Asymptotic locations of bounded and unbounded eigenvalues of sample correlation matrices of certain factor models -- application to a components retention rule
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2024
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866910521294323712 |
|---|---|
| author | Akama, Yohji Tian, Peng |
| author_facet | Akama, Yohji Tian, Peng |
| contents | Let the dimension $N$ of data and the sample size $T$ tend to $\infty$ with $N/T \to c > 0$. The spectral properties of a sample correlation matrix $\mathbf{C}$ and a sample covariance matrix $\mathbf{S}$ are asymptotically equal whenever the population correlation matrix $\mathbf{R}$ is bounded (El Karoui 2009). We demonstrate this also for general linear models for unbounded $\mathbf{R}$, by examining the behavior of the singular values of multiplicatively perturbed matrices. By this, we establish: Given a factor model of an idiosyncratic noise variance $σ^2$ and a rank-$r$ factor loading matrix $\mathbf{L}$ which rows all have common Euclidean norm $L$. Then, the $k$th largest eigenvalues $λ_k$ $(1\le k\le N)$ of $\mathbf{C}$ satisfy almost surely: (1) $λ_r$ diverges, (2) $λ_k/s_k^2\to1/(L^2 + σ^2)$ $(1 \le k \le r)$ for the $k$th largest singular value $s_k$ of $\mathbf{L}$, and (3) $λ_{r + 1}\to(1-ρ)(1+\sqrt{c})^2$ for $ρ:= L^2/(L^2 + σ^2)$. Whenever $s_r$ is much larger than $\sqrt{\log N}$, then broken-stick rule (Frontier 1976, Jackson 1993), which estimates $\mathrm{rank}\, \mathbf{L}$ by a random partition (Holst 1980) of $[0,\,1]$, tends to $r$ (a.s.). We also provide a natural factor model where the rule tends to "essential rank" of $\mathbf{L}$ (a.s.) which is smaller than $\mathrm{rank}\, \mathbf{L}$. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2407_07282 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Asymptotic locations of bounded and unbounded eigenvalues of sample correlation matrices of certain factor models -- application to a components retention rule Akama, Yohji Tian, Peng Statistics Theory Let the dimension $N$ of data and the sample size $T$ tend to $\infty$ with $N/T \to c > 0$. The spectral properties of a sample correlation matrix $\mathbf{C}$ and a sample covariance matrix $\mathbf{S}$ are asymptotically equal whenever the population correlation matrix $\mathbf{R}$ is bounded (El Karoui 2009). We demonstrate this also for general linear models for unbounded $\mathbf{R}$, by examining the behavior of the singular values of multiplicatively perturbed matrices. By this, we establish: Given a factor model of an idiosyncratic noise variance $σ^2$ and a rank-$r$ factor loading matrix $\mathbf{L}$ which rows all have common Euclidean norm $L$. Then, the $k$th largest eigenvalues $λ_k$ $(1\le k\le N)$ of $\mathbf{C}$ satisfy almost surely: (1) $λ_r$ diverges, (2) $λ_k/s_k^2\to1/(L^2 + σ^2)$ $(1 \le k \le r)$ for the $k$th largest singular value $s_k$ of $\mathbf{L}$, and (3) $λ_{r + 1}\to(1-ρ)(1+\sqrt{c})^2$ for $ρ:= L^2/(L^2 + σ^2)$. Whenever $s_r$ is much larger than $\sqrt{\log N}$, then broken-stick rule (Frontier 1976, Jackson 1993), which estimates $\mathrm{rank}\, \mathbf{L}$ by a random partition (Holst 1980) of $[0,\,1]$, tends to $r$ (a.s.). We also provide a natural factor model where the rule tends to "essential rank" of $\mathbf{L}$ (a.s.) which is smaller than $\mathrm{rank}\, \mathbf{L}$. |
| title | Asymptotic locations of bounded and unbounded eigenvalues of sample correlation matrices of certain factor models -- application to a components retention rule |
| topic | Statistics Theory |
| url | https://arxiv.org/abs/2407.07282 |