diff --git a/tests/manual_check/f32/lnmoment_amcloud.py b/tests/manual_check/f32/lnmoment_amcloud.py
deleted file mode 100644
index d34aaecb..00000000
--- a/tests/manual_check/f32/lnmoment_amcloud.py
+++ /dev/null
@@ -1,55 +0,0 @@
-"""comparison of the errors for exp(-9/2 log^2 sigmag) for FP32 jax
-
-Note:
-    this code was used to determine the optimal function of exp(-9/2 * log^2 sigmag) used in layeropacity.layer_optical_depth_clouds_lognormal
-    The conclusion is that we should use f(sig) in the following code.
-"""
-
-import jax.numpy as jnp
-import numpy as np
-
-def gnp(sig):
-    logs = np.log(sig)
-    return np.exp(-4.5*logs**2)
-
-def f0(sig):
-    return (sig**(-4.5*jnp.log(sig)))
-
-def f(sig):
-    return sig**(jnp.log(sig**-4.5))
-
-def g(sig):
-    logs = jnp.log(sig)
-    return jnp.exp(-4.5*logs**2)
-
-
-if __name__ == "__main__":
-    arr = np.logspace(0,3,10001)
-    
-    print(len(arr[f0(arr)>0.0]))    
-    print(len(arr[f(arr)>0.0]))
-    print(len(arr[g(arr)>0.0]))
-    print(np.min(arr[f(arr)>0.0]))
-    print(np.max(arr[f(arr)>0.0]))
-    farr = f(arr)
-    print(np.min(farr[farr>0.0]))
-    print(np.max(farr))
-    
-    import matplotlib.pyplot as plt
-    fig = plt.figure()
-    ax = fig.add_subplot(211)
-#    plt.plot(arr,np.log(arr))
-    plt.plot(arr,gnp(arr))
-    plt.xscale("log")
-    plt.yscale("log")
-    plt.axhline(1.0,color="gray",ls="dashed")
-    ax = fig.add_subplot(212)
-    plt.plot(arr,g(arr)/gnp(arr)-1.0,label="exp",alpha=0.3)
-    plt.plot(arr,f0(arr)/gnp(arr)-1.0,label="sig**(-4.5*jnp.log(sig))",alpha=0.3)
-    plt.plot(arr,f(arr)/gnp(arr)-1.0,label="sig**(jnp.log(sig**-4.5))",alpha=0.7)
-    plt.xscale("log")
-    plt.ylim(-2.e-5,2.e-5)
-    plt.legend()
-    plt.show()
-
-    
diff --git a/tests/manual_check/xs/Ttyp_demo.py b/tests/manual_check/xs/Ttyp_demo.py
deleted file mode 100644
index 0fe23e54..00000000
--- a/tests/manual_check/xs/Ttyp_demo.py
+++ /dev/null
@@ -1,42 +0,0 @@
-from exojax.spec.lpf import auto_xsection
-from exojax.spec import line_strength, doppler_sigma,  gamma_natural
-from exojax.spec.exomol import gamma_exomol
-from exojax.spec import api
-import numpy as np
-
-def demo(Tfix,Ttyp,crit=1.e-40):
-    """reproduce userguide/moldb.html#masking-weak-lines
-
-    Args:
-       Tfix: gas temperature
-       Ttyp: Ttyp for line strength criterion
-       crit: line strength criterion
-
-    Returns
-       nus, xsv
-
-    """
-    
-    nus=np.linspace(1000.0,10000.0,900000,dtype=np.float64) #cm-1
-    mdbCO=api.MdbExomol('.database/CO/12C-16O/Li2015',nus, Ttyp=Ttyp, crit=crit)
-    Mmol=28.010446441149536 # molecular weight
-    Pfix=1.e-3 # we compute P=1.e-3 bar
-    qt=mdbCO.qr_interp(Tfix)
-    Sij=line_strength(Tfix,mdbCO.logsij0,mdbCO.nu_lines,mdbCO.elower,qt,mdbCO.Tref)
-    gammaL = gamma_exomol(Pfix,Tfix,mdbCO.n_Texp,mdbCO.alpha_ref)\
-        + gamma_natural(mdbCO.A)
-    # thermal doppler sigma
-    sigmaD=doppler_sigma(mdbCO.nu_lines,Tfix,Mmol)
-    #line center
-    nu0=mdbCO.nu_lines
-    xsv=auto_xsection(nus,nu0,sigmaD,gammaL,Sij,memory_size=30)
-    return nus, xsv
-
-if __name__=="__main__":
-    Tfix=2000.0
-    Ttyp=1000.
-    nus,xsv=demo(Tfix,Ttyp,crit=1.e-40)
-    import matplotlib.pyplot as plt
-    plt.plot(1.e4/nus,xsv)
-    plt.yscale("log")
-    plt.show()