【官方雙語/合集】線性代數(shù)的本質(zhì) - 系列合集

[來自2023.5.29:似乎看的人還挺多?距離筆者寫這篇notes已經(jīng)有小半年了,由于筆者是物理系的,目前正在學(xué)量子場(chǎng)論(qft),這里我稍微推薦基本比較能讓讀者建立線代思想的書籍,以及進(jìn)一步的學(xué)習(xí),當(dāng)然,這是按照物理系安排的方法來的,我猜數(shù)學(xué)人估計(jì)不會(huì)點(diǎn)進(jìn)來看筆記doge:第一個(gè)要聲明的是,這篇note里并無除此視頻以外的內(nèi)容,只不過稍微對(duì)視頻中的內(nèi)容有所解釋,沒有更多的延拓與理解,因此我推薦一些書籍:pku的那本《簡(jiǎn)明高等代數(shù)》的風(fēng)格和3b1b的風(fēng)格還是挺像的(線代部分,當(dāng)然要是想學(xué)完高等代數(shù)也并非不可),值得一看。還有一本鼎鼎大名的書《Linear algebra done right》中譯本叫《線性代數(shù)應(yīng)該這樣學(xué)》的書,盡管書的作者說這是為第二次學(xué)線性代數(shù)的學(xué)生準(zhǔn)備的,但我仍然認(rèn)為這是一本講線代的思想講的極其清楚的書籍,并且后面有講一些譜理論和一點(diǎn)算子的思想,3b1b的可視化將是你理解線性代數(shù)的利器,如果我以后給我的學(xué)生講線代,我估計(jì)會(huì)讓他們把這系列看完,然后我再給他們講done right;第二個(gè)有趣的點(diǎn)在于,右手定則在更廣義的拓展上來自于Grassmann algebra的反對(duì)稱性(感興趣的可以看梁昆淼的《微分幾何和廣義相對(duì)論》講外積那一節(jié),或者你可以在b站搜索“當(dāng)時(shí)月影已不在”,加入我們的學(xué)習(xí)群(私信即可,記得標(biāo)注自己的年級(jí)以及方向),有寫的更棒的講義,這本講義的作者就是這位up主,代數(shù)專業(yè)的,你看了后就會(huì)明白他寫的多么好了),大概先說這么多,要是后面有物理系的學(xué)生對(duì)進(jìn)一步學(xué)習(xí)感興趣的話我再稍微寫寫doge]
linear algebra(basis is the words we describe linear space):
1:linear combination:if you fix one of those scalars and let the other one change its value freely,the tip of the resulting vector draws a straight line.(an intuitive and funny explanation)

2.span: the set of all possible vectors that you can reach with a linear combination of a given vectors is called the "span" of those two vectors.
3.linear dependent:

linear independent:

I prefer this way(since it could show all the basis have the same weight):

([doge]):

4.linear transformation(input--->output "map"):
definition:T(a+b)=T(a)+T(b)
T(ca)=c T(a)
inference:

Matrix:[T(i) T(j)] (the langue we describe the linear transformation)
"shear"(“剪切”):

Matrix--vector multiplication:


Satisfying associative law:
(M?M?)x=M?(M?x) (x是向量)
geometric way to explain it:

3-d space:

(值得一看的動(dòng)漫)


5.determinant:
definition(geometrically):

PS:行列式可為負(fù):






3-d:



PS:變換矩陣的行列式為0時(shí),它的Columns must be linearly dependent

描述三維空間定向的方法(the famous"right hand rule"):



computing method(intuitive way):



algebraic operation(代數(shù)運(yùn)算):

geometric way to explain it:
the stretching or compression of a linear space by two matrices is equivalent to the effect of the "sum matrix" on the linear space

6.solve linear system equations:
geometric interpretation:

reason:
tip:column space--->row space(linear transformation)
det(M)≠o:

可通過逆向線性變換(逆矩陣)求解
notice:the reason is it's an injective map(單射)


det(M)=0,no inverse
注:該線性變換的結(jié)果為向量,即該線性變換為“非單射”(非函數(shù)),任意升維線性變換("raised" rank)都是“非單射”形式的(這里把Matrix比作函數(shù)可能有些誤導(dǎo)),如對(duì)一維向量使用“升維”線性變換將會(huì)得到一個(gè)平面


(需要在transformation后存在解, "eigenvector")


Rank("秩"的概念):

the dimension of the column space(
column space:the "span" of the linearly transformed vector)

null space("Kernel",核):the span of the vectors that becomes the zero vector after the linear transformation.
非方陣:
3 by 2:


2 by 3:

7.duality property and dot product
非方陣與投影(這里以2 by 2 matrix舉例)


duality property:




dot product is equivalent to the projection of the basis onto the dual vector (the geometric meaning of the dot product)
8.cross product(linear transformation and duality property) :
determinant的結(jié)果是一個(gè)number,因此它與一個(gè)三維到一維的線性變換有關(guān)(這里以三維到一維為例)
我們知道,三維到一維的線性變換可等價(jià)為“該向量”對(duì)dual vector的投影
同上,這個(gè)dual vector即為叉積的幾何意義

定義linear transformation






[若考慮輸入任意 "x,y,z" 向量,該計(jì)算的幾何意義為該平行六面體的體積,'x,y,z'向量對(duì)v,w平面上的投影與v,w平面面積的乘積,同樣的,它與對(duì)偶向量(長(zhǎng)度為v,w的平面面積值,不帶有物理意義,行列式)和"x,y,z"向量的點(diǎn)積等價(jià)]

9.Transformation of basis
other's language(other basis)

(PS:2,1 and -1,1 are "our language"or"our vector space".)

make 2,1 -1,1 be basis(1,0 0,1)