Possible divisions by zero

[gjcoram]

I have not checked all of these warnings, but I see that LVT and WVT are correctly flagged: they appear in the denominator of expressions:
DVT_long = - AVT * MAXA(ln(Leff / LVT),0.0,1.0e-2); DVT_wide = - AVT * MAXA(ln(Weff / WVT),0.0,1.0e-2);
but those parameters are allowed to be zero:
parameter real LVT = 1.0 from [0.0:inf) ATTR(info="LENGTH FOR LONG AND WIDE CHANNEL VTO CORRECTION"); parameter real WVT = 1.0 from [0.0:inf) ATTR(info="WIDTH FOR LONG AND WIDE CHANNEL VTO CORRECTION");

WARNING in file ekv3.va, line 258: Possible division by zero for parameter LVT
WARNING in file ekv3.va, line 258: Possible division by zero for parameter LVT
WARNING in file ekv3.va, line 258: Possible division by zero for parameter LVT
WARNING in file ekv3.va, line 259: Possible division by zero for parameter WVT
WARNING in file ekv3.va, line 259: Possible division by zero for parameter WVT
WARNING in file ekv3.va, line 259: Possible division by zero for parameter WVT
WARNING in file ekv3.va, line 264: Possible division by zero for parameter LGAM
WARNING in file ekv3.va, line 264: Possible division by zero for parameter LGAM
WARNING in file ekv3.va, line 264: Possible division by zero for parameter LGAM
WARNING in file ekv3.va, line 265: Possible division by zero for parameter WGAM
WARNING in file ekv3.va, line 265: Possible division by zero for parameter WGAM
WARNING in file ekv3.va, line 265: Possible division by zero for parameter WGAM
WARNING in file ekv3.va, line 328: Possible division by zero for parameter LA
WARNING in file ekv3.va, line 329: Possible division by zero for parameter LB
WARNING in file ekv3.va, line 335: Possible division by zero for parameter WKP1
WARNING in file ekv3.va, line 335: Possible division by zero for parameter WKP3
WARNING in file ekv3_include/ekv3_extrinsic_diodes.va, line 8: Possible division by zero for parameter NJS
WARNING in file ekv3_include/ekv3_extrinsic_diodes.va, line 9: Possible division by zero for parameter NJD
WARNING in file ekv3_include/ekv3_extrinsic_diodes.va, line 126: Possible division by zero for parameter NJS
WARNING in file ekv3_include/ekv3_extrinsic_diodes.va, line 132: Possible division by zero for parameter NJTSS
WARNING in file ekv3_include/ekv3_extrinsic_diodes.va, line 132: Possible division by zero for parameter NJTSSWS
WARNING in file ekv3_include/ekv3_extrinsic_diodes.va, line 132: Possible division by zero for parameter NJTSSWGS
WARNING in file ekv3_include/ekv3_extrinsic_diodes.va, line 138: Possible division by zero for parameter NJD
WARNING in file ekv3_include/ekv3_extrinsic_diodes.va, line 144: Possible division by zero for parameter NJTSD
WARNING in file ekv3_include/ekv3_extrinsic_diodes.va, line 144: Possible division by zero for parameter NJTSSWD
WARNING in file ekv3_include/ekv3_extrinsic_diodes.va, line 144: Possible division by zero for parameter NJTSSWGD
WARNING in file ekv3_include/ekv3_extrinsic_diodes.va, line 156: Possible division by zero for expression 1/(1.0-MJS)
WARNING in file ekv3_include/ekv3_extrinsic_diodes.va, line 157: Possible division by zero for expression 1/(1.0-MJSWS)
WARNING in file ekv3_include/ekv3_extrinsic_diodes.va, line 158: Possible division by zero for expression 1/(1.0-MJSWGS)
WARNING in file ekv3_include/ekv3_extrinsic_diodes.va, line 178: Possible division by zero for expression 1/(1.0-MJD)
WARNING in file ekv3_include/ekv3_extrinsic_diodes.va, line 179: Possible division by zero for expression 1/(1.0-MJSWD)
WARNING in file ekv3_include/ekv3_extrinsic_diodes.va, line 180: Possible division by zero for expression 1/(1.0-MJSWGD)
WARNING in file ekv3_include/ekv3_extrinsic_rc.va, line 39: Possible division by zero for parameter RBWSH
WARNING in file ekv3_include/ekv3_extrinsic_rc.va, line 45: Possible division by zero for parameter RSBWSH
WARNING in file ekv3_include/ekv3_extrinsic_rc.va, line 48: Possible division by zero for parameter RDBWSH
WARNING in file ekv3_include/ekv3_extrinsic_rc.va, line 54: Possible division by zero for parameter RSBWSH
WARNING in file ekv3_include/ekv3_idb.va, line 9: Possible division by zero for parameter IBB
WARNING in file ekv3_include/ekv3_noise.va, line 41: Possible division by zero for parameter ECN
WARNING in file ekv3_include/ekv3_noise.va, line 78: Possible division by zero for parameter KP
WARNING in file ekv3_include/ekv3_oppoints.va, line 162: Possible division by zero for parameter KP

[gjcoram]

Also, watch for ln(0) in this equation:

axetaqm2_3 = aqma * limexp(``TWO3RDS * ln(ETAQM));

since ETAQM may be zero:

parameter real ETAQM = 0.75 from [0.0:inf) ``ATTR(info="QUANTUM MECHANICAL EFFECT ACCUMULATION ETA PARAMETER");

[gjcoram]

Interestingly, the code checks for RBN==0, but not for RBWSH==0:

            rb              = (RBN  == 0.0) ? // NOTE: Check if bulk resistance per finger is defined.
                              RBWSH * 0.5 / Weff :
                              1.0 / ((Weff * 2.0 / RBWSH) + (NF / RBN));

and RBWSH is allowed to be zero:
parameter real RBWSH = 3.0E-3 from [0.0:inf) `ATTR(info="SUBSTRATE NETWORK BULK RESISTANCE PER WIDTH UNIT BETWEEN INTERNAL BULK AND EXTERNAL BULK NODES (USED AT RF4, RF AND NQS MODES)");

[gjcoram]

Also, the comments indicate WKP1 and WKP3 were intentionally given a wider range:
// LIMIT WKP1 CHANGES TO (-INF FROM [0
parameter real WKP1 = 1.0E-6 from (-inf:inf) ATTR(info="CRITICAL WIDTH OF MOBILITY WIDTH SCALING"); // LIMIT WKP2 CHANGES TO (-INF FROM [0 parameter real WKP2 = 0.0 from (-inf:inf) ATTR(info="AMPLITUDE PARAMETER FOR MOBILITY WIDTH SCALING");
// LIMIT WKP3 CHANGES TO (-INF FROM [0
parameter real WKP3 = 1.0 from (-inf:inf) `ATTR(info="SPAN PARAMETER FOR MOBILITY WIDTH SCALING");

But it seems to me a non-positive value of WKP1 will certainly result in a math error:
// WIDTH SCALING OF THE MOBILITY AROUND THE VALUE OF WKP1 PARAMETER
tmp_kpw = (ln(Weff / WKP1)) / WKP3;

[gjcoram]

The divisions by zero for KP are a little harder to track down:
WARNING in file ekv3_include/ekv3_noise.va, line 78: Possible division by zero for parameter KP
WARNING in file ekv3_include/ekv3_oppoints.va, line 162: Possible division by zero for parameter KP

ekv3.va:
KP_a = KP * (1.0 + AKP * awl);
KP_DEV = KP_a * MKP_l * MKP_w * MKP_sti;
KP_DEV_t = KP_DEV * exp(BEX * lnrT) * MKP_sti_t;
beta = KP_DEV_t * (beta_nom / beta_denom) / beta_clm_denom;
i0 = 2.0 * nq * UT2 * beta * inv_dqmip1;
ekv3_include/ekv3_oppooints.va:
tmp = i0 /( nq * UT * COX );
tau0 = pow(Leff - deltal,2.0) / (tmp);

But I must confess that the warning about ekv3_noise.va is incorrect:
omegaspec = beta / COX * UT / Leff2; // NOTE: Normalization factor for angular frequency.
OMEGA = (omegaspec != 0.0) ?
1.0 / omegaspec : 0.0;
I think VAMPyRE is not properly considering the condition in the ternary operator.