TheoremBase

Proof of Fatou's Lemma

Proof

For k∈Nk\in\mathbb{N} define gk:X→[0,∞]g_k:X\to[0,\infty] by gk(x)=inf⁡m≥kfm(x)g_k(x)=\inf_{m\ge k}f_m(x). Each gkg_k is measurable in the sense of Lebesgue Integral of a Nonnegative Measurable Function: for a∈Ra\in\mathbb{R},

{gk≥a}=⋂m≥k{fm≥a},with {fm≥a}=⋂j∈N{fm>a−1/j}∈F,\{g_k\ge a\}=\bigcap_{m\ge k}\{f_m\ge a\},\qquad\text{with }\{f_m\ge a\}=\bigcap_{j\in\mathbb{N}}\{f_m>a-1/j\}\in\mathcal{F},

so {gk≥a}∈F\{g_k\ge a\}\in\mathcal{F} by closure of the σ\sigma-algebra under countable intersections, and {gk>a}=⋃j∈N{gk≥a+1/j}∈F\{g_k>a\}=\bigcup_{j\in\mathbb{N}}\{g_k\ge a+1/j\}\in\mathcal{F}.

The sequence (gk)k(g_k)_k is nondecreasing pointwise (the infimum is over a smaller index set as kk grows), and by definition sup⁡kgk=lim inf⁡mfm\sup_k g_k=\liminf_m f_m pointwise. By Monotone Convergence Theorem, lim inf⁡mfm\liminf_m f_m is measurable and

∫X(lim inf⁡mfm) dμ=sup⁡k∫Xgk dμ.\int_X\Bigl(\liminf_m f_m\Bigr)\,d\mu=\sup_k\int_X g_k\,d\mu.

For every m≥km\ge k we have gk≤fmg_k\le f_m pointwise, so by monotonicity of the nonnegative integral (claim 1 of Linearity and Monotonicity of the Lebesgue Integral) ∫Xgk dμ≤∫Xfm dμ\int_X g_k\,d\mu\le\int_X f_m\,d\mu; taking the infimum over m≥km\ge k,

∫Xgk dμ ≤ inf⁡m≥k∫Xfm dμ.\int_X g_k\,d\mu\ \le\ \inf_{m\ge k}\int_X f_m\,d\mu.

Taking the supremum over kk and using the definition of lim inf⁡\liminf for sequences in [0,∞][0,\infty] given in the statement,

∫X(lim inf⁡mfm) dμ=sup⁡k∫Xgk dμ ≤ sup⁡k inf⁡m≥k∫Xfm dμ=lim inf⁡m∫Xfm dμ.■\int_X\Bigl(\liminf_m f_m\Bigr)\,d\mu=\sup_k\int_X g_k\,d\mu\ \le\ \sup_k\ \inf_{m\ge k}\int_X f_m\,d\mu=\liminf_m\int_X f_m\,d\mu.\qquad\blacksquare

Citations

Loading…

Dependencies

Uses0

Loading…

Comments

Log in to comment.

Loading…