡

The operator ≡ is used to define the replaceable linking state of the operators and operands dynamically like the function definition. If the operator can not be found in the hardware when it is asked to be allocated, it can be replaced with the other linking graph defined by the operator ≡. If the allocated graph is already in usage, it clones the other one for new usage.