1. 矩阵的转值定义为行和列下标相互交换,一个![](https://cdn.nlark.com/yuque/__latex/20f872dd193041899f98d8c4df7e4a36.svg#card=math&code=m%2An&id=pJuOC)的矩阵转值之后为![](https://cdn.nlark.com/yuque/__latex/1d77d3c14d4eed27f2f7c2719880302d.svg#card=math&code=n%2Am&id=RF87g)的矩阵。矩阵![](https://cdn.nlark.com/yuque/__latex/7fc56270e7a70fa81a5935b72eacbe29.svg#card=math&code=A&id=hf6W2)的转值记为![](https://cdn.nlark.com/yuque/__latex/521a9daa8c1cd6293c6e22e8e8386c42.svg#card=math&code=A%5ET&id=afGvk)。<br />两个矩阵的加法为其对应位置元素相加,显然执行加法运算的两个矩阵必须有相同的尺寸。矩阵![](https://cdn.nlark.com/yuque/__latex/7fc56270e7a70fa81a5935b72eacbe29.svg#card=math&code=A&id=LHPBf)和矩阵![](https://cdn.nlark.com/yuque/__latex/9d5ed678fe57bcca610140957afab571.svg#card=math&code=B&id=Z5NZ1)相加记为<br />![](https://cdn.nlark.com/yuque/__latex/470c3f34cde50a8a7c4ed8f3926033af.svg#card=math&code=A%2BB&id=c1967)<br />加法和转值满足<br />![](https://cdn.nlark.com/yuque/__latex/057eac6c484603c3bf5b7cb3a099e391.svg#card=math&code=%28A%2BB%29%5ET%3DA%5ET%2BB%5ET&id=ThMUX)<br />加法满足交换律与结合律<br />![](https://cdn.nlark.com/yuque/__latex/9246423af39a3b2b4ff2a8b4e1c63e9c.svg#card=math&code=A%2BB%3DB%2BA%20&id=kS10p)<br />两个矩阵的减法为对应位置元素相减,同样两个矩阵必须有相同的尺寸<br />![](https://cdn.nlark.com/yuque/__latex/f2b2365c89536857b3a326a6ae69079a.svg#card=math&code=A-B&id=rKhzt)<br />矩阵与标量的乘法,即数乘,定义为标量与矩阵的每个元素相乘。<br />![](https://cdn.nlark.com/yuque/__latex/30f767aa191cd5d261e767fd78393607.svg#card=math&code=kA&id=UHsSb)<br />数乘和加法满足分配律<br />![](https://cdn.nlark.com/yuque/__latex/f766ace42a930c26ada7e5ebd430c3c9.svg#card=math&code=k%28A%2BB%29%3DkA%2BkB&id=whfd9)<br />两个矩阵的乘法定义为用第一个矩阵的每个行向量和第二个矩阵每个列向量做内积,形成结果矩阵的每个元素,显然第一个矩阵的列数要和第二个矩阵的行数相等。矩阵![](https://cdn.nlark.com/yuque/__latex/7fc56270e7a70fa81a5935b72eacbe29.svg#card=math&code=A&id=XMh92)与![](https://cdn.nlark.com/yuque/__latex/9d5ed678fe57bcca610140957afab571.svg#card=math&code=B&id=cOVvX)相乘记为<br />![](https://cdn.nlark.com/yuque/__latex/b86fc6b051f63d73de262d4c34e3a0a9.svg#card=math&code=AB&id=a801P)<br />单位矩阵与任意矩阵左乘和右乘都等于该矩阵本身<br />![](https://cdn.nlark.com/yuque/__latex/2d7f28f8b70c9e44064f7ceeca771670.svg#card=math&code=IA%3DA%20%5Cquad%20%5Cquad%20AI%3DA&id=mABSF)<br />矩阵![](https://cdn.nlark.com/yuque/__latex/7fc56270e7a70fa81a5935b72eacbe29.svg#card=math&code=A&id=zofBE)左乘对角矩阵![](https://cdn.nlark.com/yuque/__latex/57d9d68629fa59ad15d71eddb6c9e493.svg#card=math&code=%5CLambda%20%3Ddiag%28k_1%2C%5Cdots%2Ck_n%29&id=Nxpty)相当于将![](https://cdn.nlark.com/yuque/__latex/7fc56270e7a70fa81a5935b72eacbe29.svg#card=math&code=A&id=OZrdY)的第![](https://cdn.nlark.com/yuque/__latex/865c0c0b4ab0e063e5caa3387c1a8741.svg#card=math&code=i&id=gvtwD)行的所有元素都乘以![](https://cdn.nlark.com/yuque/__latex/34c1d173d638ceb8fb5bec184c055549.svg#card=math&code=k_i&id=ktQsT)<br />矩阵![](https://cdn.nlark.com/yuque/__latex/7fc56270e7a70fa81a5935b72eacbe29.svg#card=math&code=A&id=PaJtU)右乘对角矩阵![](https://cdn.nlark.com/yuque/__latex/57d9d68629fa59ad15d71eddb6c9e493.svg#card=math&code=%5CLambda%20%3Ddiag%28k_1%2C%5Cdots%2Ck_n%29&id=E7i88)相当于将![](https://cdn.nlark.com/yuque/__latex/7fc56270e7a70fa81a5935b72eacbe29.svg#card=math&code=A&id=Iptd3)的第![](https://cdn.nlark.com/yuque/__latex/865c0c0b4ab0e063e5caa3387c1a8741.svg#card=math&code=i&id=DRxUh)列的所有元素都乘以![](https://cdn.nlark.com/yuque/__latex/34c1d173d638ceb8fb5bec184c055549.svg#card=math&code=k_i&id=G5iKR)<br />矩阵的乘法也满足结合律<br />![](https://cdn.nlark.com/yuque/__latex/cfa74e7ef93465d86542ad172ea00506.svg#card=math&code=%28AB%29C%3DA%28BC%29&id=gnliq)<br />矩阵乘法和加法满足左分配律和右分配律<br />![](https://cdn.nlark.com/yuque/__latex/1eddad4d50f6647377d3c114e5ae226c.svg#card=math&code=A%28B%2BC%29%3DAB%2BAC%20%5C%5C%28A%2BB%29C%3DAB%2BAC%20%0A&id=Dxf9e)<br /> 需要注意的是矩阵的乘法不满足交换律<br />![](https://cdn.nlark.com/yuque/__latex/4e76a57cb6f246e17c512b1e46d5a44e.svg#card=math&code=AB%20%5Cneq%20BA&id=nxjf2)<br />矩阵乘法和转值满足" 穿脱原则"<br />![](https://cdn.nlark.com/yuque/__latex/21c851cadf1b96018d007e656d589148.svg#card=math&code=%28AB%29%5ET%3DB%5ETA%5ET&id=EOTOI)<br />矩阵可以将其分块表示为<br /> ![](https://cdn.nlark.com/yuque/__latex/d9ba28f74047ce40ba5ff6f1e89ec467.svg#card=math&code=A%20%3D%20%5Cbegin%7Bpmatrix%7D%20A_%7B11%7D%20%26%20%20A_%7B12%7D%20%5C%5C%20%20A_%7B21%7D%20%26%20%20A_%7B22%7D%20%5Cend%7Bpmatrix%7D&id=CRiuN)<br />如果对矩阵![](https://cdn.nlark.com/yuque/__latex/24d01c0a944ba7f80e68ab1f46f12301.svg#card=math&code=A%EF%BC%8CB&id=WLfYD)进行分块后各个块的尺寸以及水平、垂直方向的快数量均相容,那么可以将块当作标量来计算乘积![](https://cdn.nlark.com/yuque/__latex/b86fc6b051f63d73de262d4c34e3a0a9.svg#card=math&code=AB&id=cWz25)。对于下面两个分块矩阵<br /> ![](https://cdn.nlark.com/yuque/__latex/aba326491bf201edf9c8aada7d9c3fcd.svg#card=math&code=A%20%3D%20%5Cbegin%7Bpmatrix%7D%20I_2%20%26%20%200_%7B2%2A3%7D%20%5C%5C%20%20A_%7B1%7D%20%26%20%20I_3%20%5Cend%7Bpmatrix%7D%20%20%20%5Cquad%20%5Cquad%20B%3D%20%5Cbegin%7Bpmatrix%7D%20B_1%20%26%20%20I_%7B2%7D%20%5C%5C%20%20-I_%7B3%7D%20%26%20%200_%7B2%2A3%7D%20%5Cend%7Bpmatrix%7D&id=TmPxX)<br />![](https://cdn.nlark.com/yuque/__latex/3d1cfa32a42ed6f334bc7b8d66482291.svg#card=math&code=AB%3D%20%5Cbegin%7Bpmatrix%7D%20I_2%20%26%20%200_%7B2%2A3%7D%20%5C%5C%20%20A_%7B1%7D%20%26%20%20I_3%20%5Cend%7Bpmatrix%7D%20%20%20%5Cbegin%7Bpmatrix%7D%20B_1%20%26%20%20I_%7B2%7D%20%5C%5C%20%20-I_%7B3%7D%20%26%20%200_%7B2%2A3%7D%20%5Cend%7Bpmatrix%7D%20%3D%20%5Cbegin%7Bpmatrix%7D%20B_1%20%26%20%20I_%7B2%7D%20%5C%5C%20%0AA_1B_1-U_3%20%26%20%20A_1%20%5Cend%7Bpmatrix%7D&id=C1Q24)